This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
). But it is significant that, if we leave aside the concluding formulae which are traditionally written in latinate style, all five other instances of forms of the verb escribir in the text show prosthesis. The exceptional case may therefore owe itself to scribal influence from Latin. 94 The dating of the Cid has been much discussed. In 1908, Mene´ndez Pidal proposed 1140 as the original date of composition, a view which gained universal acceptance. However, from the 1970s there have been increasingly dissenting voices and today it is generally accepted that 1207 represents the date when the present version of the poem was written down (cf. Smith [1972] 1989: esp. 37–46). Smith, p. 38, sees Per Abbat (cf. Per Abbat le escrivio, l. 3732) as the poem’s author, while others view him as just the scribe who set
I-prosthesis
105
There are just four lexical items yielding twelve tokens which have word-initial s þ C- graphies.95 Three are etymological s impura forms where the expected prosthetic vowel is not written, spirital (always in the latinizing set phrase Padre spirital), parts of (e)sperar ‘to hope, wait’ and the place name Spinaz de Can, and the fourth involves hypercorrected aphaeretic forms of espedirse ‘to take one’s leave’ (< EXPE˘TERE þ SE). However, all these lexical items except for the place name96 appear elsewhere in the same poem with the expected initial <e-> , a fact which suggests that those tokens where the vowel has not been written probably reflect scribal omission due to Latinizing influence rather than phonetic reality.97 Prosthetic vowels continued to be faithfully represented in the enormous Alfonsine corpus of texts compiled later in the thirteenth century (Kasten and Nitti 1978, 2002).98 There are few items written with word-initial s þ consonant but those that do appear are predominantly Latin words cited as such; for example, in the General Estoria I (fol. 199r, 69) we find ‘ . . . & por aquellos vasos dize el latin dela biblia sciphos. Et por aquellas mac¸anas. sperulas. ca en el latin dizen otrossi spera por rondeza o por c¸erco.’99 However, Latinisms that have been incorporated into Castilian usage appear only with graphies containing a prosthetic vowel, estulto, escola´stico, esca´ndalo, especificar. The practice of systematically representing prosthetic <e> in prose emanating from the royal court may well have been encouraged for sociolinguistic reasons as well as phonetic ones. In the culturally and ethnically diverse society of later thirteenth-century Castile there were three languages in widespread use, two of which were strongly
down the text possibly from live recitation by a performing juglar (Wright 2000: 98). The sole extant manuscript of the poem dates by general consent from the mid-fourteenth century. The exceptional items, with line numbers, are: spirital (300pv, 372pv, 1102pv, 1651pv); sperare (1194pv), sperar (1457pv), spero´ (1481pc), sperando (2239pc); Spinaz de Can (393pv); and spidios (226 pp, 1307 pv), spidies (1252 pc), where ‘pv’ = post-vocalic, ‘pc’ = postconsonantal, ‘pp’ = post-pausal. 96 Attested forms of this place name with a prosthetic vowel present do occur in other vernacular texts from the thirteenth century, however (cf. Mene´ndez Pidal 1964b: I, 41, n. 2). 97 An alternative explanation is that the omission is due to metrical constraints. However, metrical constraints do not appear to be strict in this poem since there is significant variation in the syllabic composition of its individual verses. 98 Alfonso X ‘el Sabio’ reigned 1252–84. Under his leadership, a number of scholars including Alfonso himself compiled a great range of prose texts in Castilian on historical, legal, scientific, religious and even recreational topics. 99 ‘and for those containers the Latin of the Bible says sciphos. And for those apples, sperulas, for in Latin they also say spera for roundness or for circle.’ 95
106
I-prosthesis
associated with specific religions, Arabic with Islam and Latin with Christianity. In contrast, Castilian provided a religiously neutral language that would be acceptable to all citizens (Penny 2002: 20–1). As the prosthetic vowel was a characteristic and highly familiar Castilian phenomenon which clearly distanced it from Latin, it is possible that its rapid acceptance and integration within the orthographic system that was sanctioned for the formal writings of the king and his scholars may have been in part politically motivated. From the thirteenth century onward, I-prosthesis remains a fully productive phonological rule in Castilian. Even in the Renaissance period when strong latinizing tendencies arose which militated against the continued use of prosthetic vowels in the spelling and pronunciation of words of Latin origin, Castilian was not affected in the same way as other Romance varieties. The dominant figure in linguistic matters in the later fifteenth and early sixteenth centuries was Antonio Nebrija (1464–1512), the leading Spanish Humanist scholar of his day. In various works that pioneered the restoration of Classical Latin in Spain, Nebrija considered the problem of the appropriate pronunciation of Latin and the view which he adopted was also applied to the appropriate pronunciation of Castilian. This was that ‘ası´ tenemos descreuir como hablamos y hablar como escriuimos.’100 The use of prosthetic vowels in Castilian, which had been established in all styles of writing and speech since the thirteenth century, was therefore not challenged. It was not until the early seventeenth century that supporters of the etymologizing approach to spelling and pronunciation began to raise their voices. This led to a conservative trend that developed strongly from the eighteenth century resulting in the restoration of ‘silent’ etymologically justified consonants in the standard pronunciation of more learned words, as in -ct- and –pt- in doctor, concepto (Claverı´a Nadal 1991: esp. 99–109). However, the converse process of eliminating etymologically unjustified sounds like the prosthetic [e] did not gain ground. Structural factors doubtless played a part in preserving the vowel. Castilian had developed a simple syllable structure with a maximum of two consonants, plosive þ liquid, permitted in word-initial onsets and at most one consonant in word-final codas. Abandonment of the prosthetic vowel would have resulted in this long-established simplicity at word edges being destroyed. Since there were no other structural changes under way which would have similarly created greater syllabic complexity at word edges, the chances of vowel prosthesis being abandoned were slight (cf. Sampson 2005, and also standard French in 4.4.3.1 below).
‘Thus we should write as we speak and speak as we write’, Ortographı´a [1517] 1977: Principio segundo, f. 3v. The same doctrine appears in his Grama´tica espan˜ola (1492: I, chs 5 and 10). 100
I-prosthesis
107
In eastern varieties of the Peninsula, the thirteenth century likewise brings for the first time numerous texts written in a vernacular orthography, and these reveal the generalized use of prosthetic vowels. The earliest known prose text in vernacular (albeit with various Latin interpolations) is the Homelies de Organya` which dates from about 1200 and contains over a dozen examples of prosthetic vowels, e.g. les espines, senes escarn, no estaue, al seu espirit, as against the formulaic latinate la sancta scriptura which appears twice (Molho 1961). Later prose writings, whether technical or creative in character, similarly point to the generalized incidence of prosthesis although scribal inconsistency can obscure the evidence. For instance, in the Libre de Evast e Blanquerna by Ramo´n Llull (1233–1316) dating from 1285 but surviving in fourteenth-century manuscripts only, we find a good deal of variation; for example, Nastasia estava, per espos, Tota aquella nit estech but tota aquella nit stech, per spos, que stegues which all occur in the same chapter (19).101 Nonetheless, the widespread presence of an unetymological initial <e> whatever the nature of the immediately preceding phonological context indicates that prosthesis has been established as a regular phonological process, and, just as in Castilian, it remains productive to the present day. The quality of the prosthetic vowel is [‰-] in standard Catalan, which is based on the eastern variety of Barcelona (Badia 1981: 160). It is also realized as [‰] in Roussillonnais and in Mallorcan and parts of other Balearic islands (Fouche´ 1924: 134–5; Wheeler 2005: 250), whereas in western varieties it appears as [e-] or often as [a-],102 which is also the usual outcome in Valencian. Further west, in Aragonese, the reflex is [e-], as in the dialects of Gistaı´n (Mott 1990) and Bielsa (Badia 1950). Beyond the Peninsula, the Catalan variety spoken in Alguer (Alghero) in north-west Sardinia also preserves the rule of prosthesis systematically but the vowel used has taken on the phonetic value [a] as part of a more general change affecting all occurrences of earlier unstressed [e], for example asto´mak, aspera´, askrivı´ ‘stomach, (M), SPERA¯RE, SCRI¯BERE (Blasco Ferrer 1984). The to hope, to write’ < STOMACHU treatment of more recent loans in Alguere`s from Italian shows prosthesis to be still as productive here as in Peninsular Catalan (cf. 1.2), stitico > astı´tik ‘constipated’, scarlattina > askarlatı´na ‘scarlet fever’. Turning to western varieties, we find that vernacular texts in Galician-Portuguese also begin to appear in the thirteenth century.103 However, it is only from the middle of that century that the use of vernacular for documents of a nonliterary character gains ground. Exceptionally, there are two isolated non-literary 101
The forms cited are taken from the edition by Galme´s (1935). Badia (1981: 160) claims that [a-] is the normal reflex in western dialects, but Veny (1987: 130) is more circumspect, ‘La /e/ a`tona inicial del romanc¸ primitiu, especialment formant part dels segments es- i en-, tendeix a confondre’s en una [a].’ 103 The earliest known cantiga, by the Portuguese poet Joa´n Soares de Paiva, has in fact been dated to 1196 (Lorenzo Va´zquez 2003: 161). But this forms a rare if not unique example of a vernacular work pre-dating 1200. 102
108
I-prosthesis
texts which were written much earlier in the century, the Notı´cia de Torto (a draft of a notarial document dating from 1205–11) and the will of King Afonso II of Portugal from 1214. In these, there is just one form that offers a possible site for I-prosthesis to occur, namely sten < STENT ‘may they be’, which occurs in the royal will in post-pausal position. The absence of a prosthetic vowel here may be attributed to Latin influence, since a preliminary draft in Latin was very probably composed prior to the writing of the vernacular version.104 In contrast, the growing body of vernacular non-literary texts that appear from the middle of the century do provide evidence of prosthesis although it is less systematically attested than in the Castilian of the period. This may reflect a greater retentiveness of the conservative tradition of using latinizing spelling amongst scribes or it could also be due to the relatively weak phonetic realization of the prosthetic vowel in certain varieties. Examples of the inconsistent representation of prosthesis in non-literary writings from Galicia are este scrito ‘this text’, i stauel,105 firme i stauel (twice) in a text from Portomarı´n (1255) where no prosthetic vowels are indicated, whilst in a 1283 document from Temes there is systematic representation, Santo Esteua´o´, a carta estando sempre and i escriuj. Many texts offer evidence that lies between these extremes, prosthesis being represented in some but not all possible contexts. Thus, we find en esta carta scriptas (‘written (f.pl.) in this document’), pera aquesto specialmente but i escreui in a text of 1265 from Betanzos, and in two texts from Sobrado a carta ste ‘let the (this) document be’ twice but i escriuj (1281), and para aquisto spec¸ialmente but ffize escriuir, Eu Johan Paris escreuj (1300).106 Literary texts of the thirteenth century indicate prosthesis much more consistently, in part because latinizing scribal traditions were less pervasive here than in formal legalistic writings, and in part because these texts were often founded on vernacular literary genres such as, notably, Occitan troubadour verse where prosthetic vowels were regularly indicated. Thus, in the Cancioneiros prosthetic vowels are usually represented except occasionally in post-vocalic contexts, and prose works of creative literature similarly attest to generalized prosthesis.107 104
In fact, the use of vernacular in this text is curious since King Afonso subsequently made other wills in 1218 and 1221, these being written not in vernacular but in Latin. He died in 1223. 105 The conjunction here and in the other Galician examples is written using a conventional abbreviation which we have resolved. 106 All the examples are taken from the collection of early Galician documents in Maia (1986). 107 For example, estou (verse-initial), mal estar and en un estrado in the Cancioneiro da Ajuda, XCV 20, CXL 10, CLXX 17, respectively (Carter 1941); and vejo estar (post-vocalic), as estrellas (post-consonantal), esperanc¸a (post-pausal in a list) in the fourteenth-century Vida de Sancto Amaro (Klob 1901).
I-prosthesis
109
In the adjacent Asturo-Leonese area, non-literary vernacular texts appear from the later twelfth century and already in one of these, dated 1171 from Sahagu´n, several prosthetic forms are found, namely the proper names Esteuan Rocha (three times, once with Roca) and de Espinel twice (Staaff 1907: text 1).108 In the thirteenth century, when many more vernacular texts become available, prosthetic vowels commonly appear although orthographic practice can vary a good deal from scribe to scribe, as happens elsewhere in Romania continua109 (cf. the striking scribal variation in medieval Tuscan documents, see 4.3.3). The variation is increased by the growing politico-cultural influence from Castilian in the Asturo-Leonese area which leads to literary production in local varieties becoming increasingly linguistically hybridized and unrepresentative of spoken usage. In the period leading up to the present day, the fate of I-prosthesis differs across these varieties of the Peninsula. It has usually continued to be productive in the more conservative linguistic areas of Galician and north-western dialects of Portuguese, whereas in other types of Portuguese and in some dialects of AsturoLeonese, progressive phonetic weakening has taken place in prosthetic vowels which has even led to their deletion sometimes, thereby restoring surface wordinitial /s/ þ consonant sequences. It is not clear whether this later difference in the treatment of prosthetic vowels reflects in any way the earlier linguistic rift that has been noted between Galician and NW Portuguese dialects on the one hand, and the more innovating Asturo-Leonese and other Portuguese dialects on the other hand (Lindley Cintra 1963a, 1963b).110 Looking more closely at the data, we find that in the more conservative varieties the prosthetic vowel is widely found although individual high-frequency forms may show aphaeresis. For instance, in the Galican dialect atlas (ALGa I, 2, maps 245–253), inflexional forms of verb estar are recorded as being realized without the original initial syllable in a number of peripheral varieties spoken in NW La Coruna, the extreme NE of Lugo including NW Asturias, a few points in far SW Pontevedra and one isolated point in the south of Ourense, whereas 108
The clear indication of prosthesis in this text is matched by prosthetic forms appearing in earlier Latin texts, e.g. de sanctu Isteban in a document dated 996 also from Sahagu´n (Mene´ndez Pidal 1964a: }59,2). 109 Thus, amongst the documents emanating from the Leonese monastery of San Esteban de Nogales, there is a text dating from 1267 scribe where the scribe Aparicio writes que studioron, Rodriguez scudero with no prosthesis indicated, while Gonc¸alvo Migueliz in a 1275 text freely uses prosthetic vowels, especial mientre, fiz escriuir, escriui, escriuanos (Staaf 1907: texts 89, 90). 110 Cf. ‘Cette diffe´rence entre le Portugal [du nord et surtout du nord-ouest] peuple´ de longue date, essentiellement renferme´ et conservateur, et le Portugal reconquis et repeuple´ pendant les XIIe et XIIIe sie`cles, essentiellement innovateur, a du point de vue linguistique une importance sur laquelle je crois qu’on n’insistera jamais assez’ (1963b: 72).
110
I-prosthesis
in central Galician varieties the vowel usually remains present. However, it is uncertain just how lexically generalized such aphaeresis is in these peripheral varieties. Retention of the prosthetic vowel is also found widely in non-standard Portuguese varieties spoken in the adjacent Minho area of north-western Portugal111 and, as of the late nineteenth century, in Beira Alta (Leite de Vasconcellos 1970: 87). Amongst the more innovating varieties, perhaps the most familiar example of the loss of prosthetic vowels comes in standard Portuguese where forms such as estar, espac¸o, escuta [ʃtaɾ], [’ʃpasu], [’ʃkutN] are found (standard Portuguese is based on educated usage in the central part of Portugal including Lisbon and Coimbra). This development is evidently of fairly recent date as it is not found at all in Brazilian Portuguese (cf. Mateus and d'Andrade 2000: 45). The first stage in the weakening of the prosthetic vowel appears to have been part of a more general process affecting unstressed /e/ which led it to adopt a short, centralized, high-mid value [i] by the end of the eighteenth century (Teyssier 1980: 78). Subsequently, the prosthetic vowel and other types of unstressed [i] continued to weaken further and since the early twentieth century they have come to be deleted altogether in the usage of growing numbers of standard speakers. Phonetically, the result may be complex word-initial onsets that include not only restored s impura sequences but also new hypercomplex sequences which often override the sonority sequencing generalization, e.g. telefone [tlfPn] ‘telephone’, merecer [mɾseɾ] ‘to deserve’, desprevenir [dʃpɾvniɾ] ‘to fail to provide’ (Mateus and d'Andrade 2000: 44). The development here can be compared with the pre-stress vowel reduction and loss in northern Italian and Rheto-Romance varieties which had direct relevance for a different type of prosthetic vowel usage there (see 6.1.1). Amongst non-standard varieties in Portugal, the prosthetic vowel has also undergone similar deletion in a number of areas. In the Algarve in the south, loss is widely reported (Hammarstro¨m 1953: 140; Maia 1975-8),112 while in the north-east frontier area between Portugal and Spain, there is systematic deletion Cf. Santos Silva (1961: 315) who states that prosthetic e- ‘e´ geralmente pronunciado’ in Minhoto dialects. However, at the end of the nineteenth century, deletion was also evidently found in these dialects especially with forms containing standard est- (Leite de Vasconcellos 1970: 86). The preferential deletion here in initial est- sequences resembles the circumstances reported for the dialect of the Verı´n valley, located in SE Galicia just north of the Portuguese border, by Taboada (1979: 77), see below. 112 Hammarstro¨m (loc. cit.) states that where deletion of unstressed /‰/ occurs, a phonetically syllabic consonant results. His description of certain Algarve dialects, which draws on earlier detailed investigations by Armando de Lacerda and himself, indicates a possible syllabic value not only for the initial pre-palatal fricative of words like espina but more generally for all types of consonant which precede (or sometimes follow) unstressed [‰] irrespective of context. 111
I-prosthesis
111
in the essentially Leonese dialect of Miranda and the nearby dialects of Rio d’Onor and Guadramil (Herculano de Carvalho 1958; Leite de Vasconcellos 1929a, 1929b).113 Delicate variability between major weakening and deletion is also found in the transitional variety of the commune of Sabugal which lies on the frontier adjacent to the province of Salamanca (Maia 1977). And in the Verı´n valley, located in SE Galicia just north of the border, the prosthetic vowel is reported to be realized as a very weak schwa, estrume [‰s’tɾume] ‘dung’ < ˘ PPA, and especially before st it may be STRĀMEN, estopa [‰s’topa] ‘tow’ < STU imperceptible with the result that the sequence [st] forms a complex s impura onset (Taboada 1979: 77). The similarity between this development and the aphaeresis reported for some Minhoto and Galician varieties (see above and n. 111) is striking. Further to the north-east, deletion of the prosthetic vowel was noted for Asturian varieties in the later nineteenth century, as in spinu, streitu, scalera (Munthe 1887: 23, 72). In the unfortunately unidentified varieties concerned,114 the loss of the prosthetic vowel had presumably occurred fairly early on since spinu has developed like other two-syllable words such as vinu in that it preserved its final [u], whereas trisyllabic words like camin(u), padrin(u) regularly lost their final vowel. However, this of course presumes that when apocope began to operate, the prosthetic vowel had already come to receive the same degree of (secondary) stress as that of the etymological initial vowel in words like camin(u). More recently, loss has been reported for the Asturian varieties of Babia and Laciana, [sku’ðjeʃa] < SCUTE˘LLA ‘bowl’, as part of a general weakening process affecting unstressed pre-stress /e/, although it is also observed that often prosthesis does now occur (Alvarez 1985: 210). This may in part reflect the growing influence of Castilian on Asturian over the past century as a result of powerful centralizing factors such as the introduction of obligatory education (in Castilian), the rise of the media, and national conscription. One expected consequence of this would be some phonological convergence favouring the retention and even restoration of prosthesis. Significantly, the standard bable prescribed by the Asturian Academy (founded in 1909) and now promoted as an official regional language in the statute of autonomı´a for Asturias in 1981 contains solely prosthetic forms; for example, espeyu, esfollar, estraordinario, esquı´, espontaneu
113
Forms cited include staka “stake”, spiga “ear of corn”, skila “cowbell”, striga “handful of flax”, skrencˇa “parting of the hair”. Here, the initial consonant is apico-alveolar as in Castilian and it can be weakly articulated, notably in parts of the verb star the opening syllable of which Herculano de Carvalho phonetically represents as [sta]. 114 Some uncertainty reigns as to which varieties are affected. For instance, Zamora Vicente (1967: 148) cites a form speillo < SPECULUM “mirror” for the dialect of El Bierzo in NW Asturias. However, in his dictionary of this variety Garcı´a Rey (1934) gives no indication at all of the occurrence of non-prosthetic forms.
112
I-prosthesis
which appear in the official Normes ortogra´fiques (third edition, 1990) promulgated by the Academy. 4.4.2.1
Summary
For most of its history, Ibero-Romance as a whole has maintained and generalized the use of I-prosthesis. The only exceptions have been certain Asturo-Leonese dialects and the more innovative types of Portuguese including the standard variety. In standard Portuguese, the disappearance of prosthetic vowels at a phonetic level is a recent phenomenon caused by the general weakening of unstressed vowels. Phonologically, however, there may still be grounds for postulating a rule of prosthesis (cf. 1.4). In Asturo-Leonese, certain varieties evidently did abandon the use of I-prosthesis but more recently these have been increasingly subject to influence from Castilian leading to possible restoration of the process.
4.4.3
G A L LO - RO M A N C E
The early history of I-prosthesis in Gallo-Romance closely resembles that of Ibero-Romance. Prosthesis evidently began to operate in all varieties and it went on to establish itself as an unconditioned process in most types of GalloRomance during the period up to the twelfth century. In the period from the twelfth century onward, however, there has been major change in its incidence. For it has been wholly abandoned as a productive process in a large number of varieties including standard French. This development occurred at different times and for different reasons in the varieties affected. Today, I-prosthesis remains productive but, it seems, in certain types of Occitan only. In contrast, a number of Gallo-Romance varieties and especially those spoken in the north-east followed a quite distinct path of change from early on. In these, use of the prosthetic vowel was abandoned early on and the non-prosthetic alternant was generalized. Subsequently, in Walloon dialects of the far northeast, there was further change whereby a new alternation arose in s impura forms, with /sC-/ in post-vocalic contexts and an epenthetic sequence /sVC-/ elsewhere (where ‘V’ represents the epenthetic vowel). The epenthetic vowel typically has a high front quality /i/ or /y/; for example, (epenthetic) one supe`ne [Pnsyp¡n] ‘a thorn’, on oˆte sutaˆve [Pno:tsyta:v] ‘another shed’ vs (non-epenthetic) de`s spe`nes [d¡sp¡n] ‘thorns’, `e staˆve [¡sta:v] ‘in the shed’ (Remacle 1948: 41). The circumstances of the exceptional developments in north-eastern Gallo-Romance are explored below (4.4.3.3). Looking more closely at the ongoing development of prosthesis, we may begin by recalling that Gallo-Romance inscriptions in the Imperial period show the same
I-prosthesis
113
relative lack of evidence for the process as was noted for Ibero-Romance (4.1.2). However, early medieval Latin documentation from the Merovingian period offers some signs that prosthesis was becoming established, although direct representation occurs much less frequently in texts written by the more educated. Amongst the few individuals at the upper end of the scale of learning is Gregory of Tours (539–94) who became metropolitan bishop of that city in 573. In his writings there are no direct examples of prosthetic vowels, but there are indirect indications of their existence such as the frequent interchange of prefixal and simplex verbal forms attributable to hypercorrection, e.g. spoliarent expoliarent, spectat (= EXSPECTAT) populus ‘the people await’ rather than ‘look at’, spiravit ¼ EXSPIRAVIT ‘he died’ rather than ‘he breathed’ (Bonnet 1890: 148; cf. 4.1.1 above). However, legal documents written by scribes of a lower level of education often provide more revealing evidence of contemporary speech habits. A significant number of such texts emanating from the royal chancellery during the seventh and early eighth centuries have survived and these contain various prosthetic forms, some with initial such as istabilis (657–673 AD) but much more commonly with <e> , estipbulacione (682 AD), estudiant (709 AD), esperare (716 AD). ‘Internal’ prosthesis is also attested in prefixal forms, conestructus (695–711 AD), supraescripthis (709 AD), etc. These are complemented by numerous examples of hypercorrect forms akin to those noted above for Gregory of Tours, such as structus = INSTRUCTUS (693 AD), strumentum (691, 697 AD) and extromento (716 AD) corresponding to INSTRUMENTUM (Vielliard 1927: 102–3). A further source of information comes in inscriptions on Merovingian coins. Coins could be struck not only by the royal mint but also by other authorized institutions such as the palatine school and churches. Examples drawn from Prou (1892) are ESCOLA RE[GIA] MONE[TA] item 704 from Paris (cf. SCOLA RE[GIA] item 705 of same origin), ESPANIACO item 1980 from Corre`ze (cf. SPANIACO item 1981 of same origin), where the vowel is represented as <E> , and there are also cases using which was commonly adopted in Merovingian script to represent the mid front vowel /e/: ISCOLA RI[GIA] item 76 from Paris, ISPIRADUS item 496 from Rennes, and ISTEPHANUS MUNI[TA] item 1330 from Geneva. Though difficult to date precisely, these inscriptional forms go back mainly to the seventh century. However, a couple of cases from Chalon-sur-Saoˆne (Saoˆne-et-Loire) definitely date from the second half of the sixth century, although they both involve abbreviations: EPISCOPUS ESTN (item 163) and EPISCOPUS ESTNU (item 164) where the name ESTEPHANUS is doubtless being indicated (cf. Prou 1892: liv). Substantial change occurred during the eighth century in the written representation of prosthetic vowels. The Carolingian reforms, whose earliest stages date from the middle of the century, led to the restoration of a more Classicalstyle orthographical system and pronunciation for the Latin used in official state
114
I-prosthesis
and Church matters. The effect was that attested examples of prosthetic vowels became ever rarer. The analysis carried out by Pei (1932) on forty-seven documents appearing in Tardif (1866) revealed the following: Date of document 700–17 750–70
Total documents 11 11
Possible cases of prosthesis 28 19
þ Prosthesis Prosthesis 10 4
18115 15
And in Pei’s analysis of twenty-five texts dating from the following period up to 812, the scribal suppression of indications for prosthetic vowels was found to be almost complete.116 However, when texts written in a vernacular-based rather than a Latin-based orthography begin to appear, a clearer picture becomes available of the incidence of I-prosthesis in spoken usage. The two earliest texts, both from northern GalloRomance, date from the ninth century but these only offer one possible site, in une spede (Eulalie, l. 22) where no prosthetic vowel is indicated (spede < SPATA ‘sword’). The generally accepted explanation for the absence of a prosthetic vowel in this form is that at that period there was still alternation between prosthetic and nonprosthetic forms, the latter occurring post-vocalically (Fouche´ 1966: 695; cf. 4.2.1). It is only from the eleventh and twelfth centuries when substantial numbers of vernacular writings become available that prosthesis is systematically attested in all phonological contexts. Textual evidence for this appears a little earlier in langue d’oc than in langue d’oı¨l. In the subsequent development of I-prosthesis, there are important differences between northern and southern Gallo-Romance. The two linguistic blocks will therefore be treated separately. 4.4.3.1
Langue d’oı¨l
The evidence from vernacular texts indicates that I-prosthesis had probably been generalized in most varieties by the end of the eleventh century.117 With few Of this subgroup, one text (c. 700, or earlier) has ten potential sites for prosthesis, only one of which directly attests the vowel. If this text is excluded, the statistics for this subgroup would of course show a significantly higher proportion of attested prosthetic forms. 116 Just one apparent case appears: estrumenta (775 AD). However, the initial vowel here may reflect the prefix of the Latin etymon INSTRUMENTA. 117 The last text still showing any signs of alternation is the Vie de Saint Alexis, written probably in the second half of the eleventh century. The earliest manuscript is L (first found in the German monastery of Lamspring) which dates from the first quarter of the twelfth century. However, only one originally s impura lexeme is found without a prosthetic vowel, spuse, which always occurs in post-vocalic position, ma spuse, ta spuse, 115
I-prosthesis
115
exceptions, texts of the twelfth century consistently contain forms with the prosthetic vowel indicated irrespective of phonological context, whether in verse or prose works and creative or official writings. Thus, the mid twelfthcentury Anglo-Norman Oxford Psalter shows generalized prosthesis: post-consonantal in tutes esteilles (< STE¯LLAS), il establit (< STABIL-¯IVIT) in Psalm 148; postpausal in sentence-initial espeirent (< SPE¯RANT), establis (< STABIL-¯ISCE) in Psalm 9; and post-vocalic in tu establiras (< STABILI¯RE-HABES) in Psalm 17 and la meie esperance (< SPE¯RANTIA) in Psalm 83 (Michel 1860). Similarly, the earliest legal charters in vernacular typically show just prosthetic forms; for example, a charter dated 1191 from Maubeuge (de´p. Nord) has soit estaule (< STA¯BILE) and et escrit (< SCRI¯PTU(M)).118 A rule of unconditioned I-prosthesis appears therefore to have been established, just as in Ibero-Romance. The presence of this rule is also confirmed by the fact that whenever loanwords originally containing s impura were adopted during the medieval period up to the thirteenth century they systematically underwent I-prosthesis: Frankish (6th–8th cent.) skum > escume (e´cume) ‘foam’; speut > espiet (e´pieu) ‘spit’; Norse (10th–11th cent.) skipa þ -er > e´quipe(r) ‘(to) crew’; stafn > estrave (e´trave) ‘stem’; Middle Dutch (11th–13th cent.) skope > escope (e´cope) ‘bailer’ ; stapel > estappe (e´tape) ‘stage’;
sparanjan > esparnier (e´pargner) ‘to spare’ streup- > estreu (with suffix change > e´trier) ‘stirrup’ stafnbord > estambor (e´tambot) ‘sternpost’ Sten-hu¯s ‘stone house’ > E´tainhus (place name) splissen > espisser (e´pisser) ‘to splice’ staeye > estaie (e´tai) ‘prop’
As the bracketed forms from modern French indicate, these items were subject to the same deletion of pre-consonantal /s/ that occurred in native words inherited from Latin. Voiced [z] which appeared before voiced consonants was deleted in most varieties of langue d’oı¨l by the beginning of the twelfth century, and voiceless
etc. Other relevant items in the text indicate that the prosthetic alternant has been generalized, as in s’espethe rather than ** sa spethe ‘his sword’ (< SPA¯TA). The full text of this charter appears in Woledge and Clive (1964: 54–5) and Sampson (1980: 138–9). 118
116
I-prosthesis
[s] which occurred before voiceless plosives119 disappeared in the later twelfth or thirteenth century (Pope 1952: }377, Fouche´ 1966: 861–2). Only in eastern dialects was /s/ preserved pre-consonantally. In the late medieval period a major change occurred. Prosthesis began to be abandoned as a productive process. No clear signs of this development however are to be found in many literary works of the fourteenth and fifteenth centuries. Thus, in the verse of Charles d’Orle´ans (1394–1465), just prosthetic forms such as l’estude, mon esperit, on escript are found. Similarly, prosthetic vowels are consistently indicated in the verse of Franc¸ois Villon (1431– d. after 1463); the only example of a non-prosthetic form is le roy Scotiste ‘the Scottish king’ referring to James II of Scotland (Testament l. 365), although later in the same poem there is d’Ecossoys ‘of Scots’ (l. 1216). In literary prose of popular type, a similar picture emerges with little evidence of the loss of productivity for I-prosthesis. For instance in Les Cent nouvelles nouvelles of the mid-fifteenth century, there are no examples of forms in <st-> ; just one lexical item special(e)ment with <sp-> although especial and especialement are also found; and only three items with initial <sc-> (= [sk-]) scabelle, scandale, scribe (Dubuis 1996). Otherwise, forms with a prosthetic vowel such as esclandre, espirituel, estable (adjective and noun) consistently appear. Indications of the change in the status of prosthesis come predominantly from non-literary writings of a technical or professional type and to a lesser extent from highly cultivated literary prose. This is because one of the factors triggering the change120 lies in the phonological treatment of learned Latin borrowings and these are to be found most commonly and with ever-increasing frequency in such writings. The first serious moves towards the displacement of Latin in favour of French in formal domains, such as the law and the royal chancellery, took place during the course of the thirteenth and fourteenth centuries. But to enable French to operate in these domains, significant elaboration of linguistic resources was necessary which was achieved through the wholesale incorporation of Latinisms.121 Further Latinisms came in the many works of translation that were commissioned by the monarchs Jean II (Le Bon) 1350–64 and Charles V (Le Sage) 1364–80 (Monfrin 2001). For instance, items such as scandaleux, 119
It appears that [s] may have weakened and been lost before [f] earlier than before other voiceless obstruents. Evidence for this comes from English borrowings, feast, espouse, descry (< OFr. feste, esposer, descrier) with [s] preserved but defeat, effort (OFr. desfaite, esforz) with [s] deletion. The earlier weakening of [s] before [f] is perhaps due to a greater assimilatory tendency operating in a sequence of two voiceless fricatives [sf]. 120 The causation of the abandonment of I-prosthesis is explored in the next subsection below. 121 On ‘elaboration’ see Haugen (1972) and, for French, Lodge (1993: 118–52). For the importance of the fourteenth century for Latinisms, cf. ‘Le XIVe sie`cle est ve´ritablement l’e´poque ou` se constitue le vocabulaire savant’ (Brunot 1966: 566).
I-prosthesis
117
sphe´rique, ste´rile, stoı¨cien are first found in the translations by Nicole Oresme (Taylor 1965: 735). As these examples indicate, the spelling adopted for such loanwords usually respected the orthography of the Latin original except in their final part which had to conform to French morphological structure. Hence, s impura words were written with initial <s> þ consonant rather than reflecting the possible contemporary French pronunciation with a prosthetic [e-], unlike the situation in Castilian where the spelling of cultismos normally reproduced vernacular pronunciation. From the later thirteenth century, the etymologizing tradition increasingly prevailed over the previously dominant phonemic type for writing the vernacular, to a large extent because the band of praticiens or professional legal clerks who wrote down the texts came from a Latin-writing tradition. In the burgeoning bureaucracy of the French state, the number and linguistic influence of these clerks grew significantly. Unfortunately, there is no clear indication of how Latinisms like scandaleux, ste´rile were actually pronounced in the fourteenth century. It has been claimed that all such words containing an initial <s> þ consonant sequence were systematically articulated with a prosthetic vowel until the sixteenth century.122 But this seems questionable since by the start of that century the use of non-prosthetic forms was undoubtedly very well established amongst educated speakers, as we shall see. Instead, it is likely that such speakers had come to be increasingly influenced by the spelling of the many prestigious Latinisms that entered higher-register French from the thirteenth century onwards. As a result, whether for affectation123 or to achieve a greater perceived authenticity, non-prosthetic pronunciations for words written with initial <s> þ consonant were doubtless used by literate fifteenth-century speakers and perhaps even some fourteenth-century speakers when pronouncing words like scandaleux, ste´rile. The publication of the important work by Erasmus on the pronunciation of Latin and Greek in 1528, which advocated a spellingbased approach, served to give definitive endorsement to this practice.
122
Cf. ‘where an initial consonant cluster of this type whatever its source came to be used in actual speech before the sixteenth century, it assumed an [e] on-glide’ (Hope 1971: II, 585). The emphasis is in the original. 123 A fine example of the affected use of Latinisms is cited by Rickard (1976: 112–13). It is an anonymous letter dating from c. 1450 in which the writer apologizes for not being able to attend a wedding. It begins Maistre magnifique et eminent en faculte´ de prospicue eloquence, aourne´ de rethorique, tout preambule de recommendacion presuppose´, vueille´s sagacement concepvoir que . . . and contains the non-prosthetic Latinism speculant. As Rickard notes, the sort of overblown latinate style affected here was later to be the butt of pastiche by Rabelais in the celebrated episode of the Limousin scholar in Pantagruel published 1532 (ch. 6).
118
I-prosthesis
In the first half of the sixteenth century, therefore, it seems probable that prosthesis was no longer operating as a truly productive process for many if not most speakers of the nascent standard variety of French. Strong direct evidence for this comes in the earliest detailed analysis of French published in 1530 by the Englishman John Palsgrave. His description, which reflects educated usage of the first decade of the sixteenth century, contains three sections or ‘books.’ In the first of these, where he deals with pronunciation, Palsgrave focuses attention on words that are well established in the lexicon of the emerging standard language. The prosthetic vowel is present in all of them, but they fall into two groups depending on the phonetic realization or non-realization of the pre-consonantal <s> appearing in the orthographic form of the words. With items such as escrı´pre, estudie´r, esco´le, espe´e, estoı´lle, he indicates (Book I, ch. 25) that the first consonant is not pronounced. The other group emerges in chapter 43, where words are identified whose orthographic pre-consonantal <s> is pronounced. These include escabea´u, escla´ve, espe´ce, espere´r, estime´r in which modern French has continued to preserve pre-consonantal [s].124 However, a third group of words showing a quite different treatment of etymological initial [s] þ consonant is also apparent in Palsgrave’s account although no direct mention is made of it in the section on pronunciation. In this group, vowel prosthesis does not occur. Occasional examples appear by chance in the first book; for instance, sphe´re is reported to be pronounced ‘sfere’ (ch. 23) and scovlptu´re as ‘scouture’ (ch. 26). But it is in the substantial dictionary of French that occupies more than 80 per cent of the overall text that the widespread existence of nonprosthetic forms really emerges, e.g. scrupuleu´x, (je) specu´le, specta´cle, statio´n, stı´le. This third category consists overwhelmingly of learned words, many of which were doubtless viewed as high-register forms that were not part of the core lexical structure of French of the period. Nonetheless, their phonetic treatment indicates that prosthesis can no longer have been a productive process amongst users of the emerging standard variety. We may once again compare the phonetic circumstances here with those in Castilian where learned forms have continued to be subject to prosthesis, cf. the counterparts to the forms just cited: escrupuloso, (yo) especulo, especta´culo, estacio´n, estilo. In the following year 1531, the first grammar of French compiled by a Frenchman for French readers (albeit written in Latin) was published. Its author Jacques Dubois (1531: 57–8) formally distinguishes the same three categories of pronunciation for words containing etymological initial s þ consonant as those identified rather less clearly by Palsgrave:
124
The ambiguous phonetic value of orthographic pre-consonantal <s> is only resolved in 1740 when the third edition of the Acade´mie dictionary formally abolished the letter in words such as esclat, feste where it does not represent [s].
I-prosthesis
119
(i) with prosthetic [e-] and with deletion of etymological [s]: e´pine, e´tude, etc. (ii) with prosthetic [e-] but with retention of etymological [s]: espe´rer, espoir, etc. (iii) without prosthetic [e-] and with retention of etymological [s]: scribe, station, etc. The distribution that is indicated here corresponds almost exactly to that found in modern standard French. The only significant exceptions occur in the words espace and espe`ce where the sibilant <s> is reported to be silent, and in escabeau for which the pronunciation scabeau is given and escabelle which is cited along with the alternative form scabelle (modern French has escabeau, escabelle both with initial [¡sk-]). In formally identifying the category (iii), Dubois is confirming Palsgrave’s observation that prosthesis no longer operates with learned loanwords in the ‘best’ usage of French. Thereafter, during the course of the sixteenth century the use of prosthesis was evidently abandoned by all speakers who sought to conform to the norms of the crystallizing standard variety of French. In established items of vocabulary, the prosthetic vowel [e-] became lexicalized, but neologisms and especially Latinisms that began with /s/ þ consonant were no longer subject to prosthesis. Acceptance of this model of pronunciation could however be obstructed for some standard speakers as a result of interference from their own local Gallo-Romance variety. This was particularly evident with speakers from southern France since I-prosthesis remained fully productive in almost all varieties of langue d’oc at that time. Thus, in his Traite´ de la conformite´ du language franc¸ois auec le grec (1565), Henri Estienne reports that in Dauphine´ and Languedoc many users of the standard variety said estatuts, estatue, espectacle, espacieux (Thurot 1881 [1966]: I, 216). Southern writers also sometimes carried over traces of their native use of prosthesis in their works, as with the memorialist Brantoˆme (1540–1614) who was from south-west France and used numerous non-standard prosthetic forms in his writings such as escrupule, espectacle, espacieux, este´rille ‘sterile’ and even hypercorrect stase ‘ecstasy’ (Lalanne [1880] 1970).125 Against the background of the general abandonment of prosthesis, the treatment of Italian borrowings in the sixteenth century appears anomalous at first sight. For of the forty-three borrowings beginning with [s] þ consonant in standard Italian which are first attested in French in this century, no fewer than thirty-five emerge with a prosthetic vowel, e.g. escompte, escopette, espadon, 125
Huguet (1925-67) provides evidence of widespread polymorphism with forms in (standard) /sC-/ (regional) /esC-/ in sixteenth-century writers, for example (e)scorpion, (e)scrupule, (e)specifier, (e)special, (e)sp(h)e`re, (e)spirituel, (e)spacieux, (e)spectacle, (e)splendeur, (e)spongieux, (e)statut, estable stabile, (e)stile.
120
I-prosthesis
estafier, cf. It. sconto, scoppietto, spadone, staffiere (Hope 1971: I, 187–93). However, the French forms do not reflect the continued productivity of prosthesis in the sixteenth century. Rather, the presence of the prosthetic vowel can be attributed to the action of two factors (Sampson 2004a). The first of these concerns the mode of transmission of the loanwords. Before entering the standard language, it appears that many Italianisms passed through a southern French linguistic filter. Particularly affected were technical words connected with professional activities and trade which were acquired, orally in most cases, by French people interacting with Italians. Southern French speakers were particularly likely to be involved in this linguistic interchange for obvious geographical and commercial reasons.126 Borrowings with initial s impura onsets were therefore liable to I-prosthesis since this rule was still normally operating in most varieties of langue d’oc. The resulting forms were then diffused northwards into the standard variety with a lexicalized initial [e], alongside similarly prosthetic native Occitan forms like escalier and escargot which also entered standard French in the sixteenth century. The other factor explaining the anomalous appearance of prosthesis in Italianisms relates to the precise linguistic source of these words. It is customary for linguists to cite standard Italian forms as the basis for the French loanwords but it is likely that many items did not derive directly from standard Italian but instead came from, or via, non-standard Italo-Romance varieties spoken in Piedmont where not only was there a long-term French military presence but also vowel prosthesis has enjoyed continuing currency in the local speech (cf. below, 4.4.5).127 A prosthetic vowel may therefore have been present in the original form of certain Italianisms that passed into French. By the beginning of the seventeenth century, the rule of I-prosthesis had certainly ceased to be productive amongst standard French speakers. In a century when French linguistic purism was taken to new heights, even writers such as Chifflet and Hindret128 who are specifically concerned with identifying ‘correct’ pronunciation for speakers aspiring to master the standard variety make no direct 126
Reflecting the strong commercial and artisanal presence of Italians, Lyon for example was known as the ‘French Florence’ at this time. 127 A curious problem arises in the treatment of Italianisms in two celebrated dialogues by Henri Estienne, both dating from 1578, which attacked the excessive Italianization of the language found at the royal court. Pastiching the affected overuse of Italianisms, Estienne has forms such as spurquesse ‘dirt’, spaceger ‘to stroll’, straque ‘tired’. Most, like these examples, have no prosthetic vowel. The probable reason for this is that Estienne was basing the Italianate forms he cites on words appearing in the literary standard form of Italian where I-prosthesis was moribund by the 1570s (Sampson 2004a). 128 In the introduction to his detailed 1687 work, the Breton Jean Hindret identifies seventeen features of ‘bad’ pronunciation found ‘a` la Cour aussi bien qu’a` Paris’, a further sixteen are attributed to ‘la petite Bourgeoisie de Paris’, and eleven others are
I-prosthesis
121
reference at all to prosthetic vowels—a fact which suggests that amongst such speakers the inappropropriate use of prosthesis was not seen to be a problematic issue. Instead, writers who do allude to the (unacceptable) use of prosthesis usually direct their remarks to provincials especially those from the south of France and, even more particularly, to Gascons whose pronunciation was evidently often affected by local non-standard usage and was consequently held up for censure. Thurot (1881 [1966]: I, 216) cites two cases, Me´nage (1675) who criticizes Gascons for saying estomacal, estupide, and estatuts, and Dumas (1733) who claims that Gascons ‘font entendre un e devant le s initial’ in words like estile. In non-standard varieties of the langue d’oı¨l (leaving aside the special case of eastern dialects), there has also been a general retreat from the use of I-prosthesis though it is not clear how rapidly it occurred across the different varieties. Already by the end of the sixteenth century, prosthesis seems to be on the way to abandonment amongst the less educated from Paris and its environs even though there is only ex silentio evidence for this assumption. Thus, the journal of Jean He´roard which provides a record of the speech of the young dauphin, later to be Louis XIII, in the first decade of the seventeenth century contains many examples of informal and substandard use. But in the few contexts where prosthesis would be possible, there is no indication of its presence: une scabele (August 1605), la sphaere and une sphaere (July 1606), bon sculteur (April 1608) (Ernst 1985).129 Later in that century, literary pastiche of linguistic features characteristic of substandard pronunciation came to enjoy great popularity, for instance in Cyrano de Bergerac’s Le pe´dant joue´ (1654) and Molie`re’s Don Juan (1665), and in the anonymous mazarinade known as Agre´ables confe´rences of the mid-seventeenth century (Deloffre 1999).130 However, amongst the thirty-four features of substandard pronunciation and grammar identified by Lodge (1996) as forming particular objects for condemnation or derision from writers of the period, the inappropriate use of I-prosthesis does not figure at all.131 There seems described as being characteristic of ‘gens de Provinces’. Nowhere, however, is mention made of prosthesis. 129
The only form showing prosthesis is escluseau ‘kind of mushroom’ (= e´cluseau, now archaic but included in the dictionary of Littre´). This is clearly not a learned word but part of the popular wordstock. The prosthetic vowel here has of course been lexicalized. 130 Though generally viewed as anonymous, the work is tentatively attributed by Deloffre to L. Richer, the author L’Ovide Bouffon. 131 For example, Jean Gareau in Le pe´dant joue´ has usage with prosthesis that is identical to that of the standard language, l’espoisseur, qui espousit, grande espe´e, l’escriture but du scandale (not ** de l’escandale). The ten letters making up the Agre´ables confe´rences have just three contexts involving words with etymological /sC-/: (i) . . . pour avoir de´ Comedian, de´ Murissian, de´ Stature ´e de´ bilboquette pour boutre dans son Palai (letter 1);
122
I-prosthesis
good reason to believe therefore that by the end of the seventeenth century at the latest the influence of the prestigious standard variety was steadily undermining the use of I-prosthesis amongst the inhabitants of the capital and the area immediately surrounding it, and also in other Paris-influenced urban centres of the langue d’oı¨l except those of the north-east where prosthesis never fully developed (see below). Speakers of non-standard varieties of langue d’oı¨l who lived at some remove from Paris and other regional urban centres doubtless preserved the use of I-prosthesis for rather longer. However, moves towards the abandonment of I-prosthesis in regional varieties gain ground particularly from the nineteenth century onward when various centralizing forces began to exercise ever greater sociolinguistic influence promoting standard French. These include the introduction of mass education, military conscription, the establishment of an increasingly intrusive bureaucracy, the development of speedy and effective means of transport, and the rise of mass media. These influences were to have the same destructive effect on the productive use of I-prosthesis in growing numbers of langue d’oc varieties as well (see below 4.4.3.4, and cf. Italian 4.3.3). Unfortunately, clear contemporary data on the status of prosthesis in individual non-standard langue d’oı¨l varieties are not available before the later nineteenth century. Until then, we only have rather imprecise characterizations such as that found in the ‘dictionary’ by Desgranges (1821) in which numerous forms are cited showing prosthesis with etymological s impura words, e.g. escandale, escorpion, espatule, espectaque, estupide, estation. Desgranges merely observes that this is how ‘les gens du peuple’ pronounce such words, but no indication is given of where these speakers came from. However, it is of interest that Nisard (1872: 271) calls attention to the practice amongst the less educated of Paris and its environs of using the prosthetic vowel /e-/, ‘L’e est attire´ par les consonnes combine´es sc, sp, st: escrupule, escorie, escorbut, espectacle, espe´cial, estyle, estatue . . . Cette prothe`se est maintenue dans la prononciation de nos compatriotes du midi. Mais partout le peuple a pour elle plus ou moins de penchant.’ It seems likely that the massive immigration from the provinces to the cities of France that occurred during the nineteenth century goes some way in explaining the presence of the prosthetic vowels observed by Nisard. Nonetheless, (ii) je visme de belles statuses toutes dose´es (letter 10); (iii) avec des grands escritiau (letter 10). Again, usage of the prosthetic vowel here is just as in standard French. In (iii), the vowel has clearly been lexicalized as in standard French e´criteau, while cases (i) and (ii) show the learned form stature (= statue). Item (ii) also has hypercorrect [z] for [r], the development of intervocalic [r] > [z] being a stigmatized and widely pastiched substandard feature of the sixteenth and seventeenth centuries. The use of the forms appearing in (i) and (ii), rather than ** des Estature(s), ** belles estatuses, once more suggests clearly that there was no longer any productive rule of prosthesis in the substandard variety or varieties of langue d’oı¨l being pastiched in the text.
I-prosthesis
123
the observation suggests that the abandonment of the rule of I-prosthesis was a prolonged process in northern France even amongst urban speakers. In more rural areas, the abandonment was slower still and forms with a prosthetic vowel are still reported for certain varieties in second half of the twentieth century. For example, Schortz (1998: 69) cites [¡n ¡staty] ‘une statue’ for Senneville-surFe´camp in Normandy. However, it is not clear whether I-prosthesis has remained a genuinely productive process in this variety and others like it. It may be that it is only in established lexical items that we find a prosthetic vowel which has now become lexicalized, whereas new words containing word-initial s impura onsets no longer undergo prosthesis. 4.4.3.2
Causation of the loss of I-prosthesis in French
It is curious that whereas the background to the rise of I-prosthesis has been the subject of considerable investigation by linguists, far less attention has been given to the circumstances leading to its subsequent abandonment in standard French and other Gallo-Romance varieties. In fact, only two types of explanation appear to have been proposed. In the more familiar one, the key factor undermining prosthesis is taken to lie in the prestigious spelling-based system of pronunciation which Erasmus had proposed in 1528 for Latin and Greek (e.g. Pope 1952: }653). It is assumed that by extension this spelling-based approach to ‘good’ pronunciation was adopted by the educated classes when articulating learned words in French, and from there it spread to the bourgeoisie before finally reaching the mass of the population, especially in Paris and other urban environments where there was a strong cultural presence. In this new system of pronunciation, prosthetic vowels had no place since they were not orthographically represented. In the other explanation, it is suggested that French prosthesis may have ceased to be productive as a result of the regular deletion of preconsonantal [s] in the langue d’oı¨l during the later medieval period as a result of which O.Fr. espine, espee were transformed into [epin], [epe] (Posner 1996: 290–1). This change would have caused the resulting word-initial sequence [e] þ consonant- to be in some sense ‘stranded’ from etymological [s] þ consonant-, thereby destabilizing the previously more transparent productive rule of prosthesis. Neither explanation appears entirely acceptable, however. As we have seen, there are clear indications that prosthesis had already been effectively abandoned in the usage of ‘good’ speakers of French well before 1528 when the celebrated work of Erasmus was published. The Erasmist reforms can therefore be viewed at most as a powerful force confirming existing practice amongst the educated. More generally, there are difficulties in interpreting the abandonment of
124
I-prosthesis
prosthesis in late medieval French as, in Labovian terms, a purely ‘top-down’ change.132 For it seems very doubtful whether a change initiated by a small if prestigious minority of speakers could have been diffused so completely down through the population of Paris and environs in the timescale envisaged such that it left so little discernible trace by the end of the sixteenth century, particularly in view of the sociolinguistic circumstances of the sixteenth century when perhaps as many as 90 per cent of the population of France were illiterate (Glatigny 1989: 18).133 The other explanation, which interprets the loss of prosthesis as a ‘bottom-up’ development, also encounters problems. First, there is a significant time-lag between the loss of pre-consonantal [s] which is generally recognized to have occurred by the later twelfth or the thirteenth century (Pope 1952: }377; Fouche´ 1966: 861–2; Zink 1986: 122–3) and the abandonment of vowel prosthesis which probably got underway from the late fourteenth or early fifteenth century. Second, when French speakers were confronted with learned or other foreign words beginning with s impura, we might have expected them in most cases to have incorporated them by adapting the complex onset to a sequence compatible with existing patterns in the language.134 It is therefore surprising that these speakers should instead have relicensed a heterosyllabic onset sequence which had been prohibited for centuries. In fact, it seems likely that no single factor brought about the abandonment of I-prosthesis in northern France. Rather, this development was probably prompted by the action of several quite separate factors operating in conjunction with one another. First, a major ‘top-down’ factor was the adoption for prestige reasons of a spelling-based pronunciation by literate speakers when they articulated learned loanwords. As was noted above, this mode of pronunciation was already current amongst the educated in the fifteenth century and received powerful endorsement from Erasmus’s influential work published in 1528. Vernacularized pronunciations of Latinisms certainly continued to occur amongst less educated speakers into the sixteenth century but they enjoyed no prestige and even became the butt of humour.135 A modern parallel may be found in pastiche pronunciations of recent Anglicisms in French such as meeting [met~¡´]. For discussion of ‘top-down’ and ‘bottom-up’ change, see Labov (1994: 78 and 2001: 272–5). 133 Unfortunately, it is not made clear whether this figure relates to individuals able just to read or those able both to read and write, nor is there any indication of possible geographical variation in literacy levels. 134 Compare what happens amongst modern Andalusian speakers of Spanish where preconsonantal [s] has been deleted in many if not most varieties. Here, s impura loans are usually adapted to [eC-], just like native words with etymological s impura (cf. Sampson 2005). 135 Occasional present-day French words like quolibet [kolib¡] ‘jibe’ (< QUO LIBET), cancan [kA ~ kA ~ ] ‘(slanderous) gossip’ (< QUAMQUAM) recall vernacularized pronunciations of Latinisms. The writer E´tienne Tabourot (1547–90) from Dijon provides entertaining and sometimes 132
I-prosthesis
125
Second, various formal factors also appear to have played a role. Amongst these, the deletion of pre-consonantal [s] may have been of significance but this is uncertain, as we have seen. More important seem to be broader changes taking place in syllable structure in the later medieval period. For, although a number of changes occurred in popular speech which resulted in greater simplicity of syllable structure, notably the progressive deletion of coda consonants [r], [k], [t], and (word-final) [s] during the fifteenth and sixteenth centuries,136 various other developments were taking place which led to greater complexity in syllable structure. Amongst these was the ongoing incorporation of loanwords such as est, ouest (first attested in twelfth century) from English, lest from Dutch (thirteenth century) along with the many learned words such as gest (thirteenth century, later geste), laps, pact (fourteenth century, later pacte) which increased the complexity in codas especially. But more significant was the gradual abandonment of schwa. Fouche´ (1969: 509–27) describes in some detail the stages of this long-running process which, depending on phonetic context, operated from preliterary times up to the late seventeenth century when word-final schwa ceased to be used in ordinary speech, as in la port(e). Schwa deletion was unlikely to have been at first a characteristic of educated usage, however. It probably represented a feature originating in popular speech so that there is every chance that in informal usage amongst the less educated it already enjoyed some currency before the sixteenth century when a number of grammarians call attention to the phenomenon (Thurot 1881–3: I, 162–4). As a result of schwa deletion, not only would more complex codas have developed, as in words like arme, triste, farce, porche, more complex onsets would also have arisen. Numerous observations by grammarians of the sixteenth and seventeenth centuries indicate that word-initial sequences of fricative þ [‰] þ consonant were particularly susceptible to deletion of schwa (cf. Fouche´ 1969: 526). This is confirmed by the appearance of attested forms such as stier ‘a measure of grain’ < SEXTA¯RIU(M), squenie ‘a smock’ < Middle High German sukenıˆe, and the demonstratives c’t, c’tte, c’tui (also spelt st, ste, stui) in sixteenth-century writings (= modern French setier, souquenille, cet, cette, and archaic cestui). The deletion of schwa in such items doubtless played a role in re-establishing /sC-/ sequences as possible onsets in the usage of less
obscene pastiches of such pronunciations, e.g. DUCUM EST AMOR RUS COELI AQUILAE VITAM represented as Du con est amoureux celui a` qui le vit tend. The representation of SI CUM STIPE ´ tu es is of special interest as there is no sign of a prosthetic vowel. The reason TU ES as Si constipe for this appears to be that such vernacular pronunciations only reflected letters that were actually present in the Latin form; examples from Tabourot 1970 [1588]: 45, 47. 136 Such changes clearly tie in with a general Romance tendency to unblock syllables through the weakening of coda consonants, as has long been recognized by Romanists, e.g. Granda (1966) and Sala (1976: 21–50).
126
I-prosthesis
educated speakers. The presence of a /sC-/ initial sequence in demonstrative forms was particularly important since these occurred with high frequency and would have familiarized the new onset type for these speakers. Indeed, although spellings like st, ste for cet, cette and (presumably) the corresponding pronunciation were condemned by some sixteenth-century grammarians such as Jacques Peletier (1550: 200), st and ste went on to establish themselves as acceptable variants of the demonstrative adjective for use in informal conversation by educated speakers of the seventeenth and eighteenth centuries, as is noted by many grammarians (Thurot 1881–3: I, 210).137 In this way, both ‘top-down’ and ‘bottom-up’ factors conspired to re-establish complex onset sequences of the form /sC-/ across the different social subgroupings in Paris and its environs and also, doubtless, in other langue d’oı¨l varieties that were subject to the same factors. Thereafter, any remaining varieties (excluding those of the east and north-east) which preserved I-prosthesis came progressively to abandon the process as a result of growing influence from the prestigious standard French language. 4.4.3.3
An exceptional case: eastern and north-eastern dialects
In Wallonia, eastern Lorraine and Romance-speaking Alsace, I-prosthesis failed to become established as it did in varieties spoken elsewhere in the langue d’oı¨l area.138 During the early medieval period, prosthetic vowels appear to have developed but at some stage in the later pre-literary period (i.e. prior to the twelfth century) prosthesis was abandoned as a process. Subsequently, a further development came to affect words beginning with initial /sC-/ sequences in the varieties of central and eastern Wallonia but not those found in western Wallonia or in Lorraine and Alsace. This was the appearance of a rule of epenthesis whose effect was to create a new type of sandhi-conditioned alternation that still operates today, as in spe`ne sipe`ne, su¨pe`ne ‘thorn’ (< SPI¯NA). The non-epenthetic alternant appears after a vowel-final word and the other alternant elsewhere, e.g. de`s spe`nes [d¡sp¡n] ‘some thorns’ but avou’n sipe`ne [avunsip¡n] ‘with a thorn’ or post-pausal sipe`nes [sip¡n] ‘thorns’ in Lie´geois.
137
For instance, Thomas Corneille (1625–1709) writes, ‘Dans le discours familier on prononce st homme, ste femme, et ce seroit une affectation vicieuse de dire cet homme, cette femme, quoy que dans la chaire on doive prononcer ainsi ces mots’ (1687: II, 164), and Pierre Restaut (1696–1764) similarly observes, ‘Cet se prononce st, & cette comme ste. Ainsi, quoiqu’on e´crive cet oiseau, cet honneur, cette femme, il faut prononcer stoiseau, sthonneur, ste femme’ ([1730]1773: 449). 138 To the west, prosthesis regularly operated in Picard, except in Rouchi which is spoken around Valenciennes in French Hainaut (Gossen 1970: }47).
I-prosthesis
127
The epenthetic vowel usually has a high front quality. It appears as [i] in the majority of central varieties including that of Lie`ge as we have seen, but [y] is found in dialects of the extreme east (e.g. Herve and Verviers) and also in a narrow belt in the extreme south of the epenthetic area. More unusually, a lowered front unrounded value [e] or [¡] occurs in parts of the north-west and far south-west of the epenthetic area, and a short front schwa-like rounded vowel [] is reported in a few varieties in the extreme north-west and south (cf. Map 2). The predominant choice of a high front quality accords with the principle of minimal saliency (1.6). The expected quality for an epenthetic vowel, namely [‰], was not possible in these dialects since existing instances of schwa were themselves in the process of being systematically deleted, as we shall see. The quality adopted by the epenthetic vowel was based on that of a more stable existing vowel type. High front vowels are the least salient; the selection of [i] or, less expectedly, [y] is therefore understandable. The mid front quality found in a small minority of varieties would seem to represent a localized development of earlier [i], for reasons which are not immediately obvious. The historical background to the abandonment of prosthesis and to the later appearance of epenthesis in eastern and, particularly, north-eastern varieties is not fully understood. There are significant numbers of texts from the areas concerned dating from the second half of the twelfth century onward which might be expected to shed useful light. However, it is generally accepted that they fail to give a faithful picture of local usage, as they are very often marked by linguistic influence from other varieties of northern French that enjoyed cultural prestige and, in particular, the variety used in the area centred on Paris.139 One consequence of this outside linguistic influence has been the frequent scribal use of prosthetic vowels at a time when they were doubtless not in general use in local speech. Thus, in one of the first literary prose writings, the Psalter Commentary composed between the 1160s and 1180s for Laurette of Alsace, we find many forms contining an initial <e> ; for instance, l’estoile, la veraie esperance, saint Estevene, des espeies, et estroites occur in just one psalm (Gregory 1990: psalm 36). In other texts of this century ostensibly prosthetic forms appear with similar frequency.140 More 139 This has been established for Walloon as a result of the close textual analyses carried out by Remacle (1948) and others. 140 For instance, the translation of the sermons of St Bernard dating from the end of the twelfth century has numerous cases although a minority of non-prosthetic forms do occur, especially involving learned words. Thus, sermon 26 has li espouse, des estoiles, ki espirs, ceste estroite, ki escrite est but est stoile, plus splendianz, parmanable splendor, bele spirituel, si studioet, la celeste sperance, etc. (Gregory 1994). In the verse Poe`me moral of c.1200, ‘normalement, il y a prothe`se de e devant s þ cons.’ according to Bayot (1929: lxx), as in bone esperance (l. 410), n’i espargniez (l. 1287). As regards the dozen non-prosthetic forms
128
I-prosthesis
revealing of local usage in Walloon are non-literary vernacular documents. These first appear in 1236 and they become available in significant numbers from the second half of the century (Boutier 1995: 291), although often they too betray linguistic influence from central forms of langue d’oı¨l. In view of the limited usefulness of much textual evidence, use must also be made of internal linguistic evidence to clarify likely patterns of pronunciation. In particular, the treatment of forms originally containing prefixal [es-] < EX- is suggestive. These forms have systematically undergone aphaeresis and their onsets have become formally indistinguishable from those of s impura words, e.g. (dialect of Neuweiler, Lorraine) SCA˘NDULA > [åadr], EX-CALDA¯RE > [åad¡] (Horning 1887: 66) and (dialect of Lie`ge) STE¯LLA > steuˆle (siteuˆle), EX-TE˘NDERE > stinde (sitinde) (Haust 1933). The aphaeresis here is part of a more general deletion process affecting word-initial unstressed etymological [e-] in these dialects (cf. similar changes in Old Occitan, 4.4.3.4). Evidence of this development is present in the 1236 charter where we find two forms
I-prosthesis
129
where the original fricative of word-initial [s]C- sequences underwent phonetic change of some sort, the previously operative rule of change may not apply to lexical items containing [s]C- which have been borrowed more recently. Thus, in the Alsatian dialect of Ranrupt, original [s]C- > [x]C- but in later lexical loans containing [s]C- this sequence remains unchanged, [sk‰l¡t] ‘skeleton’, [skr¡] ‘secret’ (Aub-Bu¨scher 1962). In Wallonia, comparable phonetic adaption of etymological SC- sequences has also occurred, giving [h-] or [ʃ-] (cf. ALW I: map 32 e´cume). However, [sk-] onsets have later re-appeared through borrowing. The reasons for the abandonment of I-prosthesis in these north-eastern varieties of Gallo-Romance remain uncertain. However, they were geographically peripheral dialects spoken in transition areas between Romance and Germanic where a substantial proportion of the population has long been bilingual to varying degrees. As Germanic languages have always readily permitted word-initial s impura onsets, the abandonment of the prosthetic alternant may well have been partly promoted as a result of interference between the two language systems. This view finds some support from other cases where prosthetic vowels have been lost in Romance areas exposed to linguistic interference from Germanic or other languages (cf. 4.3.1 Balkan Romance, 4.4.4 RhetoRomance). In eastern and central Walloon, following the abandonment of I-prosthesis and the re-establishment of [s] þ consonant sequences as licensed onsets (including [sk-] in borrowings), these underwent further change through epenthesis. The earliest reported attestation of this development is the place-name Sitiers (< STIRPUS) which appears in a document dated 1211 (Boutier 1995: 295). Further examples are found in the following century in the Registre des E´chevins de Revin (Revin lies in the north of de´partement of Ardenne), sepale = French e´paule c. 1350 ‘shoulder’ (< SPA¯T(U)LA) and sekevins = French ´echevins from 1363 ‘deputy mayor’ (< Frankish *skapin ‘judge’; cf. Old High German scaffin, mod. German Scho¨ffe ‘juror’); Bruneau (1913a: 389), Remacle (1948: 40–2). Epenthesis appears to have arisen as a means of resolving the problem of syllabifying the heterosyllabic s impura word-initial sequences which had been re-established. A crucial factor in the use of epenthesis rather than some other repair strategy, however, lies in a broader development that occurred in medieval Walloon. This was the weakening which regularly affected vowels in unstressed syllables and led first to vowels taking on a schwa-like quality. Subsequently, further weakening caused widespread deletion, but in contexts where an unsyllabifiable consonant sequence would arise, the schwa vowel was strengthened and assigned a more salient quality (see Map 2 for the various qualities found). The result has been the appearance of alternation between forms where deletion was able to occur and forms where deletion of the vowel has been blocked and strengthening has taken
130
I-prosthesis
place.143 This alternating pattern is still to be found across central and eastern Walloon varieties to the present day, as in: with strengthening pol tchu¨voˆ ‘for the horse’ por lu¨ vinde ‘in order to sell it’
without strengthening lu¨ tch’foˆ ‘the horse’ i l’mindje ‘he is eating it’
These appear in the south-eastern dialect of Bastogne (about 70 km south of Lie`ge) where [y] is used as the realization of the strengthened vowel (Francard 1981). As the examples indicate, proclitic forms as well as lexical words could be affected. And since weakening to schwa never occurred in absolute word-initial position, the alternating vowel is only found after the initial consonant of the word concerned: l’ ! lu¨, tch’foˆ (with regressive assimilation of [v] > [f] following voiceless tch) ! tchu¨voˆ, c’nu¨che ! cu¨nu¨che ‘to know’, etc. It seems likely that the same pattern of alternation was applied to s impura forms like scole ‘school’. As a result, in contexts where a consonant-final word preceded in the same phonological phrase or where there was a preceding pause, a new alternant was created through the insertion of the same high vowel [i] or [y] that had developed in forms which had experienced regular schwa strengthening. And, just as the new high vowel was located after the word-initial consonant with forms like tchu¨voˆ, so too epenthesis was adopted for s impura forms like scole ! su¨cole rather than using prosthesis (** u¨scole). Alternations of the type illustrated above are certainly still in evidence in present-day Walloon, but the genuine productivity of the rule of epenthesis with s impura forms is equivocal. Francard (1980: 197, n. 49), who bases his observations specifically on the dialect of Bastogne, states that ‘cette re`gle est toujours productive’ and cites alternations such as scole su¨cole. But it is also noted (loc. cit.) that ‘les mots franc¸ais d’introduction re´cente (scarole, scapule`re, scarlatine) ne sont plus re´gis par cette re`gle,’ i.e. they have no epenthetic alternants. This would appear to indicate that the rule operates only with a closed set of established lexical items and hence is no longer truly productive. 4.4.3.4
Langue d’oc
In southern Gallo-Romance, the development of I-prosthesis appears to have followed the same path during the Middle Ages as it did in all langue
143 There is a clear affinity between the Walloon situation described here and one type of prosthesis found in the nearby Picard speech area. In the latter case, regular deletion of schwa has occurred in many proclitic forms and in unstressed initial syllables of lexical items, e.g. those beginning in prefixal re-. To enable syllabification a prosthetic vowel, [a] or some type of mid front vowel [e], has been introduced (cf. sections 6.1.4 and 6.1.5).
I-prosthesis
131
d’oı¨l varieties except those of the east and north-east. During the course of the pre-literary period (i.e. pre-eleventh century), the great majority of Occitan varieties evidently acquired a generalized rule of prosthesis for s impura words and by the end of the medieval period the rule seems to have become established throughout the langue d’oc area with the exception of three zones: a compact central area lying SE of Clermont-Ferrand and two bands of peripheral dialects, one located in the far east of Provence and the other adjacent to the central Pyrenees. Outside these zones, the rule continued to be productive and has remained so in many varieties up to the present day. However, in recent times with the ever more invasive presence of standard French, the integrity of the rule has been increasingly undermined in the varieties where it had continued to operate. Looking more closely at historical developments, we find that at the beginning of the literary period the earliest extant texts indicate that there may still have been residual alternation between prosthetic and non-prosthetic forms in some varieties. Alternation certainly seems to be indicated in the two oldest extant literary works although both of these are in verse and hence would be subject to metrical constraints likely to influence the inclusion, or omission, of unstressed vowels at word edges. Even so, it is notable that non-prosthetic forms are found solely after a word ending in unstressed /-a/.144 Thus, in the Chanson de Sainte Foi d’Agen (c. 1060) there are five non-prosthetic forms attested, e.g. la spina (l. 56), umbra streins (l. 62), and all occur after a preceding word-final /-a/; these contrast with sixteen prosthetic forms which appear in other phonological contexts. Similarly, in the Boeci of c. 1100 (Schwarze 1963: 11–15) a preceding unstressed /-a/ provides the only context which fails to trigger a prosthetic form, riqueza star (83), a ferma schala (l. 149), etc., with just one anomaly auia escript (l. 205). The many charters and official documents of the twelfth century from Quercy and Albi show a comparable situation, with non-prosthetic forms regularly being used after the article la but not elsewhere (Grafstro¨m 1958: }21). Thus, in an Albigeois text dated 1120 (Brunel 1926–52: text 20), ella strada appears (96) although prosthetic forms are also found, post-vocalically in so escriot (66) and post-pausally in Esteves (93).145 The question therefore arises whether I-
If a preceding [a] is stressed, prosthesis is found as in fa estar (l. 162). The same document also has a non-prosthetic form in II.as. stairadas (56–7). Grafstro¨m (loc. cit.) suggests that stairadas ‘measures of grain’ may owe its lack of a prosthetic vowel to haplology in syntagmas like tres s(es)tairadas (< SEXTARI-ATAS). However, reflexes of non-derived SEXTARIUS frequently appear in early texts without a prosthetic vowel; cf. Brunel (1926-52: text 34) from Quercy where forms such as II. steirs, I. steir (11), IIII. sters (21) although VI. sesteirs (9) is also found. 144 145
132
I-prosthesis
prosthesis at this period was still becoming generalized in all post-vocalic contexts, or whether generalization of prosthesis had already occurred but subsequently aphaeresis had operated on prosthetic forms preceded by words ending in unstressed /-a/. The aphaeretic treatment of la gleisa < EC(C)LE˘SIA (8, 34) in this text (and many others) together with other attested cases of aphaeresis like la spleita, sa spleita ‘farmstead’ (< EXPLICITA) in a text of c. 1140 from Quercy (Brunel 1926-52: text 34, 10, 13, 14) suggest that the latter interpretation may be more appropriate. However, it remains unclear why specifically unstressed final /-a/, and not other final unstressed vowels, should provoke aphaeresis. If a tendency for aphaeresis to operate was starting to get under way, the impression is that it was soon abandoned in most (though not all) varieties and, crucially, in the developing Occitan literary koine` (cf. Wunderli 1969: 54). In the later Middle Ages and into the early modern period, two major tendencies affecting prosthesis appear to have operated in the langue d’oc. On the one hand, in the great majority of varieties I-prosthesis remained (or became established) as a fully productive rule operating in all linguistic contexts including post-vocalically, just as in Castilian and many other varieties of Ibero-Romance, cf. 4.4.2). Thus, Mushacke (1884: }10) reports that prosthesis is regularly attested in medieval non-literary texts from Montpellier and that it was still operative in the nineteenth century. Similarly, the mainly fifteenth-century Thesaur del hospital de Sant Sperit from Marseille shows widespread use of prosthetic forms, notwithstanding the final word of its title. But occasionally prosthesis is not represented in forms following a word ending in [s] and also sometimes post-pausally. Otherwise, items lacking a prosthetic vowel represent either latinisms or graphies influenced by the traditional non-prosthetic scribal usage of Provence (see next paragraph); cf. Glessgen (1989: 272). On the other hand, a minority of Occitan varieties experienced aphaeresis leading to the loss of the rule of prosthesis. Cases of aphaeresis of prosthetic and etymological [e-] can be found in medieval documents. Texts from the far eastern area of Provence are notable in this respect,146 and in the medieval Occitan variety Wunderli (1969: 52–4) notes widespread aphaeresis in Occitan Bible translations dating from the twelfth to sixteenth century most of which are of Provenc¸al origin. Zufferey (1987: 210) also reports aphaeresis in the fourteenth-century Guiraud chansonnier (MS f) which was probably composed in the Arles area (e.g. speransa, star, stage, spauen), and also in medieval documents from the arrondissement of Digne (AlpesMaritimes), in two epic poems preserved in a compilation made by a notary from Apt (Alpes-Maritimes), and in some manuscripts of the Vie de Saint Honorat. Ronjat (II: }451) notes that aphaeresis can be found in modern varieties of parts of the Pyrenean area and eastern Aquitaine. Here, interestingly, aphaeresis occurs in prosthetic words when they are preceded by a feminine singular definite article. No rationale is offered for this phenomenon, however, and its possible link with the linguistic conditions found in the earliest Occitan texts, noted above, remains unclear. 146
I-prosthesis
133
used by the Waldensians in south-eastern France, a similar loss of the vowel has been reported.147 The present-day dialects of this area continue to provide evidence of aphaeresis. Thus, the ALF pts 889 (Menton) and 890 (Fontan, Breil) show non-prosthetic reflexes for SCA¯LA, SPI¯NA, STE¯LLA, STRICTA (maps 436 e´chelle, 476 e´pine, 494 e´toile, 524 (trop) e´troite, and this is confirmed in the more recent detailed dialectological study of Alpes-Maritimes by Dalbera (1994) which found aphaeresis to be regular in the varieties of the Roya valley down to Menton.148 The close proximity of this region to Liguria in north-western Italy is highly significant, as the varieties spoken in the latter area likewise show evidence of early aphaeresis (4.4.5). As a result, delicate interplay has occurred in these transitional dialects between the conflicting patterns of prosthetic vowel use that developed in medieval Occitan (where use of the vowel was generally productive) and Ligurian. The fluid political history of the region heightened the interplay. Nice and the territory which it controlled passed to the County of Savoy in 1388 but this reverted to France in 1860, while Tende in the upper Roya valley passed to Savoy in 1626 and became part of France once more in 1947. Some equally peripheral varieties of Gallo-Romance lying further to the north in the valleys of Vaud also continue to have aphaeretic forms (Wu¨est 1995: 439), although in the adjacent area of Piedmont prosthetic forms were reported to be usual in certain varieties at the end of the nineteenth century.149 Nauton (1974: 53, 223) reports a zone in southern Puy-de-Doˆme and north-western Haute-Loire where words beginning with unstressed a- and e- have been systematically affected, as in [tsatar~¡] ‘we will buy (= French ache`terons)’. As a consequence of this change, previously occurring prosthetic vowels have been deleted, as in sko`la ‘school’. Other varieties also undergoing aphaeresis are reported in parts of the
147
‘Sembra assente la protesi dinanzi a S þ consonante, anche in situazione etimologica’ (Cornagliotti 1995: 470). The examples cited indicate that the absence of prosthetic vowels is due to the action of aphaeresis: EXTRA¯NEU(M) > strang, EXPRESSA¯ RE(M) > scarnidor. ME˘NTE > spresament, Germ. skirn- þ -TO 148 , *SCU ¯ RIU(M) emerge with a prosthetic vowel, [skola], [eskiro‰] in However, SCOLA Fontan, although Menton has non-prosthetic forms reported for these items (ALF 441 ´ecole, 450 ´ecureuil). Ronjat (II, }321) provides a detailed overview of localities showing aphaeresis and generalization of the non-prosthetic alternant. 149 Morosi (1888) records forms such as [ej’ka:lo], [ej’piNo] [ejly’Ja] < SCA¯LA, SPI¯NA, EXLUNGIA¯RE ‘ladder, thorn, to move to a distance’ for the dialect of Prali (Germanasca Valley). However, a little to the south, the dialect of Bobbio and Villar-Pellice (Pellice Valley) had undergone aphaeresis to give [’ste:la] etc. and further down the same valley at Torre Pellice aphaeresis was also general, although lexicalized prosthetic forms remain in the more isolated mountain areas, such as [es’te:la], [es’pina]. The motivation for the drift towards aphaeretic forms in these Piedmontese dialects is not clear.
134
I-prosthesis
Pyrenean area and eastern Aquitaine (Ronjat II: }451). The reasons for the occurrence of aphaeresis in all these varieties, however, remain uncertain. While the two opposing tendencies affecting the occurrence of I-prosthesis were operating across different varieties of the langue d’oc, a further force for change began to make itself felt from the late medieval period onward, namely the influence of standard French. The increasing pressures of centralization in the crystallizing French state led to ever greater use of standard French in the south. Socially more advantaged speakers from urban areas were generally the first to be affected by this new linguistic presence while speakers from rural areas, less exposed to standardizing influences, have maintained the integrity of their Occitan speech longer. An obvious source of tension for speakers of the many Occitan varieties where prosthesis functioned as a fully productive rule was the absence of prosthetic vowels in standard French pronunciation. Such was the sociolinguistic pressure to conform with standard usage that sixteenth-century southern speakers were already adapting to the new norm when using standard French, at least in their writings. Nonetheless, traces of prosthetic usage still appear not infrequently in the works of southerners such as Monluc and especially Brantoˆme (cf. 4.4.3.1). Less clear is how standard French was actually spoken by southerners whose native Occitan variety had a productive rule of prosthesis. Occasional comments, usually censorious, are made by sixteenth- and seventeenth-century grammarians on this question but these shed little clear light on contemporary patterns of usage. Thus, Henri Estienne writes of the ‘vice’ of saying estatuts, espectacle, espacieux, etc. and of hypercorrecting estomach (sic) to stomach, and he attributes this pronunciation to speakers from Dauphine´, Languedoc, and ‘ces quartiers la`.’ Also, as noted earlier, the grammarians Me´nage and Dumas single out Gascons for their ‘inappropriate’ use of a prosthetic e- as in estupide, estile (Thurot I: 216). In more recent times, the increasing exposure of southern speakers to the standard language has tended progressively to undermine the use of I-prosthesis in the langue d’oc area. In the regional French found there, the incidence of prosthetic vowels has been reduced although indications of their continued presence are given in twentieth-century studies. For instance, Brun (1931: 38) noted their occurrence in words like esculpture, espe´cial, escrupule as ‘bien connu et toujours persistant’ in Marseille, and a little later Se´guy ([1950] 1978: 18) reported that speakers of the franc¸ais re´gional of Toulouse were still commonly saying estatue, estylo. Standard French influence has also served to undermine the rule of prosthesis in Occitan varieties where it had previously been productive. Thus, in the Gascon dialect of Donzac there has been recent loss of productivity in prosthesis: ‘The Fr. clusters made up of /s/ plus consonant were avoided in the past when word-initial by a prosthetic [e-]; hence Donzacais /espektaty´r/ ‘spectateur’ (15th cent. Fr. word), /eskele´to/ ‘squelette’ (16th cent.). In more recent loans like /spesjalı´sto/ ‘spe´cialiste’
I-prosthesis
135
(19th cent.), /standardiza´/ ‘standardizer’ (20th cent.), such initial consonant clusters have become permissible sequences’ (Kelly 1973: 74). A complex pattern with prosthesis seems to exist in the dialect of Notre-Damede-Sanilhac in the Dordogne (Marshall 1984). Long-established s impura words have undergone unconditioned prosthesis and subsequently pre-consonantal [s] > [j], as in [ei’salP] [ei’pinP] [ei’kolP] < SCALA, SPI¯NA, SCHOLA. Speakers evidently view the initial vowel of these words now as part of their underlying form. More recent loans beginning with /sC-/, however, are subject to conditioned prosthesis with partial palatalization of the sibilant, as in [¡ʃtP’ty] ‘statue’, [¡ʃk¡’l¡tP] ‘skeleton’ (post-pausal citation forms) vs [ynPʃtPty] ‘a statue’, [ynPʃk¡’l¡tP] ‘a skeleton (f.sg.)’ (post-vocalic) where [ynP] is the feminine singular definite article. Speakers are reportedly conscious of the alternation in such words. In other varieties, however, I-prosthesis may remain a fully productive process irrespective of context. This is the case in the modern standardized forms of both Provenc¸al and Languedocian (based on usage in Toulouse). For the former, Coupier (1995) cites examples such as escrupule, esculta ‘to sculpt’, escri ‘script’, escrable ‘scrabble’; espountanie´u, espoutnik, espirau ‘spiral’, esfe`ro; estati ‘static’, estatut, estatuo, estendardisto ‘switchboard operator’. For the latter, contemporary newspapers in Languedocian offer cases such as Escandinaus, estagi (= Fr. stage), estatisticament, especialista, estructura, estereotip alongside the forms belonging to the inherited lexicon such as escala, espatla, escriure, estrech, estat.
4.4.4
R H E TO - RO M A N C E
Although individual varieties of Rheto-Romance may show considerable differences in their phonetic history, there is a high degree of similarity in their treatment of I-prosthesis. The facts suggest that prosthesis did occur in the Late Latin of the Rheto-Romance speech area, though it is unclear whether it ever came to operate unconditionally. In the course of the medieval period, however, aphaeresis evidently operated and unstressed word-initial [e-] in prosthetic and other forms was deleted. As a result, no modern variety of Rheto-Romance contains forms showing I-prosthesis nor are there attestations of such forms in surviving texts. We can therefore only infer their possible existence in pre-literary times from two types of persuasive internal linguistic evidence. On the one hand, the common treatment of s impura forms and prefixal words beginning with EX-, e.g. STE¯LLA, EXTE˘NDERE > (Upper Engadinish) ˇstaila [’ʃtajla], ˇstender [’ʃtender] (Walberg 1907), is consistent with an interpretation that s impura forms underwent vowel prosthesis in Late Latin before losing the vowel again later on. On the other hand, the relatively high level of preservation of word-final consonants in Rheto-Romance suggests that in the early Middle Ages I-prosthesis would probably have remained a productive process just as in other types of Romance where
136
I-prosthesis
final consonants were equally retentive, such as Ibero-Romance and Gallo-Romance. As we have seen, these showed a strong tendency to make systematic use of I-prosthesis (4.4.2, 4.4.3). The lack of any attestation for prosthetic vowels indicates that they were deleted well before the end of the medieval period. In Friulian, where vernacular texts become available from the fourteenth century, the earliest documents contain only non-prosthetic forms. Thus, a private letter dated c. 1300 contains (post-vocalic) ti scriv, ti scrif and (post-consonantal) par scriviti ‘to write to you’, and the fourteenth-century ballad Piruc¸ myo` doc¸ inculurit contains a strit ‘in dispute’ and the phrase dut stoy ardit ‘I am all full of life’ five times in the refrain (Sampson 1980: texts 67, 68). Substantial texts in the vernacular of the Grisons only date from the sixteenth century (and, for Ladin, the eighteenth century; Kuen 1995: 62), and by then no direct trace of earlier prosthetic vowels remains. For example, in Jakob Bifrun’s translation of the New Testament into Engadinish published in 1560, we consistently find non-prosthetic forms being used in all phonological contexts: (post-consonantal) schi fu¨s sto pussibel dalg scriver ‘if it had been possible to write it’, (post-vocalic) la sanchia scritu¨ra ‘the holy scripture’ (post-consonantal and post-vocalic) chi saien stoˆs stampoˆs ne scrits ‘that may have been printed or written (m.pl.)’ all of which appear in the Prologue, and (postpausal) Steˆd dimeˆ cun vos flaungs schintoˆs ‘Stand therefore (having) your loins girt about’ in Ephesians 6. 14. Given the relative lack of medieval vernacular documentation in RhetoRomance, it is difficult to identify the reasons for the early abandonment of Iprosthesis. Certainly, in a Romance perspective the general preservation of Latin word-final consonants in the Grisons and the Ladin area might be expected to have encouraged the maintenance of prosthesis there, at least as a sandhi process operating when original s impura forms occurred post-consonantally or postpausally. Nonetheless, this has not happened. A possible factor to account for this might be sought in the linguistic interference that resulted from the increasingly widespread bilingualism with Germanic during the Middle Ages in the Grisons and the Ladin area, particularly as Germanic has always permitted word-initial /sC-/ sequences. Indeed, it has been observed that where one language community exerts strong cultural pressure on another, ‘moderate structural borrowing’ including the introduction of new syllable structure constraints can and does occur (Thomason and Kaufman 1988: 75).150 In view of the cultural and linguistic
150 The interference between Germanic and Grisons Rheto-Romance has been much studied, notably in the classic study of Weinreich (1953). In this work, some aspects of phonological interference are addressed but unfortunately these relate exclusively to phonemic systems. Nothing is said on areas such as syllable structure or sandhi phenomena.
I-prosthesis
137
pressure long exerted on Rheto-Romance varieties by northern Italian and especially Germanic in both of which word-initial s impura onsets were licensed, appeal to linguistic interference from these outside sources would seem to offer at least a partial explanation for the abandonment of prosthesis. A parallel may be found in the retreat from I-prosthesis in other more peripheral Romance areas similarly exposed to intense outside linguistic interference, Balkan-Romance (4.3.1) and north-eastern Gallo-Romance (4.4.3.3).
4.4.5
N O RT H E R N I TA LO - RO M A N C E
The history of I-prosthesis in northern Italo-Romance is problematic. As northern Italy is a non-peripheral area in western Romance, we might expect that the use of prosthetic vowels with s impura forms generally extended during the Middle Ages and commonly led to the establishment of an unconditioned rule of I-prosthesis, as was the case in Ibero-Romance and most of Gallo-Romance. However, this evidently did not happen. The earliest vernacular texts, which appeared in the later medieval period, already suggest that in general I-prosthesis had failed fully to take root as a phonological rule, although some cases of (typically contextually conditioned) prosthetic vowels do continue to occur in certain medieval texts, especially from the north-west of Italy. Complicating our understanding of the use of I-prosthesis in this area has been the gulf that has often existed since the medieval period between actual spoken usage and written conventions which have generally ignored prosthetic vowels even when it seems highly probable that in reality they were present. Even in more recent times, although brief mention is occasionally made by linguists of the existence of prosthetic vowels in certain northern Italian varieties, no formal account seems to have been attempted of their incidence. Against this background, it is particularly difficult to obtain a faithful picture of the history of I-prosthesis in northern Italo-Romance. The philological evidence relating to the use of prosthesis in the Imperial period in northern Italian varieties is equivocal. Inscriptions from CIL V, which covers Cisalpine Gaul, contain just seven cases (three pagan and four Christian) showing vowel prosthesis according to Prinz (1938: 106). However, although the total of seven is modest, it is the same as, for example, the Iberian Peninsula where subsequent generalization in the use of I-prosthesis has occurred. Evidence of a more revealing nature comes from early medieval texts, for here we find markedly fewer indications of prosthesis than in comparable texts from early medieval Spain or France (cf. 4.4.2, 4.4.3 above). For instance, in the set of official papyri written between 445 and 700 and edited by Tja¨der (1955), there are very few instances of prosthetic vowels or of aphaeresis in texts composed in northern Italy. All the relevant examples of prosthesis occur in the signatures (i.e. formal
138
I-prosthesis
depositions) of witnesses and are matched elsewhere by corresponding forms where prosthesis or aphaeresis has not been indicated. Prosthetic forms appear in two northern texts from Ravenna both of which date from about 600: expathario and ØÆÆæ[ . . . ] (= ispatar[ . . . ]) < SPATA¯RIU(M) ‘weapon bearer’ in papyrus 16 (both are post-pausal), and sipraiscripta < SUPRA-SCRI¯PTA twice in papyrus 20. The latter, which seems to show internal prosthesis, occurs in the signature of the sixth and final witness where interestingly other signs of less Classical-style usage are also found, such as donacionis as against the spelling donationis appearing in other signatures. There is just one example of aphaeresis, in non amplius spectata (papyrus 8, Ravenna 564 AD), where the final word derives from EXSPECTA¯TA (cf. 4.1.1).151 Prosthesis is even more poorly attested in the corpus of thirty-two legal charters dating from 568–774 which come from Lombard northern Italy.152 They contain only one example of a prosthetic vowel, and cases of aphaeresis are equally rare with just two examples occurring (Politzer and Politzer 1953: 2, 11). The circumstances in this northern subset of Lombard legal documents are thus in marked contrast to those apparent in the corresponding subset composed in Tuscany where prosthetic vowels are widely indicated, as we have seen (4.3.3). The relatively low level of attestation for prosthetic vowels in early medieval northern Italian texts is striking and difficult to explain. It may reflect superior and more widespread knowledge of Classical Latin-style spelling conventions amongst local scribes. Alternatively, it could indicate that already the use of prosthesis in speech was already becoming more limited in the geographical areas or social circles that the scribes came from, though the reasons for this are not clear. Perhaps a combination of these two explanations comes closest to the truth. Significant numbers of vernacular texts start to appear in northern Italy from the thirteenth century onward and these reveal significant variation in the incidence of I-prosthesis. Regional differences are evident, particularly between Piedmont in the north-west where a productive rule of (contextually conditioned) prosthesis has become established and varieties of the centre and east where the incidence of prosthetic vowels appears more restricted and indeed in some varieties I-prosthesis may already have been abandoned altogether. Complex sociolinguistic factors were to lead to further differences. In particular, the development of prestigious latinizing written linguistic models, much influenced 151
In a detailed linguistic study based on the documents in Tja¨der (1955), Carlton (1973: 203–6) reports eight examples of prosthetic vowels. However, five of these are found in a text composed in Rome in the early seventh century (papyri 18–19), the relevant form being ab Istefano five times. Such data are clearly not helpful in determining linguistic patterns in northern Italy. 152 The documents appear in Schiaparelli (1929-33).
I-prosthesis
139
by literary Tuscan, and their increasing top-down diffusion acted against the use of I-prosthesis since this linguistic phenomenon enjoys at best a marginal status in any of the models. Looking at the different outcomes of I-prosthesis in northern Italy, we may begin with the north-west. Here, there seems to have been a rapid abandonment of prosthesis in Liguria, but in Piedmont the process has continued to be productive. The earliest substantial vernacular text from the latter area is the Sermoni subalpini dating from about 1200 and containing in all twenty-two sermons.153 These have seventy-seven direct examples of prosthetic vowels such as en escrit (19, 46), la sainta Escritura (19, 17), avem esperanza (11, 59), tote le especie (16, 67), sore le espine (17, 23), dis l’espos (8, 120), qui estan (9, 266 and 277), where the references are to the edition of Clivio and Danesi (1974). If sections of text written in Latin are excluded, there are only three cases where prosthesis was possible but is not indicated; all involve learned latinizing forms, speciosa (8, 54), spiritali (12, 73), spiritalment (6, 11). In the prosthetic examples, the vowel is indicated as being of a mid quality, either [e] or perhaps centralized [‰].154 As might be expected, the prefixal vowel [e-] is consistently represented, esforcer (16, 83), esgarder (4, 4), esteigner (5, 129), esveiller (3, 43), < EX-FORTIA¯RE, EX-ward-A¯RE, EXTINGUERE, EX-VIGILA¯RE. Thereafter, prosthetic vowels are only directly indicated on rare occasions in Piedmontese texts. Isolated examples are found in fourteenth- and fifteenthcentury writings (Clivio 1971: 338 n. 11) but the practice of representing the vowel orthographically generally disappears, perhaps as a result of a combination of factors: the preferential use of a Latinizing spelling, the phonologically conditioned nature of the prosthesis (post-consonantal only), and later on the influence exerted by standard literary Italian. However, even if the vowel ceased to be represented in writing, it evidently remained in productive use in spoken usage although it was never extended in distribution to include more than just postconsonantal contexts. The evidence of the Sermoni subalpini is therefore curious since prosthetic vowels appear in post-vocalic contexts too. This exceptional characteristic of the Sermoni may reflect influence from adjacent Gallo-Romance varieties where, as we have seen, I-prosthesis had come to operate in all phonological contexts by the end of the twelfth century. Despite the near-total absence of indications of I-prosthesis in later medieval and early modern texts, linguistic studies of modern Piedmontese show that Iprosthesis remains a productive process not only in the koine` based on Turinese but also across almost all local varieties in Piedmont. The phonetic value of the prosthetic vowel varies a good deal, appearing as [‰] in the koine` and [a], [A], [N]
On the basis of internal phonological and lexical evidence, Danesi (1976: 99) identifies the south-west of Piedmont as the place of composition of the sermons. 154 In the sermons, there is just one example of the graphy
140
I-prosthesis
etc. in other varieties. However, it is represented uniformly in spelling as <e¨> following its somewhat belated (re)introduction to written Piedmontese thanks to spelling reforms dating from around 1930. In modern usage, I-prosthesis with s impura forms continues to be contextually determined, operating only when the preceding word is consonant-final. Thus, in the koine` we find la ste`ila ‘a star’ but set ¨este`ile ‘seven stars’, mi i scrivo ‘I write’ but ti it e¨scrive ‘you (sg.) write’ and pe¨r e¨scrit ‘in writing’ (Brero 1971: 23). However, in contexts where the preceding consonant is a rhotic or a glide, i.e. a maximally sonorous and vowel-like segment, the appearance of the prosthetic vowel is reported to be optional ‘without any ascertainable conditioning factor, other than perhaps the speed at which one is talking,’ as in [‰n ’f¡r (‰)s’pPrk] ‘a dirty iron, knitting-needle’ (Clivio 1971: 338). No prosthetic vowel appears in post-pausal contexts; for instance, [le ’trPp ‰str¡jta] ‘it’s too tight (f.sg.)’ but [le ’trPp # str¡jta] l’e´ tro`p . . . stre`ita where an intervening pause blocks the use of a prosthetic vowel (Clivio loc. cit.). Citation forms are therefore always non-prosthetic. This distributional limitation, if it directly descends from original patterns of usage in Late Latin, would offer support to the three-stage view of actualisation in I-prosthesis (4.1.4). The same distributional pattern found with s impura forms is adopted in the Piedmontese koine` by words containing complex heterosyllabic onsets created by later syncope of an initial unstressed vowel, e.g. [doi ’pnas] ‘two tails’ but [ses ‰p’nas] ‘six tails’ < *DUI / SEX P(I)NN-ACIOS (Clivio 2002: 161; cf. also 6.1.4).155 And predictably, in the Piedmontese koine` as elsewhere in Romania continua, the evolution of the prosthetic vowel has been directly related to that of prefixal [e-]. Thus, the same pattern of alternation is found in forms such as zbaru¨e´ ‘to frighten’ as against gat e¨zbaru¨a` ‘frightened cat’ (< EX-PAVORA¯RE), with an initial vowel in post-consonantal contexts (Aly-Belfa`del 1933: } 45). In other Piedmontese varieties, differences exist in the details of the application of I-prosthesis. For instance, in the transitional Piedmontese-Ligurian dialect of Cairo Montenotte the prosthetic vowel may sometimes be omitted following a word ending in a sonorant as in [a suN ’ʃtPja ’li] ‘I (fem.) have been there’ where [’ʃtPja] < STA¯TA (Parry 2005: 95). In the northern Piedmontese dialect of Viverone spoken NE of Turin, phonetically the difference between prosthetic and
155
This is not true for all Piedmontese varieties. For example, in the variety spoken in Cairo Montenotte, I-prosthesis remains a productive process typically involving a front vowel [i], [e], or [¡], as in ra sc-pala ‘the shoulder’ but in isc-pala ‘on one’s shoulder’ where sc- indicates [ʃ] (Parry 2005: 95–8). The prosthesis arising from the later creation of complex word-initial onsets involves [a-] which has now been lexicalized in the forms concerned, e.g. amsuria [am’surja] ‘scythe’, avge` [av’q¡] ‘to watch over’ (< MESSO¯RIA, ¯ RE). VIGILA
Ð
I-prosthesis
141
non-prosthetic forms is reported to be more evident when the preceding conditioning word is a grammatical particle (article or clitic pronoun) as in l’e¨skule´ ‘the schoolboy’, kwatt e¨skule´ ‘four schoolboys’ but elsewhere skule´.156 In the Valsesia in NE Piedmont, the occurrence of the prosthetic vowel has been limited to just contexts where a proclitic article precedes, as in l’aspo´s ‘the husband’, n’asku¨ ‘a shield’ < SPO¯NSU(M), SCU¯TU(M) (Spoerri 1918: } 119). Elsewhere in northern Italo-Romance, I-prosthesis has tended to be less retentive than in Piedmontese. It has widely lost ground in more formal registers and amongst more educated speakers, but nonetheless it appears to have remained in productive use in many areas as a low-level sandhi feature in more informal registers. In the north-east of Italy, the earliest surviving Venetian texts from the late thirteenth and fourteenth century provide evidence of the retreat of prosthesis. Notarial documents, which form the majority of non-literary writings, show generalized loss. For example, in a deposition dating from 1299 there are no examples of the vowel even post-consonantally, e` stado, per spensarie, ai scrito, da Meo speciale (Stussi 1965: text 19). A rare instance of a prosthetic vowel being represented in writing comes in a private text, a mercantile record of accounts from 1307, where per Istefanotto occurs (Stussi, text 42). Yet, in the same text we also find per Stefanoto. Various explanations are possible for the scribal variation here: (i) that the prosthetic vowel was a current feature in speech but not normally indicated in writing except through inattention, (ii) that its occurrence was unusual in speech and occurred only (variably) in post-consonantal contexts, (iii) that the individual concerned may have come from another part of Italy where vowel prosthesis was normal at this time, e.g. Florence, and would have called himself Istefanotto, so that a writer might sporadically choose to represent this pronunciation.157 The last two possibilities are perhaps more likely than the first. At all events, moves towards the general abandonment of I-prosthesis in Venetian appear to be well under way by the later Middle Ages. The reasons for this remain unclear. Here, as in other northern cities, we may suppose that an important factor in helping to hasten the abandonment, at least in written usage and presumably in more formal spoken registers too, in later medieval centuries was the growing influence of the emerging standard written Italian language where prosthetic vowels (and prefixal [e-]) had only a very marginal status. Data from Nigra ([1901]1973: 253). The relative salience of the prosthetic vowel depends on the degree of syntactic cohesion between the s impura form and the preceding word. When the s impura form follows a semi-lexical rather than proclitic consonant-final word, the prosthetic vowel (“vocale copulativa” or link vowel, as Nigra terms it) is reported to be scarcely perceptible. This explains the superscript representation. 157 A counterpart can be found in the practice followed in the correspondence of the Datinis when they refer to Francesco Datini’s assistant Stoldo di Lorenzo di ser Berizo. His name appears variably as Istoldo or Stoldo depending on context (see also 4.3.3). 156
142
I-prosthesis
However, there may already have been internal forces in Venetian that were undermining the process of I-prosthesis. We can only speculate on what these might have been.158 Beyond the city of Venice and ‘good’ usage there, however, I-prosthesis seems to have survived as a productive process in at least some local forms of speech of the Veneto. Thus, in the transitional area between western Veneto and eastern Lombardy the form [nel es’p¡tʃ] ‘in the mirror’ is reported for pt. 248 (Limone) in AIS 4, map 675, where a prosthetic vowel appears characteristically in postconsonantal position.159 In Lombardy and Emilia-Romagna, a similar picture emerges. In the former area, latinizing practice and (from the thirteenth century) influence from the emerging Tuscan-based literary language coloured written vernacular usage amongst the cultivated elites of major cities like Milan and this doubtless affected educated spoken usage too.160 Already in the earliest surviving vernacular texts from Lombardy there are no direct indications of prosthetic vowels. Thus, in the 2440 line poem known as the Sermone composed in 1274 by the Milanese poet Pietro da Barsegape` (< Basilica Petri), no prosthetic vowels are directly indicated and there is widespread aphaeresis. For example, we find in scoso ‘in the bosom’ (l. 1200), con spade (l. 1372), per scampare ‘to escape’ (l. 1997), se lagaven scortegare ‘they let themselves be flayed’ (l. 2080), where the relevant etyma are (Langob.) skauz (REW 7986), SPA¯THAS, EX-CAMPA¯RE, EX-CORTICA¯RE, respectively. The only graphy that suggests the possible presence of prosthesis is the form inestabile ‘unstable, variable’ (l. 386). There is a clear resemblance between this item and graphies found in other regions dating from Late Imperial and early medieval times where ‘internal’ prosthesis appears as in INISTANTE for INSTANTE (cf. 4.1.1). It may therefore be that we have here a rare residual example of the early tendency to restructure sequences of consonant þ s impura by vowel insertion. If so, the implication is that the tendency was abandoned in Lombardy more rapidly across word boundaries
158
For instance, it may be that the greater levels of preservation of word-final vowels in the Veneto area was significant as this would statistically have favoured the use of nonprosthetic alternants. However, the contemporary circumstances in popular medieval Tuscan usage (cf. 4.3.3) would prove problematic for such a hypothesis. 159 Pt. 248 stands out in bold relief, since nearby locations all have forms with no trace of a prosthetic vowel; [alʃp¡tʃ] for pts. 427 (Baura, Ferrara) and 443 (Tizzano), and [intalsp¡tʃ] for pt. 423 (Parma). 160 Cf. Lurati (1988: 499), ‘Almeno dal Duecento il milanese non ha cessato di avvicinarsi al modello fonetico italiano, abbandonando progressivamente tratti locali.’ In turn, the enormous linguistic influence of Milan on usage elsewhere in western Lombardy over many centuries doubtless played some part in the disappearance of the rule of Iprosthesis in varieties of this area (1988: 489–91).
I-prosthesis
143
than word-medially.161 A similar situation is found in the works of the poet Bonvesin da la Riva (d. c. 1315), also from Milan. For example, his De Sathana cum Virgine contains no examples of prosthesis and many instances of aphaeresis, no’m lassa in stao ‘does not let me be’ (l.14), strabello e in grand splendor ‘very fine and in great splendour’ (l. 115), e trop stragrand desnor ‘and very great dishonour’ (l.199).162 The fifteenth-century prose text Elucidario, although of less certain geographical background but certainly from the Milan region (Gasca Queirazza 1995: 103), likewise has forms such as mete in scrigo ‘put in writing’ (Prologue), caze´ del prumer stao ‘he fell from the first state’ (I, 36), per scampare (II, 13), illi in scampay ‘they have escaped’ (II, 29), con spagurose contegne ‘with fearsome appearances’ (III, 12), but cases of non-aphaeresis are found which may reflect learned influence, e.g. per esperienza (I, 87), cotale ustruminti (II, 9).163 In eastern Lombardy, there are comparable indications of the abandonment of I-prosthesis in written usage. A fourteenth-century verse composition from Bergamo on the Ten Commandments contains graphies such as in sperzur ‘in perjury’ (l. 21), se trova scripto (l. 31), no posemo stare (l. 83), in scifi ‘in vileness’ (l. 91), s’ol strasinava ‘he dragged him’ (l. 92) (Lorck 1893), and a fifteenth-century document on notarial practice includes per stipulatio´, lasara` star, del spectabel miser lo zudes ‘of the respectful lord judge’ (Tomasoni 1985). And the two earliest known vernacular prose texts from Brescia which date from 1393 and 1412 contain forms such as gom scome´s ‘we began’, tu fos stath ‘you had been (subj.)’, de queli plagi spander el to sanc ‘from those wounds to shed your blood’, in special (Bonelli and Contini 1935). To the south, the situation in the peripheral usage of Cremona and Mantua is broadly comparable. Cremonese is not well attested until the later Middle Ages, and when texts do become available from the mid fifteenth century, the representation of their phonology and morphology is already ‘fatalmente condizionata dal modello toscaneggiante’ (Grignani 1980: 55). Thus, in surviving financial accounts of building works, there is no sign of prosthesis and numerous instances of aphaeresis: per spaltar el stadiol ‘to floor the wooden loggia’ (IV, 27), per stange ‘for bars’ (IV, 21), per spazadurra ‘for sweeping’ (V, 12). Mantua, in contrast, has texts in volgare from the thirteenth century. Five letters written between merchants survive, the earliest dating from about 1282. These provide evidence of the abandonment of prosthesis, in stanfortin ‘in stamford (type of prized woollen fabric originating in Stamford)’ (I, 21), per sbriga` (II, 10), in quator stanforti (I, 23). A little later, in the first decade of
161 Keller (1934: 86), it may be noted, suggests a possible etymon in þ instabile rather than IN-STABILE for this word. This seems less plausible. 162 Examples from the edition of Contini (1941). All the readings cited concur with the reading of Go¨kc¸en (1996). 163 Examples drawn from the edition by Degli Innocenti (1984).
144
I-prosthesis
the fourteenth century, a vernacular version of a medieval Latin encyclopaedic work was composed by the notary Vivaldo Belcalzer. Here too, prosthesis is absent and aphaeresis normal, as in the section on thunder where forms are found such as col so sforzoso impet squarza e fend le nuvolie ‘with its violent blast it shreds and splits the clouds’, el splendor, and adus sonor, strepit e fragor ‘it brings din, uproar and shattering’ (Schizzerotto 1985: 13–18, 37–8). However, at an informal spoken level there is evidence to suggest that Iprosthesis remained productive, albeit contextually determined. For instance, in a compact group of varieties spoken around Trento a prosthetic and prefixal vowel [e-] has continued in use, occurring exclusively in post-consonantal contexts, as in per escriver ‘to write’, per estrada ‘on the way’, per escome´sa ‘as a wager’, son estrac ‘I am tired’, l’aven espaventa` ‘we frightened him’, piat espars ‘spilled dish’ (Tomasini 1951). The distribution here is therefore directly comparable to that found widely in Piedmont. In Emilia-Romagna, a similar picture to that of Lombardy emerges. Textual evidence suggests a relatively early retreat from I-prosthesis in more formal usage, as in a Bolognese proclamation dated 1294 which contains the forms d’onne condizione e stato and en scripto twice (Frati 1900: 249). In this text, there are other unmistakable signs of Tuscanizing influence, as in forms like trovato, cuocho.164 Bolognese texts from the following century offer further apparent evidence of the abandonment of I-prosthesis and are also marked by Tuscan influence. There are three business letters dating from 1320–50 which each contain the formulaic phrase averme per scluxa` (= per scusato) ‘to excuse me’ (Stella 1969),165 while a mid fourteenth-century medicinal work contains e schiva ‘and avoids’, grogo scropollo ‘one twenty-fourth of an ounce of saffron’, toi stercho ‘take (imper.) droppings’, and post-pausal scorc¸e de mele granate ‘pomegranate peel’ (Longobardi 1994). There are no instances of I-prosthesis in either text. The available evidence therefore appears to suggest that the rule was steadily losing productivity in more educated, formal usage during the Middle Ages in this area of northern Italy too. Though the original motivating force leading to the elimination of I-prosthesis in the formal registers used by the more favoured literate classes of society are unclear, the later influence of the Tuscan-based literary standard language doubtless contributed to the non-prosthetizing tendency.
164 The evident influence of Tuscan has been noted in an even earlier document dated 1260 which survives from Imola (Bertoni 1908–11). 165 Cf. the Tuscan counterpart appearing in the slightly later Datini letters, e.g. Francesco (17.4.1397) tera’mi per ischusatto ‘you will excuse me’, Margherita (23.11.1398) che voi l’abiate per ischusato ‘that you excuse it’.
I-prosthesis
145
However, amongst the less uneducated majority of Emilia-Romagna, there is some evidence to suggest that the use of I-prosthesis with s impura forms continued in post-consonantal contexts. Particularly significant are more recent reports by linguists of varieties in which the continued occurrence of conditioned prosthesis is noted (usually rather perfunctorily, unfortunately). For example, in the late nineteenth century, Gorra (1890: 153) mentions that in the variety of Piacenza the prosthetic vowel [a-] still appeared with s impura forms in post-consonantal contexts. In the variety of Travo, spoken some 27 km south of Piacenza, [a-] continues to operate as a prosthetic vowel before s impura forms occurring post-consonantally, as in [ern astrak] ‘they (m.pl.) were tired’ (Zo¨rner 1989: 64). And in the variety of Grizzano Morandi (located 40 km south of Bologna), the structure of clitic verb phrases such as [al ti’stofen] ‘they (f.) are boring you’ vs [at’v¡den] ‘they (f.) see you’ indicates the operation of prosthesis, this process occurring when a consonant-final clitic precedes an s impura form (Loporcaro 1998). However, here as in other northern Italian varieties, the action of I-prosthesis has overlapped with U-prosthesis which applied later in the Middle Ages to words containing new heterosyllabic word-initial sequences that had developed in various parts of northern Italy, especially Piedmont and Emilia-Romagna (see Chapter 6). The result was that normally a common vowel type was generalized for both categories of prosthesis; for instance, in Grizzanese, the prosthetic vowel [i] in forms such as [al ti’vde:ven] ‘they (f.) saw you’ which was triggered by U-prosthesis is identical to that of [al ti’stofen]. As a consequence, this later prosthetizing process has served to mask to some extent the operation of the change specifically brought about earlier on by I-prosthesis. In sum, it seems that I-prosthesis has operated in an unbroken way from Late Latin times across a range of varieties of northern Italo-Romance, and especially those of Piedmont and Emilia-Romagna. However, there is no solid evidence to indicate that it ever came to apply systematically in other than post-consonantal contexts. The factors which limited its generalization to other contexts remain unclear. From the later Middle Ages onward, a pattern not unlike that in later medieval Tuscan appears to have developed, whereby the use of I-prosthesis was progressively abandoned in more formal written registers while in informal spoken registers and in the usage of the less educated majority it continued to operate widely as a conditioned process. This pattern has continued until recent times, but it remains to be seen how far the prosthetizing lects will be affected in the future by the various sociolinguistic forces which have been increasingly promoting familiarity with the standard variety in Italy where of course I-prosthesis is effectively absent.
5 A-prosthesis 5.1 Introduction A-prosthesis appears to have arisen somewhat later chronologically than I-prosthesis, which was discussed in the preceding chapter. Nevertheless, there is evidence to show that it had become an active phonological process in some varieties of Romance before the end of the first millennium AD. Its typical effect has been to introduce a low vowel [a-] (hence our term ‘A-prosthesis’), although occasionally a different vowel quality may develop. Illustrative examples appear in Figure 5.1. Sardinian
Arom.
ar'riðe ar'riu
ar'ri:ði ar'ri:u
a'rîdire1
a'rîu 2
‘river’
ar'ram ar'roðo ar'rui
ar'ra:mu or'r :ða or'ruβiu
a'raru< RĀRU(M)3 a'roat a'roı˘bu
‘branch’ ‘rare’ ‘wheel’ ‘red’
c
RŎTA
RŬBEU(M)
Gasc.
e
RĪDERE RĪVU(M) RĀMU(M)
‘to laugh’
FIGURE 5.1. Romance examples of A-prosthesis Sources: Gascon: ALG pt. 697NE (Bare`ges); Sardinian: AIS pt. 959 (Baunei) and Bo¨hne (2003); Aromanian: Papahagi (1974)
At first sight, A-prosthesis differs significantly from both the other major categories of prosthesis found in Romance. For whereas I-prosthesis and 1 Papahagi (1974) also reports a variant infinitival form [ari’de˘are] which presumably derives from paroxytonic RIDE¯RE. It will be recalled that the Classical Latin infinitive form was also paroxytonic (RIDE¯RE) but that this form underwent widespread restructuring in Late Latin to proparoxytonic RI¯DERE, hence Fr. rire, Ital. ridere, etc. Whether the modern paroxytonic Aromanian form shows faithful preservation of the Classical infinitive stress pattern or whether it represents a later restructuring back from the Late Latin proparoxytonic form is uncertain. 2 Here, as in subsequent transcriptions of Aromanian, [u] serves to indicate a short non-syllabic vocalic off-glide. In all Aromanian varieties except Farserotic, which is spoken in northern Greece, word-final unstressed –u is non-syllabic after simplex consonants but syllabic after consonant groups, e.g. aspargu [as’pargu] ‘I sprinkle’; cf. Daco-Romanian sparg (Caragiu Mariot¸eanu 1977: 176–7). 3 There appears to be no reflex of Latin RA¯MU(M) in Aromanian.
A-prosthesis
147
U-prosthesis (to be discussed in Chapter 6) can be seen to represent transparent restructuring processes which serve to simplify complex heterosyllabic wordinitial syllable onsets, no such interpretation seems possible for A-prosthesis since it operates on words whose initial syllable originally contained a simple onset composed of a single consonant. Indeed, the development here might even be seen as destroying simplicity since it led to the replacement of a word-initial syllable containing an optimal CV structure by a new onset-less syllable containing a coda: CiV->VCi-CiV- (where ‘Ci’ indicates a given type of consonant),4 although the coda consonant has subsequently been eliminated in some, but not all, of the Romance varieties concerned. However, as we shall see, the structural basis to A-prosthesis probably has more similarities to that of the other two categories of Romance prosthesis than appears to be the case.
5.1.1
I D E N T I F I C AT I O N
Before examining the nature and historical development of A-prosthesis in more detail, we may address the problem of how genuine cases of this phenomenon in Romance can be safely identified. In many instances, the presence of A-prosthesis appears clear and uncontroversial. Thus, in Gascon, the historical derivation of items like RI¯VU(M) > [ar’riu] ‘river’ cited above would seem to provide an indisputable instance, since not only is there addition of a word-initial [a-] to the etymological form but precisely the same modification is found occurring systematically in other words similarly beginning with etymological R-. Since the process has operated irrespective of grammatical properties of the words > concerned, e.g. their syntactic word-class, (cf. RE˘M > [ar’reN] ‘nothing’, ROTA ˘ ¯ [ar’roðo] ‘wheel’, RUBEU(M) > [ar’rui] ‘red’, RIDERE > [ar’riðe] ‘to laugh’, etc.), it may be assumed to have a phonological basis just like the two other major categories of Romance prosthesis. However, as has been noted (1.7), it is not uncommon for a new word-initial [a-] to arise in Romance as a result of non-phonological factors. These include (i) prefixation; (ii) morpheme-boundary reinterpretation where the final vowel [-a] of determiners, especially reflexes of UNA, ILLA, or IPSA in noun phrases, is interpreted by speakers as belonging to the start of a following (feminine) noun; (iii) analogical remodelling, as in Sp. avispa ‘wasp’ (< VE˘SPA) which was re-formed on
4
In Optimality Theory terms, A-prosthesis would thus cause a violation of two constraints: ONSET and, in medieval varieties at least, NOCODA.
148
A-prosthesis
the model of abeja ‘bee’ (< APICULA); and (iv) lexical borrowing from another variety where words with initial [a-] were widespread. A well-known example of (iv) is to be found in the numerous loans from Arabic into Ibero-Romance, the great majority of which were nominals borrowed with the Arabic definite article a(l)- that was subsequently interpreted as part of the stem.5 However, undoubtedly the most statistically significant of these non-phonological sources of new words containing an unstressed initial [a-] has been type (i) involving the prefix AD-. Prefixal forms of this sort (typically verbs) have continuously been created across Romania continua from Roman times, as in AD-RIPA¯RE ‘to reach the shore’ ) ‘to arrive’>(Fr.) arriver, (Port., Sp., Cat.) arribar, (Gasc.) arriba`, and right up to the present day as in (It.) ap-prezzare ‘to value’, (Fr.) ap-parenter ‘to ally, link’, (Rom.) a-la˘pta ‘to suckle’, etc., where the prefix has increasingly come to serve as a marker of transitivity. Although prefixal forms can often be identified on morphosyntactic, semantic, or philological grounds, there can sometimes be difficulty even so in distinguishing between cases where a word-initial [a-] has arisen through prefixation and those where it can be attributed to the phonological process of A-prosthesis. This is particularly true for Romance varieties such as Gascon where not only are there forms containing a non-etymological initial [a-] which can be plausibly attributed to prosthesis (i.e. purely phonologically conditioned), but there has also been widespread use of prefixation with [a-] < AD-. Thus, the appropriate interpretation of verbs such as arrauba` ‘to steal’ and arroustı` ‘to roast’ poses problems, as their initial [a-] could be taken to be the result of either A-prosthesis or prefixation. For this reason, evidence provided by verbal forms or by transparently deverbal derivatives containing [a-] needs to be evaluated with special care before safe conclusions can be drawn over whether they have undergone true A-prosthesis. A rather different type of problem is posed by, for instance, northern Italian forms such as R(E)CIPERE ‘to receive’ > (Bolognese) arzaver [ar’tsavv‰r], R(E)-CORDO ‘I remember’ > (Romagnolo, dialect of Forlı`) arcord [Nr’koNrt],6 where there has been insertion of a new word-initial vowel in forms containing etymological R-. There is an evident similarity between these examples and the cases of A-prosthesis noted above, but in reality the circumstances here are quite different. Crucial for the introduction of the new initial vowel in the Italian forms has been the action of syncope affecting the original pre-tonic vowel, this being the hallmark of what we shall term U-prosthesis (cf. Chapter 6). In contrast, vowel 5
The lateral of the definite article was regularly assimilated to a coronal stem-initial consonant in Arabic (e.g. ar-ruzz ‘(the) rice’, an-nafıˆr ‘(the) trumpet’, az-zaˆ’uq ‘(the) mercury’ > Cast. arroz, an˜afil, azogue). This may have played some modest role in obscuring the identity of the article to non-native speakers. 6 Data here and elsewhere in this paragraph are drawn from Coco (1970) for Bolognese and Schu¨rr (1919) for the dialect of Forlı`.
A-prosthesis
149
prosthesis is not found in these Romance varieties in words which contain etymological initial R- if there has been no syncope of the vowel immediately ‘wheel’ > (Bolognese) following this consonant; for example, RE¯TE(M) ‘net’, ROTA raid [rajd], roda [’ro:da], (Forlisano) reda [’rejdN], [’roNdN]. If it is assumed that true A-prosthesis is a process that is phonologically conditioned by the presence of a given type of word-initial simplex consonantal onset, typically /r-/, then the restricted type of prosthesis which is found solely before complex word-initial consonantal onsets arising from syncopated forms like R(E)CIPERE in Bolognese and other Romance varieties would not be interpreted as a case of A-prosthesis. These are best treated under a separate heading, U-prosthesis. Finally, further difficulties remain with certain types of Romance such as Gascon where a new word-initial vowel [a-] may appear both in words which have undergone syncope of the original pre-tonic vowel and in words where there has been no syncope, as in (Gasc.) arceber ‘to receive’ and arriu ‘river’, respectively. Two interpretations for such forms are possible: (i) they reflect the action of two distinct (though sometimes overlapping) categories of prosthesis, which need to be treated separately; (ii) despite appearances to the contrary, the prosthetic vowel in words such as arceber is due to the same process of A-prosthesis that gave rise to arriu, the assumption being that the development of unstressed initial RE- in Gascon was RE- > arre- > arr(e)- > ar-, i.e. A-prosthesis occurred first and then syncope of the medial vowel operated. The latter interpretation has generally been accepted.7 The main justification advanced for this is that the syncope of pre-tonic vowels in the opening syllable of words is unusual in the history of Gascon, cf. nebout [ne’ ut] < NEPO¯TE(M) ‘nephew’, bezi [be’zi] < VICINU(M) ‘neighbour’ (ALG 855, 993). It is therefore argued that a syncopated transitional stage RECIPERE > **r(e)ceber would seem unlikely. In contrast, word-medial unstressed vowels in pre-tonic position are often susceptible to deletion, as in VASS(E)LL-ITTU(M)>bajle´t ‘valet’, VESP(E)R` t ‘truth’ (Scho¨nthaler 1937; Palay A¯TA > brespado ‘afternoon’, VER(I)TA¯TE(M) > berta 1971). As this development would of course also affect forms like *ARR(E)CIPERE, interpretation (ii) might appear the more plausible of the two. However, as we shall see (6.1.4), there is a good deal of evidence from other varieties of Romance, notably in northern Gallo-Romance and Rheto-Romance dialects, to suggest that syncope in initial unstressed syllables operated earliest and most forcefully in words beginning specifically with RE-. It is not impossible therefore that the same syncopating development affected Gascon too but failed to be generalized to other types of word-initial syllable, hence VICI¯NU(M) > bezı` and not **bzı`. If this
7
For example, by Sarrieu (1902: 429), Millardet (1910: 121), and Bec (1968: 177).
150
A-prosthesis
were so, interpretation (i) would appear more plausible. With the data currently available, however, it is not possible to determine which of the two interpretations is correct.
5.2 5.2.1
A-prosthesis: early developments GEOGRAPHICAL DISTRIBUTION
A-prosthesis has occurred, either systematically or sporadically, in a wide range of Romance varieties. In the western area of Romania continua, all forms of IberoRomance appear to have been influenced but not in a systematic way.8 North of the Pyrenees, however, prosthesis has operated widely and systematically in Gascon (see Map 3) and the process has remained productive in some varieties until modern times, but there is little evidence of it operating to any significant degree elsewhere in the langue d’oc area. Further to the east, A-prosthesis has occurred in certain types of Rheto-Romance, especially in varieties of the Engadine in eastern Switzerland and in Friulian. Elsewhere in Rheto-Romance, its presence appears to be at best limited to a small subset of forms. In contrast, prosthesis has operated in a wide-ranging way in many types of Sardinian, notably those of the south (see Map 7). The circumstances in Sicilian and southern Italian varieties of the mainland, however, are less clear. Although significant numbers of forms with initial [a-] have appeared, the evidence suggests that phonologically based prosthesis may not have developed as a genuine native process. We explore this question more fully in a special subsection directly below. As one goes further northward on the Italian mainland, indications of A-prosthesis become ever more sporadic, so that central and particularly northern varieties of Italo-Romance show scant evidence of the phenomenon. There is, however, one notable exception which presents itself in some northern Tuscan dialects. Finally, varieties of Balkan Romance have also experienced A-prosthesis, especially those spoken south of the Danube. Although there is evidence therefore of A-prosthesis having operated across a wide range of Romance, in almost all cases the process has been carried through in an incomplete way only. Full implementation leading to an unconditional rule Meyer-Lu¨bke (1890: }383) indicates that A-prosthesis before words in /r-/ is regular in Catalan. However, this view has found little support in subsequent diachronic studies of Catalan. Moll (1952: }105) and Badia (1981: }66), for example, give no indication that prosthesis affected words containing etymological R-. In synchronic studies such as the optimality-based presentation by Wheeler (2005), Catalan reflexes of etymologically rhotic-initial words are consistently assigned rhotic-initial underlying forms and no constraint is invoked to yield a preferred prosthetic output. 8
A-prosthesis
151
of A-prosthesis (either still fully productive or fully productive at earlier stages leading to subsequent lexicalization of initial /a-/ in the words affected) appears to have occurred in the history of just two types of Romance. These are, on the one hand, southern and central Sardinian and, on the other, Gascon or more accurately the varieties spoken in the Landes in the west and in the Pyrenean area in the south.9 However, in each of the Sardinian and Gascon areas concerned the systematic operation of A-prosthesis has widely been undermined in more recent times, so that only in very few varieties has the process been able to establish itself in a permanent way comparable with that found, for example, with I-prosthesis in Ibero-Romance. 5.2.1.1 A-prosthesis in southern Italy?
It is difficult to determine to what extent A-prosthesis ever really became established as a phonological process operating on native lexical forms in the varieties of southern Italy. In his monumental study of Italo-Romance, Rohlfs (1966: }164) notes that the word-initial rhotic R- is pronounced with a strengthened articulation in a wide sweep of southern dialects10 and he goes on to report that this strengthened rhotic is often pronounced with a prosthetic vowel, suggesting that a variable process of prosthesis has developed. Examples cited include: (Sicilian) arre`sta ‘onion peelings’, arraggiu ‘ray’ (though in }278, Rohlfs states that this is a clear northern borrowing), arrigordu ‘memory’ (deverbal), arriposu ‘rest’ (deverbal), arrı´sicu ‘risk’ (deverbal from RESECA¯RE (?)), arrugna ‘mange’, arre`nniri ‘to give up’, arririri ‘to laugh’; and (Calabrian, especially southern varieties) arramu ‘branch’, arre` ‘king’, arre`jari ‘to support’, arrisi ‘laughing stock’, arruina ‘ruin’, arrumbu ‘roar’, arruffianu ‘pimp’. However, other accounts of southern Italian dialects fail to confirm the presence of the phonologically conditioned rule of A-prosthesis as indicated by Rohlfs. For instance, the classic study of Sicilian by Schneegans (1888: 61–3) suggests that forms with initial [a-] like those cited by Rohlfs do occur but that On the basis of just the isogloss for ALF 1158 rien (< RE˘M) ‘nothing’, Rohlfs (1970: map 1, isogloss 3) indicates that almost all the dialects to the south of the Bassin d’Arcachon and to the west of the Garonne are affected by prosthesis, with forms like arre` appearing. One locality just to the east of the Garonne and also the Val d’Aran in the Pyrenees likewise reportedly have prosthetic forms. However, Bec (1968: 175) points out that prosthetic forms deriving from RE˘M are also found in many varieties where prosthesis does not otherwise occur, so that evidence based just on this one item may not be too revelatory. Isogloss 16 of Bec’s first general phonetic map for the more limited Pyrenean zone that he examines in detail defines the incidence of A-prosthesis for the reflexes of RI¯VU(M) ‘river’. 10 ‘In vaste zone del Mezzogiorno la r iniziale viene pronunciata con un forte appoggio della voce (come rr-),’ loc. cit. 9
152
A-prosthesis
they actually show just morphological conditioning. It is claimed that these forms represent, on the one hand, verbs or deverbal forms with the familiar ItaloRomance prefixal compounding of AD þ RE-, or, on the other hand, feminine nouns where the new initial [a-] arose through the recutting of morphological boundaries; for example, (l)a rrugna ‘the mange’ (with regular initial rhotic strengthening) > (l’)arrugna. No evidence of purely phonologically conditioned prosthesis is reported. A similar finding appears in the account of De Gregorio ([1890] 1993: 131–2). Pirandello (1891)11 in a detailed description of his native dialect of Agrigento adds further support to the contention that prosthesis in Sicilian is not a regular phonological process. He states that general word-initial consonant strengthening is certainly to be found in this south Sicilian dialect, particularly of [r-, b- d- d´-] (p. 30), but, as in Schneegans’s account, it is noted that the many cases of forms which have acquired a new initial [a-] developed either from verbs which almost always go back to etymological RE- (expanded by prefixal AD-), or from feminine nouns with morphological recutting (p. 23). The later derivative work of Ducibella (1934: 222–5) indicates widespread strengthening of initial [r-] in Sicilian dialects but makes no mention of the lexical items affected undergoing prosthesis, although an isolated form arriju ‘I laugh’ is reported from the central Sicilian variety of Caltanissetta (p. 375), this presumably representing the result of analogy with the many verb forms beginning with arr- < prefixal RE-.12 Studies of speech patterns in the mainland of southern Italy are equally unsupportive on the question of phonologically conditioned prosthesis. For example, Falcone (1976: 48) makes no mention of prosthesis when reporting on the reflex of etymological R- in Calabrian, merely noting that there is strengthening of initial rhotics in forms like rrota, rrosa within a broad area of southern Calabria reaching northward to a rough line from Brancaleone (prov. of Reggio Calabria) on the east coast to Gioia Tauro (also prov. of Reggio Calabria) on the west coast. Also, in his detailed study of the speech of the town of Altamura situated in Puglia some 50 km SW of Bari, Loporcaro (1988: } 94) calls attention to some significant data. First, strengthened initial etymological R- is found in just two lexemes [rr Pbb] ‘things’ (St.It. roba) and [rrei], (pl.) [rrIi] ‘king(s)’, the strengthening of the rhotic in these cases representing either the last residue of an earlier general strengthening across central and southern Italy or the result of borrowing from the dialects spoken further south where strengthening was regular. He intimates that the former is more probable. Second, for the many
11
This is the same Luigi Pirandello who was later to become one of the great twentiethcentury European playwrights. 12 Ducibella merely remarks (loc. cit.), ‘though found in some popular songs, [arriju] is no longer a common form.’
A-prosthesis
153
verbal forms in arr- (paralleled by many verbs in abb-, all-, etc.), such as [arrUb’bw¡i] ‘to steal’, [arr‰kUrd¡i] ‘to recall’ (= St.It. rubare, ricordare), the strengthening and the presence of the initial vowel [a-] is attributed to the common use of prefixal AD- as well as to reinforcement by the extended syntactic use of prepositional AD in pre-infinitive contexts.13 Finally, though they are an imperfect source of information for anything but lexicalized forms, dictionaries provide little in the way of confirmation for the existence of phonologically based prosthesis. Thus, Piccitto (1977–2002) records forms such as rrana, rre´, rriccu, rrota, rrussu for Sicilian, and the only forms cited with initial arr- are verbs where prefixation rather than prosthesis seems to be the principal process at work, e.g. arrifa`ri ‘to re-do’, arru´mpiri ‘to bankrupt’, with arrı`riri varying with rrı`diri ‘to laugh’. In his dictionary of Calabrian, Rohlfs (1977) includes a certain number of entries with arr- but these are typically verbs and feminine nouns, arra´ggia ‘rage’, arruina ‘ruin’. On the other hand, many items such as rimu ‘oar’, rocca, rota, rini ‘back’ appear with no initial vowel. For Abruzzo and Molise, entries which begin with arr- in Giammarco (1968–79) are overwhelmingly verbs, arraffa` ‘to snatch’, arre`nn‰ ‘to give back’, arrubba` ‘to rob’ with occasional feminine nouns, arraggı¨o´n‰ ‘reason’, whereas many forms etymologically beginning with R- are cited with no initial vowel: rre`, rro`bb‰, ro´t‰, rin‰ ‘back’, etc. For Neapolitan, Rohlfs (1966: }164) cites some ostensibly prosthetic forms from the 1873 dictionary of D’Ambra such as arrissa ‘fight’, arrobba, arre´quia ‘peace’, arrecietto ‘shelter’, arraggia ‘rabbia’, as well as the verbs arre`jere ‘to support’, arresponnere ‘to reply’, but these too are typically (prefixal) verb forms or feminine nouns. On the other hand, D’Ambra also records numerous forms which lack a prosthetic [a-] such as russo ‘red’, rota ‘wheel’, rine ‘back’, rajo ‘sun , RE¯NES, RA˘DIU(M), RE¯MU(M)), but it is indicated ray’, rimmo ‘oar’ (< RU˘SSU(M), ROTA (p. 305) that word-initial [r-] (< R-) is typically strongly articulated and preceded by [a] in popular speech, with only rre ‘king’ and rrobba ‘possessions, object’ not showing a preceding [a].14 There is therefore rather mixed evidence as to whether true phonologically based A-prosthesis ever operated on native rhotic-initial words in southern Italian varieties in a comparable way to what we find in Gascon or southern Sardinian. An interpretation in conformity with the available facts might be that, following the general strengthening of word-initial rhotics, morphological factors may at first have acted as the main driver for the insertion of initial [a-] but that,
13 Loporcaro (1988: n. 35) notes the widespread use of a < AD when speakers of that area cite an infinitive form in isolation and also when an infinitive is used as a subject in a phrase, e.g. [je bbrytt a pperd] ‘it is unpleasant to lose’ (= St. It. e` brutto perdere). 14 The special phonological status of these two lexical items here may be compared with the data from Altamuran discussed above.
154
A-prosthesis
by analogy with the many rhotic-initial forms affected by the change, other words also containing an initial rhotic sporadically came to undergo the same development. Verbs such as RI¯DERE ‘to laugh’ would probably have been affected in this later incomplete generalization of initial [a-]. A subsequent important factor which further promoted the insertion of initial [a-] appears to have been, again sporadically, later lexical borrowing. Insertion of [a-] occurred whether the borrowings into these southern dialects were learned words, or whether they came from other types of Italo-Romance, especially the literary standard language, or from other Romance varieties altogether (cf. learned arre´quia < REQUIEM, arraggiu < St.It. raggio, arruffiano < French ruffian). Borrowings containing a different word-initial voiced consonant show similar strengthening of the consonant and may likewise undergo the insertion of an initial [a-]. Echoing the remarks of Pirandello noted above, Rohlfs (1966: }}150, 153, 156) identifies this latter development especially with the forms beginning with [b-], [d-], and [d´-].15 In sum, although native lexicon shows apparent signs of A-prosthesis in a number of southern Italian and Sicilian varieties, this may well have been morphologically conditioned at first. Phonologically conditioned A-prosthesis represents a rather later (though never systematic) development which seems to have been brought about in large part by the arrival of lexical borrowings which created new phonological conditions word-initially.
5.2.2
C H RO N O LO G Y
There appears to be no evidence of A-prosthesis in the Imperial period. Roman grammarians, whose reference point is the standard variety of Latin pronunciation, make no mention at all of it. However, as was noted in the previous chapter, no metalinguistic observation is made on the earlier novelty of I-prosthesis either until the seventh century, long after it had gained widespread usage. More significant therefore is the total absence of any indication of A-prosthesis in Roman inscriptions, even in proper names where scribal orthographic conservatism would be less likely to operate. We may therefore assume that if A-prosthesis was present in the speech of certain people during the Imperial period, it represented at most a phenomenon of restricted incidence, geographically and
15 Significantly, etymological initial B-, D-, and (palatalized) G- had evolved to [v-], [ð-] > (weak) [r-], and [j-] in native forms so that the borrowings would have had unfamiliar initial consonants. Southern speakers evidently realized them as geminates, perhaps partly as a result of hypercharacterization and partly because these consonant qualities usually appeared only in geminates in word-medial position.
A-prosthesis
155
socially. In fact, it is only late in the first millennium that the earliest positive indications of the development appear. In Ibero-Romance, the evidence provided by slate inscriptions from Visigothic times (i.e. sixth to early eighth centuries) is unrevealing. Thus, in contrast to forms such as iscrip[si], isperabi, ispe (= spe ‘hope’) and istare which clearly indicate the action of I-prosthesis, words with etymological initial r- show no signs of A-prosthesis. No examples are forthcoming even from words of Germanic origin, which would be more likely to betray prosthesis since their latinized forms would probably have been rather less subject to conservative Latin spelling practices; cf. the proper names Ranulf[us], Recaredo, Reccesuindi (Vela´zquez 2000). The first examples of A-prosthesis from northern Spain come only in documents of the tenth to twelfth century, e.g. the name Aramirus rex ‘King Ramiro’ from a Riojan manuscript of 976, aretundo (< ROTU˘NDU(M)) ‘round’ from a 1055 Riojan manuscript, and arroturas (< RUPTU¯RAS) ‘ploughings’ dated 1137 from On˜a in northern Castile (Mene´ndez Pidal 1964a: 193–4).16 For Aragonese, early instances include arripera < RIPARIA ‘river bank’ from 1042, arretundo ‘round’ (eleventh cent.), Arramon ‘Ramo´n’ from 1119 (Alvar 1953: 53), while early Navarrese documents offer eleventh-century toponymic evidence, notably with the present-day location of Riezu which appears as Arriezu (1054), Arriec¸u (1055), Arriezo (1060), Ariec¸u (1060) in texts from the monastery of Irache (Saralegui 1977: 65). From varieties spoken north of the Pyrenees, various personal names appear with prosthetic in Gascon texts: personal names such as Aregemundo in a text from the Gironde dated 990, Arreinaldo from 1026–30 (also Gironde), Arremon from an Armagnac document of the late eleventh century (Luchaire 1879: 209; 1881: texts 40, 52), and place names Arramos dated 1010 (= Ramons, a commune in the canton of Orthez) and Arribaute dated 1105 (= Rivehaute, in the canton of Navarrenx), cited by Bec (1968: 176). However, there appear to be few if any safe cases of A-prosthesis in the body of early Languedocian texts from Toulouse and Albi appearing in the edition of Brunel (1926-52).17 In the examples occurring in
Mene´ndez Pidal (1964a: 193) also cites the item aredoma ‘flask’ dated 996 from Sahagu´n in east Leo´n (cf. modern Castilian redoma). For the same item, the DCECH reports a form arrotoma dated 942. However, in neither source is an attempt made to offer an etymon for the word, which is evidently seen as being of uncertain origin. In his etymological dictionary of Portuguese, Machado (1977) likewise leaves open the etymology of the cognate Portuguese form redoma. It is therefore unclear whether the alternation in the forms arredoma (arrotoma) / redoma reflects the action of prosthesis or aphaeresis. 17 On the basis of a close study of the texts from Toulouse and Albi which date from the eleventh to the early thirteenth centuries, Grafstro¨m (1958: }22) concludes that they offer ‘aucun exemple irre´futable de a prothe´tique’. 16
156
A-prosthesis
the different texts and sometimes in individual texts, there is variable use of the graphies
Echoing the view of Millardet (1910: 121), Ronjat (1932: }252) dates A-prosthesis in Gascon to ‘probablement’ between the fifth and eighth centuries. (This clear statement mysteriously eludes Chambon and Greub 2002: 479). Bec (1968: 176–7) sees the development as occurring in the fifth to sixth centuries, and Chambon and Greub (2002: 482) propose ‘avant ca 600,’ adding (p. 489) that it may perhaps even go back to earlier than 511. Dinguirard (1979: 38–9) too hints that the early stages may date back into Imperial times. 18
A-prosthesis
157
preceded the loss of -N-. In view of these considerations, it is perhaps safer to leave chronological questions open. In southern Italy, the appearance of a prosthetic (or, more precisely, a quasiprosthetic) vowel [a-] goes back less far into the medieval period and there are also frequent examples of generalization to contexts other than initial [r-] (see also 5.2.1 above and 5.2.4 below). As expected, almost exclusively verbs or deverbal forms are affected, confirming the predominantly morphological basis to the process there. Thus, La regola salernitana or De regimine sanitatis liber (c.1300) from Campania has the probable prosthetic dı´giate arrecordare ‘you must remember’ (l. 561) alongside clearly prefixal forms like t’assicuri ‘you take care’ (l. 608) (Altamura 1977) and the Neapolitan Libro de la destructione de Troya (thirteenth century) contains numerous examples, con grande arrecuordo ‘with a great sense of duty’ (ch. 16), et arrobare ‘and to plunder’ (ch. 30), and with nonrhotic base forms, ammacare chisto barbaro ‘would that this barbarian . . . ’ (ch. 2), perzo abesognava ‘therefore it was necessary’ (ch. 30) (De Blasi 1986). Similarly, the later Cronaca di Partenope (fourteenth century) also written in Neapolitan shows generalized occurrence of [a-] accoro (= coro), arreposare, abruciare (Altamura 1974: 51), as does the Libro di Sidrac (mid-fifteenth century) from Salento where there appear forms such as (si nde) arrecorda ‘remembers (it)’, arrobare ‘to rob’, ammanca ‘diminishes (3rd sg.pres.)’, allapidato ‘stoned (p.pt.)’, abisogno ‘need’ (Sgrilli 1983: 100). Further north in Italy, phonologically based A-prosthesis occurred in northern Tuscan dialects but the earliest evidence for it dates only from more recent times. Nieri (1902: 95) cites an example from 1835, le su iragioni ‘their reasons’ but nothing earlier. Internal evidence suggests that the prosthetic vowel must have developed before the simplification of geminate rhotics in the dialects concerned (cf. 5.2.4). As this change was certainly pre-nineteenth century, the implication is that prosthesis took place sometime in or before the early modern period. In other Romance varieties where A-prosthesis occurred, dating is equally problematic for want of textual evidence. For Rheto-Romance, there is little surviving documentation from the medieval period for the Grisons. However, substantial texts survive from the sixteenth century, one of which is the translation of the New Testament into a form of upper Engadinish by Jakob Bifrun. This contains numerous examples of prosthetic forms, such as araig ‘king’ < RE¯GE(M), arains ‘back’ < RE¯NES, arait ‘net’ < RE¯TE(M), aram ‘branch’ < RA¯MU(M), araschun ‘reason’ < RATIO¯NE(M), aroba ‘property’ < (Germanic) rauba, arumper to break’ < RO¯MPERE, arir ‘to laugh’ < RI¯DERE, arespuonder ‘to reply’ < RESPO¯NDERE, suggesting that the process did not represent a recent innovation (Gartner 1912; Fermin 1954). For Friulian, there are vernacular texts going back to the later medieval period and these suggest that A-prosthesis enjoyed some currency at this time although the prosthetic vowel is by no means consistently
158
A-prosthesis
indicated.19 For instance, texts from the fourteenth century contain forms such as arecivir ‘they received’ < *RECIP-¯IRUNT in a 1355 administrative charter from Cividale, aronch ‘terrace’ < RU˘NCU(M) in a 1355 charter from Gemona, aribuelo ‘Ribolla (kind of wine)’ in 1395 charter from Cividale, aresons ‘reasons, statements’ < ¯ NES in a legal charter dated 1387–94 from Udine (D’Aronco 1960). Further RATIO medieval examples are cited in the etymological dictionary by Zamboni et al. (1984–7) including: (fourteenth century) aro`vul ‘oak tree’ < RO¯BUR, arro`gol ‘regis (M), and (fifteenth century) ara´s ‘turnips’ < RA¯PAS, aro`mpi ter of canons’ < ROTULU ˘ ‘to break’ < RUMPERE. For the Balkans, there is almost no surviving material of any substance written in Latin or early Romance during a period of almost a thousand years extending up to the early sixteenth century, and in the variety where A-prosthesis has been most fully exploited (Aromanian), substantial written evidence dates only from the eighteenth century. All that is available is a handful of Daco-Romanian words and phrases which appear transliterated in Slavic texts between the late tenth century and the early sixteenth century. In the collection of these assembled by Miha˘ila˘ (1974), there are no forms showing prosthesis; instead, we find just items such as: ripi ‘banks’ (< RI¯P-¯I) dated 1428 from Moldavia, Ra˘tundul proper name ‘Round’ (< ROTU˘NDU(M) þ article) dated 1476 from Wallachia, and Valea Rrea (with a doubled graphy for the rhotic) for the place name ‘the Bad Valley’ (< RE˘A) dated 1510 from Walachia. These would suggest that prosthesis was not active as a process in Daco-Romanian. However, they shed little light on the medieval situation in Romance varieties spoken south of the Danube where prosthesis is now particularly in evidence. It is not until the eighteenth century that Aromanian writers, all of them from Albania, provide textual evidence of prosthetic vowels, e.g. arra˘du ‘I laugh’, arra˘ma˘sa˘tura˘ ‘remainder’ (Capidan 1932: 351). Accordingly, whether A-prosthesis in Aromanian is of comparable antiquity to its apparent counterparts elsewhere in Romania continua remains unclear. Given the common basis to I-prosthesis across Romance, we may wonder whether it is safe to assume that documented cases of A-prosthesis in Romania continua also share a common origin in Late Latin. It is certainly true that significant chronological differences exist across the varieties concerned in respect of the earliest attestations of A-prosthesis. Even so, general resemblances can be observed in the formal characteristics of this development, both in its origins and its subsequent vicissitudes. It may therefore be that ultimately A-prosthesis has its origins in a low-level and sporadic tendency that was already present locally in
19
Thus, in a private text dated to the end of the fourteenth century and probably composed by Simone de Vittore from Cividale, we find io l’areprı´nt . . . yestri rinprindut ‘I rebuke him . . . to be rebuked’ with variant reflexes, prosthetic and non-prosthetic, to inflexional parts of the same verb RE-PR(EH)E¯NDERE.
A-prosthesis
159
Late Latin, but we cannot exclude the possibility that in some Romance varieties it arose as a later, independent linguistic process.
5.2.3
S T RU C T U R A L P R E C O N D I T I O N S TO P RO S T H E S I S
A series of special structural features are associated with A-prosthesis. First, as we have seen, this phonological development appears most commonly to have operated on words which begin with etymological R-, as in the examples given above in Figure 5.1. In fact, in all Romance varieties which have experienced A-prosthesis, words with initial R- have always been affected. Sometimes cases of A-prosthesis may also be found in words which begin with other types of consonants, e.g. annuod‰ ‘knot’ < NO¯DU(M) in certain Calabrian varieties (Rohlfs 1966: }161), addaino ‘fallow deer’ in Neapolitan (D’Ambra 1873: s.v. adda`ino).20 However, the occurrence of [a-] in such forms is evidently far from systematic in southern Italian varieties and probably owes itself to later and rather exceptional factors. In fact, it is notable that in those Romance varieties figuring above in Figure 5.1, Aprosthesis has not only been widespread but has seldom affected words containing initial etymological consonants other than R-. For example, in the dialect of Bare`ges (ALG pt. 697 NE) the following forms are found: with no prosthetic [a-] > lu¨o [l P] ‘moon’ (ALG 1010) NOVU (M) > nau [’nau] ‘new’ (ALG 903) ¯ TU(M) > mut [myt] ‘silent’ (ALG 883) MU ¯ NA LU
h
with prosthetic [a-] > arrou`y [ar’rui] ‘red’ (ALG 1593) ˘ NDU(M) > arredou ` n [arre’ðu˜N] ‘round’(ALG 1087) ROTU ¯ TA > arrousada [arru’zaða] ‘dew’ (ALG 1432) ROS-A ˘ BEU(M) RU
The presence of an initial rhotic seems therefore to represent a crucial precondition for A-prosthesis to get under way. A further important characteristic directly associated with A-prosthesis is strengthening. This affected original word-initial R- independent of, and prior to, the development of a prosthetic vowel and it was a necessary though not sufficient
20 Seemingly comparable cases of prosthesis involving initial nasals are also found in Sardinian. For example, immo`i ‘now’, inno`i ‘here’, innui ‘where’ appear in the Campidanese dialect of Sestu and, synchronically, the initial vowel may be interpreted as the result of prosthesis operating on underlying /mmo`i/, etc. in all but post-vocalic contexts (Bolognesi 1998: 393). However, from a diachronic viewpoint these forms clearly derive from prefixal etyma, IN-MO¯DO, IN-HO¯C, IN-U˘BI (cf. also Logudorese ino`ke, ino`ge ‘here’, inu`e ‘where’, REW 4159, 9028). They may therefore be excluded from consideration here. The background to the word-initial vowel found in certain Italian forms such as ignudo, ignocchi, Iddio is examined in 5.2.7.
160
A-prosthesis
condition for A-prosthesis to occur. In view of its significance for A-prosthesis, a closer look at the nature and development of rhotic strengthening is appropriate. The process of strengthening involved the initial rhotic taking on added duration which led to its being identified with its geminate counterpart -RRwhich had previously appeared in intervocalic position only. Chronologically, strengthening does not seem to be a development dating from Imperial times since Roman grammarians make no reference at all to speakers giving wordinitial R- a special or anomalous articulation in relation to the pronunciation of simplex R in other phonological environments. It is presented as a simplex alveolar trill (cf. Kent 1945: 59; Allen 1978: 32–3). But there is evidence to indicate that strengthening of initial R- and its resultant identification with medial geminate -RR- occurred across most though perhaps not all the Latin-speaking world in the medieval period. Most persuasive is the fact that strong word-initial rhotics continue to exist in many present-day Romance varieties, whether or not a prosthetic vowel has subsequently come to establish itself in the words concerned. Thus, in all Ibero-Romance varieties word-initial rhotics have evolved in a parallel way to the medial geminate rhotic -RR-, as in Castilian ramo [’ramo] < RA¯MU(M) ‘branch’ and carro [’karo] < CA˘RRU(M) ‘cart’ (both with a multiple-trilled alveolar rhotic) but caro [’kaɾo] < CĀRU(M) ‘dear’ (with a single ballistic alveolar tap).21 North of the Pyrenees, Gascon likewise has a strongly trilled realization for the reflex of etymological word-initial R-. This pronunciation has a long ancestry and continues to be widely found today (Millardet 1910: 128; Bec 1968: 173–5). Elsewhere in southern France, strengthening of original initial R- and its identification with the medial geminate -RR- are indicated for a number of modern varieties, e.g.
Catalan shares this pattern with Castilian (cf. Wheeler 2005: 24–34). However, Portuguese (the standard European variety from the later nineteenth century) and some varieties of Latin American Spanish, notably Puerto Rican Spanish (A´lvarez Nazario 1991: 695–7), have more recently restructured the phonetic basis of the two-way rhotic system. Instead of a ‘strong’ or trilled vs ‘weak’ or tapped contrast (both items being alveolar), there is now a uvular [R] vs alveolar [ɾ] contrast, respectively. Throughout the literary period and doubtless already in the pre-literary period too, the contrast was neutralized in word-initial position in favour of just the ‘strong,’ and later the uvular, realization. Evidence for this comes in the frequent use of the graphies
A-prosthesis
161
Agen and Toulouse (Se´guy 1950: 20) and Arles, Toulon, Aix and Marseille (Coustenoble 1945: 92–5), suggesting once again that the strengthening of the word-initial rhotic represents an early development. Further north, in the langue d’oı¨l, it seems that a comparable pattern operated widely in the medieval period. The etymological word-initial R- was consistently identified with the strong intervocalic rhotic -RR- and shared a common path of evolution which was distinct from that of the weak rhotic -R-. This arrangement with the strong rhotic appearing in word-initial position continued into the sixteenth century, as is borne out by metalinguistic remarks of early contemporary observers of French such as La Noue ([1596] 1623: 298) who writes in this connection: ‘L’r quand elle est au commencement d’un mot se prononce rudement et avec toute la vigueur qu’elle a.’ Thurot (1881: II, 269–70) and Reighard (1985: 315) assume from this remark that a uvular articulation [R] had established itself as the realization of the formerly strong rhotic whilst the reflex of the original simplex weak -R- was probably still coronal. However, such a conclusion is rather unsafe as La Noue’s observation only indicates that in the variety of French he is describing the reflexes of original geminate -RR- and initial R- had merged as a strong rhotic and that this continued to be phonetically distinct from the reflex of simplex -R-. Both rhotics might still of course have been coronal.22 In fact, it is only later that the strong rhotic is known for sure to have taken on a uvular realization which was ultimately to be extended to all phonological contexts. The process of uvularization first got under way amongst speakers of the standard variety, it seems, in the course of the seventeenth century and the establishment of a single type of (uvular) rhotic doubtless took some time to become generalized thereafter amongst such speakers.23 Residues of the two-way pattern remain to the present day in some French varieties, e.g. that of Arles (Coustenoble 1945: 92–5). In Italo-Romance, strengthening of R- has occurred widely in southern varieties, and initial rhotics have usually remained strong here whether or not prosthetic vowels have come to establish themselves. This is especially evident in the most southern varieties spoken in Sicily, southern Calabria, and the Salento 22
As we have seen above, this is the case in Castilian Spanish and Catalan where there are two contrasting types of rhotic, strong [r] < R-, -RR- and [ɾ] < -R-, both of which have remained coronal to the present day. 23 An anecdotal indicator of the spread of the uvular pronunciation amongst the highest social classes of the later seventeenth century comes in a letter written by Charlotte Elisabeth, Princess Palatine (1652–1722) known as ‘Madame’, the wife of Philippe Duke of Orle´ans or ‘Monsieur’ who was Louis XIV’s brother. In one of her letters written in German dated 14th July 1718, she recalls that in contrast to Louis XIV who pronounced r ‘clearly’ (presumably as a coronal), all his children made use of a different realization. By way of illustration, she represents their pronunciation of Paris as ‘Pahi’, which suggests a uvular value (Goudeket 1964: 299).
162
A-prosthesis
peninsula (De Gregorio [1890] 1993: }152; Rohlfs 1956: II, 526, 1966: }164, and 1977: 565). Further north, strengthening is less in evidence, but in Tuscany various northern dialects underwent this development although direct traces of it have been lost more recently as a result of the general simplification of geminate rhotics (cf. 5.2.4).24 Further evidence for Tuscan rhotic strengthening in earlier times appears in Corsican which is itself Tuscan-based (see Map 6). Here, strong initial rhotics are general in the centre and south of the island except for Bonifacio in the extreme south.25 A similar picture appears in Sardinian, where a strongly trilled rhotic continues to be regularly found in word-initial position.26 In Balkan-Romance, the situation less clear, due in large part to the lack of written evidence for this type of Romance until comparatively recent times. Adding further to the uncertainty has been the general neutralization of strong vs weak consonantal contrasts in pre-literary times, e.g. in Daco-Romanian and > rog(u) ‘I ask’ with an Aromanian TE˘RRA > ¸tara˘ ‘land’, SE¯RA > seara˘ ‘evening’, ROGO identical rhotic. However, three types of evidence suggest that word-initial rhotics were strengthened in early Balkan Romance too. First, occasional cases are reported of a strong trill being used word-initially in modern varieties. Thus, Rosetti (1978: 536) reports such a pronunciation in northern Ardeal (NW Romania) and in Maramures¸ (N Romania), and Papahagi (1974: 1044) indicates such a pronunciation in Farserotic Aromanian spoken in Albania, as well as in the DacoRomanian variety of Poiana Sibiului (central Romania). Second, in philological data for Daco-Romanian from the sixteenth century and Aromanian data from the eighteenth century, scribal practices indicate that initial rotics were strongly articulated. Thus, in the sixteenth-century Psaltirea Hurmuzachi there are numerous forms written with a double Cyrillic graphy <ææ> (=
Medieval Tuscan texts also sporadically show strengthened word-initial rhotics (cf. Rohlfs 1966: }164). 25 In the linguistic atlas ALEIC, maps 107 (un nasone rosso) and 158 (abbiamo riso) for example yield forms such as (pt. 30, Ghisoni) [un ’na:zu r’r Pssu], (pt. 37, Cavro) [un ’na’z P:ni r’r Pssu] and (pt. 22, E´visa) [a’¡mu r’ri:zu], (pt. 32, Bocognani) [a’w¡mi r’ri:su]. 26 Cf. ‘Das anlautende r ist im Sardischen ein starkgerollter alveolarer Vibrant, wie im Spanischen’ (Wagner 1941: }74).
A-prosthesis
163
aurıˆre, (Daco-Rom.) urıˆ ‘to hate’, with stressed centralized [t_], as against *(AU) FER-¯IRE > afirire, (Daco-Rom.) feri ‘to guard’, with stressed front [i]. When a stressed front vowel follows word-initial R-, we find the same centralization, e.g. RI¯VU(M) > (Arom.) arıˆu, (Daco-Rom.) rıˆu ‘river’, and not ** (a)riu. It thus seems reasonable to conclude that original word-initial rhotics did in fact undergo early strengthening resulting in added duration.27 Rheto-Romance is problematic in this context since, unlike the other Romance varieties which have experienced A-prosthesis, it offers no cases of strong wordinitial rhotics in modern varieties nor is there philological evidence for previous strengthening in word-initial rhotics. However, the development of prosthetic forms in certain varieties during the medieval period suggests the likelihood of earlier strengthening, direct evidence for which has disappeared along with the general simplification of geminates in Rheto-Romance in more recent times. Finally, we may note that in certain varieties which had experienced strengthening of initial R- another word-initial rhotic may also arise as a result of later independent change. Where this happens, a contrast usually develops between primary etymological initial R-, which remained articulatorily strong, and the new secondary word-initial [r-] which was realized at first as a simplex consonant and often stayed so. This has occurred in many southern Italian varieties, where etymological word-initial GR- and, less widely, D- have both developed to give [r] (M) ‘big’ > (Sicil.) ro`ssu, (Cal.) ruossu; DE˘NT-¯I which has remained weak: e.g. GROSSU ‘teeth’ > (Sicil.) rendi, (Neapol.) riend‰ (Rohlfs 1966: }153, 185; AIS 107, 184). Such forms with secondary [r-] normally do not undergo A-prosthesis. In Gascon too, a secondary initial rhotic developed in a number of western and southern dialects, when the group FR- passed to [hr-] (as part of the general Gascon change F- > [h-]) before the group simplified to [r-] probably from around the thirteenth century. There was later strengthening of the secondary r- in some though not all dialects (cf. ALG 157 freˆne, 285 froment) but, even so, A-prosthesis failed to operate in these words (cf. Millardet 1910: 123, Bec 1968: 175 n. 2). However, more recent loanwords from French containing word-initial r- have sometimes been aligned with prosthetizing forms, as in arre´gre`t [arregr¡t] ‘regret’ in certain Gascon varieties which have experienced regular A-prosthesis (Bec 1968: 179; see also 5.2.5 below). This change may well reflect hypercharacterization by Gascon speakers as they articulate prestigious loanwords from the standard language. It thus resembles the word-initial consonant gemination found in loanwords in southern Italian dialects (cf. 5.2.1).
This has been called into doubt by Jungemann (1955: 276) on the basis of the claim made by a linguist of his acquaintance familiar with Aromanian that, in the modern language, there is no strengthening. However, this of course does not rule out the possibility that at earlier stages this variety had a rule strengthening word-initial /r-/. 27
164
A-prosthesis
5.2.4
Q UA L I T Y O F T H E P RO S T H E T I C VOW E L
In the earliest attestations of A-prosthesis, the vowel was regularly represented as , and indeed many Romance varieties such as those spoken in Gascony have preferentially retained this quality up to the present day. So, we may plausibly assume that, usually, the new vowel at first adopted a low quality [a], although subsequently regular phonetic change might modify this quality in individual varieties of Romance. However, more rarely there have been alternative outcomes. In Sardinian dialects, vowel copying is found instead whereby the prosthetic vowel has taken on a quality determined by that of the vowel present in the following syllable. In medieval texts there is a discernible tendency for the prosthetic vowel to appear as <e> before front vowels,
28 For this word, Solmi provides the more detailed definition ‘prestazioni di lavoro agrario, dovute dai sudditi al pubblico potere, e continuate poi lungamente appunto col titolo di roadia.’ 29 The stressed vowel is recorded without indication to length and with a voiceless sibiliant [s]. Both these phonetic outcomes are unexpected.
A-prosthesis
165
with the vowel [a-] occurring only when the following vowel is low or front whilst [o-] is retained when a back vowel follows. A further and quite different outcome presents itself in dialects of the Lucca area in northern Tuscany and in a number of varieties spoken in central and southern Corsica (Corsican being Tuscan-based). Here, a high front prosthetic vowel [i-] has been reported. Pieri (1890–92: 124) cites forms such as ho irotto ‘I have broken’, diventa irosso ‘it’s becoming red’, date iretta ‘pay attention’, while Nieri (1902: 95) adds further information, noting that it is mainly, but not exclusively, verbs in the ‘parlar volgar’ that take this prosthetic [i-]; cf. non irende ‘not to give back’ (= St.It. non rendere), le su iragion ‘its justifications’, Irifanni ‘Re-do some’ (= Rifanne). It seems impossible to interpret this development as the result of vowel copying, nor can it be attributed to regular sound change operating on an earlier initial [a-]. For Lucchese dialects do in fact show numerous examples of forms where an initial [a-] has been inserted, almost exclusively in verbs originally containing the prefix RE-, such as ariposare and aritornare. (We may note that geminate rhotics were systematically simplified at some more recent period in the history of Lucchese.30) However, the initial vowel [a-] in these forms has been lexicalized and preserved unchanged. They can therefore be of no relevance in the present connection. Furthermore, the examples containing initial [i-] indicate that this type of prosthesis has occurred irrespective of phonological context, i.e. when post-pausal, post-vocalic, or post-consonantal, although it has not operated in a consistent way across all potential lexical items. Comparable data to those occurring in Lucchese are also found in central and southern Corsica (see Map 6). In the rich array of materials presented in the linguistic atlas of Corsica (ALEIC), a number of forms with a prosthetic vowel appear, e.g. (1370) unn irre`mi 31 ‘you (sg.) are not rowing’ and (1371) un irre`mu ‘an oar’ (cf. Latin RE¯MA¯RE ‘to row’, RE¯MU(M) ‘oar’). Likewise, there are (153) irridi ‘he laughs (post-pausal)’ (just at Conca in SE Corsica) and (155) unn irriðerajˆu ‘I will not laugh’, although at two localities (E´visa, Bocognani in central Corsica) there are rare instances of prosthetic [a-] (respectively, unn arriðerajˆu, unn arriðarajˆu ‘I will not laugh’), all of these forms deriving from parts of (Late Cf. Nieri (1902: 95) who states: ‘il doppio erre non si sa pronunziare’, hence tera ‘land’, etc. This constraint against geminate rr applies also to contexts where rafforzamento fonosintattico regularly applies. Nieri cites: ‘Se restate e non se rrestate, Piu` robba e non Piu` rrobba’ (italics are ours). 31 In cases where the ALEIC reports a number of slightly different forms for a given lexical item from one locality to another, for convenience we cite just one of the forms provided that the phonetic variation is not of immediate significance for our purposes. Thus, as well as unn irre`mi (from Piana and E´visa in the west centre down to Coti and Conca in the south), we find at other southern localities unn irrammi (Petreto), un irre`mi (Propiano), unn irre`mmi (La Monacı`a), as well as un ire`mi for Lucca. 30
166
A-prosthesis
Latin) RI¯DERE ‘to laugh’. In the light of the prosthetic Lucchese form irotto cited above, the responses for reflexes of parts of the Latin verb RU˘MPERE ‘to break’ are of special interest. However, in map 160 (Mi son rotti gl’incisivi ‘my front teeth have been broken’), no prosthetic form is indicated for any Corsican variety. But the ALEIC also includes a few localities on mainland Italy including Lucca (pt. 54) and the form cited here is [mi s¡n i’rotti i d’d¡nti da’vanti] with a prosthetic [i-]. Similarly, a prosthetic form [i’rompe l’kardo] with [i-] is reported for Lucca but not for any Corsican variety in map 993 Rompi il riccio ‘Break the chestnut husk’. 5.2.4.1 Selection of vowel quality
No serious attempt appears to have been made so far in order to provide a systematic account of the factors guiding the varying selection of vowel quality for this category of prosthesis. To arrive at such an account, a useful point of reference lies in the more recent cross-linguistic findings of general phonologists investigating vowel epenthesis, which were considered above in section 1.6. It was noted there that five factors can play a role in assigning a quality to an incipient epenthetic vowel, this vowel at first being brief and of indeterminate quality (typically schwa-like) in accordance with the principle of minimal saliency. Since word-initial schwa was not found in any of the early Romance varieties affected by this category of prosthesis, a permissible quality needed to be established. In central Sardinian dialects, we have seen that strategy (v) was adopted whereby the new prosthetic vowel was shaped by vowel harmony, resulting in the vowel taking on a quality determined by the vowel in the following syllable. In the great majority of cases, however, a low-quality [a] has been selected and a combination of factors appear to have played a determining role in this. Articulatory and perceptual factors alone would suggest that the prosthetic vowel might adopt a coronal (i.e. front) quality, since the original word-initial consonant which the vowel came to precede, the rhotic [r], is coronal. A front vowel of minimal saliency would then be expected, typically [i] (strategy (iv); cf. also 4.1.3). However, we can understand the preferential adoption of a low value [a] if phonological considerations are also invoked in the way proposed by Rose and Demuth (2006) in their analysis of vowel epenthesis. Their suggestion is that the place feature of a consonant is only copied by a prosthetic or more generally an epenthetic vowel when that place feature is phonologically distinctive in the language concerned. Now, the coronality of the rhotic /r/ is not a distinctive feature in Latin or early Romance, as there was no other non-coronal rhotic or non-coronal liquid in Latin. This being so, there was no copying of place feature in the prosthetic vowel which preceded the rhotic. Instead, a vowel was adopted whose quality is likewise not specified for place, namely [a], as in the representation given below. The circumstances here thus differ from those with I-prosthesis, since the coronality of the voiceless fricative /s/ was distinctive in Latin and
A-prosthesis
167
continued to be so into Romance since it contrasts with the labiality of the other voiceless fricative /f/. Its coronality was therefore copied for I-prosthesis, as we saw. σ
a
r
Root
Root
C Place
C Place
V Place
no specification for Place
FIGURE 5.2.
Agreement of the prosthetic vowel in respect of non-specified Place feature
Further guiding the selection of a low-quality [a] was doubtless the expansion of verb forms originally containing the prefix RE- but subsequently augmented with AD-, which yielded a rich array of forms in [ar-] in many forms of Romance, as in AD-RESTARE > arrestare, arreˆter, etc. However, given that this morphological factor would have typically operated on specific words only, namely verbs and deverbal derivatives whose first original syllable was unstressed, it would seem to be of secondary importance. Thus, the adoption of the quality [a] for the prosthetic vowel appears to have been determined by a combination of strategies (ii), (iii), and (iv), as identified in 1.6. This leaves the problem of accounting for the outcome [i] in certain northern Tuscan and central-southern Corsican varieties. As there is no independent evidence of a regular sound change in these varieties whereby word-initial [ar] > [ir-], nor is there any possibility of interpreting the place feature of coronality in the initial rhotic /r/ as being distinctive, the selection of a high front quality [i] is curious. However, certain data suggest a possible rationale for the development here. In particular, it is significant that in the Lucchese dialects affected, the prosthetic vowel [i] is found not only in rhotic-initial words but also in words beginning with complex word-initial onsets (excluding consonant þ liquid). Thus, Nieri (1902: 95) states that prosthetic [i] occurs with s impura words ‘quasi costantemente’ in the speech of the less educated, and especially peasants, such as irrespective of phonological context. In addition, words with initial [ts] zio ‘uncle’, zappa` (= St.It. zappare ‘to hoe’) also tend to have a prosthetic [i] in the
168
A-prosthesis
speech of the less educated, giving [it’tsio] for zio, etc. But Nieri adds that amongst the very least educated of speakers (‘l’infimo volgo’) the affricate [ts] may often be simpified to [s] and that, where this happens, no prosthetic vowel appears as the onset is no longer complex. The implication of the data here is that the high vowel [i] has come to be generalized as the prosthetic vowel used for resolving problematic complex word-initial onsets. To account for this, a chronologically based interpretation suggests itself as perhaps the most likely. Wordinitial rhotics may be assumed to have undergone the same strengthening in Lucchese that affected all southern Italo-Romance varieties but only at a later stage, as a result of the change diffusing northwards as a sporadic and localized development. But by the time that strengthened rhotic onsets [rr-] became generalized in popular usage, I-prosthesis affecting s impura forms would already have been long established, just as it had become with ordinary speakers elsewhere in Tuscany (cf. 4.3.3). The principal function of I-prosthesis, it will be recalled, was to resolve the problem of complex heterosyllabic word-initial onsets, but over time its scope may have been extended in Lucchese to handle not just s impura sequences but other complex word-initial onsets that developed, such as Thus, for forms like rrende(re) < RE˘NDERE, it was the prosthetic [rr-] and [ts]. vowel [i] which was already in regular and productive use that was inserted.32 Such a chronologically based interpretation is also compatible with the facts of Gascon where both A-prosthesis and I-prosthesis have similarly occurred but with differing outcomes, giving [a-] and [e-], respectively. Here, the evidence suggests that A-prosthesis arose as a process at a time when I-prosthesis itself was still in an early stage of its generalization, i.e. at the end of the Empire and in the early Middle Ages. The absence of an established default prosthetic vowel may have led to each of the two concurrent processes being less readily identified so that the quality assigned to each type of prosthetic vowel was determined independently. Finally, the facts of Campidanese in southern Sardinia also need explaining since here too both categories of prosthesis operated, giving [a-] and [i-]. However, it is striking that I-prosthesis never really became fully established in Campidanese as it did in the northern half of Sardinia. The implication is therefore that as A-prosthesis began to take root in medieval Campidanese, Iprosthesis was ceasing to enjoy any significant measure of productivity. The selection of vowel quality [a] for the prosthetic vowel in rhotic-initial words was therefore unconstrained by existing patterns of prosthesis and was determined instead according to strategies (ii), (iii), and (iv), as discussed above.
32
Subsequently, Lucchese has undergone regular simplification of geminate [rr]; cf. n. 25. This development, however, must postdate prosthesis in rhotic-initial words.
A-prosthesis
5.2.5
169
AC T UA L I Z AT I O N
In the Romance varieties where A-prosthesis occurred, the evidence indicates that at the outset it was triggered by the presence of a word-initial rhotic (cf. 5.2.3). The special status of the rhotic as a triggering context is shown by the fact that (a) in all Romance varieties experiencing A-prosthesis words in etymological R- have always been affected, and (b) words originally beginning with a segment other than R- are more sporadically affected. The data from Romance suggest that the relative susceptibility of words to the implementation of A-prosthesis operated along the following parameter: more susceptible, chronologically earlier
# r-
less susceptible, chronologically later
#C [+ son]-
FIGURE 5.3.
#C [+ vcd]-
#C-
# V-
Parameter showing the actualization of A-prosthesis in Romance
This parameter reflects prosthetic outcomes which may not have arisen solely as a result of phonological conditioning native to the variety in question. Thus, southern Italian data are included although the circumstances of A-prosthesis there are special (cf. 5.2.1). rhoticGASCON C. SARDINIAN ENGADINISH S. ITALIAN AROMANIAN
þ þ þ þ þ
FIGURE 5.4.
any sonorant þ þ
voiced consonant þ / þ
any consonant þ
any segment þ
Implementation of A-prosthesis across Romance varieties
At the phrasal level, it is difficult to demonstrate whether A-prosthesis regularly came to operate in a staged way or whether the development occurred as a single across-the-board development irrespective of context. With a staged actualization, there would at first have been alternation between prosthetic and nonprosthetic forms depending on phonological context—non-prosthetic forms would be more likely in post-vocalic contexts whereas prosthetic forms would be expected elsewhere. Alternation would then give way to the subsequent generalization of the prosthetic alternant, as appears to have happened with Iprosthesis in most varieties of western Romance (cf. 4.1.4). For A-prosthesis, however, the early available textual data are inconclusive. Although some modern
170
A-prosthesis
Gascon varieties now show contextually conditioned alternation, this may reflect the result of deletion of a prosthetic vowel in post-vocalic contexts rather than a continuation of original alternation. This is the view of Sarrieu (1904: 509–10) who reports that in the Gascon variety of Bagne`res-de-Luchon there is a regular absence of [a] in words beginning with arr- when a vowel-final word precedes in ) there are alternating close-knit syntactic phrases. Thus, for arroˆzo ‘rose’ (< ROSA realizations such as: ‘rose’ ‘two roses’ where u¨ = [ ] ‘a rose’ ‘a bunch of flowers’ h
(post-pausal) arroˆzo (post-consonantal) du¨es arroˆzes but (post-vocalic) yo (or yu) rroˆzo u¨m buke´d de ’rroˆzes (or d’arroˆzes)
Similarly, for the modern Gascon varieties of the Comminges in the central Pyrenean zone, Bec (1968: 175) states that the prosthetic vowel /a-/ only appears in postpausal and post-consonantal contexts. This leads him to characterize the vowel as ‘un simple adjuvant articulatoire,’33 although no comment is made on its historical development. While it is ultimately unprovable whether the distribution here represents the result of more recent change or the preservation unchanged of an earlier alternating pattern that existed more generally, there is suggestive evidence in favour of the latter view which comes from the pattern of alternation reported by Sarrieu (1903: 319) for historically s impura forms, which was presented in 4.1.4 above. Here, a directly comparable pattern of synchronic alternation is found: espyo´ ‘thorn’ du¨es espye´s ‘two thorns’ where u¨ = [ ] era spyo´ ‘the thorn’; yo/yu spyo´ ‘a thorn’ h
but
(post-pausal) (post-consonantal) (post-vocalic)
Given that alternations of this type only appear consistently with forms that underwent prosthesis and that aphaeresis is rare,34 we may suspect that the two Gascon varieties concerned may well have remained at the first stage of prosthetic actualization and never gone on to generalize use of the prosthetic vowel in all contexts. It therefore seems not unlikely that A-prosthesis developed originally as a staged process, with post-vocalic contexts being the last to be incorporated. And
33
The interpretation of the prosthetic vowel here is reminiscent of Martinet’s proposal to view schwa in modern French as a ‘lubrifiant’ (1969: 216). 34 Sarrieu (loc. cit.) notes one other context where aphaeresis can occur, namely in words beginning with vowel þ Nasal þ Consonant, e.g. (a)nna´do in buno ’nna´do ‘good year’. The deletion here is evidently less normal than in the two historically prosthetizing contexts and appears typically in allegro speech styles only (‘ne s’accomplit gue`re que dans la parole assez rapide’).
A-prosthesis
171
in view of the similarity with the likely actualization of I-prosthesis (cf. 4.1.4), the possibility also exists that A-prosthesis may have developed preferentially in postconsonantal contexts first of all before spreading to post-pausal contexts. However, some limited counter-evidence to this view comes from recent experimental work on the articulation of consonants at the beginning of units at different levels of the phonological hierarchy. The studies of Fougeron (2001) and Keating et al. (2003) on modern French and certain non-Romance languages have revealed that in general terms the articulation of a consonant is more forceful and of greater duration at the beginning of higher-level prosodic units than at lower-level units, i.e. more forceful in post-pausal, intonational phrase-initial position than wordinitial or syllable-initial.35 This would suggest that R- in post-pausal position might have been the first to experience the strengthening that was evidently indispensable for vowel prosthesis to occur. If so, the order of actualization of A-prosthesis might have been post-pausal > post-consonantal > post-vocalic. However, in the present limited state of our knowledge of pronunciation in early Romance, this interpretation must remain speculative.
5.2.6
C AU S AT I O N
If, as appears to be the case, A-prosthesis typically arose in Romance as a process preferentially affecting forms with etymological initial R- which had undergone preliminary strengthening, attempts to explain the etiology of this category of prosthesis must address two problems. First, the factors leading to word-initial rhotic strengthening need to be accounted for; second, there is the question of why the presence of a strengthened word-initial rhotic in particular should have triggered prosthesis in certain varieties of Romance. The causation of initial rhotic strengthening has attracted relatively little direct attention. Indeed, only one proposal seems to be of relevance here and even this is of uncertain appropriateness. The proposal was developed by various structuralist linguists, notably Martinet (1952/1955), Weinrich (1969), and Hall (1964), and more recently it has been taken up again by Cravens (2002). Since the structuralist view is presented in slightly different ways by its various proponents, we will just outline a broadly consensus account of it. The crucial assumption made is that, in early Romance, phrase-medial word boundaries had no direct phonetic significance such that when word-initial consonants appeared within phonological phrases they varied allophonically in respect of strength in the same way as
Fougeron (2001: 123), for example, found that in the speech of her four French informants the duration of the lingual occlusion of the test consonants was ‘usually very long in IPi position’, where ‘IPi’ stands for Intonational Phrase-initial. 35
172
A-prosthesis
their word-medial counterparts. Thus, a post-vocalic allophone such as the rhotic in UNA RANA ‘one frog’ had a relatively weak realization identical to that found when the consonant was intervocalic within a word, e.g. in CARA ‘dear (f.sg.)’. However, a relatively strong realization appeared when the consonant was not post-vocalic, that is, when it occurred either post-pausally or post-consonantally within a phonological phrase (irrespective of word boundaries); for example, a strong rhotic would appear in TRES RANAS ‘three frogs’ and SUBRIDERE ‘to smile’ (post-consonantal), and also in RANA EST ‘It’s a frog’ (post-pausal). It should be noted that ‘post-consonantal’ here includes only those cases where the rhotic is syllable-initial; in forms like TRES where the rhotic was post-consonantal but the second member of a complex onset, the rhotic evidently had a weak articulation. Latin also has geminate consonants including a geminate rhotic RR as in TERRA ‘land’ which only appeared in intervocalic position. This at first remained distinct from the strong variant of the simplex consonant, but, as we have seen, little by little the latter became increasingly identified with the original geminate in many forms of Romance. Thus, schematically, we may represent the pattern as follows, using ‘R’ as the strong allophone and ‘r’ as the weak allophone of simplex /r/. usage in early Romance post-vocalic post-vocalic post-consonantal post-consonantal post-pausal geminate
word-medial word-initial word-initial word-medial
r r R R R R(R)
after later, regional generalization of strong variant word-initally r R R R R R
From this hypothesized starting-point, the assumption is that in western Romance varieties and in some, but not all, eastern Romance varieties the strong variant was later generalized to all word-initial contexts. Different explanations for this later development have been proposed. Hall and Cravens note that, in a substantial number of phrase-medial contexts, word-initial consonants which appear to be post-vocalic in later Romance probably underwent strengthening through rafforzamento fonosintattico (RF) in earlier stages.36 For instance, in a phrase such as Spanish y piedras ‘and stones’ which derives from ET PE˘TRAS, the
36
See Loporcaro (1997a) for a detailed review of the historical development of RF. In broad terms, RF has occurred either (i) through the assimilation of an original word-final consonant to the initial consonant of a following word provided the words are closely syntactically linked, e.g. ET ME¯ > Italian e me [em’me] ‘and me’; or (ii) in certain
A-prosthesis
173
word-initial consonant would be strengthened as a result of the preceding wordfinal consonant becoming assimilated to it, [et p-] > [e pp-]. Clearly, the effect of RF would be to increase the frequency of the strong variant at the beginning of many apparently post-vocalic words in early medieval Romance. Hall (1964: 555) offers some statistical evidence that points to the predominance of the strengthened allophone of word-initial consonants in western Romance. Using an Old Occitan text as a basis, he calculates that out of the total of word tokens beginning with a consonant, 64 per cent would have been in contexts where strengthening would be expected (post-pausal, post-consonantal, and post-vocalic in RF contexts). The suggestion therefore is that it was the statistical predominance of the strong variant in word-initial position within phonological phrases that led to its subsequent generalization. Attractive though the argumentation may at first sight seem to be, the structural account encounters some difficulties. As Pensado (2006) has noted, there is little actual philological evidence for many of the developments postulated. Hall (1965: 552) cites forms such as Cast. bravo ‘bad-tempered’ < PRA¯VU(M) or Old Sard. gruke ‘cross’, gurtellu ‘knife’ (< CRU˘CE(M), CULTE˘LLU(M)) as possible indicators, in that they are claimed to represent lexicalized forms showing exceptional generalization of the weak word-initial allophone. However, given the supposed universality in earlier times of the word-initial alternation between weak and strong consonants, we might perhaps expect to see rather more such cases where the presence of the weak variant is indicated word-initially, especially in medieval texts written by less educated scribes. Furthermore, the structuralist account takes as basic that word boundaries had no phonetic correlate within phonological phrases (cf. Weinrich 1969: }57). In the light of recent phonetic studies, however, this assumption is shown to be questionable since evidence has been found that consonants do display phonetic differences in French and other languages depending upon their location in relation to the boundaries of prosodic domains. As we saw above (5.3.2), experimental research has demonstrated that consonants appearing at the beginning of prosodic units tend to be more strongly articulated the higher up the prosodic hierarchy the unit in question is (syllable, word, accentual phrase, intonational phrase). Hence, a given consonant is typically more weakly articulated syllableinitial than when it is word-initial, and so on. The typical occurrence of phonetically stronger consonantal realizations word-initially serves to undermine to some degree one of the key assumptions of the structuralist vision.
Italo-Romance varieties, when a short stressed word-final vowel appears before a syntactically linked, consonant-initial word, e.g. Italian canto` male [kan’t P m’ma:le]‘he sang badly’.
174
A-prosthesis
Finally, although the interpretation advanced by the structuralists offers some insights into the possible rationale for strengthening for obstruents, it is less helpful in accounting for circumstances with sonorants. This is tacitly recognized by Hall and Weinrich who focus on obstruent change.37 Certainly, some supportive evidence comes from Ibero-Romance where word-initial R- has generally been strengthened and evolves like geminate -RR- (cf. 5.2.3 and n. 19). However, obvious counterevidence comes, for instance, from Castilian where L-, N- do not palatalize in the same way as geminate -ll-, -NN-, as in lago, nombre (< LA˘CU(M), NO¯MINE) but silla, can˜a (< SE˘LLA, CANNA).38 Similarly, Gascon initial L- patterns like weak > mulo ‘millstone’ simplex -L- rather than strong -LL-, e.g. LU¯NA > lu¨o and MOLA ˘ but SELLA > se`ro ‘saddle’ (Scho¨nthaler 1937). In eastern Romance too, the dialects of southern Italy offer problematic evidence. For, although the systematic strengthening of word-initial sonorants presupposed by the structuralists widely affects the rhotic R-, it seldom operates on the other types of sonorant.39 The structuralist rationale therefore offers some potentially useful insights but it is beset with various difficulties when used to account for the strengthening of word-initial rhotics. In trying to identify the factor(s) triggering rhotic strengthening in early Romance, the possibility needs to be considered that the development may not have had a single common causation across all Romance varieties. Instead, certain factors may be shared, but additional and more localized factors may also have operated yielding comparable results. A few tentative observations may be advanced here. To The title of Hall’s 1964 paper makes this clear. Weinrich (1968: 52, n. 14) does make a brief remark on sonorant development when he states that the presence of ‘spontanous initial doubling’ in the rhotic of Cast. la rana is unconnected with RF. Doubling is not due to syntactic phonetics, i.e. sandhi, it is claimed. Indeed, it is viewed not as lengthening but rather as an ‘intensification’ (Intensivierung) of the consonant’s articulation, but no explanation is offered for the change. 38 Martinet (1956: 283–4) considers the failure of initial L- to develop to palatal ll- just like the medial geminate -LL- does, and he attributes it to the need to avoid homonym clash with the palatal ll- which had arisen from initial PL-, CL-, FL-. Nothing is said about the failure of N- to give palatal n˜- in Castilian just as -NN- did, although such an initial palatalization is found in C. and E. Asturian and Leonese dialects, e.g. the Asturian saying Quien n˜on diga n˜on, n˜abos, n˜avaya, a mio´ tierra que n˜on vaya ‘Whoever does not say n˜on “not”, n˜abos “turnips”, n˜avaya “(clasp) knife”, let him not go to my land’ (Zamora Vicente 1967: 130). 39 Cf. Rohlfs (1966: }}159, 161) where the near absence of strengthening of initial L- and N- is reported for southern dialects. Initial M- however undergoes widespread strengthening (}160). Martinet (1964: 282, n. 58) briefly claims that the strong initial rwidely found in southern Italy is the result of analogy with forms where the initial rhotic was strengthened through RF. Why other initial sonorants were not similarly affected is not discussed. 37
A-prosthesis
175
begin with, it seems appropriate to assume that consonant allophones at the beginning of prosodic units tend to be stronger than ones occurring within such units, as has been shown in recent experimental phonetic investigation to which reference has already been made. Word-initial consonants might therefore be expected to develop a stronger articulation than their simplex medial counterparts. From this common basis, western Romance varieties may well be expected to have been affected by the factors identified by structuralist linguists, particularly as wordfinal consonants were much more retentive in the west of Romania continua and their presence was important in promoting strengthening of a following wordinitial consonant. However, it remains unclear why rhotics should have shown greater susceptibility to such strengthening in comparison with other types of sonorant. Rhotic strengthening in eastern Romance varieties is no less problematic to account for. Italo-Romance is especially curious since southern varieties regularly show it but central varieties do not normally do so. Here, the possibility exists of some localized influence from the long-standing adstratum language Greek. In Ancient Greek, initial rhotics were always voiceless [3] and their medial allophonic counterpart was geminated [33] (Allen 1987: 43). It seems not inconceivable therefore, that generations of bilingual Latin-Greek speakers may have carried the allophonic patterning of rhotics occurring in Greek over to Latin, the result of which could have been the identification of the initial Latin R- with the medial geminate rhotic -RR-. As only rhotics show this patterning in Greek, we have at least a possible contributory factor helping to explain the exceptional strengthening in this consonant in southern Italian dialects. In the other major eastern Romance area, Balkan Romance, the factors underlying the strengthening of the rhotic R- are unfortunately even more problematic and they remain enigmatic. Much work is still to be done therefore on clarifying the background of initial consonant strengthening in Romance. A little more attention has been paid by romanists to explaining the origins of the process of A-prosthesis. Various approaches have been used, some more plausibly than others. We may identify three broad types which will be reviewed in turn: phonetic, phonological, and contact influence from pre-Roman languages. (a) Phonetic-based accounts of A-prosthesis have mainly addressed the circumstances of its occurrence in Gascon. The earliest example of this approach was presented by Millardet (1910: 128–9) who claimed that prosthetic [a] had arisen merely as an anticipatory vowel that foreshadowed the vowel following initial R-.40 Ronjat (1932: 54) rejected this as ad hoc and claimed that word-initial R- may have been relatively long when in post-pausal or post-consonantal ‘La voyelle de la syllabe primitive initiale a pe´ne´tre´ partiellement a` travers l’[r-] . . . Ce fragment de voyelle, s’infiltrant a` travers l’[r], a transpire´ a` l’initiale, et colore´ en [a] par l’[r], il a dans la suite forme´ syllabe’ (Millardet 1910: 129). 40
176
A-prosthesis
contexts and that it was this longer variant that ‘split’ into a sequence of syllabic and non-syllabic rhotic [rr] before becoming arr-.41 Unfortunately, it is not made ’ entirely clear what might have caused the lengthening of the rhotic and its subsequent ‘splitting.’ Nonetheless, the significance of the prior lengthening of initial R- is clearly perceived, as is the likely creation of alternation between prosthetic and non-prosthetic forms in the early development of A-prosthesis. In his structuralist examination of A-prosthesis in Gascon and Ibero-Romance, Jungemann (1956: 285–7) subscribes to Ronjat’s account, suggesting that the creation of a prosthetic vowel before strengthened /rr-/ is comparable with I-prosthesis before the ‘liquid s’ of word-initial s impura sequences where a similar inherent vocalicness is attributed to an original consonant (cf. 4.1.5). A rather more plausible phonetic basis to the development of a prosthetic vowel can be found, however, if certain phonetic properties inherent to the rhotic trill R- are considered more closely. Of particular significance appears to be the very precise articulatory control required for the production of a trilled alveolar /r/, as recent experimental investigation has revealed. Sole´ (2002a, 2002b) describes in some detail the physiological and aerodynamic demands made when pronouncing this sound-type, noting that ‘the conditions for initiating tonguetip trilling involve muscle contraction of the tongue to assume the position, shape and elasticity requirements, and a sufficient pressure difference across the lingual constriction’ (2002a: 656). The complexities are such that the articulation of a strongly trilled [r] can pose difficulties for some speakers.42 This is borne out by the findings of Hammond (2000) for Latin American Spanish. In an acoustic and perceptual analysis of the speech of ninety-five native speakers from nine different countries, it was found that the informants frequently did not realize the strongly-trilled rhotic normally found in word-initial position with multiple tongue-tip vibrations. Instead, a wide range of articulations were reported which included various pre-aspirated sequences such as the pre-aspirated trill [hr] and tap [hɾ]. Comparison can be made between these findings and those reported in the investigation by Ladefoged and Maddieson (1996: 219) of the articulation of word-initial /r/ by two Italian speakers. Here, it was discovered that when these speakers articulated the words rana and rosso, there was a short approximant or 41 ‘dans un mot tel que rei < rege, isole´ ou prononce´ apre`s un mot finissant par une consonne, la cons. relativement longue se soit scinde´e en rr devenue ensuite arr-, ce qui ne se produisait pas quand r e´tait entre voyelles dans un groupe e´troitement lie´ tel que lo rei < (il)lu rege; puis il y aura normalisation de l’une ou de l’autre des formes alterne´es’ (Ronjat 1932: 54). 42 Cf. ‘Tongue-tip trills involve a complex production mechanism requiring finely tuned neuromotor adjustment of various parameters—positioning of the articulators, shape, articulator mass, stiffness and aerodynamic conditions—which accounts for the difficulties lingual trills present to inexperienced (e.g. foreign learners) and immature (e.g. children) speakers’ (Sole´ 2002b: 352).
A-prosthesis
177
vowel-like sound of about 50 ms in duration preceding the lingual contact with the alveolar ridge. A similar transitional phase was noted for one speaker at the end of the articulation of the initial rhotic. The results from these studies are suggestive, for they indicate a tendency for the realization of the articulatorily complex trilled rhotic [r] to be adapted by at least some speakers when it is in word-initial position. And the adaptation appears to involve the introduction of some sort of on-glide, either a voiceless copy of the rhotic or an indeterminate vocalic element. That such an on-glide might subsequently come to be interpreted as an independent vocalic segment and assigned a quality permissible in the language concerned, i.e. become established as a prosthetic vowel, does not seem impossible (cf. 1.6 above). (b) Romanists have not often appealed to phonologically based rationales when seeking to account for A-prosthesis. However, phonological theory can provide a useful basis for understanding the development of this process. As a starting point, we can note that after the word-initial rhotic R- underwent strengthening, effectively it can be seen to have become a geminate. Now, geminates in syllableonset position are unusual in language, though they are found. In Romance, there are certainly a number of examples. Thus, in northern Italian varieties cases have arisen as a result of syncope in word-initial syllables such that two identical consonants came together, e.g. SEXA¯(GI)NTA > [’ssa:Nta] ‘sixty’ in Bolognese, and TITT-¯INOS > [’tti:N] ‘nipples’ in Valestra, prov. Reggio Emilia (Malagoli 1934: 84). In varieties of standard French too, comparable syncopated forms such as [ppa], [mmA~ ], [ttalR] for papa, maman, tout a` l’heure may be heard. A further source of post-pausal geminates in French is provided by the clitic pronouns en and elided l’, as in: [nn] avez-vous d’autres, [ll] avez-vous vu ?43 Beyond Romance, they occur for example in Pattani Malay (Hajek and Goedemans 2003). When they arise, such onsets are theoretically anomalous for most phonological models. For instance, in moraic theory, geminates are viewed as mora-bearing but it has often been assumed that syllable onsets (and hence word-initial onsets) are not mora-bearing, so that the appropriate characterization of onset geminates continues to be controversial in this model of description.44 In government phonology, geminates are not licensed as onsets. Accordingly, it might be expected that, as with complex onsets containing any sequence except consonant þ liquid, an 43
I am indebted to Yves Charles Morin (p.c.) for the French data; cf. also Morin (1979). Already in the early days of moraic theory, certain phonologists reported linguistic data suggesting that onsets may be mora-bearing rather than weightless, e.g. Davis (1988, 1990). Since then, a number of languages have been described where there are word-initial geminates showing properties associated with the presence of moraicity, notably relevance in determining stress assignment. This has led to the growing recognition that onsets can be mora-bearing (cf. Davis 1999; Ham 2001; Topintzi 2006). 44
178
A-prosthesis
onset geminate would be interpreted as #vC|C- where the first part of the geminate forms part of the rhyme of a syllable whose nucleus ‘v’ may be phonetically unrealized. Such a representation recalls the circumstances discussed earlier for I-prosthesis, where the appearance of a prosthetic vowel was viewed phonologically as the filling of a nuclear slot motivated on general theoretical grounds. The broad tendency from pre-Classical times into Late Latin whereby syllable onsets were simplified to consist of just simplex consonants or obstruent þ liquid clusters would have actively promoted prosthesis in forms containing a word-initial geminate. Accordingly, in Romance varieties where wordinitial rhotics had undergone sufficient strengthening to be identified with the medial geminate -RR-, we might expect on theoretical grounds that prosthesis would be a possible consequence. (c) Although deep scepticism has reigned for some time amongst the great majority of romanists as to the relevance of pre-Roman languages in locally shaping Romance phonological patterns, it seems not too far fetched to suggest that such languages may have exercised some background influence in promoting the use of a prosthetic vowel before rhotics in the Latin speech of certain communities that, for many generations, may have been bilingual or diglossic, with Latin and a pre-Roman language in coexistence.45 In areas of Romania continua where pre-Roman languages long continued to exist alongside Latin, the confirmed presence in these languages of strengthened word-initial rhotics or of prosthesis before initial rhotics may well have lent some support to any internally motivated tendencies affecting initial rhotics. Appeals to pre-Roman linguistic influence have typically been made by scholars concerned with the origins of A-prosthesis in Gascon and Ibero-Romance. It was noted that most, though not all, of the Romance varieties affected were spoken originally in or near the Pyrenean area in close proximity to the Basque country which in Roman and medieval times was a good deal larger than it is today (it will be recalled that all the modern varieties of Ibero-Romance descend from forms of Latin spoken in the north of the Peninsula). Now, in early times it appears that Basque, a pre-Roman language of uncertain origins, had a prohibition not only on word-initial consonant clusters but also on word-initial rhotics. Another pre-Roman language, Iberian, which was formerly spoken mainly in the eastern coastal area of the Peninsula from Almerı´a up to modern Languedoc in SW France (Correa 2004: 38), evidently had a similar prohibition on word-initial 45 We use the term ‘pre-Roman’ in preference to ‘substratum’ since by convention the latter term refers typically to chronologically earlier languages which have ceased to exist as living languages, such as Iberian or Etruscan. In contrast, Basque and Greek have continued to be used, albeit by fewer speakers over time, in the western Pyrenean area and in southern Italy respectively.
A-prosthesis
179
rhotics. These constraints are apparent from the near-total absence of forms with word-initial
46 Cf. Correa Rodrı´guez (2004: 40) who notes that in the extant inscriptions of Iberian the two rhotics in the language only occur when preceded by a vowel so that they were impermissible in absolute word-initial position. Similarly, Echenique Elizondo (2004: 73) reports the typological coincidences between reconstructed ancient Basque and Iberian, and she identifies amongst the phonological similarities the ‘ausencia de /r-/ inicial’. 47 Rohlfs (1970: 150) only mentions the existence of [e-], citing derivations such as , RI¯PA, RO ¯ MA > errege, errota, erripa, Erroma. However, Gavel (1920: 207) who was RE¯GEM, ROTA describing Basque usage in France identifies varying outcomes determined partly by vowel copying. Thus, [e-] appears when [e] or a rounded vowel followed the rhotic, [e-] or [a-] when the low vowel [a] followed, and [e-] or [i-] when the high vowel [i] followed. Thus, the vowel [e-] appears to operate as the default but with the possibility that it may be modified through copying the height of a following unrounded vowel, raising before [i] or lowering before [a]. The reasons behind the selection of [e] as the default quality are not clear. 48 Other scholars have simply left open the possibility of substratum influence. For instance, Wu¨est (1979: 106) prudently suggests for Gascon that ‘une influence du substrat basque n’est pas exclue’.
180
A-prosthesis
that between Basque and early Romance can lead to the borrowing of prosodic features and the loss or addition of syllable-structure constraints. This certainly appears to be consistent with the possibility that Basque may have reinforced any internal tendencies within Latin to create a prosthetic vowel. Elsewhere in Romania continua, the possibility that influence from pre-Roman languages may have played a role in promoting the use of a prosthetic vowel in rhotic-initial words does not appear to have received serious investigation. This is understandable, since our knowledge of the pre-Roman languages of Sardinia, the Alps, and the Balkans is limited and inconclusive. In sum, although the causation for A-prosthesis remains poorly understood in many respects, it is possible to recognize a number of factors which may have contributed in varying degrees to the development—phonetic, phonological, and contact-based. Unfortunately, due to the general lack of research by romanists into the etiology of this category of prosthesis, the relative significance of the individual factors that have been considered remains uncertain. It is to be hoped that future investigation will help to clarify this question and also perhaps reveal further relevant contributory factors.
5.2.7
A S T RU C T U R A L LY R E L AT E D D EV E LO P M E N T : T H E I TA L I A N F O R M S I G N U D O , I G N O C C H I , E TC .
A special instance of vowel prosthesis that appears akin to A-prosthesis is found in a small set of words occurring in medieval Tuscan and, for some items, still in the archaizing register of literary Italian as well as sporadically in other central and northern varieties of Italo-Romance: e.g. ignudo ‘naked’, ignocchi ‘small dumplings’, ignucca ‘knuckle’, ignuno ‘nobody’, igno`mmero ‘elbow’. The common characteristic is the presence of a palatal nasal in word-initial position prior to the appearance of the vowel [i-], although in most cases the source of the palatality of the nasal is unfortunately not clear. Thus, for example, Rohlfs (1966: }161) derives the form ignudo from ignudare a putative variant of isnudare < EX-NUDA¯RE ‘to strip naked’, but elsewhere the possibility of its coming from gnudo < niudo < *nludu < NU¯DULU is aired (}323).49 A further case is provided by one isolated form involving etymologically initial [d-], namely Iddio ‘God’ < DE˘U(M). Almost all of these are directly paralleled by outcomes
49 Ignocco (pl. ignocchi) may derive from Langobardic *nukka ‘knuckle’ through metaphor; ignucca would represent the direct semantic continuator of the Germanic etymon. Ignuno presumably goes back to NE(C)-U¯NU(M), cf. niente < NE(C)-E˘NTE (DELI s. v. niente) or perhaps less plausibly NE-GE˘NTE (Rohlfs 1968: }499). The source of the palatal nasal in igno`mmero < CU˘BITU(M), cf. standard Italian gomito, is uncertain.
A-prosthesis
181
lacking word-initial [i-], nudo, gnocchi, nucca, gomito50 and Dio which now represent the normal form for these items in standard Italian. The unexpectedness of the initial vowel is shown by the evolution of words of similar structure such as NU¯BILU(M) > nuvolo ) nuvola ‘cloud’, NU˘CE(M) > noce ‘nut, and DE˘CEM > dieci ‘ten’ in which no such development has occurred. To explain the appearance of the initial vowel in the ign- forms, it needs to be recalled that intervocalic palatal sonorants in early Romance typically had a geminate realization. This is still the case in central and southern varieties of Italo-Romance, cf. standard Italian vigna ‘vineyard’ [’viJJa], figlia ‘daughter’ [’fiLLa]. Usually the palatal nasal was only found intervocalically but when, unusually, it came to occur word-initially the normal geminate pronunciation was maintained. In the forms developing an initial palatal nasal, vowel prosthesis seems to have been adopted therefore as a strategy to enable syllabification to occur. The selection of the vowel quality [i] was evidently determined by the palatal nature of the following consonant, i.e. strategy (iv) as identified in 1.6. Also relevant perhaps was the presence in medieval Tuscan of an existing and frequently occurring prosthetic vowel [i] (cf. 4.3.3 and 5.2.4). The appearance of the initial vowel [i-] in Iddio can likewise be attributed to the prior development of a strengthened consonant [dd] word-initially. Indeed, in standard Italian the plosive in the word Dio (as well as in its associated plural and feminine forms dei, dea, dee) remains unique in systematically having a d’di:o] geminate realization when preceded by a vowel, e.g. senza Dio [’s¡ntsa as against senza dita ‘without fingers’ [’s¡ntsa ’di:ta], although the reasons for this pronunciation are not certain.51 However, the presence of a geminate initial consonant was doubtless of key importance in triggering the appearance of a prosthetic vowel in order to enable syllabification to take place. The selection of the vowel quality [i] owes itself to similar factors as those identified for forms beginning with ign-, assimilation to the following coronal consonant [dd] and the existing presence of [i] as a general prosthetic vowel type.52 50
Italian nessuno ‘nobody’ derives from a different etymon NE(C)-IPSU-U¯NU(M). For instance, Rohlfs (1966: }153) attributes the gemination to the effect of rafforzamento fonosintattico in the phrase SOLUS DEUS, with the strengthened alternant subsequently being generalized. However, Skytte (1975: 273, n. 46) views the gemination as the result of assimilation in the phrase (ILLE DEUS >) il dio > iddio, with once again later generalization of this alternant (cf. Maiden and Robustelli 2000: 12). 52 If the origin of the prosthetic vowel lies in the phrase ILLE DEUS (cf. preceding footnote), a further possibility is that the quality of the vowel merely continues that of the initial syllable of ILLE. The outcome with initial [i] can be explained as a result of the frequent raising of initial unstressed [e] > [i] in Tuscan, a development which is seen in cipolla, virtu`, signore, etc. and which also affected proclitics such as di and the clitic pronouns mi, ti, si. The vowel of the article [il] evidently was also affected, although 51
182
A-prosthesis
The factors underlying the development of a prosthetic vowel in this idiosyncratic subset of forms show obvious similarities to those motivating A-prosthesis in rhotic-initial forms in early Romance. In both cases, the appearance of a strengthened or geminate word-initial consonant has been the catalyst for change. However in view of its wide-ranging nature across Romance, A-prosthesis with rhotic-initial forms probably got under way chronologically rather earlier than the much more localized development which we have just been considering.
5.3
A-prosthesis: later developments
From the later Middle Ages onward, there have been a number of notable changes in the incidence of A-prosthesis across Romance. In some varieties where prosthesis had established itself as a process, the broad trend has been for progressive reduction in its productivity. More rarely, other varieties have seen the incidence of A-prosthesis maintained or even enhanced.
5.3.1
VA R I E T I E S S H OW I N G S I G N I F I C A N T R E G R E S S I O N O F A - P RO S T H E S I S
Regression has occurred notably in three Romance varieties, Rheto-Romance, Gascon, and Catalan. In Rheto-Romance, there is evidence that certain varieties may have developed a productive phonological process of A-prosthesis. These include especially those spoken in the Upper Engadine and to a lesser extent in Friulian. In both cases, the more recent decline in the productivity of the process has evidently been extreme. In Gascon, it also appears that in the medieval period A-prosthesis had established itself in most if not all varieties as a fully productive process. However, from the sixteenth century onward, the use of prosthesis has experienced a steady regression so that now it continues to operate as an active process in just a relatively small subset of Gascon dialects. Finally, in Catalan, although A-prosthesis does not appear to have ever become established as a systematic phonological rule during the medieval period, it is striking that most varieties have seen a considerable retreat from the use of prosthetic forms. We outline below the main lines of development in each of these three areas of Romance.
southern Tuscan varieties like those of Siena and (medieval) Arezzo preserved el, cf. Rohlfs (1968: }130, 1967: }414).
A-prosthesis
183
5.3.1.1 Rheto-Romance
For Rheto-Romance, the available textual evidence indicates that it was Upper Engadinish in the eastern Grisons that became most subject to A-prosthesis in the Middle Ages. Unfortunately, our knowledge of medieval developments in this variety is poor since no substantial documents have come down to us from earlier than the sixteenth century when texts such as the Bible translation in Engadinish by Jakob Bifrun appeared. However, these texts reveal a widespread use of Aprosthesis suggesting that the process was well established by that time (cf. 5.2.2 above). In the earliest detailed scholarly review of all Rheto-Romance by Gartner (1883: }92), allusion is certainly made to the appearance of prosthetic [a-] in forms beginning with etymological R- in certain contemporary varieties, but perhaps surprisingly the only data to be provided relate to Dolomitish. For Upper Engadinish, Gartner (loc. cit.) simply states that ‘a pre-posed a . . . seems to have been exceptionally common’. The retrospective nature of this observation is borne out by the use of illustrative examples drawn solely from sixteenth- to early eighteenth-century texts rather than forms found in contemporary usage. Further revealing data come from subsequent studies of Engadinish varieties. Thus, Walberg (1907: }102) reports that A-prosthesis is frequent in the Upper Engadinish dialect of Celerina-Cresta (ara´m ‘branch’, aræn˜ts ‘back’, arikr ‘to laugh’ < RA¯MU(M), RE¯NES, RI¯DERE), although it is noted that, in the majority of cases, non-prosthetic forms are also found in variation with the prosthetic ones. No indication however is given of the factors governing the variation. More recent descriptions of Engadinish suggest that A-prosthesis has at best a marginal status. The grammar of (Lower) Engadinish by Arquint (1964) contains no mention of prosthetic forms; only items such as rai ‘king’, roba ‘goods, possessions’, ruina ‘ruin’, ru´mper ‘to break’ are cited. The Engadinish-German dictionaries of Peer (1962) and Bezzola and To¨njachen (1976) offer fuller data and there is some evidence of prosthetic forms still being in use, but they appear to have limited currency. Forms cited include arains (f.pl.) ‘back’, arait ‘net’, ared ‘productivity, diligence’ (deverbal form from re´der < RE˘NDERE), arir ‘to laugh’, arisch ‘root’, aro¨v ‘entreaty, request’ (deverbal form from ROGA¯RE), aru´vi ‘dew’, but non-prosthetic variants of these are also noted (rain, rait, red, rir rı´er, risch, ro¨v, ruvi) though with no indication of where one or other form appears. In Lower Engadinish, the counterpart forms are regularly non-prosthetic. It is also notable that, for a substantial number of lexical items beginning with etymological R-, a non-prosthetic form alone has now apparently established itself as the only acceptable realization in the semi-standardized form of modern Upper Engadinish. Thus, words such as rai raig ‘king’, ram ‘branch’, raz ‘ray’, rouda ‘wheel’, rer ‘rare’, ru´mper ‘to break’. The move away from the use of Aprosthesis in Upper Engadinish may reflect in part the result of dialect levelling in the Engadine. Its general absence in Lower Engadinish and indeed other varieties
184
A-prosthesis
of the Grisons would give this process high salience which would be likely to undergo elimination in any moves toward dialect levelling. For Dolomitish, Gartner (1883: }92) cited as prosthetic forms ar P:sˇk ‘frog’, arube´ ‘to steal’ from the variety of Badia,53 and arı´k ‘rich’, arı´de ‘to laugh’ from the transitional Alpine dialect of Erto.54 In his more detailed study of usage in Erto, however, prosthesis is stated to be generally ‘rare’ in Badia and Enneberg, but verb forms in ar- were in evidence particularly in Erto, Fassa, and Gardena, and more rarely in Vigo (Lower Fassa), Buchenstein, Ampezzo (Tyrol), and northern Veneto (Gartner 1892: 201, n.1). More recent studies of Dolomitish indicate that the scope of prosthesis has been much reduced. Thus, Elwert (1972: 118) states that it is rare in Fassa and that the only surviving traces are in forms containing etymological prefixal RE- where the prosthetic vowel has been lexicalized, e.g. [arne’¡r] ‘to drown’ < RE-NECA¯RE. These forms, however, do not represent true examples of A-prosthesis. Rather, they illustrate U-prosthesis which occurred after the initial vowel had been syncopated (see Chapter 6). In forms with initial R- where the following vowel did not undergo syncope (as in Gartner’s examples), there is no sign of a prosthetic vowel in Fassan, e.g. [’ri:va], [rut], [’r P:da] ‘slope, ). Kramer (1977: 174) echoes this finding belch, wheel’ (< RI¯PA, RU¯CTU(M), ROTA for all the other major varieties of Dolomitish. The implication is therefore that A-prosthesis was at best of marginal status in Dolomitish and that in more recent times it has ceased totally to be productive. Finally, in Friulian there has been a comparable retreat from A-prosthesis. In later medieval texts, as we have seen, there are frequent attestations of prosthetic forms although the evidence does not suggest that a categorical rule of prosthesis ever developed in any Friulian variety. However, where prosthetic forms did arise it appears that in more recent centuries the non-prosthetic alternants have progressively become re-established as the sole occurring form. In Gartner (1883: 184–5), no prosthetic forms are cited as reflexes of rauba ‘possessions’, ‘wheel’ is reported to have outcomes showing but in a few localities ROTA A-prosthesis, e.g. [a’ru¡d¡] in Paluzza and Tolmezzo in the north, [ra’u¡d¡] in
53 Neither of these forms is unproblematic. The verb arube´ could well be prefixal, and the etymology of ar P:sˇk is uncertain (REW 1329 relates it to *BROSCUS ‘toad’ blended with Low Latin RUSPUS or ROSPUS ‘toad’ of unknown origin). Only limited traces of a prosthetic vowel have remained in these items. For the first one, AIS pt. 314 (Colfosco in Badia) has [l awro:ʃk] whilst pt. 305 (S. Vigilio di Marebbe) has [a’r P:ʃk] (map III, 453). Only these two localities have prosthetic reflexes. For the second, no form is cited for pt. 314 and none of the surrounding localities recorded have prosthetic reflexes. 54 Erto lies approximately 15 km NE of Belluno. In this study, the existence of numerous non-prosthetic forms, rik, ride, r¡jt ‘net’, rey¡jve ‘to receive’, etc. is also indicated (p. 341).
A-prosthesis
185
Cormons and [ra’ueda] in Gorizia in the south-east, the latter two forms undergoing metathesis in the first syllable [ar-] > [ra-].55 In certain modern Friulian varieties, words showing evidence of A-prosthesis remain in use but these are not numerous and their originally prosthetic vowel has usually now been lexicalized. (M), aracli ‘prop to They include items like arue`de ‘wheel’, arue`z ‘bunch’ < ROTEU support plants’ (Vanelli 1984: 282, n. 4).56 However, these prosthetic forms are by no means generalized across Friulian. For instance, the AIS (map VI, 1227) cites prosthetic forms for ‘wheel’ at just four points,57 the form in the emerging standard Friulian variety being rue`de. Furthermore, it is suggestive that many of the prosthetic forms cited in Zamboni et al. (1984–7) are described as archaic, e.g. ara`di ‘to shave’, arefuida` ‘to reject’, aribo`la ‘seething’, ariceu ‘to receive’, arodaˆr wheelwright’, aromonda` ‘to prune’, aro`dolo ‘roll’, aro`se ‘rose’, aro`nc ‘terrace’ (as against modern ra`di, refudaˆ, ribue`le, rice´vi, ruedaˆr, remondaˆ, rul, rose, ronc). The evidence therefore points to a general retreat from A-prosthesis and to the absence in any modern Friulian variety of a genuinely productive rule of vowel prosthesis. 5.3.1.2
Gascon
Although the evidence from medieval texts suggests that A-prosthesis enjoyed considerable productivity across most if not all the Romance varieties of Gascony, systematic studies of modern usage indicate that comparatively few varieties have maintained prosthesis as a active process. In the study by Bec (1968), which took into account the use of A-prosthesis in a number of words with initial etymological R-, it was found that the eastern frontier ran down a little to the east of Boulogne-sur-Gesse and St Gaudens before swinging slightly to the south-west to pass a little to the east of Cane´jan in the Val d’Aran and Bagne`res-de-Luchon in the Pyrenees. To the west of this line, the varieties which have conserved the earlier status quo most faithfully lie in the region closest to the Pyrenees and hence most remote from outside influence.58 Even here, though, there has been a noticeable diminution in the productivity of A-prosthesis in more recent times. 55
Reflexes of ROTA in Friulian which developed an initial [a-] have generally been assumed by scholars to have undergone phonologically conditioned vowel prosthesis. However, it is not impossible that morphological factors have played a role through the . recutting of grammatical boundaries, as in ILLA ROTA > ILL’AROTA 56 Curiously, neither of the two latter forms appear in Zamboni et al. (1984–7). 57 In the north-east and east, [un¡ ar’w¡d¡], (pl.) [ar’wedes] at pt. 319 (Cedarchis, Arta); [ar’jod¡ ], (pl.) [ar’jodis] at pt. 329 (Travasans); [aru’ede], (pl.) [aru’edis] at pt. 348 (Sant’Odorico); and in the south-east, [ar’j Pda], (pl.) [ar’j Pde] at pt. 378 (Montona). 58 Cf. Millardet (1910: 121) who identifies the Landes and Basses-Pyre´ne´es as ‘le dernier refuge de l’ancienne prothe`se’.
186
A-prosthesis
Rhotic-initial learned words and French borrowings in particular have very commonly resisted nativization through prosthesis, e.g. rriðe`w ‘curtain’, rrujo`le ‘German measles’, rre´fula` ‘to repress’ (Fr. rideau, rougeole, refouler). Furthermore, cases of native Gascon words failing to undergo expected prosthesis have been observed. Thus, Bec (1968: 179) reports the findings of Lalanne dating from the mid twentieth century59 which indicate the presence of cases of polymorphism arr- rr- in native words of western Gascon dialects, and this has been found with some lexical items even in dialects of the extreme SW of Gascony where Aprosthesis may be expected to be faithfully preserved. In Pyrenean dialects of Gascon further to the east, such as that of Bagne`res-de-Luchon, words with initial r- borrowed from French evidently continued to undergo prosthesis until at least the early twentieth century, e.g. arru¨ðe`w ‘curtain’, arrande´bus ‘appointment’, arramuna` ‘to sweep a chimney’ (Fr. rideau, rendez-vous, ramoner). However, it is not clear how productive A-prosthesis has remained as a process up to the present day. In the Gascon variety spoken in the Bethmale Valley, situated a little further east of Bagne`res-de-Luchon and to the SE of St Gaudens, however, significant regression in the use of prosthesis has been reported.60 A detailed investigation of this variety carried out in 1931 revealed that prosthesis had already been largely abandoned (Scho¨nthaler 1937: }74). Prosthetic forms appeared only rarely as lexicalized items in the local speech, the examples cited being arre´ ‘nothing’ (< RE˘M), and two verbs arrapa´ ‘to seize, snatch’ (< Germ. rapo¯n) and the denominal arrama´ ‘to support (beans) with branches’ ( RA¯MU(M) ‘branch’).61 An important factor in undermining the productivity of A-prosthesis, and particularly so in more eastern and northern varieties of Gascon, has
59
The relevant work by Lalanne is an undated roneotyped study published in StVincent-de-Paul and entitled L’inde´pendance des aires linguistiques en Gascogne maritime. It has not been possible for the present writer to consult this work directly. 60 The dialectal situation in the Bethmale Valley has received a good deal of attention. In addition to coverage in the ALG pt 790S and in the monograph by Scho¨nthaler (1937), it was the subject of close sociolinguistic examination more recently by Helfenstein, Keller, and Kristol (Wu¨est and Kristol 1993: 83–108). The focus of attention here was the continued use of Gascon amongst the inhabitants of the Valley. 61 The preservation of prosthetic arre´ in Bethmale finds parallels in many other Gascon varieties where A-prosthesis has been abandoned; cf. fn. 9. Bec (1968: 175) links the widespread maintenance of a prosthetic vowel in this item with the need for a fuller phonetic form to give emphasis to an otherwise brief monosyllable. For the two verbal forms, there is a strong possibility that prefixation with a- < AD- is involved rather than prosthesis. Further prosthetic examples from Bethmale are provided by the ALG, pt. 790S, e.g. arra´t ‘rat’ (map I, 3). Scho¨nthaler (1937: }119) reports only the non-prosthetic form rrat.
A-prosthesis
187
undoubtedly been contact influence from Occitan and more especially standard French. Already in the nineteenth century, Luchaire (1879: 208) recognized the significance of this influence. After observing that there was a ‘repugnance’ for initial [r-] amongst contemporary Gascon speakers which led them to strengthen the consonant and precede it with a prosthetic vowel, he noted that this adaptation was all the more likely and consistent, the more rural the variety was and the less influenced it was by French, and also the more geographically distant it was from the Garonne and hence from the Occitan of Languedoc. Scho¨nthaler (1937) identifies as likely sociolinguistic routes of French influence, school, the Church, military service and commercial contacts. The Gascon of the Aran Valley has been shielded in some measure from influence from standard French, since this area has never formed part of the French state. However, other contact influences have operated increasingly on the speech of the inhabitants, this time from Catalan and Castilian. The absence of Aprosthesis in these Ibero-Romance varieties has had a similar negative impact on prosthetic usage as that experienced in Gascony. We may note that an equally negative impact can be found on the southern side of the Pyrenees in varieties of Aragonese, e.g. Rafel i Fontanals (1980) presents forms like [rre’ðono] ‘round (m. sg.)’, [’rrato] ‘rat’ and [’rrweða] ‘wheel’ for the Benasc Valley and Badı´a Margarit (1950) cites [rre’ðiɾ] ‘to laugh’, [’rramo] ‘branch’, [’rrio] ‘river’, [’rrojo] ‘red’ for the variety of Bielsa. Although Schädel (1908: 151) reported that A-prosthesis still appeared regularly in Aranais, the decline in productivity of this process has been considerable over more recent years. In fact, Bec (1968: 181) confirms that with the exception of the Canéjan Valley in the NW of the Aranais-speaking region, the process of A-prosthesis is clearly regressing. The special circumstances in the Canéjan Valley owe themselves evidently to its more northerly, isolated location close to other prosthetizing areas in Gascony and to the stronger pastoral tradition there which fosters regular contact with those Gascon areas. More recently, attention has been called by Coromines (1990: 41) to diastratic variation and lexical layering in the use of A-prosthesis in Aranais. Diastratically, the more cultivated a speaker is the more likely (s)he is to suppress A-prosthesis; one result of this may be hypercorrection, as in ratja ‘wild oats’ < arratja < (AVE¯NA) ERRA˘TICA. Lexically, the use of A-prosthesis by speakers is more likely with well-established words, but less usual with words perceived as being foreign or not part of the more familiar native lexicon. 5.3.1.3
Catalan
In modern Catalan which is based on the educated usage of Barcelona, there is very little evidence today of A-prosthesis. However, in more peripheral varieties of Catalan, prosthetic forms are still reported to be fairly frequent. For example, in Alguere`s (Algherese) which is spoken in NW Sardinia having been introduced
188
A-prosthesis
into the island in the fourteenth century by settlers who appear to have come predominantly from eastern parts of the Catalan-speaking area,62 items like the following are found (Blasco Ferrer 1984: } 77): Alguerès [arra’k PVt] [arras’tat] [arru’ba] [arru’veV] [ar’res]
Standard Catalan record restat rubar rovell res
‘memory’ ‘remained’ (p.p.) ‘to steal’ ‘rust’ ‘something’
etymon (deverbal) RE-CORD RESTA¯TU(M) (Germ.) raubo¯n RUBICULU(M)63 RE¯S
Also, in the variety of Catalan used in the Roussillon, cases of prosthesis are reported as still occurring in the early twentieth century, as in errebe´n [‰rr‰’ en] ‘suddenly’ (< REPE˘NTE) and various verb forms such as erregar [‰rr‰’ªa] ‘to water’ (< RIGA¯RE), errollar [‰rruL’La] ‘to form a circle, surround’ (< ROTULA¯RE), though such verb forms may also reflect influence from prefixal a- < AD-.64 However, in verb forms the prosthetic vowel only appears when the following syllable is unstressed, hence [’rreª‰] ‘(s)he waters’ (< RIGAT) etc. represents the only occurring realization. Also, even the prosthetic verb forms have non-prosthetic variants, e.g. [‰rr‰’ªa] [rr‰’ªa], although the circumstances for the use of one rather than the other are not made clear. The evidence therefore points to the lingering, albeit much diminished, presence of A-prosthesis in this variety.65 In the light of these pieces of data, it seems reasonable to infer that in earlier times A-prosthesis may have enjoyed a good deal of currency across most if not all the Catalan-speaking region. From the late medieval period onwards, however, there has been a progressive retreat from the use of this phonological process, most significantly amongst speakers of the emerging standard language in the Barcelona area.
62
The question of the precise origins of the early settlers in Alguer (Alghero) has been the subject of much controversy. The earlier presumption that most came from the Barcelona area has been largely rejected. The general consensus now is that the settlers probably came from various parts of the Iberian Peninsula but that, predictably perhaps, areas nearer to the sea were more represented, e.g. Barcelona and environs, Valencia, Roussillon, and the Balearic Islands (Blasco Ferrer 1984: 4–5; Veny 1987: 102–5). 63 As etymon, REW 7348 proposes a form derived from RUBI¯GO ‘rust’. 64 Data for this Catalan variety are drawn from Fouche´ (1980a: 206). 65 As Fouche´ (1980a: 206) observes of the contemporary situation with prosthesis, ‘il s’en faut de beaucoup qu’elle [= la prosthe`se] soit aussi fre´quente que la prosthe`se de a dans le domaine gascon.’
A-prosthesis
5.3.2
189
VA R I E T I E S S H OW I N G M A I N T E NA N C E O F A - P RO S T H E S I S
Unlike Gascon and Engadinish where A-prosthesis has experienced notable regression, Campidanese Sardinian has largely preserved the productivity of the process up to modern times, although it appears that some limited undermining of it has occurred more recently. But while prosthesis has remained in large measure an active process, a notable change has occurred in the quality adopted by the prosthetic vowel. The earliest texts written in Campidanese, dating from the twelfth century, indicate that the quality was determined by vowel copying (cf. 5.2.4). However, in the pattern which is now operative, a low-quality [a-] is systematically used. This pattern appears to have originated in the south of the island, doubtless with its focal point in the usage of the influential town of Cagliari from where it has spread progressively northward, reaching as far as Milis in the west of Sardinia but, in the east, only up to Barbaricino and Ogliastra where the earlier vowel-copying pattern has been partly or completely retained. The circumstances surrounding the development of the generalized use of prosthetic [a-] are not clear. The strong influence on Cagliari from mainland Italy, and especially from Tuscany, in the thirteenth and fourteenth centuries may be of significance but this remains speculative.66 A further aspect of the rise of Aprosthesis with generalized [a-] is that as it diffused northwards from southern Sardinia, it appears to have coincided with the displacement of I-prosthesis. As a result, in modern Campidanese we find that I-prosthesis is no longer productive,67 whereas in northern varieties, such as Logudorese and Nuorese, I-prosthesis has remained productive but A-prosthesis has never become established. Finally, in modern times it appears that the productivity of A-prosthesis in Campidanese has ceased on occasions to be complete. In addition to one native word which is reported mysteriously not to show prosthesis, namely [r¡i] ‘king’,68 the recent influx of loanwords from Italian with an initial rhotic may no longer be systematically pronounced with a prosthetic [a-] by some speakers. Thus, Bolognesi (1998: 42–3) notes that while items like radio normally surface with a prosthetic vowel giving [ar’raðiu], A-prosthesis can be variable in words Cf. Wagner (1951: 52) ‘In Cagliari, la lingua toscana era, nei secoli XIII e XIV, talmente diffusa che . . . intacco` fortemente il sardo della capitale e della pianura.’ One can hypothesize that the generalized use of prosthetic [a-] would represent a simpler, more transparent pattern to Tuscan-influenced speakers and would find a counterpart of sorts in the common occurrence in Tuscan of verbal forms in (prefixal) arr-. 67 Blasco Ferrer (1984: 210) attributes the regression of I-prosthesis in Campidanese to Italian influence. 68 Bolognesi (1998: 43) attributes this anomaly to the fact that the kings (or their nearest equivalents) in medieval Sardinia were known as judges. Yet, this would scarcely explain the lack of prosthesis. 66
190
A-prosthesis
such as [’r Pð¡u] ‘rodeo’, [rikja’mau] ‘called up for military service’ (< richiamato). The development here is reminiscent of what we saw earlier for Gascon, with the more prestigious pronunciation patterns of a standard variety serving to undermine native prosthetizing usage.
5.3.3
VA R I E T I E S S H OW I N G E N H A N C E M E N T O F A - P RO S T H E S I S
A-prosthesis has gained ground in two Romance varieties, southern Italian and Aromanian. The circumstances of these two cases are however rather different. 5.3.3.1 Southern Italian
It will be recalled that there is some uncertainty as to whether genuine phonologically based A-prosthesis did develop as a process in southern Italy in the Middle Ages. Certainly, there are attested forms which have taken on an initial [a-] but it seems that these have arisen primarily as a result of morphological rather than phonological factors. However, from the later medieval period onward, a significant numbers of words can be found containing an initial [a-] which can justifiably be viewed as phonologically conditioned. Examples are: (Neapolitan) abbusso ‘box tree’, addaino ‘fallow deer’, Adda`vete ‘David’, aggente ‘people’, all of which are cited in D’Ambra (1873). Ultimately these items derive from BU˘XU(M), Late Latin DA¯MU(M), (Hebrew proper name) David, GE˘NTE(M), but they entered southern Italian usage as later borrowings from standard Italian. This is clear as in southern Italian varieties, the original word-initial voiced plosives B-, D-, and G- (in palatalizing contexts) had typically developed to [v-], [ð-](> [r-]), and [j-] respectively (Rohlfs 1966: }} 150, 153, 156). As a result, Italian borrowings like these which contained initial [b-], [d-], and [q-] were subject to nativization. However, in an apparent attempt to reproduce these unfamiliar word-initial consonant types appropriately, more educated speakers appear in some cases to have hypercharacterized the ‘correct’ standard Italian pronunciation and given rise to a geminate realization which was then subject to A-prosthesis. Voiceless obstruents have not been affected in the same way since they had usually remained unchanged in word-initial position in native word-forms. Sonorants too, other than r- whose general tendency to strengthen in southern Italo-Romance has been noted, do not seem to be generally affected. However, a tendency to strengthen etymological initial M- has been noted for Neapolitan and other southern varieties, mmaliddittu ‘cursed’, mmerda ‘excrement’, mmesca ‘mixture’, mmira ‘aim’, mmorra ‘group, band’,69 with occasional
Ð
D’Ambra (1973: 230) says of m-, ‘Questa lettera si pronunzia sempre con forza, e assai spesso si raddoppia in capo alla parola.’ The frequency, and hence familiarity, of word69
A-prosthesis
191
instances of prosthesis as in ammaturo ‘ripe, mature’ and ammalamente ‘in a bad way’ (D’Ambra 1873). 5.3.3.2
Aromanian
As has been noted earlier, there is no direct textual evidence of any substance for Aromanian until the eighteenth century. However, the lack of attested examples in Daco-Romanian in texts from the late Middle Ages onward provides suggestive, though not decisive, evidence that the incidence of A-prosthesis may have been at best modest across all types of Balkan-Romance in earlier times. Nonetheless, its incidence has increased south of the Danube and more particularly in more southerly Aromanian varieties,70 although in a more marginal way in Megleno-Romanian (Rosetti 1978: 403, 415). The process is attested in eighteenth-century texts written with Greek lettering by Aromanian writers living in Albania, e.g. arra˘du ‘I laugh’, arra˘ma˘sa˘tura˘ ‘remainder’ (< RI¯D(E)O, *RE-MANS¯ RA) cited by Capidan (1932: }207), and it has come to enjoy considerable ITU productivity more recently. For example, in addition to the items in Figure 5.1, ) ‘river, I fight, I we may cite arıˆu, alumtu, acumpıˆru (< RI¯VU(M), LU˘CTO, COMPARO buy’, although lexical items of comparable structure such as rogu, lapte, cupa˘ (< ˘ PPA) ‘I ask, milk, goblet’ are reported with no prosthetic vowel , LACTE, CU ROGO (Papahagi 1974). Vowel prosthesis has evidently not come to apply categorically therefore, but it is widespread and clearly occurs preferentially before an etymological word-initial rhotic. This would seem to indicate that forms with initial rhotics formed the starting point for the development of A-prosthesis here as elsewhere. The relative frequency with which A-prosthesis has operated on words beginning with different word-initial segment types has been closely studied so that parameters can be identified. Three proposals are set out in Figure 5.5 below.
initial [mm-] in Neapolitan has been significantly increased through the independent development to [mm-] of prepositional or prefixal IN þ voiced labial consonant, IN MORTE > mmorte, IN-VITA¯RE > mmitare, etc. Some uncertainty surrounds the use of A-prosthesis in Farserotic. Rosetti (1978: 415) claims that it is found in this variety whilst Giese (1965: 299) asserts the opposite. Both base their claims on the earlier study of Aromanian by Capidan (1932) which is unfortunately not entirely clear on this point. Capidan (1932: }206) also reports that in Farserotic speech the rhotic /r/ has more recently begun to move from an alveolar to a back articulation except in word-medial position. Usually, a velar realization is found but with female speakers a uvular pronunciation may occur. 70
192
A-prosthesis (Capidan)
r- n- s- l- m- p- f- g-
(Giese)
r-
l-
(Schlösser)
r-
l-
n-
higher frequency
mf-
u- v- z- k- γ- j-
k-
h-
v-
d-
i-
s-
s + cons.
lower frequency
very rare
FIGURE 5.5. Parameter showing relative frequency of A-prosthesis in Aromanian Sources: Capidan (1932: }27), Giese (1965: 299), Schlo¨sser (1985)
The first parameter comes from the detailed data presented in the classic study on Aromanian by Capidan which were established on the basis of items cited in (unspecified) dictionaries of this variety. The second parameter is claimed to draw on Capidan’s, but it is not only less detailed but also changes the relative position of certain word-initial segment types. The third by Schlo¨sser results from a more recent description of the modern Aromanian variety of Metsovon located in the Pindos Mountains of northern Greece. A major point which emerges from such studies is that A-prosthesis is not categorical in words beginning with any initial consonant type, even the rhotic [r-] which is by far the most frequent trigger of prosthesis. Furthermore, for certain initial consonants it is unclear how appropriate it is to assume that A-prosthesis is genuinely operative. For instance, Capidan includes word-initial [i-] in his parameter but no data are advanced in support of this claim. The only real candidate which figures in the major dictionary of Aromanian by Papahagi (1974) appears to be airate ‘revenue, income’ (< Turkish iraˆd). The origin of the initial [a-] here is unclear, however. Similarly, Schlo¨sser postulates A-prosthesis in forms beginning with [f-], [v-], and [s þ cons.-] although the evidence for each of these is limited to a single example which in each case is open to reinterpretation. But although doubt surrounds certain aspects of the data, one clear and important development with A-prosthesis in southern Aromanian has been the extension of its operation, albeit in a limited way, to include vowelinitial words, especially those beginning with [u-], as in aungu, aumbra˘, auu˘a˘ (< U˘NGO, U˘MBRA, U¯VA) ‘I grease, shade, grape’. The preferential application to vowel-initial forms beginning with [u-] may be connected with the fact that it is a high vowel and hence the least sonorous and salient of initial vowel types in Aromanian.71 The reasons for the partial generalization of A-prosthesis in southern Aromanian are not clear, although Rosetti (1978: 386, 415) identifies some potentially
Capidan (1932: }27) also includes the high front vowel [i-] in his parameter. However, as we have seen, no reliable data appear to support this claim. 71
A-prosthesis
193
relevant factors. These relate to morphosyntactic developments in early BalkanRomance which may have created the conditions for the phonological effects noted. Two of the factors find widespread parallels elsewhere in Romance, namely the appearance of feminine singular pre-nominal proclitic forms ending in [-a], such as ILLA > [a], and the presence of substantial numbers of verbs with prefixal AD- (cf. 1.7). In addition to these, there is the rise of the syntactic pattern whereby infinitives are introduced by the preposition AD (cf. southern Italian, 5.2.1.1). In all of these contexts, it is suggested, speakers may have reinterpreted the morphological structure of individual lexical items, the result of which might be either aphaeresis or A-prosthesis. The former outcome is found lexicalized across all (M) > Daco-Rom., Arom. noatin Romanian varieties in items such as ANNOTINU ‘one-year-old lamb’. But whereas Daco-Romanian has tended to engage more generally in aphaeresis, southern Aromanian has preferentially used prosthesis. The reasons for this and for the recent striking growth in its productivity remain uncertain, however. If this interpretation of the genesis of A-prosthesis in southern Aromanian is correct, the circumstances bear some resemblance to what was seen in southern Italian (5.2.1.1), where an originally morphologically conditioned process of vowel prosthesis has subsequently given rise to phonologically based A-prosthesis.
6 U-prosthesis U-prosthesis represents chronologically the third and final of the major categories of vowel prosthesis that have operated in Romance. Like I-prosthesis, it arose as a process that served to simplify complex word-initial onset sequences. But, in contrast to it, U-prosthesis led to the appearance of a vowel that has been rather more variable in quality although a low value [a] has predominated. To reflect the sometimes unspecific quality of the prosthetic vowel and to avoid confusion with the category of prosthesis considered in the previous chapter (A-prosthesis), for convenience we have adopted the term ‘U-prosthesis’ for the category of prosthesis under consideration here. Some illustrative examples may be cited.
RE-
(prefixal) (Germ.) likkon LEVA¯RE ¯ TEM NEPO MINA¯RE
Valsesia (N Piedmont)
Novellara Bologna (Emilia) (Romagna)
ar´an’te: al’ke: -3 an’vP: am’ne:
ar’meter al’k¡:r al’v¡:r an’vo: mn¡:r
artsin’t ¡:r al’k¡:r al’dA:m4 an’vAwd mn¡:r
Celerina (Upper Engadine) algUr’de:r1 -2 al’ve:r 5 YN’gu‰ta mne:r
‘to lick’ ‘to raise’ ‘nephew’ ‘to drive’
Sources: Spoerri (1918); Malagoli (1910-13); Mainoldi (1967); Coco (1970); Walberg (1907).
U-prosthesis began to operate as an active phonological process in Romance in the course of the medieval period, building on the output of a prior syncope the details of which we consider below in 6.1.2. More recently, however, there have
1
Reflexes of three different etyma containing prefixal RE- are represented in this row. Columns 1 and 3 contain reflexes of RECENTA¯RE ‘to rinse’, in column 2 the etymon is REMITTERE ‘to put back’, and in column 4 there is the reflex of RECORDA¯RE ‘to recall’. 2 A form of comparable phonological shape is [al’ger] ‘to melt, liquify’ < LIQUA¯RE. 3 Cf. the derivationally related [al’qÐ er] ‘light’ < LEV-IA¯RIU(M). 4 Deriving from LAETA¯MEN ‘dung’. 5 Reflex of NE(C)-GU˘TTA ‘not a drop’ ! ‘nothing’ (cf. French ne . . . goutte).
U-prosthesis
195
been some developments affecting the productivity of U-prosthesis within many of the linguistic varieties where it had arisen. It will therefore be appropriate once again to split our treatment into two broad chronological sections reflecting this difference. The dividing line lies very approximately at the outset of the early modern period (sixteenth century) when written evidence for Romance first becomes available for certain varieties affected by U-prosthesis.
6.1 Rise of U-prosthesis: early developments 6.1.1
GEOGRAPHICAL DISTRIBUTION
U-prosthesis has occurred particularly in the central zone of Romania continua. In the lateral areas, i.e. Ibero-Romance and Balkan Romance, its effects have been much more limited; indeed, even when lexical items developed complex wordinitial onsets resembling those that became subject to U-prosthesis in central varieties of Romance, they have remained unaffected. For instance, forms such as [’psoN] < PERSO¯NA ‘person’ with complex initial onsets have arisen in more recent times in Portuguese but as yet no indications have been reported of their being subject to prosthesis. The only circumstance under which U-prosthesis can be found operating in lateral areas came about as a result of certain monosyllabic grammatical forms becoming procliticized and undergoing vowel weakening. A familiar example involves personal pronouns such as ME¯, TE¯, and SE¯ which developed into satellites before a verbal host, as in ME¯ VIDET ‘(s)he sees me’, and subsequently emerged with a prosthetic vowel. Prosthesis of this type has occurred in several Romance varieties such as Catalan and Romanian (see 6.1.4.3). Within the central zone, U-prosthesis has been widely in evidence in northern Italo-Romance, operating on both lexical and grammatical items (see Map 4). Piedmont and Emilia-Romagna have been focal areas for the development. In contrast, varieties from Liguria show few if any signs of U-prosthesis and likewise most of Lombardy and the Veneto have been little affected, although in more peripheral areas of these regions there are varieties in which U-prosthesis has occurred due to influence from adjacent prosthetizing varieties. Italo-Romance varieties spoken further south, including the Florentine-based standard variety of Italian, are scarcely involved, although U-prosthesis has taken place in northern transitional varieties of Lunigiana, particularly near the border with Emilia. In Rheto-Romance, Friulian varieties offer no indications of U-prosthesis, and Ladin has experienced prosthesis only in lexical items beginning with prefixal RE-, as in [arba’se] ‘to lower’ < RE-BASSA¯RE in the varieties of San Vigilio and Arabba (Kramer 1977: 174). However, in the Grisons of Switzerland the process has been productive with lexical and grammatical items, especially in eastern varieties (see Map 4). The focal area here appears to be the Upper Engadine. Further to the east
196
U-prosthesis
of the Upper Engadine, the incidence of U-prosthesis diminishes: in the Lower Engadine it has enjoyed a significantly lower degree of productivity, and in the far east of the Grisons the varieties of the Val Mu¨stair (or Mu¨nstertal) reportedly offered just a lone example in the mid twentieth century, namely [ar´ajn’tar] ‘to rinse’ < RECENTA¯RE, although at earlier periods the phenomenon may have been rather more in evidence (Schorta 1938). U-prosthesis also becomes increasingly rare the further westward from the Upper Engadine one goes. In both Subselvan and Surselvan, it is almost entirely absent in lexical forms though isolated exceptions can appear, probably as a result of borrowing. In the varieties of Surmeir spoken in an area not far westward from the Upper Engadine, just two lexical forms showing U-prosthesis are reported by Grisch (1939): [ar’vjokt] (M) and[ar´an’tar] ‘to rinse’ < RECENTA¯RE. The latter item ‘vault’ < *REVOLVITU may be compared with the Surmeiran [rN’´ajnt] ‘brand new’ < RECE˘NTE(M) where the normal outcome for lexical items of this phonetic shape is found, namely without prosthesis.6 Proclitic personal pronouns have also been affected by prosthesis in Surmeiran, although more recently these forms have been increasingly confined to literary usage by non-prosthetic counterparts (Haiman and Beninca` 1992: 127). In Gallo-Romance, U-prosthesis has also occurred. It appears widely in northern varieties, from Touraine and Anjou in the west across to Walloon and Lorraine in the east, and, as in northern Italy and the Engadine, it has affected both lexical and grammatical items. However, in the standard language, which developed principally from varieties spoken in the central area of the langue d’oı¨l, there are few if any indications of the presence of U-prosthesis in the many formal descriptions that have been carried out, although cases of this phenomenon may be found in more informal registers used by speakers of the Paris region.
6.1.2
S T RU C T U R A L P R E C O N D I T I O N S TO U - P RO S T H E S I S
Crucial to the rise of U-prosthesis was a sound change whose effect was to weaken and delete an unstressed vowel in a word-initial syllable containing an onset. This change we may refer to as ‘syncope of pre-tonic initial vowels’ or SPIV. SPIV finds a counterpart in a fairly widespread but sporadic tendency in Late Latin for an initial unstressed vowel, especially [e], to undergo syncope in forms whose initial
6
In the Surselvan varieties of Domat, Trin, and Flem (Flims) found west of Chur, the only prosthetic form reported by Rupp (1963: }98) is once again [ar´an’ta:] ‘to rinse’ < RECENTA¯RE, indicating the often special and exceptional nature of this lexical item in RhetoRomance varieties.
U-prosthesis
197
unstressed syllables were composed of obstruent þ V þ [r]-, where V represents an unstressed vowel. For example: D(I)RE¯CTU(M)
> (Fr.) droit, (Sp.) dreito, drecho,7 (It.) dritto, (Rom.) drept ‘right’
(REW 2648) > C(I)RITARE>(Fr.) crier, (Sp., Port.) gritar, (It.) gridare ‘to shout’ (REW 6967)
QUIRITA¯RE
However, despite possible appearances to the contrary, this early change is probably not directly connected with our development which only got under way chronologically a good deal later and also operated in a much more geographically circumscribed area. The scope of SPIV has also been more extensive since the change operated on word-initial syllables whose initial unstressed vowel could have a wide range of possible flanking consonants rather than just a preceding obstruent and a following rhotic. The effect of its operation was that in many linguistic varieties a set of complex word-initial onsets previously unknown in Romance was created (again unlike the Late Latin change which merely created more cases of already existing obstruent þ rhotic onsets). The new onsets could be of rising, level, or falling sonority such as [fn-], [vz-], [zqÐ -], respectively; for instance, [’fn¡ʃtra] < FINE˘STRA ‘window’, [vziJ] ‘citizen’ < VICI¯NU (M), [zqÐ Ykr] ‘axe’ < SECU¯RE(M) in the Upper Engadinish dialect of Celerina (Walberg 1907). When SPIV operated on verbs, it created morphophonemic alternation in stems, as in the Piedmontese dialect of Viverone where we find [t¡J] vs [tJi] ‘I hold’, ‘to hold’, [bejv] vs [’bvuma] ‘I drink’, ‘we drink’ < TE˘NEO, ˘ MUS,8 such alternation being widely found in Rheto-RoT(E)NE¯RE; BIBO, B(I)B-U mance, northern Italian, and northern French varieties. Sporadically, it can happen that a new complex word-initial onset is later simplified through the deletion of the initial consonant, as in [’zJa] < [’bzJa] (< *BISO¯NE-AT) ‘it is necessary’ and [Jus’si:] < [kJus’si:] ‘to know’ (< COGNO¯SCERE) from the varieties of Cevio and Villette in the northern Lago Maggiore area (Salvioni 1886), [ʃty] < [fʃty] ‘twig’ (< FESTU¯CU(M)) in the Engadinish variety of Celerina. Similarly, reflexes of [’vJi:re] < VENI¯RE ‘to come’ with later deletion of the initial [v] are not uncommonly found, e.g. [JI:kr] in Celerinese,9 [Ji:] in the Emilian variety of Novellara (Malagoli 1910-13) and also in the Piedmontese variety of Valsesia (Spoerri 1918). However, reductive onset-restructuring of this type is not general across the sweep of varieties experiencing SPIV. Interestingly,
7 The form dreito occurs in the twelfth-century Asturian Fuero de Avile´s and the phonetically Castilian form drecho appears in various Old Spanish texts. 8 Data from Nigra (1901). 9 Cf. gnir [Ji:r] in Bifrun’s sixteenth-century translation of the New Testament into Upper Engadinish.
198
U-prosthesis
in certain Romance varieties where complex heterosyllabic onsets have arisen as a result of a sound change other than SPIV, more systematic cases of restructuring through deletion or epenthesis can be found. For example, in Gascon, complex word-initial sequences such as [hr-], [hl-] developed from earlier FR-, FL- after the regular change F- > h- had operated and these have been subject to restructuring across most dialects.10 In contrast, complex word-initial onsets created by SPIV in Romance have normally been retained with the possibility of their later being subject to U-prosthesis In view of the importance of SPIV for the operation of U-prosthesis, we may usefully sketch some of its formal characteristics. In particular, we consider the constraints to which syncope was subject in respect of the quality of the original initial unstressed vowel and the nature of the original consonantal context surrounding the vowel.11 6.1.2.1
Significance of the quality of the initial vowel
SPIV appears to have been especially favoured by the presence of the mid front vowel [e] < I, E¯, E˘, AE, whereas back vowels and the high front vowel [i] < ¯I were much less susceptible to syncope and the low vowel [a] was seldom involved at all. For example, in the Upper Engadinish variety of Celerina: TENE¯RE VITE˘LLU(M)
but
FARI¯NA ¯ RE FI¯LA
*POTE¯RE ¯ NE(M) COMMU ¯ RA¯RE DU
> > > > > > >
[tJær] ‘to hold’ [’vd¡] ‘calf ’ [fa’riJa] ‘flour’ [fIler] ‘to spin’ [pU’dær] ‘to be able’ [kU’mn] ‘common’ [dY’rer] ‘to last’ (Source of data: Walberg 1907).
The available data from all Romance varieties affected by SPIV suggest a parameter of vowel susceptibility to syncope as represented below in Figure 6.1. The 10
Thus, FRIGIDU(M) ‘cold’ gives (with deletion) [rret] or [rr] in SW dialects but (with epenthesis) [he’ret] or [he’retʃ] in most Hautes-Pyre´ne´es and W Arie`ge dialects, while FLAMMA ‘flame’ emerges as (with deletion) [’lamo] in W. Gers dialects, (epenthesis) [ha’lamo] in W. Arie`ge, (prosthesis) [eh’lamo], [eh’lam], [eh’lm] in SW dialects. Only rarely are [hr-], [hl-] found unchanged: [hr-] appears in a few dialects of W. Arie`ge, e.g. [hretʃ] ‘cold’ (pt. 790SE, Couflens); and [hl-] occurs patchily in Pyrenean dialects, e.g. pts 692, 694, 689SE, 790SE (maps ALG I, 157 freˆne; III, 1017 froid; III, 700 flamme). 11 A more detailed review of SPIV in northern Italo-Romance and Rheto-Romance appears in Mayerthaler (1982).
U-prosthesis
199
location of the high front vowel [i] in relation to the rounded vowels here is motivated by the findings of Malagoli (1910–13, 1934, 1954) for a number of Emilian varieties. a
i o u
less
FIGURE 6.1.
e more
Parameter indicating susceptibility of vowel types to SPIV
The parameter for vowel quality in SPIV correlates well with parameters that can be postulated for vowels undergoing weakening and deletion in other types of unstressed syllable. Thus, for word-final unstressed vowels, [-a] has been by far the most retentive of vowel types across Romance. Furthermore, in IberoRomance it is notable that final [-e] (< E˘ E¯ I ¯I)12 is the vowel which has been most susceptible to deletion, cf. Span. mar, ayer ‘sea, yesterday’ < MA¯RE˘, HE˘RI¯ but muro, curo, cera ‘wall, I care for, wax’ < MU¯RU˘(M), CU¯RO¯, CE¯RA˘. Cases of SPIV operating on [a] are occasionally found in certain types of Romance. However, they usually appear in varieties where syncope has remained largely restricted to lexical items whose initial unstressed syllable was of the form #obstruent þ V þ [r], that is, contexts where syncope would result in complex word-initial onsets that were already licensed and had been since Latin. Syncope therefore merely increased the incidence of these onsets. An example is provided by the Surselvan variety of the Tavetsch valley, in the far west of the Grisons: FARI¯NA PAR(I)E¯TE(M) ¯ TRU(M) TARA
> > >
[’fri:nN] ‘flour’ [pr‰jt] ‘wall’ [’tra:dNr] ‘auger, drill’
Here the new onsets are identical to etymological ones found in [frun(t)] (M), PRA¯TU(M), TRA˘BE(M) ‘forehead’, [praw] ‘meadow’, [tra:f] ‘beam’ < FRONTE (Caduff 1952). And, as the parameter predicts, SPIV in this variety also acts on other types of initial unstressed vowel in comparable contexts, as in: TERRE¯NU(M) ¯ NA CORO
12
> >
[’træjn] ‘(land) free of snow’ [’krunN] ‘crown’13
Deletion of final [-e] was only carried through and lexicalized in medieval Castilian when a single consonant preceded, hence PA˘RTE(M) > parte ‘part’. Also, the reflex of original final -¯I was restored by analogy during the later Middle Ages in first singular preterite verb forms such as hice, vine ‘I did, I came’; the phonetically regular forms hiz, vin are attested in early Castilian texts. 13 Sporadically, the context for syncope has been generalized to #obstruent þ V þ liquid-, thus affecting forms containing a lateral, e.g. [plu’´æjn] ‘chick’ < PULLI-CE¯NU(M)
200
U-prosthesis
Finally, the scope of SPIV has sometimes been affected by other sound-changes either when these modify another vowel type to [e] (feeding SPIV) or when they change original [e] to another quality (bleeding SPIV). In the former case, two factors have been important in early Romance: dissimilation or, more rarely, prefixal influence. Unstressed initial ¯I has sometimes passed to [e] when it appeared in words whose stressed vowel was ¯I. For example, FI¯NI¯RE evidently developed to [fe’ni:r(e)] in certain Emilian varieties (prov. Reggio) before becoming [fni:r] in the dialect of Valestra, [’fni:re] in that of Lizzano, though nondissimilated [fi’ni:r] is found in Novellarese (Malagoli 1934: 85); Old French and Old Castilian fenir show the first stage of the same development.14 A further case of dissimilation often led to initial unstressed [o] passing to [e] in words whose stressed vowel was also [o] in Late Latin. For example, medieval Tuscan has forms such as secorso ‘help’, serocchia ‘sister’, or sicorso, sirocchia with later raising (< SUB˘ RSU(M), SORO ¯ R-CULA) and other Italian varieties show similar developments, e.g. CU medieval Paduan serore ‘sister’, remore ‘noise’ (< SORO¯RE(M), RUMO¯RE(M)).15 In Engadinish, we find the same dissimilation in SORO¯RES > *SERORES > sruors ‘sisters’ ‘sister’ where the first vowel is stressed as against the singular form sour < SOROR and so preserves its back round quality.16 In French, a substantial number of forms show a comparable outcome, as in Old French serors ‘sisters’, esperon ‘spur’ < Frankish spo¯ron-, secort ‘(s)he helps’ < SUCCU˘RRIT, selonc ‘according to’ < SUB LONGU (M).17 However, an alternative interpretation has been advanced by Holmes (1935) for the development seen in these and other French cases where initial unstressed [o] > [‰]. It is proposed that rather than first involving a dissimilatory stage [o] [o´] > [e] [o´], the initial unstressed [o] simply weakened to [‰]. Some evidence for this view comes from variant Old French forms such as menaie (beside monaie) ‘money’ < MONE¯TA, dementres (beside domentres) ‘while’ < DUMINTERIM, ferasche (beside forasche) ‘alien’ ! ‘unsociable’ < FORA˘STICU(M), quemencier (beside comencier) ‘to begin’ < CUMINITIA¯RE where weakening has taken place despite the fact that there is no following dissimilating stressed [o]. In some of these exceptional cases, alternative explanations are available that still postulate a but forms such as [pa’liw] ‘marsh’, [ku’lu:r] ‘colour’, [ka’li:ra] ‘heat’ < PALU¯DE(M), COLO¯RE (M), CAL-U¯RA indicate that this is by no means a regular process. 14 Fouche´ (1969: 456) attributes the development of unstressed ¯I > [e] in early French in forms such as fenir and devin < DI¯VI¯NU(M) to the action of a regular change in Late Latin whereby long vowels in unstressed syllables undergo shortening. This view has not won general acceptance however. 15 For further examples from Italo-Romance, see Rohlfs (1966: }330), Mayerthaler (1982: 157–8). 16 These ‘standard’ orthographical forms in Engadinish are realized in Celerinese as (sg.) [sokr], (pl.) [sru‰rs]. 17 Cf. Fouche´ (1969: 455).
U-prosthesis
201
first stage of unstressed [o] > [e]; for example, dementres may well have experienced early prefixal influence from DE-, and ferasche may reflect the influence of FE˘RUS ‘wild, uncivilized’ (cf. REW 3432). However, there remain a number of cases such as quemencier which cannot be explained through dissimilation. A possible scenario that accounts for the data while reconciling the two opposing views is that, in conformity with the parameter in Figure 6.1 above, weakening of unstressed initial vowels operated at first with [e] (including [e] < [o] through dissimilation) before spreading sporadically to other vowel types. The preferential involvement of initial unstressed non-assimilated [o] in this later generalization (as in quemencier) may owe itself to the incomplete implementation of dissimilation of [o - o´] > [e - o´] in the langue d’oı¨l as in corone querone ‘crown’ < CORO¯NA, for example. The presence of variants with initial unstressed [o] and [‰] (the latter from dissimilated [e] < [o]) in items such as corone querone may well have helped to motivate a more general weakening of initial [o] > [‰] in forms not containing a stressed vowel [o]. Adaptation of an initial unstressed vowel to [e] may also be effected through association with an established prefix. Particularly important was RE- which enjoyed considerable productivity for lexical derivation in medieval Romance. Its influence resulted in a number of cases of remodelling, e.g. ROTU˘NDU(M) > *RETU˘NDU(M) ‘round’ > Spanish redondo, Old Tuscan retondo, Upper Engadinish arduond (Celerinese) [ar’du‰nt]; RADE˘NTE > *REDE˘NTE ‘near to’ (literally, ‘shaving’) which gives the now literary Engadinish form ardaint (Celerinese [ar’dænt]).18 In contrast to these instances of change which feed SPIV, other developments have caused bleeding. One such is the lowering of etymological [e] to [a] when it precedes a rhotic. This is found fairly widely and appears to explain nonsyncopated forms such as [sa’ræJ] ‘clear weather’, [ta’ræJ] ‘free of snow’, [ma’r¡nda] ‘snack’ (< SERE¯NU(M), TERRE¯NU(M), MERE˘NDA) in the variety of Celerina in which SPIV has otherwise been conspicuously active, although early dissimilation may also be involved. We may compare [’pri‰v‰l] ‘danger’ < PERI¯CULU(M) where original pre-tonic [e] evidently underwent no quality change as a high vowel followed and SPIV could therefore operate. In other varieties, however, there has been no lowering of [e] and SPIV has systematically occurred before rhotics too, e.g. [sraJ], [’mraNda] in Bolognese (Coco 1970).
18 Further examples of SPIV being fed through formal alignment with prefixal RE- are found in Piedmontese, e.g. in the northern dialect of Castellinaldo, HOROLOGIU (M) > *RELOGIU > [ar’lPqÐ e] ‘public clock’, ROBU¯STU(M) > *REBUSTU > [ar’byst] (Toppino 1902–5). As is apparent, U-prosthesis has operated after SPIV in both these forms, just as in the Engadinish examples.
202
U-prosthesis
6.1.2.2
Surrounding consonantal context
Weakening and deletion of an initial unstressed vowel have taken place most often in contexts where the resulting word-initial onset would consist of two consonants, i.e. in words with the original shape # CVCV-. However, various constraints existed which governed the incidence of SPIV and these determined which of the complex onset sequences potentially created by SPIV were licensed in the individual Romance varieties concerned. The constraints related in particular to the inherent sonority of the flanking consonants and to their place and manner of articulation. Thus, in the dialect of Valsesia (N. Piedmont) despite the wide application of SPIV there appears to have been a constraint blocking syncope if a complex initial onset [qÐ ] þ consonant would result; hence [qÐ a’nu] ‘knee’, [qÐ a’le] ‘to freeze’ < GENU˘CULU(M), GELA¯RE (Spoerri 1918). In other dialects, the constraints blocking syncope may be weak to the extent that even sequences of two identical consonants may be licensed, giving a strong or geminate onset, and thereby overriding the OCP (obligatory contour principle) which operates widely in phonology and prevents identical adjacent segments or features from arising through deletion within a morpheme.19 Cases of identical consonants coming together after syncope and being retained as a geminate are found notably in varieties from Emilia-Romagna, a region where SPIV has been particularly intense: e.g. Piacentino [zzeæı˜] < SEX-¯INU(M) ‘a monetary unit’ (Gorra 1890: 142), Novellarese (W. Emilia) [’ssa:Nta] ‘sixty’, [bbu] ‘drunk (p.pt.)’ < SEXAGINTA, ¯ TU(M) (Malagoli 1910–13: 158). BIB-U In contexts where deletion would have created a word-initial onset containing more than two consonants, stronger constraints have operated. Usually, new onsets of no more than three consonants have been permitted. However, in general such onsets have been licensed only if they contain a sibilant consonant which was followed by an obstruent, e.g. [’sptʃ ¡:r] ‘to mirror’ (= St.It. specchiare) < SPEC(U)LA¯RE, in the Novellarese (Malagoli 1910-13: 109), [msti:r] ‘trade, profession’ < MINISTE˘RIU(M) in Bolognese (Mainoldi 1967). An exception to this arrangement, found in some but not all varieties undergoing SPIV, concerns forms beginning with a sonorant. Most frequently affected by far were forms beginning with RE-, a prefix whose widespread use in word formation we have already noted. The special significance of the word-initial sequence RE- for U-prosthesis is explored more fully further below (6.1.4). Examples of the creation of three consonant onsets may be cited from two varieties, those of Celerina and Bologna, in both of which SPIV has been highly active as a process. The OCP was first enunciated by William Leben in his 1973 doctoral thesis to account for a widely found constraint against adjacent syllables with identical tones within a morpheme. It has subsequently been extended to cover prohibitions on identical adjacent segments or segmental features within a morpheme. 19
U-prosthesis
VESTI¯RE ¯ CU(M) FESTU SEPT(I)MA¯NA MI(NI)STE˘RIUM DIS-GRA¯TIA SPERA˘NTIA
Celerina [fʃtikr]>[ʃtikr] [fʃty]>[ʃty] —20 — — [’ʃpræntsa]
Bologna [vti:r] (with later [s] deletion) — [’stm¡:na] [msti:r] [’dz grA:tsja]
[var’gw‰Ja] [tam’p¡sta] [saN’glu‰t] [kra’vu‰sta]21 —
[var’gaJJa] [tim’p¡:sta] [sin’qÐ at] [kar’p¡:r] [sar’paNt]
203
‘to dress’ ‘twig’ ‘week’ ‘profession’ ‘disgrace’ ‘hope’
but ˘ NDIA VER(E)CU ˘ TEMPESTA(S) ˘ TTIU(M) SINGLU CREPA¯RE SERPE˘NTE(M)
‘shame’ ‘storm’ ‘sob’ ‘to burst’ ‘snake’
SPIV with later U-prosthesis R(E)-TENE¯RE /VENI¯RE R(E)-COMMENDA¯RE ¯ SCERE R(E)-COGNO
[art’Jær] [arkUman’der] —22
[ar’vJi:r] [arkman’d¡:r] [ark’Josser]
‘to restrain, return’ ‘to recommend’ ‘to recognize’
It appears that constraints blocking the formation of complex onsets of three or more consonants were most subject to violation in varieties from Emilia-Romagna, so that word-initial onset sequences of considerable complexity became permissible during the course of the medieval period. Thus, in the one surviving version of the sixteenth-century poem Pulon Matt or ‘Mad Paul’ written in the Romagnolo dialect of Cesena, apparent cases of complex onsets abound (Bagli 1887). Some occur after a preceding vowel-final form, thereby allowing ready syllabification, as in u Rbgon ‘the Rubicon’ (III, 19, 8) and bona mdsina ‘good medicine’ (II, 20, 8) which can be syllabified as urb|gon, bo|nam|dsi|na. However, certain cases suggest that onsets of considerable complexity came to be permissible in this variety by the sixteenth century, though more usually perhaps in more informal registers: u bon Flpon Pulon rspos 20
‘the good Filippone’ ‘Paul replied’
I,18,5 I,59,3
In Rheto-Romance, HEBDOMAS or the variant *HEBDOMINA usually provide the forms designating ‘week’; Celerina dialect has the reflex [’¡gvna]. 21 Meaning ‘crust, scab’. The etymon CREPA¯RE seems to have been crossed with CRU˘STA to yield this reflex. The verb CREPA¯RE gives [kra’per] where the [p] rather than [v] suggests possible influence from krap ‘rock’, according to Walberg (1907: 84). 22 No reflex is cited for Celerinese by Walberg (1907). However, the form arcugniouscher appears in the sixteenth-century translation of the New Testament in Upper Engadinish by Jakob Bifrun.
204
U-prosthesis
tant dsdgnos un bssbij tant cstun ha dstne`
‘so disdainful (f.pl.)’ ‘a whisper’ ‘so many questions’ ‘has destined’
II,5,3 III,61,1 IV,4,5 IV,26,6
The presence of such complex word-initial onsets is matched by the occurrence in the poem of words with word-final codas of comparable complexity, as in prepausal destr ‘right’ (I, 15, 7), u mdesm ‘the same’ (I, 36, 2). To conclude, SPIV operated in a variable, parametrized way and came to create word-initial onsets of considerable diversity. Varieties of Piedmontese, Emilian, and Romagnolo in Italy, and eastern varieties of the Rheto-Romance spoken in the Grisons, especially Upper Engadinish, were particularly susceptible to this type of syncope. In turn, these are the varieties in which U-prosthesis has been most in evidence, reflecting the direct correlation between SPIV and U-prosthesis.
6.1.3
C H RO N O LO G Y
Dating the rise of U-prosthesis is problematic. In Rheto-Romance, for example, it is certainly attested in the first substantial vernacular texts from the Grisons but these only go back to the sixteenth century. For example, in Bifrun’s translation of the New Testament published in 1560 we find: (tu¨ t’) alguordas ‘you (sg.) , (es sto) alvo ‘(it) rose up’ < (EST STATUM) LEVA¯TU remember’ < (TU TE) RECORDAS (M); and in the sixteenth- or seventeenth-century poem Susanna has (t’) almeintast < (TE) LAME˘NTAS ‘you (sg.) complain’, presumably via the stage (TE) *LEME˘NTAS which developed as a result of assimilation.23 For the prosthetizing northern varieties of Gallo-Romance, there is available philological evidence from the Middle Ages but U-prosthesis does not appear to have been operative before early modern times. Old Picard texts show no sign of the change but it is clearly indicated in Middle Picard documents. For instance, in Des Fill’ qu’al n’ont point grament d’honte, a text from the Valenciennes region dating from the end of the sixteenth century, forms appear such as erligieuse, s’ertourno, which correspond to standard French religieuse, se retournait, and ercran ‘tired’ which goes back to old Picard recreant (Flutre 1977: 38).
23
A parallel form [le’me:Nta] ‘lament’ (cf. also [leN’te:rna] ‘lantern’) is reported for the rural speech of Novellara (Emilia) by Malagoli (1910–13: 108). Different dates of composition have been proposed for the Susanna, a verse text originating from Bergu¨n. The earliest manuscript dates from the early seventeenth century, and Ulrich (1885–6) and Lutta (1923: }16) take this as the century of composition, whereas Decurtins ([1900]1983–6: vol. 5) attributes the text to the sixteenth century.
U-prosthesis
205
In northern Italy, there is also substantial medieval textual evidence but it likewise offers little direct evidence of U-prosthesis until the early modern period. For instance, there are no indications of its presence in the Piedmontese sermons written in vernacular spelling in the twelfth century, although these clearly show I-prosthesis (cf. 4.4.5). In the sixteenth-century poem Pulon Matt from Cesena in Romagna, however, there are prosthetic forms such as (l’era dsorta) uvsin ‘(he was by chance) near’ < VICI¯NU(M), (dfat) armaner ‘(in fact) to stay’ < REMANE¯RE.24 The philological data directly attesting the U-prosthesis, therefore, suggest that the development only got under way by the end of the Middle Ages at the earliest. Another possible indicator for dating comes from relative chronology. As U-prosthesis presupposes the prior operation of SPIV, by establishing the approximate chronology of the latter we can at least provide a terminus a quo for U-prosthesis. Unfortunately, the philological data necessary for dating SPIV are once again of limited usefulness. For, on the one hand, there is a lack of extant documentation from the Middle Ages for many of the varieties concerned, and on the other hand, where medieval texts are available they often prove to be uninformative as a result of influence from Latin or other Romance varieties not affected by SPIV. For instance, in Bolognese where SPIV has been a very productive process, there are three private letters dating from the first half of the fourteenth century between members of the Bentivoglio family from Bologna (Stella 1969). These contain a number of lexical items which might be expected to provide us with revealing insights. However, the forms which occur
The forms appear in, respectively, IV, 12, 2; III, 66, 6. The lengthening of [v] is explained by the presence of a short preceding stressed vowel. Bolognese developed a stressed vowel pattern of [V:] or [VC] creating greater syllable isochrony. In coda-less syllables with a short vowel, lengthening of a following consonant has occurred to form a coda (Coco 1970). 26 A Bolognese medical text dating from the mid-fourteenth century shows similar Tuscan influence (Longobardi 1994). As the editor concludes, ‘E` dunque bolognese il nostro testo, con influenza del toscano letterario.’ 24 25
206
U-prosthesis
Other evidence can be invoked to shed light on the relative chronology of SPIV, however. Reflexes of NEPO¯TE(M) such as [an’vAwd] in Bolognese and [an’vo:] in Novellarese indicate that SPIV postdates the regular lenition of intervocalic voiceless obstruents, since etymological -P- must have remained intervocalic for long enough to enable voicing to occur.27 All other Romance varieties subject to SPIV likewise show the effect of lenition in syncopated reflexes of Latin forms similar in structure to NEPO¯TE(M). For example, in Engadinish, there are forms such as [al’gwa:r] or [al’ge:r] ‘to melt’ < LIQUA¯RE which likewise show that syncope preceded the voicing caused by lenition.28 As lenition of voiceless plosives was a productive process that doubtless extended into the early Middle Ages in northern Italy and Alpine regions, SPIV can safely be assumed to be a medieval development. However, there is further evidence that SPIV may only have operated at a fairly advanced stage in the Middle Ages, in some areas at least. In the varieties of Valsesia (N Piedmont) and Voghera (S Lombardy, some 60 km west of Piacenza) where SPIV has occurred widely, there are forms such as [lja:m] ‘dung’ < LAETA¯MEN.29 These indicate that before SPIV began to operate, lenition had not only voiced the original intervocalic -T- but had gone so far as to delete it, the probable path being [le’ta:me] > [le’da:me] > [le’a:m(e)] > [lja:m]. Had SPIV occurred before the final stage, a sequence [(a)ld-] would have developed word-initially just as has happened elsewhere in northern Italy, e.g. Bolognese [al’dA:m]. Now, in all probability the deletion of [d] (< -T-) was not carried through before the end of the first millennium. Even in the very rapidly evolving Romance varieties of northern France, the evidence suggests that this change was only implemented by about the eleventh century (Fouche´ 1966: 600). We may therefore tentatively conclude that SPIV first began to operate, at least in northern Piedmontese, at some stage near the end of the first millennium at the earliest. In other varieties experiencing SPIV, however, its chronology may of course have been somewhat different. A final possible indicator may be briefly noted although it is of arguable usefulness. It builds on the relation of SPIV to the palatalization of preconsonantal [s] and [z] which occurred as a regular development across all
27 Lenition is a general development in western Romance, i.e. northern Italo-Romance, Rheto-Romance, Sardinian, Gallo-Romance, and Ibero-Romance. It affected all intervocalic obstruents and led to changes such as -B-> [v], -P- > [b] > [v], -PP- > [p]. 28 The first form is found in the standardized form of Engadinish as cited in the dictionaries of Peer (1962) and Bezzola and To¨njachen (1976), the second form appears in the Upper Engadinish variety of Celerina (Walberg 1907). It will be recalled that RhetoRomance varieties derive their word for ‘nephew’ from the nominative NE˘POS, giving [neiv], [n¡kf], etc. Such forms were not of course affected by SPIV. 29 Data from Spoerri (1918) and Maragliano (1976).
U-prosthesis
207
Rheto-Romance, giving forms such as [’ʃtæla] ‘star’ < STE¯LLA and [’ʃtc¸ela] ‘ladder’ < SCA¯LA in Upper Engadinish (Celerina). It also operated, but rather less consistently, in a number of far northern Italo-Romance varieties, notably those of northern Piedmont and Lombardy, Trento and the canton of Ticino, and parts of Romagna (cf. Rohlfs 1966: }188).30 Now, in those varieties that underwent this palatalization and also experienced SPIV, it is notable that any pre-consonantal sibilants created by SPIV have failed to undergo palatalization; for example, [stIL] ‘thin’ < S(U)BTI¯LE(M) and [ster] ‘to dry’ < S(I)CCA¯RE in Upper Engadinish. Preconsonantal S-palatalization must therefore have ceased to be a productive process before SPIV arose. Given that the palatalization of pre-consonantal sibilants remained productive long enough into the medieval period to affect borrowings such as [ʃprUm] ‘spur’ < early Germanic spo¯ro(n)-, it would appear that SPIVonly started to apply at some stage well into the Middle Ages.31 A crucial consideration, however, is the date at which pre-consonantal [s] palatalized in southern Germanic itself. For if it predated the borrowing of items like spo¯ro(n)(> sp(e)ron-) into early Romance, we would clearly not be able to use evidence from such items to draw any conclusions on the chronology of pre-consonantal sibilant palatalization in Rheto-Romance. Unfortunately, there is some disagreement amongst Germanists over the chronology of pre-consonantal [s] palatalization. The first stage of the development is generally agreed to involve the palatalization of initial [sk-] > [ʃ-] (presumably via a stage [ʃc¸-]), but the dating of this and later palatalizations of sibilants in other pre-consonantal contexts is controversial. For some, the first stage dates only from late Old High German at the earliest, i.e. the tenth or early eleventh centuries, and palatalization in other contexts, e.g. in sp-, st-, sn-, sm-, is assumed to have occurred by the end of the thirteenth century. The other interpretation, which suggests a much earlier dating, is based largely on the consistent distinction made in Old High German and early Middle High German texts of the two graphies <s> and
30
E.g. the reflexes of STE¯LLA ‘star’: [’ʃt¡la] Ornavasso (NE Piedmont), [’ʃt¡:la] Germasino (N Lombardy), [’ʃt¡la] Predazzo (N Veneto) (AIS 2, 362; pts. 117, 222, 323). 31 REW 8130a. Cf. It. sp(e)rone, Old French esperon > ´eperon, Occ. esporo´.
208
U-prosthesis
definitely accomplished by the thirteenth century, may therefore have involved only a small increment in palatality. If this was true, early Germanic loans into Rheto-Romance such as spo¯ro(n)- might already have contained a partly palatal initial sibilant. The use of such forms in attempts to determine the chronology of SPIV in Rheto-Romance therefore would seem to be inappropriate.32 We may therefore rely on just the data emerging from forms affected by western lenition, e.g. Bolognese amvod, to conclude that SPIV probably represents a change that was initiated close to the end of the first millennium at the earliest. U-prosthesis would then have got underway at some later stage.
6.1.4
AC T UA L I Z AT I O N
After SPIV had operated, the pattern of implementation of U-prosthesis in the Romance varieties concerned has been complex with a good deal of variation from region to region. Only an outline of the broad characteristics of actualization will therefore be possible. Two factors in particular have guided the process, and these have interacted in sometimes delicate ways. They are: (i) the nature of the prosodic domain in which the word appears, and (ii) the internal structure of the complex word-initial onset that was created by SPIV. We may consider these factors in turn, focusing attention first on developments in lexical items. Prosthetizing contexts which involve monosyllabic proclitic forms, e.g. M(E) VIDET ‘(s)he sees me’, are examined separately below. 6.1.4.1 Nature of the prosodic domain
U-prosthesis in lexical items appears to have been triggered particularly in contexts where there was a preceding, closely syntactically linked word which ended in a consonant. A preceding consonant-final determiner or clitic provided the most favoured prosthetizing context. U-prosthesis has also occurred in postpausal contexts, though this has happened less commonly with heterosyllabic onsets beginning with an obstruent; in contrast, it is usual with onsets beginning with a sonorant, especially a liquid. Generally, prosthesis seems not to have occurred in post-vocalic contexts within close-knit phrases, though cases can be
32 Proponents of the first interpretation, which assumes a first stage dated tenth to eleventh century, include Bach (1965: 175), Tschirch (1969: 19) and Young and Gloning (2004: 110). The classic work setting out the interpretation in favour of an earlier dating is Joos (1952). My thanks go to Martin Durrell for his invaluable guidance in this complex chapter of German phonological history.
U-prosthesis
209
found where this has taken place with sonorant-initial onsets, a development which led to lexicalization of the prosthetic vowel. The following basic stages seem to have occurred in a pattern of implementation that resembles what happened with I-prosthesis and, less certainly, with A-prosthesis (4.1.4, 5.3.3): post-consonantal ! post-pausal ! post-vocalic The extent to which individual Romance varieties have progressed along this path of actualization varies a good deal. In ‘common Piedmontese’ based on Turinese, for instance, certain types of heterosyllabic word-initial onset (namely, obstruent-initial) show U-prosthesis in post-consonantal position only (Clivio 1971, 2002: 160–1). Indeed, if there is a hesitation or partial pause between the preceding consonant and the onset, a prosthetic vowel may fail to appear even in that context, e.g. (post-vocalic) sensa dne´ ‘without money’, (phrase-internal, post-pausal) tant . . . dne´ ‘much . . . money’ but (post-consonantal, without pause) tant ¨edne´ ‘much money’, where e¨ indicates a schwa. In the Piedmontese variety of Viverone, the occurrence of prosthesis is even more nuanced. Nigra (1901: 253) reports that when a consonant-final determiner or clitic precedes, there is regular prosthesis which yields a vowel with the quality [N]. But when some other consonant-final form precedes within a noun phrase, the prosthetic vowel is weak and scarcely perceptible, as in l’a-vzin [lNv’ziN] ‘the neighbour’ but sett aksu-i [’set Nk’sui] ‘seven buns’. Complex onsets whose initial consonant was a sonorant have usually undergone U-prosthesis not only in post-consonantal contexts but also post-pausally in Piedmontese. As a result, citation forms for words which had contained a sonorant-initial onset sequence generally have a prosthetic vowel. Post-vocalically, however, prosthesis has been carried through only incompletely. A prosthetic vowel may be absent when a vowel-final proclitic form precedes, particularly when the initial consonant of the complex word-initial sequence created by SPIV was not a liquid. Whether this reflects elision of an earlier prosthetic vowel or the failure of prosthesis to occur at all is unclear. As an example of this pattern, the dialect of Monferrato (SE of Turin) shows forms like in anvud ‘a nephew’ but me¨ nvud ‘my nephew’ < N(E)PO¯TE(M). However, onsets with an initial liquid appear to maintain the prosthetic vowel in this dialect even after a vowel-final proclitic: aj hoˆ arcunsı` ‘I have recognized’ < R(E)-COGNOSC-¯ITU(M) (Nebbia 2001). For the other major syncopating zone of northern Italy, Emilia-Romagna and adjacent border areas, Malagoli (1910-13: 111) reports the regular presence of a prosthetic vowel in Novellarese (Emilia) before complex onsets beginning with the sonorants [r, l, n] but only in post-consonantal and post-pausal contexts, since elision (or non-implementation of prosthesis) is apparently found in postvocalic contexts. A similar pattern is described for the variety of Pontremoli in the Lunigiana, located in the northern transitional area where Tuscany abuts onto Emilia and Liguria (Maffei Bellucci 1977: 46–7). Further to the east, in the variety
210
U-prosthesis
spoken in Piacenza at the end of the nineteenth century, U-prosthesis was reported to have been generalized to all contexts and fully lexicalized in forms beginning with liquid þ consonant and less commonly with nasal þ obstruent, whereas other types of word-initial heterosyllabic onset appear not to be subject to prosthesis, except post-consonantal s impura sequences (Gorra 1890: 143, 153). In the Rheto-Romance of the Grisons, those varieties which experienced SPIVand subsequent U-prosthesis have lexicalized the prosthetic vowel [a] in words whose onsets began with a liquid and, sometimes but not often, with a nasal also. Thus, Lutta (1923: }}121, 123) cites the following forms for the dialect of Bergu¨n, where the prosthetic vowel is maintained even when a vowel-final auxiliary precedes: post-pausal [al’de:r]
post-vocalic ‘to manure’ [iL ’pro: ¡ al’do:] [ar’tʃ ¡gv‰r] ‘to receive’ [¡l P ar’tʃi:] [amna’tʃe:r] ‘to threaten’ [¡l P am na’tʃo:]
‘the field is < LAET-A¯RE, -A¯TU(M) manured’ ‘he has received’ < RECIPERE, *-¯ITU(M) ‘he has < MINACIA¯RE, -A¯TU(M) threatened’
Finally, in those varieties of northern Gallo-Romance where there was categorical deletion of schwa in unstressed syllables, U-prosthesis has come to operate where syllabification of a heterosyllabic consonant sequence would otherwise be impossible. Thus, U-prosthesis has not applied in post-vocalic contexts since the opening consonant of a heterosyllabic onset can be resyllabified as the coda of the preceding syllable, but in post-consonantal contexts this is not possible and prosthesis therefore offers a solution. Post-pausally, although prosthesis would also be expected, the data prove to be a little less clear-cut and the appearance of prosthetic vowels may not be categorical. The following examples are cited from the Picard variety of Mesnil-Martinsart (Flutre 1955) and they indicate that prosthesis here has occurred irrespective of the quality of the first consonant of a heterosyllabic word-initial onset. post-pausal 33 [ mne s vAk] ‘to lead one’s cow’ [ertire] ‘remove ! (imp.pl.)’ e
post-consonantal [i f ll emne] ‘it is necessary to lead it’ [i n ersa˜n pwq sq p¡r] ‘he does not look like his father’
33 The data in Flutre (1955) are a little unclear in places over the incidence of post-pausal U-prosthesis. Forms are cited in post-pausal contexts both with and without a prosthetic vowel indicated. For instance, beside the prosthetic example [emne s vAk . . . ] which we cite from p. 36, there is non-prosthetic [mne qn vAk . . . ] on p. 83, both post-pausal, and in the glossary the citation (hence post-pausal) form is given as mne´ (non-prosthetic). However, it is explicitly stated: ‘Quand un mot commence par plusieurs consonnes et qu’il est a`
U-prosthesis [elve s mq] ‘to raise one’s hand’ [edmq] ‘tomorrow’
211
[klik el pPrt] ‘try the door knob’ (to see if someone’s in) [ty n edma˜n pwq] ‘you do not ask’
post-vocalic [o mno˜] ‘we lead’ [i rsa˜n sq p¡r] ‘he looks like his father’ [(´ em ) sy lv¡] ‘I got up = (I) raised (myself)’ [sa˜ mq dma˜de py lo˜] ‘without asking me for more of it’ Although I-prosthesis and later pre-consonantal [s] deletion had eliminated original s impura onsets in this and the other Gallo-Romance varieties concerned, new onsets with this structure reappeared in later loans and as a result of schwa deletion, and these too have been subject to U-prosthesis since they have continued to be interpreted as being heterosyllabic. For example, once again in the Mesnil-Martinsart variety, there are forms such as [ʃe stasjo˜] ‘the stations’ but [al proʃqn estasjo˜] ‘at the next station’. Also systematically affected by the same rule of prosthesis have been proclitic pronouns and other monosyllabic grammatical forms which had formerly contained the vowel schwa. These forms have come to display the same morphophonemic alternation, e.g. with the feminine singular definite article [klik el pPrt] ‘try the door knob (to see if someone’s in)’ but [dq l kur] ‘in the courtyard’. In a more detailed coverage of prosthesis in another Picard dialect, that of Vimeu, Auger (2001) indicates a comparable pattern, leading in a parallel way to the same pattern of regular morphophonemic alternation. In addition, however, it is noted that the incidence of a prosthetic vowel decreases the higher up on the prosodic hierarchy the potentially prosthetizing context is found. Thus, if an unlicensed onset is found within the prosodic word, i.e. a lexical word plus affixes and associated clitics, the prosthetic vowel [e] appears categorically. Within a phonological phrase, prosthesis is almost categorical but occasional exceptions occur, whilst the incidence of prosthesis in intonational-phraseinitial and especially utterance-initial position drops considerably. In the corpus of data examined by Auger, prosthesis in utterance-initial position drops to less
l’initiale de la phrase, ou que le mot pre´ce´dent ne finit pas par une voyelle sur laquelle ces consonnes puissent appuyer, il y a de´veloppement d’un e prosthe´tique ou e´penthe´tique, plus ou moins nettement articule´’ (pp. 35–6). It is added (p. 36) that, ‘C’est en particulier le cas lorsqu’un mot commence par le pre´fixe r- (franc¸. re-) suivi de consonne.’ Yet, the citation form given in the glossary for even these prefixal items is non-prosthetic, e.g. for the reflex of RE-SIMILA¯RE ‘to resemble’ we find rsa˜ne´ (p. 226) rather than [ersa˜ne]. Some doubt therefore hovers over the incidence of post-pausal prosthetic vowels in this dialect.
212
U-prosthesis
than one half of the possible cases where it might be expected to have occurred, echoing in part the evidence from Flutre’s description (cf. n. 33).34 This finding suggests that in at least some forms of medieval Romance a further path of actualization relating to prosodic domains may have operated for U-prosthesis: clitic phrase ! phonological phrase ! intonational phrase ! utterance Some support for the existence of this path of actualization can be found elsewhere in Romance. For instance, in certain Romance varieties (e.g. Catalan) U-prosthesis has only occurred within the clitic phrase, indicating that it is the primary locus for prosthesis on the prosodic hierarchy (see 6.1.4.3). Also, Clivio (1971: 338) cites Piedmontese forms showing prosthesis to be just optional rather than obligatory in post-consonantal forms occurring in prosodic domains higher than the clitic phrase. 6.1.4.2 Internal structure of the word-initial onset
As some of the data in the previous subsection have indicated, the actualization of U-prosthesis has also been directly conditioned by the sonority profile of the complex word-initial onsets created by SPIV. In broad terms and directly in line with the SSG (cf. 1.7), onsets of rising sonority have generally been less subject to prosthesis, whereas onsets of level and, particularly, falling sonority have been more subject. In onsets of a given sonority slope, the sonority distance between the successive consonants will also play a role. The greater the sonority distance in rising sonority onsets, the less the incidence is of U-prosthesis; the greater the distance is in falling sonority onsets, the greater the incidence of U-prosthesis.35 Thus, we have at one extreme those sequences of rising sonority composed of stop þ glide or liquid, where the sonority distance is large. U-prosthesis is not triggered by such onsets except in very special cases. And at the other extreme, there are onsets of falling sonority composed of glide or
34
The corpus of Auger is unfortunately entirely based on fictional dialogue appearing in novels by modern Picard writers. Whilst such data may replicate linguistic usage fairly accurately, some doubt remains as to just how authentically they represent natural speech. Flutre’s data were based on usage in live speech. Nonetheless, both sources point to the less than categorical nature of prosthesis in utterance-initial position. 35 This is in conformity with generally preferred syllable structure patterns as presented by Vennemann (1988). For complex onsets, the preference is for fewer rather than more constituent consonants and for those consonants to be arranged with maximally rising sonority. Thus, strategies for breaking up complex onsets may be expected to be more systematically used, the more such onsets deviate from the preferred structure.
U-prosthesis
213
liquid þ stop, where there is a large sonority distance: these have systematically triggered U-prosthesis. The data can be represented in broad terms36 using a parameter on which prosthesis in individual dialects may be variably located (Figure 6.3). stop+gl./liq. obs.+son. rising sonority
less
obs.+obs. son.+son. level sonority
son.+obs. gl./liq.+stop falling sonority
more (gl. = glide, liq. = liquid, obs. = obstruent, son. = sonorant, nas. = nasal)
FIGURE 6.3.
Relative susceptibility of sample context types to U-prosthesis
We may briefly review the relevant Romance data that underpin this parameter before going on to consider the probable path of actualization that it indicates. Maximally rising-sonority onsets containing stop þ liquid usually fail to undergo U-prosthesis in any variety, as in [pla:] ‘to peel’, [fra:] ‘to shoe a horse’ < PILA¯RE, FERRA¯RE in Valsesiano (N Piedmontese). Such onsets, it will be recalled, had existed in Latin, and phonological theory generally views obstruent þ liquid sequences as forming tautosyllabic or ‘true’ onsets. Elimination of such onsets would not therefore be expected. The very rare exceptions concern the coronal sequences [tl-], [dl-] which were both impermissible in Latin. Reflexes of ¯ RE(M) ‘pain’, for example, reportedly show U-prostheTELA¯RIU(M) ‘loom’ and DOLO sis in some Piedmontese varieties.37 In other types of obstruent þ sonorant onset sequence where the sonority rise is relatively smaller (especially obstruent þ nasal) or in onsets where there is level sonority, the susceptibility to prosthesis has been variable. It has been greater in Piedmontese, for instance, than in Emilian or Romagnolo varieties. Thus, onset sequences such as [fn-], [dn-] (obstruent þ nasal), and [vz-] (obstruent þ obstruent with level sonority) generally show a prosthetic vowel in appropriate phonological contexts in Piedmontese:
36 A much finer-grained schema would be required for a detailed and accurate analysis. Thus, the present simplified parameter fails to distinguish onsets such as [ml-] and [rm-], both falling here under ‘son. þ son.’. However, the former is often found without prosthesis in northern Italian dialects, e.g. Bolognese [mlAN] ‘melon’ < MELO¯NE(M), whereas the latter has regularly undergone prosthesis. 37 For Viveronese, Nigra (1901) reports [tl-] as providing a prosthetizing context. In his detailed inventory of onsets triggering U-prosthesis in Piedmontese, Telmon (1975) includes [dl-] and cites the form [ad’lo:r] ‘pain’ < DOLO¯RE(M) for the dialect of the Andorno valley. However, no mention is made here of [tl-].
214
U-prosthesis
with prosthesis (post-consonantal) > [dez ‰f’noj] ‘ten bulbs of fennel’ D(E)NA¯RIU(M) > [taNt ‰d’ne] ‘much money’ VICI¯NU(M) > [dez ‰v’ziN] ‘ten neighbours’ ˘ CULU(M) F(E)NU
without prosthesis [tre f’noj] ‘three bulbs of fennel’ [’seNsa d’ne] ‘without money’ [tre v’ziN] ‘three neighbours’
However, in Emilian-Romagnolo it appears that comparable prosthesis has not regularly occurred.38 There are few onsets of rising or level sonority, the initial element of which was a sonorant. The only type that has passed into Romance appears to have contained [m] as the first element and for these the results are variable, just as they are in onsets of falling sonority beginning with [m]. Prosthesis has occurred but in some varieties it has been limited at most to post-consonantal contexts, as in ‘common’ Piedmontese based on Turinese [mluN] ‘melon’< MELO¯NE(M), [mni’ze] ‘garbageman’ < MINU¯TIA þ -A¯RIU(M) (Clivio 1971) and Piacentino (Emilian) [mloN] ‘melon’, [mny:d] ‘tiny’ < MINU¯TU(M) (Bearesi 1982). In other varieties of Piedmontese and Emilian, however, it has been extended to post-pausal contexts (hence citation form) as well, e.g. Castellinaldese [am’ruN] ‘melon’ (Toppino 1902-5), Monferrino [am’n¡stra] ‘soup’ < MENE˘STRA, [am’ne] ‘to lead’ < MINA¯RE (Nebbia 2001), and sporadically and variably in Bolognese [(a)m(b) ra˚uz] ‘amorous’ < AMORO¯SU(M) but [’mraNda] < MERE˘NDA (Mainoldi 1967). Finally, in Upper Engadinish, U-prosthesis appears to have been lexicalized in some ‘threat’ < MINA¯CEA39 as against [mnYkt] forms only, e.g. Celerinese [Im’natʃa] ‘tiny’ < MINU¯TU(M), [mne:r] ‘to lead’ < MINA¯RE ‘to lead’. Falling sonority onsets have been much more susceptible to prosthesis across all varieties where SPIV has operated. Particularly affected have been onset sequences beginning with a sonorant. U-prosthesis has always occurred at least The Piedmontese data are drawn from Clivio (1971: 336, 2002: 161). For EmilianRomagnolo, Loporcaro (1998) indicates that for the Emilian variety of Grizzana Morandi, which lies 40 km south of Bologna, onsets of rising sonority or level sonority form true onsets, at least within clitic phrases, so that they do not give rise to prosthesis. Thus, [at ’kJPsen] ‘they (f.) know you’, [at pke:ven] ‘they (f.) were pecking you’, rather than nonprosthetic **[al ti k’JPsen] [al ti p’ke:ven], as against [al ti v’de:ven] ‘they (f.) saw you’ with prosthesis before an obstruent-initial onset with falling sonority. 39 This form may owe its initial vowel to morphological recutting rather than Uprosthesis, ILLA MINA¯CEA > ILL’ AMINACEA. However, as initial [a-] usually gives [a] in Upper Engadinish, e.g. APRI¯LE(M) > [a’vriL] ‘April’, AMI¯CU(M) > [a’mix] ‘friend’, the emergence of a high front vowel in [Im’natʃa] would need explanation. It is significant that the regular outcome of what is certainly a prosthetic vowel in other nasal-initial onset sequences is a high vowel, e.g. [YN’gu‰ta] ‘nothing’ < NE(C)-GU˘TTA (the rounded quality of the initial prosthetic vowel of this form is presumably due to labiovelar influence from the following velar consonants and rounded stressed vowel). It may therefore be that either (i) prosthesis 38
U-prosthesis
215
in post-consonantal contexts, but where the initial sonorant is a liquid, prosthesis is invariably found in both post-consonantal and post-pausal contexts (e.g. in Novellarese), and it may have been generalized in all contexts and become fully lexicalized (e.g. in Rheto-Romance varieties spoken in the Upper Engadine). For instance, in the Engadinish dialect of Celerina, the prosthetic vowel is now ‘to receive’ < RECIPERE and [al’ge:r] ‘to lexicalized in forms such as [ar’tʃæv‰r] ¯ melt’ < LIQUARE. With sonorant-initial onsets of falling sonority which begin with a nasal, there has been widespread variation depending on the quality of the nasal. Where the initial nasal is coronal, U-prosthesis has typically occurred in postconsonantal and post-pausal contexts as in Novellarese (Emilian) [an’v¡:r] ‘to snow’ < NIV-A¯RE and Monferrino (Piedmontese) [an’vu:d] ‘nephew’ < NEPO¯TE(M), and sometimes it has been generalized and fully lexicalized, as in Upper Engadinish [YN’gu‰ta] ‘nothing’ < NE(C)-GU˘TTA, [YJ’djyn] ‘nobody’ < NEC-U¯NU(M). However, prosthesis has been much less systematic when the nasal was bilabial [m], in conformity with the pattern already seen for such onsets when they are of level or rising sonority.40 The incidence of U-prosthesis before falling sonority onsets beginning with an obstruent has been more variable across different varieties. In Upper Engadinish, the evidence suggests that it has never occurred. For example, it seems that syncopated forms such [vd¡] ‘calf’ < VITE˘LLU(M) have always been non-prosthetic,41 and following the abandonment of I-prosthesis in Rheto-Romance, original s impura onsets that (re)appeared in forms such as STRA¯ME(N) ‘straw’, SPI¯NA ‘thorn’ have behaved as other obstruent-initial falling-sonority onsets like [vd¡] and similarly failed to undergo U-prosthesis, hence [ʃtram], [’ʃpiJa]. In northern Italian dialects, U-prosthesis may occur with obstruent-initial onsets of falling sonority in certain phonological contexts. For instance, this reportedly occurs in
did operate on MINA¯CEA although it curiously failed to do so in [mnYkt] ‘tiny’ (the view of Lutta 1923: }126), or (ii) a word-initial vowel [a] was first created in this word through morphological recutting and later the vowel was aligned with that of other words which began with unstressed vowel þ coda nasal, notably forms containing the prefix [Im-, In-] < ¯ NERE. Similar conclusions may be drawn for forms IN- such as [Im’pu‰nd‰r] ‘to use’ < IMPO like [In’dzYgra] ‘measure’ < ME(N)SU¯RA. 40 Thus, in the Emilian dialect of Novellara, where U-prosthesis consistently occurs with heterosyllabic onset sequences beginning with a liquid or [n], word-onset sequences beginning with [m] are only sporadically affected by prosthesis whether they are of rising, level, or falling sonority, e.g. [mlo:N] ‘melon’, [mn¡:r] ‘to lead’, [mzi:N] ‘half a litre’ beside [am’s¡:l] ‘missal’ (Malagoli 1910–13: }161). 41 It is noteworthy, however, that in the earliest attestations of the reflex of VITE˘LLU(M) ‘calf ’ dating from the sixteenth century, SPIV had evidently still not operated in Upper Engadinish. For example, in Bifrun’s translation of the celebrated parable of the fatted calf (Luke 15), the form uidilg appears.
216
U-prosthesis
Grizzanese (Emilian) though it may not be general in all dialects of EmiliaRomagna. Here too, where I-prosthesis lost productivity, s impura onsets have generally been aligned with other falling-sonority onsets beginning with an obstruent.42 In Piedmontese varieties, I-prosthesis normally continued to operate post-consonantally and, following SPIV and the rise of U-prosthesis, the two processes interacted. As a result, a single generalized quality was adopted for the prosthetic vowel with all types of heterosyllabic word-initial onset beginning with an obstruent. Thus, in ‘common Piedmontese’ based on Turinese, forms are found such as set e¨ste`ile ‘seven stars’ (as against la ste`ila ‘the star’) alongside set e¨pnass ‘seven tails’ (as against tre pnass ‘three tails’), both with the generalized vowel [‰]. At first sight, the actualization of U-prosthesis would appear to have involved a single parametrized process, whereby contexts on the right-hand side of our Figure 6.3 above were affected first and then prosthesis was generalized progressively to contexts further to the left, in varying degrees according to dialect. However, this may not be an entirely accurate picture of events. There is evidence which suggests the possibility that U-prosthesis may have been actualized originally along two distinct but related paths, dependent on whether the complex word-initial onset began with an sonorant or an obstruent. Various pieces of data offer some support for this view. First, in certain varieties, phonetically different types of prosthetic vowel are found in these two contexts. For instance, in the Piedmontese variety of Viverone, the presence of two word-initial vowels [N] and [€I] is reported.43 The latter appears exclusively preceding sonorants, e.g. [€Irkur’da] ‘to remember’, [€Il’va] ‘to lift’ (< RECORDA¯RE, LEVA¯RE), and the former in all other contexts containing complex word-initial onsets that are not tautosyllabic.44
42
In Grizzanese, clitic phrases such as [al ti v’de:ven] ‘they (f.) saw you’ show a prosthetic [i] triggered by the presence of the falling-sonority and hence heterosyllabic onset [vd-]. Similarly, [al ti stofen] ‘they (f.) annoy you’ where the onset contains s impura. In contrast, level-sonority onsets beginning with an obstruent such as [pk-] evidently do not trigger prosthesis, e.g. [at pke:ven] ‘they (f.) were pecking you’ rather than prosthetic ** [al ti pke:ven] (Loporcaro 1998). 43 The phonetic description of both vowels by Nigra (1901: 252) is unfortunately not entirely clear. The vowel we represent as [N] is described as ‘un suono ottuso, che sta tra i suoni di a ed e’ (‘a dull sound, situated between a and e’) which corresponds to the familiar ‘mute e’ [‰] widely found in Piedmontese varieties. Nigra emphasizes the difference between it and [€I] which he presents as very short, unstressed, closed and less distinct. 44 Just one exception concerns complex onsets involving etymological initial M- which pattern like obstruent-initial onsets, e.g. [dl Nm’luN] ‘of the watermelon’ (< MELO¯NE(M)). Also, as noted above in 6.1.4.1, the realization [N] appears only in contexts where a consonant-final determiner precedes; where other consonant-final forms precede, a much phonetically reduced realization of the vowel occurs.
U-prosthesis
217
Second, the phonetic stages by which a prosthetic vowel developed seem to have been different in the two types of context. With sonorant-initial onsets, the original syllabicity of the etymological first syllable may well have been maintained at all stages, particularly in post-consonantal and post-pausal contexts. As the original vocalic nucleus weakened, its syllabicity was probably taken over by the sonorant which in turn became syllabic. Thereafter, there would be linearization of the syllabic sonorant so that a sequence of unstressed vowel þ non-syllabic sonorant developed (cf. 1.3). Direct evidence of the intervening stage is no longer widely found but some modern Romance varieties do offer suggestive data. For example, in the dialect of Pontremoli (N Lunigiana), weakening of unstressed vowels in word-initial syllables has led to pre-consonantal sonorants appearing in word-initial onset position where they may be realized as syllabic or be linearized into vowel þ sonorant sequences, e.g. [rkun’tar] or [arkun’tar] ‘to relate’ < RE’ COMPUTA¯RE, the former of which can be found when a vowel-final form precedes; a comparable alternation also appears medially in [’nuj a kr’d‰ma] [’nuj a ’ kar’d‰ma] ‘we think’ < *CRED-E¯MUS (Maffei Bellucci 1977: 46). Similarly, in Gallo-Romance the presence of syllabic l, m (though not n)45 and especially r in word-initial pre-consonantal position following the syncope of the original unstressed vowel is reported for the Vende´en dialect of Vouvant, as in [rturna] ‘to ’ return’, [rlik] ‘remains’, [lver] ‘lever’, [mluna] ‘to hum, buzz’ (Re´zeau 1976: }11). ’ ’ ’ The first three of these forms may be compared with standard French retourner, reliques, levier whose schwa can also be syncopated, though just in post-vocalic contexts and without creating a syllabic sonorant; the fourth item which apparently derives ultimately from MASCULUS (FEW VI, 426) has no direct counterpart in standard French. Some further evidence of word-initial syllabic sonorants, especially syllabic r, appears in linguistic atlases.46 For instance, the ALCe (map 800, une reprise mal faite) has [yn rsenyr] at points 28 (Fle´re´-la-Rivie`re) and 32 (St’ Genou) in western Indre, where it is evidently the presence of a preceding consonant that has triggered the syllabic outcome in the lexical item. In the light of such data, it seems plausible that U-prosthesis with etymological word-initial
45
For example, for ‘nephew’ the variant forms reported are [nvr], [nvu] < NEPO¯TE(M) where syncope has occurred but apparently without leading to the creation of a syllabic initial consonant. 46 The ALF provides no clear evidence of syllabic consonants in potentially relevant maps (1135, 1136, 1140, 1147, 1153, 1154, 1163). However, map 1585 grelot lists at least nineteen points (mostly in Poitou-Charente) for which initial [grl-] is reported, a sequence which can scarcely be pronounced unless [r] is vocalic (my thanks to Yves Charles Morin for bringing this to my attention). More recent French linguistic atlases likewise often prove to be of limited assistance in this connection. This is because many of the potentially relevant forms that have been elicited are located in post-vocalic contexts.
218
U-prosthesis
sonorants may generally have involved a transitional stage where the sonorant was syllabic; only later was there linearization yielding a prosthetic vowel. The starting point in the special pattern of evolution found with sonorantinitial words appears to lie in lexical items beginning with the prefix RE-, which were evidently one of the first, if not the first, to be subject to SPIV and subsequent U-prosthesis.47 It is striking that all Romance varieties experiencing SPIV in sonorant-initial words have undergone syncope in RE- forms, and in some varieties of central Italy these forms represent almost the only items to have been affected. For example, Rohlfs (1966: }164) cites for the dialect of Ancona, Marche, arcava`, arfa`, ardı`, argala` ‘to extract, redo, say again, give a present’, whereas SPIV in other contexts is unusual. The forms reported for Umbrian arduna`, armane´, artira`, armette ‘to gather, remain, draw back, replace’ and for the dialect of Cortona, E Tuscany, arcapite´, arco`glie, arcuprı`, armane´ ‘to happen again, gather up, re-cover, remain’ all have etyma in RE- and they again provide virtually the only examples of syncope and U-prosthesis. Flutre (1955: 36) also calls attention to the particular importance of prefixal R(E)- for prosthesis in Picard.48 The explanation for the special susceptibility of word-initial RE- to adaptation is uncertain. However, Blevins and Garrett (1998, 2004) note some suggestive perceptual considerations which are associated with the presence of rhoticity. Rhoticity is seen as a feature with ‘elongated cues’ such as a lowered F3 that can spread over adjacent segments leading to the possibility of individuals ‘mishearing’ and then reinterpreting the original sequencing of segments which include a rhotic. These phonetic cues are more likely to spread from a rhotic into a
47 Cf. ‘Es gibt Evidenz dafu¨r, dass Synkope des Vortonsvokals und folgende Prosthese in dem Pra¨fix RE- fru¨her eingetritt und daher auch ha¨ufiger zu finden ist als in sonstigen Kontexten mit anlautendem Sonorant’ (Mayerthaler 1982: 92, n. 32). The observation is based on Rheto-Romance and Italo-Romance data. Meyer-Lu¨bke (1890: }367) likewise notes the special status of etymological RE- and assumes the creation of a syllabic rhotic as the intervening stage between etymological [re-] and later #vowel þ [r] sequences. He cites forms such as prno˜, r venir (= St.Fr. prenons, revenir) as being frequently found in a wide ’ ’ range of varieties in the west and east of the langue d’oı¨l. Loriot (1984: 190) also postulates an intervening stage with a syllabic rhotic for Picard dialects. 48 Flutre observes that, in the variety of Mesnil-Martinsart, vowel prosthesis is ‘en particulier le cas lorsqu’un mot commence par le pre´fixe r- (franc¸. re-) suivi de consonne’, e.g. in rtire ‘to withdraw’, rpartir ‘to leave again’. However, it is not quite clear whether the ‘particular’ nature of the U-prosthesis here relates to its greater consistency, its historical priority or to the relatively high statistical frequency of verb forms containing the prefix r-. Similarly in the Picard the variety of Gondecourt, the only lexical items to have systematically developed a prosthetic vowel are forms containing prefixal re- preceding a consonant, e.g. [rtirei] ‘to withdraw’ (Cochet 1933).
U-prosthesis
219
preceding or following vowel when the vowel is unstressed and short. The presence of such spreading may thus be a trigger for perceptual metathesis, leading ultimately to the possibility of restructured underlying forms on condition that enough speakers adopt and promote the resequenced forms. However, the perceptual conditions described would also provide a basis for understanding the creation of a syllabic rhotic, since the reinterpretation of a rhotic þ short unstressed vowel sequence as a syllabic rhotic might well represent the first stage of reinterpretation on the way to metathesis. This is more likely with initial unstressed [e] which, as we have seen, has had a greater tendency to weaken than other vowels in Romance (cf. 6.1.2). And it is particularly understandable in a high-frequency and hence more predictable and rapidly articulated sequence like the prefix RE-.49 Where such reinterpretation begins to occur, ‘the general perceptual pattern is for listeners to attribute lowered F3 to a postvocalic segment. Therefore, where there is a rhotic adjacent to the vowel historically, it will be analyzed as postvocalic’ (Blevins and Garrett 1998: 518). A scenario thus emerges for the creation of a prosthetic vowel in etymological RE- sequences. Unstressed [re-] > [re-] first came to be reinterpreted as a syllabic rhotic before undergoing restructuring into a sequence of vowel þ non-syllabic rhotic. We may envisage that subsequently other word-initial sonorant þ consonant sequences arising from the weakening and syncope of [e] followed the pattern of evolution established by forms in RE-. In contrast, it seems unlikely that in complex word-initial onsets created by SPIV where the initial segment was an obstruent, e.g. Emilian [’vde:va] ‘he saw’ < VIDE¯BAT, syllabicity was maintained in the obstruent after syncope had occurred. Instead, if a prosthetic vowel did develop, it was evidently created to enable syllabification to occur. In the light of the available evidence, a tentative conclusion which can be drawn is that U-prosthesis in lexical forms may have operated first on Romance words which had originally contained the high-frequency prefix RE- preceding a consonant stem, as in RE-CORDA¯RE. When SPIV got under way, the word-initial rhotic in these forms absorbed the syllabicity of the weakening vowel [e] and became syllabic. Later, it was linearized to a vowel þ [r] sequence. Other forms also came to acquire heterosyllabic word-initial onsets through SPIV, and those onsets that were sonorant-initial followed the pattern established by original REforms, similarly passing through a phase where the sonorant was syllabic. When 49
The link between the frequency of occurrence of a linguistic form and its consequent predictability, on the one hand, and between its predictability and rapidity of its articulation leading to phonetic reduction, on the other, is well known. Familiar examples are (Golden Age Spanish) vuestra merced > (standard Spanish) usted > (informal styles) (u)te´ ‘you’ (sg., polite form); (French) je ne sais pas > (informal) [ʃpP] ‘I don’t know’.
220
U-prosthesis
obstruent-initial heterosyllabic onsets developed, they too could take on a prosthetic vowel for syllabification purposes. The vowel used was generally identified with the prosthetic vowel that had emerged with sonorant-initial onset sequences. 6.1.4.3 Actualization in proclitic forms
Monosyllabic proclitic forms, such as ME¯, TE¯, DE¯, QUID>[ke], have been subject to SPIV just like lexical items and they too can surface with a prosthetic vowel. However, SPIV and U-prosthesis have not always occurred in a directly parallel way with proclitic forms and lexical forms. Three outcomes of U-prosthesis can be distinguished: (i) identical treatment of proclitic and lexical forms, (ii) similar but not identical treatment of proclitic and lexical forms, (iii) distinct treatment of proclitic forms. Examples of (i) are found in north Gallo-Romance varieties. For example, Picard has the following forms where a uniform vowel quality appears: proclitic lexical form [elve s m¡] ‘to raise one’s hand’ [el sma˜J pas¡] ‘the last week’ e [ty ka˜´ d plAʃ] ‘you (sg.) are changing [ty n edma˜n pw~¡] ‘you (sg.) are not place’ asking’ Data from the dialect of Mesnil-Martinsart (Flutre 1955) Examples of (ii) appear in northern Italian varieties where a different quality may occasionally be found in the prosthetic vowel used with proclitics. In Grizzanese (Emilian), for instance, the prosthetic vowel with lexical items is reportedly [a] but with proclitics it may be [i] (Loporcaro 1998), and in Turinese a prosthetic [i] can appear with the subject proclitic [t] ‘you (sg.)’,50 although the
The special circumstances with the 2nd sg. clitic pronoun arise from the fact that it has been alone amongst the subject clitics of northern Italian dialects to have emerged with a consonantal base-form. Turinese, for example, has the paradigm: (1st sg., 1st pl., 2nd pl.) [i], (3rd sg. and 3rd pl.) [a] (Vanelli 1984: 292). The possibility of vowel prosthesis with these forms is therefore excluded. Vanelli (1984 and 1987) offers a useful diachronic survey of subject pronoun evolution in northern Italian varieties. It is demonstrated inter alia that seemingly prosthetic proclitic forms such as am in the Romagnolo am arcord ‘I remember’, made famous as the title of a 1973 film by Federico Fellini, in reality contain a subject clitic þ object clitic but no prosthetic vowel; thus, am = a (1st sg. subj. cl.) þ m (1st sg. obj. cl.). However, the verb itself arcord (< RECORDO ) has of course undergone prosthesis following SPIV. 50
U-prosthesis
221
usual quality of the prosthetic vowel elsewhere is [‰] (Clivio 1971: 343; Vanelli 1984: 293): Grizzanese Turinese
proclitic [al ti ’vde:ven] ‘they (f.) saw you’ [it ‰m ’dize] ‘you (sg.) tell me’
lexical form ‘to remain’ ¯ N-¯I>[dez ‰mluN] ‘ten DECEM MELO melons’ REMANE¯RE>[arma’Je:]
However, in most northern Italian varieties there seems to be identity between the prosthetic vowel found with proclitics and lexical items, with [a] being the most usual outcome. For example, [a] appears for syllabification reasons as a prosthetic vowel in proclitic sequences involving the subject proclitic t ‘you (sg.)’ followed by the object proclitic g [qÐ ] ‘to him/her’ or m ‘(to) me’: e.g. Ferrarese t ag da an pum and Mantuan ti t am de ‰n pum ‘you give him an apple’(Vanelli 1984: 293). Lexical forms and object clitics likewise show the use of [a] as a prosthetic vowel in these varieties, e.g. Mantuan (lexical) arfudar ‘to refuse’, arvgnir ‘to come back’, aldam ‘dung’, alvar ‘to raise’ and (object clitic) as ved ‘is seen’ (= St.It. si vede), etc. (Cherubini 1827). Finally, cases of (iii) arise when U-prosthesis has occurred with proclitic forms whereas lexical forms have not been subject to SPIV and hence have not undergone U-prosthesis. This is found in several Romance varieties, notably in Catalan, Romanian, and certain types of Swiss Rheto-Romance. In Catalan, a range of prosthetic clitics have developed in most varieties including the standard language although not in western varieties and in Alguere`s in Sardinia. Standard Catalan has the forms em, et, es, ens [‰m ‰t ‰s ‰nz/‰ns] ‘me, you (sg.), him, himself/herself etc. (reflex.), us’ < ME, TE, SE, NOS, which serve as both direct and indirect object proclitic pronouns, e.g. em veu (a mi) ‘(s)he sees me’. Also, less certainly, the forms el ‘him (dir.obj.)’, els ‘them (m. dir. obj. & m./f. ind. obj.), and en ‘some (partitive), from there’ < ILLU(M), ILLOS / ILLIS, INDE may owe their vowel to the action of prosthesis following the deletion of the etymological initial vowel. These prosthetic forms occur just pre-consonantally. In other contexts, nonprosthetic alternants are found, [m‰], [m]; [t‰], [t]; [s‰], [s]; [nus], [ns]; [łu], [ł]; [łus], [ł‰s]; [n‰], [n] which appear respectively in enclitic position, volia veure-me ‘(s)he wanted to see me’ and pre-vocalically in proclitic position, m’ajuda ‘(s)he helps me’. Prosthetic proclitics evidently developed in two stages. First, there was weakening and loss of the vowel in contexts in which there was an adjacent vowel within a syntactic phrase. This is clearly illustrated, for instance, in the prose work Libre de Evast e Blanquerna by Ramo´n Llull which dates from the 1280s. Here, we consistently find alternation indicated between asyllabic and syllabic proclitics. Asyllabic realizations are evident in: (prevocalic) Evast s’asech ‘E. sat down’ (p. 105), enveja t’a tengut ‘envy has held you’ (p. 108); (postvocalic) que nosaltres no·ns mullem ‘so that we do not get wet’ (p. 101), ni·n volia haver ‘nor did he want any’ (p. 106), no·m do´na ‘it does not give’, que·s penedı´s ‘that he
222
U-prosthesis
should repent’ (p. 114); and fa’m considerar en la gran gra`cia que De´us m’a feta ‘it makes me think about the great grace that God has granted me’ (pp. 113–14) showing deletion both post-vocalically and pre-vocalically.51 However, in syntactic phrases where the adjacent segments were both consonants a syllabic form is indicated: la vostra . . . caritat me fa cogitar ‘your charity makes me reflect’ (p. 113), lo malalt se penedı´ ‘the sick man repented’ (p. 114). The pattern here thus resembles closely that found in modern French: tu me vois [tym vwa] ‘you see me’ and elle m’e´coute [¡l mekut] ‘she listens to me’ with vowel deletion in the proclitic pronoun, as against elle me voit [¡l m‰ vwa] ‘she sees me’ with vowel retention. The second stage of development in Catalan saw moves to eliminate the interconsonantal proclitic alternants me, te, se, etc. It seems likely that the forms containing a sonorant were first affected in this development, notably me and nos, and possibly le(s), lo(s) and ne also. In these the syllabicity of the vowel was transferred to the sonorant, giving a syllabic consonant. A possible early indication of this occurs in certain feudal documents dating from the eleventh century, et nu·ls en dedebre´ ni mal nu·ls en menare´ ‘and I will not deceive them in this nor will I bring them harm in this’ (Russell-Gebbett 1965: 76), where <en> may represent a syllabic nasal. Badia (1981: }125) also notes the attestation of proclitic pronominal <el> from the late thirteenth century even in Rossellone`s (Roussillonnais). However, in both cases it is also conceivable that the vowel is a residue of the etymological initial vowel of ILLU(M), INDE. According to Blasco Ferrer (1995: 500), it is not until the fifteenth century that prosthetic forms start to appear.52 Apparent examples are found in Rossellone`s: lany mill quatre cents y quinze es crema la sglesia ‘in 1415 there was a fire in the church’ (although the same text also contains laqual se crema ‘which was on fire’) text dated 1415, and e ens ne esposarem ‘and we will be married’ dated 1462 (Fouche´ [1924] 1980a: 43). Romanian has certain dative pronouns all of which have developed a prosthetic vowel [t_], namely ˆımi, ˆı¸ti, ˆı¸si, ˆıi (< ME, TE, SE, ILLI) ‘to me, to you, to oneself, to him/her’ and, less transparently, the masculine singular accusative pronoun ˆıl (< ILLUM). A parallel change has also occurred with the now non-standard first singular and third plural present indicative forms of the verb ‘to be’ ˆıs (< SUM and 53 SUNT). As in Catalan, U-prosthesis in Romanian has taken place when the forms concerned were used proclitically within a verb phrase where they did not immediately precede a vowel-initial clitic form (auxiliary or clitic pronoun), e.g. ˆımi place foarte mult ‘I like him/her/it very much’, ˆı¸ti uit¸i cartea ‘you (sg.) 51
We follow the edition by Salvador Galme´s (1935), using his punctuation. Unfortunately, few if any of the examples which Blasco Ferrer adduces clearly demonstrate the unambiguous presence of a prosthetic vowel. 53 Forms ˆıs and a˘s ‘they are’ are reported in the ALR to occur at various points in a broad band of territory in north central Romania extending from the region around Arad in the west across to the region around Ias¸i in the east, cf. map 1619. 52
U-prosthesis
223
are forgetting your book’ (= ‘you are forgetting to yourself the book’), but ¸ti-ai uitat cartea ‘you have forgotten your book’ mi-o da˘ ‘(s)he gives it to me’. Prosthetic forms are attested from the period of the earliest surviving texts, the sixteenth century (Densusianu 1975: 405). In addition to the forms already cited, the sixteenth-century text Privila ritorului Lucaci (1581) shows ˆıle ‘to them’ and ˆılui ‘to him’ (< ILLIS, *ILLUI) but these have only ever been found in this work (Lombard 1976). A comparable outcome is also found in Surmeiran in the Rheto-Romance of the Grisons. Here SPIV did not regularly operate, unlike the situation further to the east in Upper Engadinish where it was intense (Grisch 1939: }39).54 In Surmeiran, U-prosthesis occurred only with proclitic pronouns giving am, at, as, ans, ats, iL, iLs ‘(to) me, you (sg.), us, self (3rd pers. refl. sg. and pl.), him, them (m.)’. These forms, however, are now confined to literary use only (Haiman and Beninca` 1992: 127). U-prosthesis in proclitics in Surmeiran may represent a native development but there is the possibility that the phenomenon developed as the result of the westward diffusion of this process from the adjacent Upper Engadine area where syncope of pre-tonic vowels was particularly strong. It seems not unlikely that U-prosthesis with proclitics in all these Romance varieties first operated with forms containing a sonorant. This would have become syllabic prior to being linearized to enable syllabification to occur, and emerging finally as a sequence of default vowel þ sonorant. It is unclear whether a specific form established the use of a prosthetic alternant first of all and acted as a leading form for the other proclitics, or whether prosthesis occurred as a movement simultaneously affecting the subset of proclitics containing a sonorant before it operated on the whole proclitic system. A plausible candidate as a leading form in the former scenario might be the first person singular proclitic ME (cf. Lutta 1923: }126). Parallel cases where tightly knit groups of grammatical forms have been analogically remodelled on the first singular form are not unusual in Romance, as for example the possessive pronoun forms in French.55 54
The variety of Bergu¨n enjoys a special status as it shows clear evidence of SPIV having operated although in other respects it seems to represent a type of Surmeiran. In the light of his detailed study of this variety, Lu¨tta (1923: } 11) concludes that its basis is Surmeiran but that it later received an Engadinish overlay. 55 The possessive forms tien ‘yours’, sien ‘his, hers, theirs’ in French have been remodelled on the basis of mien. Similarly, the Portuguese possessives teu ‘your’, seu ‘his, her, their’ are based on meu; and in Rheto-Romance (Grisons) tiu, siu are formed on miu. A comparable pattern of remodelling, albeit in a more limited way, has also occurred in northern Italian dialects where the subject clitic form for the 1st sg. has widely been extended to 1st pl. and then to the 2nd pl. (Vanelli 1984: 290). Similarly, in northern Gallo-Romance, je has been extended in many varieties to 1st pl. (j’avons) and less commonly to 2nd pl. (j’avez). For the latter, the ALF at pts. 334, 336 in Orne, Normandy
224
U-prosthesis
The possibility of U-prosthesis affecting just proclitic forms is significant, for it appears to add further support to the path of actualization relating to the prosodic hierarchy which was tentatively proposed above: clitic phrase ! phonological phrase ! intonational phrase ! utterance Unfortunately, it is not possible to demonstrate whether U-prosthesis operated in principle in an identical way within the clitic noun phrase as within the clitic verb phrase. The relevant proclitics in noun phrases, namely determiners, all contained an etymological initial vowel which makes interpretation problematic (ILLE, IPSE, UNU(M), UNA).
6.1.5
Q UA L I T Y O F T H E P RO S T H E T I C VOW E L
Unlike the two other principal categories of prosthesis which operated with segmentally specific word-initial onsets, namely those composed of either s impura or a rhotic, U-prosthesis has acted on a phonetically diverse range of onset types. As a result, vowel qualities of different types have developed. However, it is evident that a low-quality [a] has predominated. This is true for Italo-Romance and Rheto-Romance, although in Gallo-Romance a mid front unrounded vowel [e] or [¡] has tended to be the most usual outcome. To try to explain the variations in vowel quality, we may recall the principle of minimal saliency which governs the initial stage of epenthetic vowel formation (1.6). According to this, a short vowel of indeterminate quality develops first of all, typically a schwa, and thereafter the vowel is assigned a quality that conforms with that of an existing vowel-type. In most, perhaps all, Italo-Romance and Rheto-Romance varieties affected by U-prosthesis, it is unclear whether schwa existed as a licensed vowel-type at the time when this process first began to operate. Where schwa was lacking in the vowel inventory, a low value [a] was typically selected for the new vowel. Two reasons can be adduced for this. First, [a] represents the most general default vowel-type in language. Second, and more importantly, it seems highly probable that the first context where a prosthetic vowel became established was in forms with a word-initial onset beginning with a rhotic and, in particular, in the numerous forms originally containing the prefix RE-. As we have seen, these have been particularly subject to SPIV and it is reasonable to assume that these forms were the ones where U-prosthesis took root first of all as their initial onsets displayed the most extreme instance of falling sonority. In the selection of the
has [´ave] for both points in map 92 (vous avez), and [´Prje] (pt. 334), [´e:rje] (pt. 336) in map 95 (vous auriez).
U-prosthesis
225
quality [a], the same factors would have operated as those that guided the choice of vowel quality in A-prosthesis (cf. 5.2.4). As other types of complex word-initial onset gradually acquired a prosthetic vowel, it is understandable that speakers would have adopted the quality already established for complex onsets beginning with a rhotic. Despite the presence of a (rhotic initial) model using the value [a] for the prosthetic vowel, sometimes a vowel of different quality has nonetheless developed in forms containing other complex initial onsets. For example, before onsets with an initial nasal consonant, parallelism with the reflex of forms containing prefixal IN-, IM- has led at times to the appearance of a high quality, [i] or [y], for the prosthetic vowel, as in imsu¨ra ‘measure’, imsu¨reˆr and imzu¨reˆr ‘to measure’, imnatscha ‘threat’, imnatscheˆr ‘to threaten’, imgiuramaint ‘improvement’, imgiureˆr, ‘to improve’, u¨nguotta ‘nothing’, u¨ngiu¨n ‘nobody’ which appear in the sixteenth-century works of Jakob Bifrun.56 Modern Upper Engadinish (dialect of ‘threat’, [In’dzygra] ‘measure’, [YN’gu‰ta] ‘nothing’ Celerina) has [Im’natʃa] whose initial high vowel may be compared with that found in prefixal forms such as [INkun’tre:r] ‘to meet’, [Im’pu‰nd‰r] ‘to impose’ < INCONTRA¯RE, IMPO¯NERE. In Italo-Romance, [a] became and has remained the usual quality for prosthetic vowels in Emilia-Romagna. But in the other main prosthetizing area, Piedmont, various adaptations have subsequently occurred locally. Telmon (1975) provides a detailed, though not always very critical, review of prosthetic vowel qualities attested across a wide range of different contemporary Piedmontese dialects. Whilst Clivio (1971, 2002) identifies just the value [‰] for ‘common Piedmontese’ (based on Turinese), Telmon reports no fewer than six types: [a, e, ‰, u, y, i]. Of these, [a] is by far the most commonly found; indeed, the others appear in only a very small number of instances in the corpus of data presented. The vowel types other than [a] appear to represent later localized developments conditioned by phonetic context. Thus, the quality [u] is reported in the forms [uv’ziN] ‘near’, [un’val] ‘avalanche’, [ur’v¡rs] ‘reverse’ < VICI¯NU(M), NIVA¯LE(M), RE-VE˘RSU(M), where the labiodental [v] has presumably helped to round and raise the vowel. In Gallo-Romance, we may assume once again that the prosthetic vowel first took on the neutral value of schwa. In certain varieties of langue d’oı¨l, this value has been maintained albeit with slight fronting sometimes to a short rounded value [] which commonly occurs as the realization for schwa. For instance, in the variety of Sainte-Jamme (Seine-et-Oise) which lies close to Paris, there appear forms such as [‰lpe:r] ‘the father’, [‰n ‰rsurs] ‘a fountain’, [‰rʃA~ ´e] ‘to change’ (cf. standard French le pe`re, une (res)source, (re)changer).57 The same outcome is
56
The Latin etyma are:
¯ RE, MINA ¯ CEA, MINACE-A¯RE, MELIORA-ME˘NTUM, ¯ RA, MENSU ¯ RA MENSU
˘ TTA, NEC-U ¯ RE, NE(C)-GU ¯ NU(M). MELIORA 57
Passy (1891).
226
U-prosthesis
reflected in ‘rural Parisian’, a spoken variety distinct from the urban franc¸ais populaire and still found in the capital in the middle of the twentieth century (Durand 1945). As in Sainte-Jamme, the prosthetic vowel appears here with both proclitic forms and lexical items: [kPm ‰´di] ‘like I say’, [av¡k ‰lp¡r] ‘with the father’, [apr¡ savwr ‰rturne] ‘after returning’ (cf. comme je dis, avec le pe`re, apre`s s’avoir retourne´). In descriptions of urban Parisian speech, such prosthetic vowels have not traditionally been reported but there are occasional indications of their presence. For instance, in her detailed transcriptions of the pronunciation of fourteen Parisian speakers all of whom lived and were raised in Argenteuil, Franc¸ois (1974) notes [alPK ‰ ʃ i paKti] ‘now I left . . . ’ (= alors, (e) je suis parti) which appears to contain a prosthetic vowel. In some Picard varieties, the closely related value [] is found. For instance, in the dialect of Gondecourt (located just south of Lille), proclitics such as articles and subject pronouns as well as the reflex of prefixal RE- developed a prosthetic schwa in contexts where an adjacent word would create a sequence of three consonants, as in the masculine singular definite article where the base form is [ʃl] < ECCE-ILLE which loses the lateral when preceding a consonant-initial noun: h
pre-consonantal: [prq ʃ ka:] ‘take the cat’, [rv¡t ʃ ka:]58 ‘look at the cat’ pre-vocalic: [prq ʃl o˜m] ‘take the man’, [rv¡t ʃl o˜m] ‘look at the man’ Source: Cochet (1933: 23) However, more generally the prosthetic vowel has adopted a mid unrounded front value of some sort, usually [e]. Thus, many Picard varieties in the de´partements of Somme, Oise, and Aisne show [e], e.g. ertourne´ ‘to return’, ercul ‘drawing back’ in the dialect of Ledieu (Loriot 1984: 186). Similarly, in dialects of Touraine prefixal re- has emerged as e´r- on a regular basis, as in e´rc¸u ‘received (p.pt.)’, e´rcounaıˆte ‘to recognize’, e´rve´ni ‘to return’.59 This development has a parallel amongst clitic elements only with the 1st sg. subj. pronoun which can variably
58
A retracted rounded vowel [o] ‘o vague’ occurs instead of [] in a couple of villages within the district whose speech Cochet is describing. 59 A similar adaptation has occurred word-medially in lexical items. Here, in cases where the deletion of earlier [‰] in [r‰] would have resulted in an unsyllabifiable sequence of three consonants, the syllabic rhotic that developed from [r‰] was therefore restructured to [er], e.g. vende´rdi ‘Friday’, be´rbis ‘sheep’, etc. It may be noted that, in Gondecourt too, schwa is likewise adapted to [e] in word-medial contexts, cf. [ gvP] ‘a horse’ vs [ʃ kevP] ‘the horse’. The data suggest that in both these varieties SPIV only systematically affected lexical items beginning with a liquid and especially those beginning with prefixal RE-.
U-prosthesis
227
appear as [e´] pre-consonantally (Davau 1979: 24, 65–77).60 Here as elsewhere, it is striking that when the prosthetic vowel takes on a mid front unrounded value, the same quality is also generally found with schwas that have experienced adaptation rather than SPIV. This is the case in varieties used in Brie.61 It is also true for the north-west and the extreme south-east of Wallonia where many varieties present the outcome [¡] in the typically proclitic monosyllables that correspond to standard French je, me, te, le, se, de, ne, que as well as in items containing prefixal re- and other lexical forms subject to SPIV, for instance the cognates of items like cerise, feneˆtre; cf. ALW I, maps 8 (cerise) and 54 (le). In these Walloon varieties, the same quality is also shared with the vowel emerging from Iprosthesis, suggesting the durable presence of a default quality for vowels required for syllabification purposes. Other qualities have also developed in the prosthetic vowels of langue d’oı¨l varieties. A low vowel [a] is found, perhaps due to the lowering effect of a following rhotic on mid front unrounded vowels; in Saintongeais, for example, the forms argardez, artenez, armacier (= St.Fr. regardez, retenez, remercier) are reported (Doussinet 1971: 406). However, Picard varieties show a remarkable diversity of outcomes. The data from map 527 of the ALP (se re´tablir ‘to recover from illness’) offer a convenient overview of the range of vowel qualities found since the great majority of Picard varieties have lexical items for this meaning which derive from forms containing prefixal RE-, namely cognates of (se) remettre, requinquer, retaper, refaire, revenir and a verb (se) retousler without counterpart in Standard French (see Map 5). In these prefixal items, there are indications of three main prosthetic vowel qualities: [a] in numerous localities of Pas-de-Calais;62 [¡] or [e] in central and southern Somme; rounded [] or [] in northern and central Oise.63 An isolated case of [o] is reported for the variety spoken in Carnin in western Nord [sorfer] = se refaire, although map 375 (relaverie) indicates a value [] or [] for the initial vowel in the same variety. A high front vowel [y] may also appear in some dialects. This has been noted in the area of Argonne (Ardennes) where, for instance, the dialect of Florent has urbe´yi ‘to look at’, urcommander ‘to recommend’, urvue ‘review’, urvuni ‘to come back’.64 60
With clitics other than je, schwa has likewise been adapted to [e] in contexts where deletion would lead to sequences of three consonants, as in [me] in [´me se tro˜pe] ‘I was wrong’ ([se] = St.Fr. suis). 61 P.c. Yves Charles Morin who was brought up in Brie. 62 Cf. the forms reported for the Arras region by Loriot (1984: 187): archuvoir ‘to receive’, armucher ‘to hide’, s’arbiffer ‘to protest’, arprinde ‘to take up again (= reprendre)’. 63 Cf. eurnier ‘to deny’, eurmembrance ‘memory’, eurposo ‘altar of repose’ found in Esne, near Cambrai (Loriot 1984). 64 The appearance of [y] has also been triggered by the presence of an adjacent labial consonant, as is the case in Picard (cf. Flutre 1977: }23). The same development to [y] has
228
U-prosthesis
6.1.6
C AU S AT I O N
There seems little doubt that syllabic factors led to the rise of U-prosthesis. In Romance varieties where SPIV occurred, the result was the creation of many new types of word-initial complex onset that remained heterosyllabic. Syllabification for these, especially in post-consonantal contexts, was impossible. Particularly affected were new complex onset types of falling sonority, the most conspicuous of which involved forms containing etymological RE- since the sonority fall here could be maximal, as in [rp-] in words like R(E)PAUSA¯RE. In such heterosyllabic onsets, the appearance of a prosthetic vowel can readily be seen to have served as a means of enabling syllabification, as in |rp- > Vr|p. In the case of the northern Gallo-Romance varieties which experienced U-prosthesis, the causation was similar. The weakening of unstressed non-low vowels, especially [e], in an initial open syllable led to their development to schwa before undergoing systematic deletion. This created sequences of word-initial consonants which were often unsyllabifiable and prosthesis served as the means for resolving the problem. In other langue d’oı¨l varieties, the same tendency for schwa deletion also occurred, but it was not systematically carried through. In standard French, for instance, in contexts where an unsyllabifiable sequence would result from schwa deletion, schwa has typically been maintained, as in deux chevaux [d ʃvo] ‘two horses’ but mille chevaux [mil ʃ‰vo] ‘thousand horses’.65 Elsewhere, as we have seen, schwa was likewise retained and in many cases it was adapted to take on a new quality [e], [¡] and less commonly [i], [y] (cf. 4.4.3.3). In varieties where schwa was retained (or strengthened) in this way for syllabification purposes, the incidence of U-prosthesis has predictably been much more limited and in some cases it has failed to operate.
operated, sporadically, with the originally initial vowel [‰] of lexical items such as the cognate of French cresson (cf. ALF map 350, pts 186, 195, 197, 166, 164, etc.). The nearest locality in the ALF to that cited by Loriot is pt. 155 (Belval-en-Argonne) which also has [y] for cresson, [kryso˜]. The quality [y] also figures for NE Walloon dialects in another form originally containing medial [e], e.g. [dyh~ ¡], [dyh¡:] corresponding to standard French descends! ‘get down’ (ALW I, 28). 65 The circumstances of the retention of schwa in standard French have given rise to a considerable literature, descriptive and theoretical. The sense of survival surrounding the more recent history of this vowel is aptly captured in the title of a paper by Walter (1990), ‘Une voyelle qui ne veut pas mourir’.
U-prosthesis
229
6.2 U-prosthesis: later developments In the period from the sixteenth century onward, there appear to have been few changes in the pattern of usage with clitic U-prosthesis. According to Vanelli (1984, 1987), in northern Italian varieties the present situation with subject clitics in the verb phrase had been established by the end of the seventeenth century. On the assumption that the object clitic system had likewise been established by this date, we may hypothesize that most developments concerning U-prosthesis with verbal clitics had been accomplished in these varieties by the seventeenth century. Similarly, the pattern of prosthesis with proclitics in Picard varieties (for which we have documentary evidence from the Middle Ages) shows no significant change since the sixteenth century when prosthetic realizations are first attested (Flutre 1977). The situation in Engadinish is, however, a little less clear. In sixteenth-century texts, U-prosthesis in proclitics is not consistently indicated. In the Bible translation of Jakob Bifrun published in 1560, for example, prosthetic forms of proclitics are not found, e.g. Per che nu faschiand stima da quellas n’s plascha da baiuer oura ‘Because, not having respect for those [fountains], it pleases us to drink from . . . ’ in the prologue, with <ns> rather than
Cf. n. 23 above. We follow the text for this verse work which appears as Una bela senchia historgia da quella sainchia duonna Susanna in Decurtins ([1900] 1983–6: vol. 5, 191–249). This reproduces the edition by J. Ulrich published in 1888. 66 67
230
U-prosthesis
set before you . . . ’ (p. 6).68 This appears to be in conformity with the pattern of actualization proposed above for U-prosthesis (6.1.4.3), and in the following century a letter dated 1634 written by Zoartz Jenatz in Upper Engadinish offers some comparable evidence. Prosthesis is indicated in am sumaglia ‘it seems to me’ which appears once post-pausally and once post-vocalically, but no vowel is indicated in l’g tractat da Isbruck s’lyaiva mauns ‘the formal statement of I. washes its hands of it’, L’g Segner s’vo¨glia acusglier ‘May the Lord wish to take to Himself ’, and Dieu n’s cusalva ‘God preserve us!’ (cf. modern Dieu ans cussalva), though curiously no prosthetic vowel is present with the first plural clitic in da nus n’s vessa ‘as far as concerns us’.69 For lexical items, an important development which has directly impacted on the incidence of U-prosthesis has been the general decline in the productivity of SPIV, for this has necessarily entailed a reduction in the number of new forms containing a phonological structure susceptible to prosthesis. The evidence suggests that the retreat in productivity has been encouraged by the growing influence of standard languages from which new lexical material including learned borrowings has usually emanated. Significantly, the varieties which have become established as standards in Romance have normally conserved pre-tonic vowels. Looking a little more closely at individual areas, we find that already in sixteenth-century Upper Engadinish such learned forms as desert ‘desert’, devot ‘pious’, legiun ‘legion’, and segret ‘secret’ appear in Bifrun’s writings. At the beginning of the twentieth century Walberg (1907) called attention to the appearance of further borrowings in the Upper Engadinish variety of Celerina where SPIV did not operate, [da’l¡t] ‘delight’, [sa’d¡la] ‘bucket’, and also noted cases where U-prosthesis would otherwise have been expected to occur, such [as r‰ba’le:r] ‘to rebel’, [r‰po’ze:r] ‘to rest’, [r‰’m¡gdi] or [rI’m¡gdi] ‘remedy’. The dictionary of Peer (1962) cites numerous other neologisms which are similarly unaffected by SPIV, e.g. resu¨mer ‘to sum up’, renascher ‘to be reborn’, medaglia ‘medal, medallion’, semester ‘semester’, tenu¨ ‘outfit, dress’, tesor ‘treasure, treasury’, genuin ‘genuine’, devisa ‘symbol’, penibel ‘painful’, pedal ‘pedal’. And we may note additional examples from recent editions of the newspaper Fo¨gl Ladin which has articles in both Upper and Lower Engadinish, such as deponia ‘disposal’, (as) retirer ‘to retire’, sedativ ‘sedative’. Furthermore, influence from outside prestigious varieties which had not experienced SPIV has sometimes led to reinsertion of a vowel which has served to restore the original unstressed initial syllable destroyed by SPIV. This was noted
Data drawn from the edition of Chiampel’s text in Decurtins ([1900] 1983–6: vol. 5, 271–96). 69 The text of the letter appears in Decurtins ([1904] 1983–6: vol. 6, 228–9). 68
U-prosthesis
231
for Lower Engadinish by Pult (1897: }166) who reported cases in the variety spoken in Sent where such restructuring appears to have occurred. The forms in the left-hand column appear in the writings of Durich Chiampel. Chiampel (16th century) tsngur dschplaschair schmaiva sngur
Sent (late 19th century) [di´u’nur] [diʃpla’ʃajr] [ʃti’me:va] [si’ Jur]
Although the forms attested in Chiampel’s writings may represent optional syncope characteristic of more informal, allegro speech (cf. Mayerthaler 1982: 100–2), the evidence points to an increased favouring of forms with a full pretonic vowel. Comparable developments have also occurred in northern Italian varieties. For Piedmontese, Clivio (1971: 338) reported that in the urban speech of Turinese the heterosyllabic word-initial clusters [fn- mn- ml- vz-] may be broken up by vowel insertion and he attributed this to the influence of Standard Italian where these clusters are not found. However, word-initial s impura sequences are not affected so that U-prosthesis remains an active process in Turinese and ‘common Piedmontese’ although its incidence has been curtailed somewhat. In Emilian-Romagnolo, a similar decline in the productivity of U-prosthesis is widely found. Malagoli (1910-13) noted the appearance of various lexical items in Novellarese (Emilia) where SPIV and U-prosthesis have failed to operate. These include borrowings such as [ro’bust] ‘sturdy’, [ri’trat] ‘portrait’, and also cases of restoration of initial unstressed vowel through Italian influence as in [ni’su: N] ‘nobody’ beside [an’suN] and [ris’poNder] ‘to reply’, [ris’p¡rmi] ‘saving’. The latter two forms can be compared with native outcomes like [ars’k¡:lda] ‘it warms’ < RE-EX-CALD-A¯RE, [ars’topja] ‘field after harvesting; field left fallow’ < ˘ PULA,70 where we see the possibility of SPIV operating to yield word-initial RE-STU sequences of [rs] þ voicless plosive prior to the operation of U-prosthesis. The greater incidence of this restructuring in urban usage is illustrated by [ro’tond] ‘round’ which had reportedly displaced the earlier form [ar’dond] except in rural speech.
The etymon is a variant of CL STIPULA ‘stubble’ (REW 8265). The form STU˘PULA is attested epigraphically and a syncopated variant STUPLA occurs in Varro’s Res Rusticae. Beyond Italo-Romance, reflexes appear in Old French estouble, Mod. Standard French e´teule ‘stubble’, and widely in other Gallo-Romance varieties. Prefixal forms in RE- are also not uncommon, e.g. (Saintongeais) re´toube ‘field covered in stubble’, (Old Occitan) restoble ‘stubble’ (FEW XII, 271–6). 70
232
U-prosthesis
In northern Gallo-Romance varieties, the decline in the productivity of SPIV and, as a consequence, a reduction in the incidence of U-prosthesis have been far less in evidence. The growing influence in more recent times from the French of the capital, both the standard variety and working-class Parisian French, has affected local speech patterns and led to the adoption of a certain number of forms where the initial unstressed vowel has been preserved. Thus, the Picard variety of Vermandois has a small number of forms such as re´jeton ‘shoot’, re´nouvelance ‘renewal’, re´piye´ ‘snack’, re´pu¨e´ ‘saturated’ which stand in contrast to items like e`rlave´ ‘to wash up’, e`rku¨le´ ‘to withdraw’, e`rpa ‘meal’, e`bzwin ‘need’, e`dvine´ ‘to guess’, cf. Standard French relaver, reculer, repas, besoin, deviner (Debrie 1987). The presence of the former items might suggest a more recent tendency no longer systematically to nativize borrowings through SPIV. However, the dictionary from which these items were taken is far from complete, so that firm conclusions are difficult to draw from the limited information which it provides. When other items of Picard data are considered such as e`ne e`rtraduction ‘a retranslation’ in the Vimeu dialect (cf. Standard French une retraduction), the impression given is that U-prosthesis continues to operate generally as a robust and productive process. This is doubtless not unconnected with the sense of loyalty to their regional usage which Picards and Walloons appear to have retained despite the linguistic influence from outside, a loyalty that is rather stronger than that found amongst speakers from most other regions of the langue d’oı¨l. To conclude, a varied picture emerges in respect of the preservation of U-prosthesis in different Romance varieties over recent centuries. In Italo-Romance, there has been a noticeable diminution in its productivity in those non-standard varieties of the north where previously it has operated regularly. Growing influence from standard Italian where the process is unknown may be seen as a major factor in this development. In a parallel way, the earlier use of U-prosthesis in Upper Engadinish has also experienced some decline which may perhaps owe itself in part to the apparent diminution in prestige of this variety in relation to other varieties of Swiss Rheto-Romance. It is significant, for instance, that the semi-official form of written Rheto-Romance which has come to be adopted for use in government regulations represents a consensus orthographic version of Lower Engadinish (or Vallader), Surmeiran. and Surselvan (Haiman and Beninca` 1992: 15–16). As a result, speakers of Upper Engadinish might well be expected to engage in dialect levelling through the elimination of this variety’s more salient phonological and orthographic characteristics including SPIV and U-prosthesis. Against this background, the fate of U-prosthesis in non-standard varieties of northern France is surprising. Despite the potentially undermining influence from Parisian usage in a nation state with one of the most strongly centralized language policies, the process appears to have retained its productivity to a considerable degree. The reasons for this perhaps unexpected outcome remain rather unclear, however.
7 Conclusion: Retrospective and Prospective Though languishing for a long time in the little-investigated pool of “sporadic” sound-changes, vowel prosthesis has emerged in this study as a wide-ranging development with a clear and systematic structural basis. It unarguably represents a type of regular sound-change and, as such, it should duly take its place within the set of regular sound-changes that are customarily identified in histories of individual varieties of Romance. Three categories of Romance vowel prosthesis have operated whose historical development we have explored in some detail. Each has its own chronological, geographical, and structural characteristics, but it is not difficult to see certain properties which they share over and above the basic defining characteristic of involving word-initial vowel insertion. In particular, it is clear that factors relating to syllable structure and, more precisely, the organization of syllable onsets have been of central importance in their genesis. In all three categories of prosthesis, the new prosthetic vowel can be seen to have arisen as a response to the presence of a complex heterosyllabic onset sequence of some sort. This is especially evident in the rise of I-prosthesis in Imperial Latin and, later on, of U-prosthesis in those varieties of central Romance which underwent SPIV (syncope of pre-tonic wordinitial vowels). In both these cases, the presence of problematic onset sequences acted as the trigger. Less clear at first sight is the affinity between these two categories of vowel prosthesis and A-prosthesis. However, in this third category of prosthesis too the trigger for the creation of a prosthetic vowel proves to have been the pressure to eliminate a heterosyllabic word-initial onset, in this case the rhotic [r-] which had undergone strengthening to become a geminate [rr-]. The shared structural basis of the three principal categories of Romance prosthesis may be expected to have led to some formal overlapping when more than one category has occurred in the evolution of a given variety. This has sometimes happened, as in ‘common’ Piedmontese based on Turinese where I-prosthesis and U-prosthesis have both operated. Here, [‰] has become established as the prosthetic vowel in both cases. However, in other Romance varieties formal differences in the reflexes of the prosthetic vowel have been maintained.
234
Conclusion
For instance, in Gascon both I-prosthesis and A-prosthesis have occurred, but whereas the former has typically yielded the outcome [e], the latter has resulted in a low vowel [a]. The data from these and other Romance varieties which have experienced more than one category of vowel prosthesis appear to indicate that the reflexes of A-prosthesis and U-prosthesis are perhaps more likely to coincide with one another than with the outcome of I-prosthesis. However, further research is required to establish more precisely the patterns of interplay between the outputs of these three categories of prosthesis in Romance. Another common property relates to the prosodic context in which vowel prosthesis first arose and became established. Given the relevance of syllabic structure and syllabification for the genesis of all the categories of prosthesis, it is not surprising that prosthetic vowels typically appear to have developed first of all in phonological contexts where permissible syllabification was not possible. This was the case in post-consonantal and post-pausal contexts since in postvocalic contexts the initial consonant of the heterosyllabic word-initial onset could be linked to the preceding syllable where it would form its coda. Only at a later stage might there be generalization of the prosthetic form to all possible contexts. Given the relevance of contextual factors in the rise of prosthesis, it is unfortunate for our purposes therefore that the focus of phonological studies has very often tended to fall on word-level forms only with few comments on phonological phenomena appearing in prosodic units higher than the word, such as the clitic phrase and the intonational phrase. This limitation of coverage is found in many descriptions of the sound structure of individual varieties, and likewise it is not uncommon even in recent times for linguistic atlases to investigate almost exclusively word-level forms.1 More detailed reports of phonetic realizations in higher-level prosodic units in different Romance varieties are needed therefore to enable more reliable inferences to be drawn about earlier patterns of actualization in Romance vowel prosthesis. A further common property relates to the process of determining the quality of the prosthetic vowel. With all categories of prosthesis, the framework which builds on the principle of minimal salience was found to provide a fruitful basis to account for the choice of vowel quality (1.6). However, the framework allows for variable outcomes, and cases of variation in the quality of the vowel triggered by a particular category of prosthesis were certainly found even in geographically close Romance varieties. For instance, the reflex of A-prosthesis in central Sardinian varieties could be determined either by vowel copying or by
1
More recent atlases which deal almost exclusively with word-level units include ASLEF, ALD, ALPI, and the various regional atlases for Gallo-Romance (ALAL, ALB, etc.). Other atlases such as ALF, AIS, ALEIC can be more revealing as the forms reported often reflect usage within phrasal contexts.
Conclusion
235
the phonological structure of the adjacent following consonant, the rhotic [rr-]. In view of the generally high degree of congruence in Romance with respect to the phonetic outcome of the vowel created by a specific category of prosthesis, anomalies like this Sardinian case invite further research. A significant Romance input could then be made to the more general ongoing debate on the form of epenthetic vowels in language. As has been noted, the organization of syllable structure has played a key role in shaping the incidence of vowel prosthesis. Two considerations have emerged from this. The first relates to the overall architecture of the syllable and the way in which it has changed in the evolution of Romance. The maximal syllable in Latin was characterized by, amongst other things, the relatively high degree of symmetry which existed between its onset and its coda in respect of sequential complexity (cf. 3.1). Each contained a maximum of three consonants, the outermost of which was always and only the fricative [s]. In the early evolution of LatinRomance from Classical times to the end of the first millennium AD, a familiar development has been the move towards the simplification of the non-nuclear elements of syllable structure. Scholarly attention to this reductive change has perhaps fallen more on evolution in coda structure. However, onsets show a comparable tendency to simplify, and the different categories of prosthesis can each readily be seen to have played an integral part in this broad development within the early Romance syllable. The apparent parallelism in the reductive changes affecting onset and coda structure invites a closer examination of syllabic evolution in order to explore to what extent adaptation in the organization of onsets and codas has operated symmetrically throughout the history of Romance and across different Romance varieties. The other aspect concerns the internal architecture of the onset itself. The role of sonority as expressed in the sonority sequencing generalization (SSG) has emerged as a factor of major importance in helping to explain the occurrence of the different categories of prosthesis. Word-initial onset sequences of rising sonority conform to the SSG and would not therefore be expected to be affected, whereas onset sequences of falling sonority do not conform to the SSG and would therefore be expected to be subject to vowel prosthesis. However, the details as to which onsets of falling sonority have triggered prosthesis often prove to be rather more complex than the SSG alone can account for. For instance, liquid-initial onsets of falling sonority have always been susceptible to vowel prosthesis and the resulting prosthetic forms in many Romance varieties have become lexicalized. In contrast, obstruent-initial onsets of falling sonority have generally undergone just the contextually conditioned prosthesis that affected all other heterosyllabic onsets. To explain this difference, appeals might be made to the relevance of the “sonority distance” between the initial and following consonant, i.e. to the degree of disparity in sonority between the successive consonants; e.g. (maximal) [rd-] down to (minimal) [vd-] and (zero) [bd-]. It might be assumed that the greater
236
Conclusion
the sonority distance in an onset, the greater its susceptibility is to vowel prosthesis. However, this criterion likewise does not always provide a satisfactory basis to account for the incidence of prosthesis. For example, in Celerinese (Upper Engadinish) no prosthesis has occurred in [vd¡] “calf ” < VITE˘LLU(M) whereas there is fully generalized and hence lexicalized prosthesis in [alman”te:r] “to lament” > LAMENTA¯RE, even though in both cases there is minimal sonority distance, fricative ! plosive, liquid ! nasal. Instead, it appears to be the inherent phonetic and phonological structure of the individual consonants in complex onsets and the relationship between them in respect of constituency and their articulatory basis that have exercised a decisive role in shaping patterns of vowel prosthesis. It has not been possible in the present work to consider in appropriate depth either of these two aspects of syllabic organization in the history of Romance, but it is evident that further work in this area would be desirable. Although some general studies on Romance syllable development have been carried out (e.g. Granda 1966, Kiss 1971, Holm 1992, Cull 1995, Sampson 2004b), the topic still awaits a more thorough investigation.2 One final shared feature can be identified for the different categories of vowel prosthesis and their historical development. This is that their productivity as phonological processes has tended to diminish over the period since the late Middle Ages and Renaissance. Exceptions can be found, notably I-prosthesis in Ibero-Romance and, more particularly, in Spanish and Catalan. However, as has emerged from the previous chapters, the unmistakeable trend in many Romance varieties has been towards the progressive abandonment of vowel prosthesis as a live process although forms with prosthetic vowels may be retained as lexicalized items. This more recent development, which previous historical accounts of Romance phonological evolution have tended to pass over in silence, owes itself in no small measure to the action of sociolinguistic factors of various types. The principal ones have already been outlined in section 1.7.5, but we may just highlight some key considerations. First, it has undoubtedly been significant that in the standard varieties which have emerged in the nation states of France, Italy, and Romania there is no operative rule of vowel prosthesis. In view of the prestige and increasing ubiquitousness of the standard variety within a nation state, especially in more recent centuries following the growth in mass literacy, communications, and other linguistically levelling forces, the gradual recession of prosthesis in non-standard varieties within these states is readily understandable. Second, from the later
2
The potential relevance of the results for phonological theory may be judged from the impact of the brief general monograph on syllable structure by Vennemann (1988). The data used for this are very largely derived from Romance.
Conclusion
237
Middle Ages onward the emerging standard varieties have commonly undergone elaboration whereby their lexical stocks were expanded, sometimes massively, in order to enable them to cover the formal and literary functions formerly performed by Latin. The source of most of the new lexis was Latin itself, a written language in which any rule of vowel prosthesis was unknown. For most emerging standard Romance varieties, the prestige associated with new learned borrowings led to a strong tendency amongst speakers to preserve the original shape of the latinisms as far as possible, a tendency which militated against the use for these new lexical items of rules of vowel prosthesis that were characteristic of the everyday vernacular. A notable exception comes with Ibero-Romance where it was mainly as a result of other more localized sociolinguistic factors, this time politico-religious in nature, that vernacular patterns of I-prosthesis came to operate on learned borrowings as well (cf. 4.4.2). Third, more recent borrowing from other Romance (standard) languages or non-Romance languages has brought new forms which may have complex word-initial onsets, e.g. [pn-], [sv-], [ft-], [ks-] as in French pneu, svelte, phtisie, Xe´re`s. The influence and prestige of the written word in modern times has been such that speakers have modified their speech habits to accommodate complex onsets of this type rather than subject them to some formerly current adaptive process such as vowel prosthesis. In the light of these and other possible sociolinguistic pressures, it is perhaps understandable that new categories of vowel prosthesis have failed to materialize in Romance from late medieval times onward and that already established productive processes of prosthesis should have experienced widespread regression. The changing fortunes of vowel prosthesis in post-medieval times offer one further example of the significance that sociolinguistic factors can have in shaping formal change in individual linguistic varieties. Romance with its wealth of surviving philological materials reaching back over many centuries provides an unrivalled testing ground for exploring the complex interplay that has occurred between sociolinguistic and structural factors in particular cases of phonological evolution. How this interplay comes to operate in guiding current and future patterns of prosthetic usage will be intriguing for later linguists to observe.
This page intentionally left blank
MAP 1.
Areas showing systematic vowel prosthesis in Romance (past or present)
MAP 2.
Epenthesis with s impura forms in Wallonia
Sources: ALW I, maps 35 e´pine, 38 e´toile, 54 le (article)
Map 3.
A-prosthesis and locations in Gascony
Sources: ALG maps 2129, 2130; Bec 1968: Carte phone´tique ge´ne´rale 1.
R.Po
MAP 4.
U-prosthesis in Italo-Romance and Rheto-Romance
Sources: AIS I 18 nipote; III 548 scure; IV 644 riposati!, 645 riposiamoci!, 737 vicino; V 892 finestra, V 954 pelare; VII 1397 siccare; VIII 1512 telaio. (cf. for SPIV, Mayerthaler 1982: 232, map)
Map 5.
U-prosthesis in Picardy
Source: ALP map 527 se re´tablir.
Map 6.
Vowel prosthesis and locations in Corsica
Map 7.
Vowel prosthesis and locations in Sardinia
Principal source: Contini 1987: vol. II, maps 52, 74
MAP 8.
Locations in the Iberian Peninsula
R.Po
MAP 9.
Locations in Northern Italy and the Rheto-Romance area
Map 10.
Locations in central-southern Italy
e R.Rhôn
Map 11.
Locations in SE France and adjacent areas of Italy
MAP 12.
Locations in Northern France
Bibliography
AGI: AIS: ALAL: ALB: ALCat: ALCB: ALCe: ALD: ALEANR: ALEIC: ALF: ALF: ALG: ALGa: ALI: ALIFO: ALJA: ALLoc:
Archivio Glottologico Italiano. Jaberg, K. & Jud, J. (eds). 1928–41. Sprach- und Sachatlas Italiens und der Su¨dschweiz, 8 vols. Zofingen: Ringier. Potte, J.-C. (ed.). 1975–87. Atlas linguistique et ethnographique de l’Auvergne et du Limousin, 2 vols. Paris: CNRS. Taverdet, G. (ed.). 1975–88. Atlas linguistique et ethnographique de Bourgogne, 4 vols. Paris: CNRS. Griera, A. (ed.). 1923–64. Atlas lingu¨´ıstic de Catalunya, 8 vols. Barcelona: Instituto International de Cultura Roma´nica. Bourcelot, H. (ed.). 1966–78. Atlas linguistique et ethnographique de la Champagne et de la Brie. 3 vols. Paris: CNRS. Dubuisson, P. (ed.). 1971–82. Atlas linguistique et ethnographique du Centre (Berry et Bourbonnais). 3 vols. Paris: CNRS. Goebl, H. (ed.). 1998. Atlant linguistich dl ladin dolomitich y di dialec vejins. 7 vols. Wiesbaden: Reichert. Alvar, M. et al. (eds). 1979–83. Atlas lingu¨´ıstico y etnogra´fico de Arago´n, Navarra y Rioja. 12 vols. Madrid: CSIC. Bottiglioni, G. (ed.). 1933–42. Atlante linguistico etnografico italiano della Corsica, 10 vols. Pisa: L’Italia dialettale. Gillie´ron, J. (ed.). 1902–10. Atlas linguistique de la France. Paris: Champion. Corse Gillie´ron, J. (ed.). 1914–15. Atlas linguistique de la France: Corse. Paris: Champion. Se´guy, J. (ed.). 1954–73. Atlas linguistique de la Gascogne, 6 vols. ToulouseParis. ILG (ed.). 1990–. Atlas lingu¨istico galego. Santiago de Compostela: ILG. Massobrio, L. (ed.). 1995–. Atlante linguistico italiano. Rome: Istituto Poligrafico e Zecca dello Stato. Simoni-Aurembou, M.-R. (ed.). Atlas linguistique et ethnographique de l’Ilede-France et de l’Orle´anais, 2 vols. Paris: CNRS. Martin, J.-B. & Tuaillon G. (eds). 1971–78. Atlas linguistique et ethnographique du Jura et des Alpes du Nord, 3 vols. Paris: CNRS. Ravier, X. (ed.). 1978–86. Atlas linguistique et ethnographique du Languedoc Occidental, 3 vols. Paris: CNRS.
252
Bibliography
ALLor: ALN: ALP: ALPic: ALR: ALW: ASLEF: CGL: CIL: CLS: DCECH: DECLC: DELI: DHLF: FEW: JFLS: Keil: LN: LRL MSLP: REW: RLaR: RLiR: ROA: RPh: TLL: ZrP:
Boisgontier, J. (ed.). 1981–86. Atlas linguistique et ethnographique du Languedoc Oriental, 3 vols. Paris: CNRS. Brasseur. P. (ed.). 1980–84. Atlas linguistique et ethnographique normand, 2 vols. Paris: CNRS. Bouvier, J.C. & Martel C.(eds). 1975–86. Atlas linguistique et ethnographique de la Provence, 3 vols. Paris: CNRS. Carton, F. & Lebe`gue M. (eds). 1989. Atlas linguistique et ethnographique picard, 1 vol. Paris: CNRS. Atlasul linguistic romaˆn, Serie noua, 7 vols. Bucharest: Ed. Acad., 1956–72. Remacle, L. (ed.). 1953–. Atlas linguistique de la Wallonie Lie`ge: VaillantCarmanne, Pellegrini, G.B. (ed.). 1972–84. Atlante storico-linguidtico-etnografico friulano, 5 vols. Padua: Univ. of Padua. Loewe, G. and Goetz, G. (eds.). Corpus Glossariorum Latinorum. 7 vols. 1888–1923. Leipzig: Teubner. Corpus Inscriptionum Latinarum. 17 vols. 1868–. Berlin: Wiedmann-De Gruyter. Chicago Linguistic Society. Corominas, J. & Pascual, J.A. 1980–91. Diccionario critico etimolo´gico castellano e hispa´nico. 6 vols. Madrid: Gredos. Coromines, J. 1980–91. Diccionari etimolo`gic i complementari de la llengua catalana. 9 vols. Barcelona: Curial. Cortelazzo, M.C. and Cortelazzo, M.A. 1999. Il nuovo etimologico DELI. Dizionario etimologico della lingua italiana. 2nd ed. Bologna: Zanichelli. Rey, A. (ed.). [1992] 1998. Dictionnaire historique de la langue franc¸aise. 3 vols. Paris: Dictionnaires Le Robert. Wartburg, W. von. (ed.). 1922– Franzo¨sisches etymologisches Wo¨rterbuch, 25 vols. Leipzig-Basel: Klopp-Zbinden. Journal of French Language Studies. Keil, H. (ed.). 1857–80. Grammatici Latini, 7 vols. Leipzig: Teubner. Lingua Nostra. Holtus, G., Metzeltin, M. & Schmitt, C. (eds.). 1988–2005. Lexikon der romanistischen Linguistik. Tu¨bingen: Niemeyer. Me´moires de la Socie´te´ de Linguistique de Paris. Meyer-Lu¨bke, W. 1935 [repr. 1968]. Romanisches etymologisches Wo¨rterbuch, 3rd ed. Heidelberg: Carl Winter. Revue des Langues Romanes. Revue de Linguistique Romane. Rutgers Optimality Archive, Rutgers University. Romance Philology. Thesaurus Linguae Latinae. Zeitschrift fu¨r romanische Philologie.
Bibliography
253
Academia de la Llingua Asturiana. 1990. Normes ortogra´fiques. 3rd edn. Oviedo: Academia de la Llingua Asturiana. Adams, J. N. 1977. The Vulgar Latin of the Letters of Claudius Terentianus. Manchester: Manchester University Press. —— 2003. Bilingualism and the Latin Language. Cambridge: Cambridge University Press. —— 2007. The Regional Diversification of Latin 200 BC–AD 600. Cambridge: Cambridge University Press. Allen, W. Sidney. 1973. Accent and Rhythm. Cambridge: Cambridge University Press. —— 1978. Vox Latina. 2nd edn. Cambridge: Cambridge University Press. —— 1987. Vox Graeca. 3rd edn. Cambridge: Cambridge University Press. Altamura, Antonio. 1949. Testi napoletani dei secoli XIII e XIV. Naples: Perrella. Altamura, Antonio (ed.). 1974. Cronaca di Partenope. Naples: Societa` Editrice Napoletana. —— (ed.). 1977. La regola salernitana. Naples: Societa` Editrice Napoletana. Alvar, Manuel. 1953. El dialect aragone´s. Madrid: Gredos. —— 1973. Estudios sobre el dialecto aragone´s. I. Zaragoza: Institucio´n “Fernando el Cato´lico”. Alvarez, Guzma´n. 1985. El habla de Babia y Laciana. [Repr. of 1947 thesis]. Leo´n: Ediciones Leonesas. A´lvarez Nazario, Manuel. 1991. Historia de la lengua espaola en Puerto Rico. Mayagu¨ez, Puerto Rico: Academia Puertorriquea de la Lengua Espaola. Aly-Belfa`del, Arturo. 1933. Grammatica piemontese. Noale: L. Guin. Andersen, Henning. 1972. “Diphthongization”. Language 48: 11–50. Andersen, Henning. (ed.). 1986. Sandhi Phenomena in the Languages of Europe. Berlin, New York, and Amsterdam: Mouton de Gruyter. Andersen, Henning. 1989. “Understanding linguistic innovations”. In L. E. Breivik and E. H. Jahr (eds.). Language Change. Contributions to the Study of its Causes. Berlin and New York: Mouton de Gruyter. Andersen, Henning. (ed.). 2001. Actualization. Linguistic Change in Progress. Amsterdam and Philadelphia: Benjamins. Anderson, James, M. 1988. Ancient Languages of the Hispanic Peninsula. Lanham: University Press of America. Arquint, Jachen C. 1964. Vierv ladin. 2nd edn. Chur: Stamparia Bu¨ndner Tagblatt. Arvinte, Vasiliu. 1980. Die Ruma¨nen. Ursprung, Volks- und Landesnam. Tu¨bingen: Narr. Ascoli, Graziadio I. 1878. “Varia”. AGI 3: 442–71. Aub-Bu¨scher, Gertrude. 1962. Le parler de Ranrupt (Bas-Rhin). Paris: Klincksieck. Audollent, A. 1904. Defixionum tabellae. Paris: Fontemoing. [repr. Frankfurt: Minerva, 1967]. Auger, Julie. 2001. “Phonological variation and Optimality Theory: evidence from wordinitial epenthesis in Vimeu Picard”. Language Variation and Change 13: 253–303. Avram, Andrei. 1990. Nazalitatea ¸si rotacismul ˆın limba romaˆnaˇ. Bucharest: Editura Academiei Romaˆne.
254
Bibliography
Azra, J.-L. and Cheneau, V. 1994. “Jeux de langage et the´orie phonologique. Verlan et structure syllabique du franc¸ais”. JFLS 4: 147–70. Bach, Adolf. 1965. Geschichte der deutschen Sprache. 8th edn. Heidelberg: Quelle & Mayer. Badia I Margarit, Antonio. 1950. El habla del Valle de Bielsa. Barcelona: Instituto de Estudios Pirenaicos. —— 1981. Grama`tica histo`rica catalana. Barcelona: Biblioteca d’Estudis i Investigacions. Baehrens, W. A. 1922. Sprachlicher Kommentar zur vulga¨rlateinischen Appendix Probi. Halle: Niemeyer. Bagemihl, Bruce. 1991. “Syllable structure in Bella Coola”. LI 22: 589–646. Bagli, Giuseppe Gaspare (ed.). 1887. “Pulon Matt. Frammento inedito di poema in dialetto cesenate”. Documenti e Studi pubblicati per cura della Reale Deputazione di Storia Patria. Vol. 2. Bologna: Regia Tipografia, 229–334. Baldinger, Kurt. 1972. La formacio´n de los dominios lingu¨´ısticos en la Penı´nsula Ibe´rica. 2nd edn. Madrid: Gredos. Banniard, Michel. 1992. Viva voce: communication ´ecrite et communication orale du IVe au IXe sie`cle en Occident latin. Paris: Institut d’e´tudes augustiniennes. Barbina, Alfredo (ed.). 1969. Concordanze del “Decameron”. 2 vols. Florence: Giunti. Barbosa, Jorge Morais. [1965] 1983. E´tudes de phonologie portugaise. 2nd edn. E´vora: Universidade de E´vora. —— 1994. “Portugiesisch: Phonetik und Phonemik / Fone´tica e fonologia”. LRL VI, 2: 130–42. Bartoli Langeli, Attilio. 1989. Storia dell’alfabetismo come storia degli scriventi: gli usi della scrittura in Italia tra Medioevo ed eta` moderna. Florence: Universita` di Firenze. —— 2000. La scrittura dell’italiano. Bologna: Il Mulino. Bartoli, Matteo. 1906. Das Dalmatische. Altromanische Sprachreste von Veglia bis Ragusa und ihre Stellung in der apennino-balkanischen Romania. 2 vols. Vienna: Ho¨lder. [Ital. version. Aldo Duro (ed.). 2000. Il dalmatico. Rome: Istituto della Enciclopedia Italiana.] Bassols de Climent, Sebastia´n. 1981. Fone´tica latina. 5th impr. Madrid: CSIC. Bauche, Henri. [1920] 1946. Le langage populaire. 4th edn. Paris: Payot. Bayot, Alphonse. 1929. Le poe`me moral. Brussels-Lie`ge: Palais des Acade´mies & VaillantCarmanne. Bearesi, Luigi. 1982. Piccolo dizionario del dialetto piacentino. Piacenza: Berti. Bec, Pierre. 1968. Les interfe´rences linguistiques entre gascon et languedocien dans les parlers du Comminges et du Couserans. Paris: PUF. Bembo, Pietro. [1525]. In Carlo Dionisotti (ed.). 1989. Prose della volgar lingua. Gli Asolani. Rime. Milan: TEA. Bertinetto, Pier Marco. 1999. ‘La sillabazione dei nessi /sC/ in italiano: un’eccezione alla tendenza “universale”?’. In P. Beninca` and A. Mioni (eds). Fonologia e morfologia dell’italiano e dei dialetti d’Italia. Rome: Bulzoni, 71–96. Bertoni, Giulio. 1905. Il dialetto di Modena. Turin: Loeschler. —— 1908–11. “Banchieri a Imola nel sec. XIII (1260)”. Studi Medievali 3: 683–9. —— 1925. Profilo del dialetto di Modena. Geneva: L. Olschki.
Bibliography
255
Bezzola, Reto and To¨njachen, Rudolf. 1976. Dicziunari tudais-ch–rumantsch ladin. 2nd edn. Chur: Lia Rumantscha. Bianchini, A. (ed.). 1987. Tempo di affetti e di mercanti. Lettere ai figli esuli. Alessandra Macinghi Strozzi. Milan: Garzanti. Biasci, Gianluca. 1998. L’evoluzione del dialetto pisano in un carteggio mercantile del XV secolo. Pescara: Libreria dell’Universita` Editrice. Biville, F. 1994. “Existait-il une diphtongue ui en latin?”. In J. Herman (ed.). Linguistic Studies on Latin. Amsterdam: Benjamins, 3–18. Blasco Ferrer, Eduardo. 1984. Grammatica storica del Catalano e dei suoi dialetti con speciale riguardo all’Algherese. Tu¨bingen: Narr. —— 1995. “Sardisch / Il sardo”. LRL II, 2: 239–71. Blevins, Juliette. 1995. “The syllable in phonological theory”. In J. A. Goldsmith (ed.). The Handbook of Phonological Theory. Oxford: Blackwell, 206–44. —— 2008. “Consonant epenthesis: natural and unnatural histories”. In J. Good (ed.). Linguistic Universals and Language Change. Oxford: Oxford University Press, 79–107. —— and Garrett, Andrew. 1998. “The origins of consonant-vowel metathesis”. Language 74: 508–66. —— —— 2004. “The evolution of metathesis”. In B. Hayes, R. Kirchner, and D. Steriade (eds). Phonetically-Based Phonology. Cambridge: Cambridge University Press, 117–56. Bloomfield, Leonard. [1933] 1935. Language. London: George Allen & Unwin. Bo¨hne, Rudolf. 2003. Il dialetto del Sarrabus (Sardegna sud-orientale). Sestu (Cagliari): Zonza. Bolognesi, Roberto. 1998. The Phonology of Campidanian Sardinian. A Unitary Account of a Self-Organizing System. HIL dissertations 38. The Hague: Holland Academic Graphics. Bonazzi, Giuliano. [1900] 1979. Il Condaghe di San Pietro di Silki. Rev. edn by I. Delogu. Sassari: Dessi. Bonelli, Giuseppe & Contini, Gianfranco. 1935. “Antichi testi bresciani”. Italia Dialettale 11: 115–151. Bonfante, Giuliano. 1955. “Il siciliano e il sardo” Bollettino del Centro di Studi Filologici e Linguistici Siciliani 3: 195–222. Bonnet, Max. 1890. Le latin de Gre´goire de Tours. Paris: Hachette. Bo¨rner, Wolfgang. 1976. Schriftstruktur und Lautstruktur. Studien zur altgalizischen Skripta. ZrPh Beiheft 155. Tu¨bingen: Niemeyer. Bottiglioni, Gino. 1920. “Saggio di fonetica sarda”. Studi Romanzi 15: 13–114. Bourciez, E´douard. 1936. “Le domaine gascon”. RLiR 12: 1–9. —— 1956. Ele´ments de linguistique romane. 4th edn. Paris: Klincksieck. Boutier, Marie-Guy. 1995. “Franzo¨sische Skriptaformen I. Wallonie / Les scriptae franc¸aises I. Wallonie”. LRL II, 2: 290–300. Brero, Camillo. 1971. Grama`tica piemonte`isa. 2nd edn. Turin: Ij Brande´. Broselow, Ellen. 1991. “The structure of fricative-stop onsets”. Unpublished manuscript. Brun, Auguste. 1931. Le franc¸ais de Marseille: ´etude de parler re´gional. Marseille: Institut historique de Provence.
256
Bibliography
Bruneau, Charles. 1913a. E´tude phone´tique des patois d’Ardenne. Paris: Champion. —— 1913b. La limite des dialectes wallon, champenois et lorrain en Ardenne. Paris: Champion. —— 1914–26. Enqueˆte linguistique sur les patois d’Ardenne. 2 vols. Paris: Champion. Brunel, Clovis. 1926. Les plus anciennes chartes en langue provenc¸ale. Paris: Champion. —— 1952. Supple´ment. Paris: A. et J. Picard. Brunot, Ferdinand. 1966. Histoire de la langue franc¸aise. I. (New edn by G. Antoine). Paris: Armand Colin. Buck, Carl D. 1904. A Grammar of Oscan and Umbrian. Boston: Ginn. Buridant, Claude. 2000. Grammaire de l’ancien franc¸ais. Paris: Nathan. Cabre´ i Monne´, Teresa. 1993. “Interfere`ncia, ortografia i grama`tica”. In R. Alemany, A. Ferrando, and L.B. Meseguer (eds). Actes del Nove` Col.loqui Internacional de Llengua i Literatura Catalanes. Barcelona: Publicacions de l’Abadia de Montserrat, vol. 3, 97–109. Caduff, Le´onard. 1952. Essai sur la phone´tique du parler rhe´toroman de la Valle´e de Tavetsch. Berne: Francke. Cagliaritano, Ubaldo. 1975. Vocabolario senese. Florence: Barbe´ra. Calderini, Aristide. 1951a. “La corrispondenza greco-latina del soldato Claudio Tiberiano e altre lettere del II sec. d. Cr. nel recente vol. VIII dei papiri del Michigan”. Rendiconti del Istituto Lombardo, Classe di Lettere e Scienze Morali e Storiche. 84: 155–66. Calderini, Rita. 1951b. “Osservazioni sul latino del P. Mich. VIII, 467–472”. Rendiconti del Istituto Lombardo, Classe di Lettere e Scienze Morali e Storiche. 84: 250–62. Camilli, Amerindo. 1929. “Il dialetto di Servigliano”. AR 13: 220–71. Campus, G. 1901. Fonetica des dialetto logudorese. Turin: Bosa. Cano, Rafael (ed.). 2004. Historia de la lengua espaola. Barcelona: Ariel. Capidan, Theodor. 1932. Aromaˆnii, dialectul aromaˆn. Studiu lingvistic. Bucharest: Imprimeria Nat¸ionalaˇ. Caragiu Mariot¸eanu, Matilda. 1975. Compendiu de dialectologie romaˆnaˇ (nord ¸si suddunaˇreanaˇ). Bucharest: Editura Academiei Republicii Socialiste Romaˆnia. ——, Giosu, S¸tefan, Ionescu-Ruxandoiu, Liliana, and Todoran, Romulus. 1977. Dialectologie romaˆnaˇ. Bucharest: Ed. didacticaˇs¸i pedagogicaˇ. Carlton, Charles M. 1973. A Linguistic Analysis of a Collection of Late Latin Documents composed in Ravenna between A.D. 445–700. The Hague: Mouton. Carnoy, Albert. [1906] 1983. Le latin d’Espagne d’apre`s les inscriptions. Repr. Hildesheim, Zurich, and New York: Olms. Carreira, M. M. 1996. “Spanish clusters: coronals, /s/ and syllable structure conditions”. In C. Parodi, C. Quicoli, M. Saltarelli, and M.L. Zubizarreta (eds). Aspects of Romance Linguistics. Washington DC: Georgetown University Press, 123–34. Carter, Henry H. 1941. Cancioneiro da Ajuda. A Diplomatic Edition. New York and Oxford: Modern Language Association of America and Oxford University Press. Castellani, Arrigo. 1952. Nuovi testi fiorentini del dugento. 2 vols. Florence: Sansoni. —— 1976. I piu` antichi testi italiani. Edizione e commento. 2nd edn. Bologna: Patron.
Bibliography
257
—— 1980. Saggi di linguistica e filologia italiana e romanza (1946–1976). 3 vols. Rome: Salerno Editrice. Cecchi, Elena (ed.) 1990. Le lettere di Francesco Datini alla moglie Margherita (1385–1410). Biblioteca dell’Archivio storico pratese, 14. Prato: Societa` Pratese di Storia Patria. Cerquiglini, Bernard. 1991. La naissance du franc¸ais. Paris: PUF. Chambon, Jean-Pierre and Greub, Yan. 2002. “Note sur l’aˆge du (proto)gascon”. RLiR 66: 473–95. Chaurand, Jacques. 1983. “Pour l’histoire du mot “francien””. In (no ed.) Me´langes di dialectologie d’oı¨l a` la me´moire R. Loriot. Fontaines le`s Dijon: Association Bourguignonne de Dialectologie et d’Onomastique, 91–9. Cherubini, Francesco. [1827] 1992. Vocabolario mantovano-italiano. Milan: Bianchi e Co. Repr. Imola: Forni. Christ, Karl. 1984. The Romans. London: Chatto & Windus. Cierbide Martinena, Ricardo. 1988. Estudio lingu¨´ıstico de la documentacio´n medieval en lengua occitana de Navarra. Bilbao: Universidad del Paı´s Vasco. Claverı´a Nadal, Gloria. 1991. El latinismo en espaol. Barcelona: Departament de Filologia Espanyola, Universitat Auto`noma de Barcelona. Clements, George N. 1990. “The role of the sonority cycle in core syllabification”. In J. Kingston and M. E. Beckman (eds). Papers in Laboratory Phonology I.. Cambridge: Cambridge University Press, 283–333. Clements, George and Hume, Elizabeth (1995). “The internal organization of speech sounds”. In J. Goldsmith (ed.). The Handbook of Phonological Theory. Oxford: Blackwell, 245–306. Clivio, Gianrenzo P. 1971. “Vocalic prosthesis, schwa-deletion and morphophonemics in Piedmontese”. ZrPh 87: 334–44. —— 2002. “Il Piemonte”. In Cortelazzo et al., 151–95. —— and Danesi, Marcello. 1974. Concordanza linguistica dei “Sermoni subalpini”. Turin: Centro Studi Piemontesi. Cochet, E. 1933. Le patois de Gondecourt (Nord). Grammaire et lexique. Paris: Droz. Coco, Francesco. 1970. Il dialetto di Bologna. Fonetica storica e analisi strutturale. Forni: Bologna. Cohen, G. 1949. Recueil de farces franc¸aises ine´dites du XVe sie`cle. Cambridge, MA: Medieval Academy of America. Collinge, N. E. 1970. “Computation and Latin consonants”. In N. E. Collinge, Collectanea Linguistica. Essays in general and genetic linguistics. The Hague–Paris: Mouton, 192–218. Collingwood, R. G. and Wright, R. P. 1965. The Roman Inscriptions of Britain. 1. Inscriptions on Stone. Oxford: Clarendon. Contini, Gianfranco. 1941. Le opere volgari di Bonvesin da la Riva. I. Testi. Rome: Societa` Filologica Romana. Contini, Michel. 1987. E´tude de ge´ographie phone´tique et de phone´tique instrumentale du sarde. 2 vols. Turin: Edizioni dell’Orso.
258
Bibliography
Corda, Francesco. 1994. Grammatica moderna del sardo logudorese. Cagliari: Edizioni Dalla Torre. Cornagliotti, Anna. 1995. “Sprache der Waldenser / Il valdese”. LRL II, 2: 467–73. Corneille, Thomas. 1687. Remarques sur la langue franc¸oise de Monsieur de Vaugelas. 2nd edn. 2 vols. Paris: Girard. Coromines, Joan. 1990. El parler de la Vall d’Aran. Barcelona: Curial Ed. Catalanes. Cornu, Jules. 1892. “E´tudes de grammaire portugaise. II. L’a prothe´tique devant rr- en portugais, en espagnol et en catalan”. Romania 11: 75–9. Correa Rodrı´guez, Jose´ Antonio. 2004. “Elementos no indoeuropeos e indoeuropeos en la historia lingu¨´ıstica hispa´nica”. In Cano (ed.), 36–57. Cortelazzo, Manlio, Marcato, Carla, De Blasi, Nicola, and Clivio, Gianrenzo P. (eds). 2002. I dialetti italiani. Storia, struttura, uso. Turin: UTET. Coˆte´, Marie-He´le`ne. 2000. “Consonant cluster phonotactics: a perceptual approach. Unpub. doctoral dissertation, MIT [# ROA 548]. Coteanu, Ion. 1981. Structura ¸si evolut¸ia limbii romaˆne. Bucharest: Editura Academiei. Cotgrave, R. 1611. A Dictionarie of the French and English Tongues. London: Islip. [Facsimile University of South Carolina Press, 1950]. Coulmas, Florian. 1989. The Writing Systems of the World. Oxford: Blackwell. Coupier, J. 1995. Dictionnaire Franc¸ais-Provenc¸al. Diciouna`ri France´s-Prouvenc¸au. Aix-enProvence: Diffusion Edisud. Coustenoble, He´le`ne N. 1945. La phone´tique du provenc¸al moderne en terre d’Arles. Hertford: S. Austin & Sons. Covarrubias, Sebastia´n. 1611. Tesoro de la lengua castellana o espaola. Madrid: Sanchez. [Facsimile Barcelona: Horta, 1943]. Cravens, Thomas D. 2002. Comparative Historical Dialectology. Italo-Romance Clues to Ibero-Romance Sound Change. Amsterdam and Philadelphia: Benjamins. Cull, Naomi. 1995. “Reconstruction of the Proto-Romance syllable”. Dans H. Andersen (dir.). Historical Linguistics 1993. Amsterdam and Philadelphia, 117–32. Curchin, Leonard A. 1991. Roman Spain: Conquest and Assimilation. London: Routledge. Cyran, Eugeniusz and Gussmann, Edmund. 1999. “Consonant clusters and governing relations: Polish initial consonant sequences”. In Hulst and Ritter (eds.), 219–47. Dalbera, Jean-Philippe. 1994. Les parlers des Alpes-Maritimes: e´tude comparative: essai de reconstruction. London: AIEO. Dalbera-Stefanaggi, Maria-Jose. 1978. Langue corse. Une approche linguistique. Paris: Klincksieck. —— 1991. Unite´ et diversite´ des parlers corses. Alessandria: Ed. dell’Orso. D’Ambra, Raffaele. 1873. Vocabolario napolitano-toscano domestico di arti e mestieri. Naples: publ. by author. [Facsimile 1996. Imola: Arnaldo Forni Editore]. Danesi, Marcel. 1976. La lingua dei “Sermoni subalpini”. Turin: Centro Studi Piemontesi. D’Aronco, Gianfranco. 1960. Nuova antologia della letteratua friulana. Udine-Tolmezzo: Aquileia.
Bibliography
259
Davau, Maurice. 1979. Le vieux parler tourangeau: sa phone´tique, ses mots et locutions, sa grammaire. Chambray-le`s-Tours: E´d. C.L.D. Davis, Stuart. 1988. “Syllable onsets as a factor in stress rules”. Phonology 5: 1–19. —— 1990. “The onset as a constituent of the syllable”. CLS 26: 71–81. —— 1999. “On the representation of initial geminates”. Phonology 16: 93–104. De Blasi, Nicola. 1986. Libro de la destructione de Troya. Rome: Bonacci. De Giovanni, Marcello. 2003. Molise. Pisa: Pacini. De Gregorio, Giacomo. [1890] 1993. Saggio di fonetica siciliana. Repr. S. Cristina Gela (Palermo): Librarie Siciliane. De Mauro, Tullio. [1963, 21970] 1993. Storia linguistica dell’Italia unita. Rome-Bari: Laterza. Debrie, Rene´. 1987. Lexique picard des parlers du Vermandois. Amiens: Universite´ de Picardie. Decurtins, Caspar (ed.). [1900] 1983–6. Ra¨toromanische Chrestomathie. 15 vols. Erlangen: Fr. Junge. Repr. Chur: Octopus Verlag. Deferrari, Harry A. 1954. The Phonology of Italian, Spanish, and French. Washington, DC: Roy J. Deferrari. Degli Innocenti, Mario (ed.). 1984. L’“Elucidario”. Volgarizzamento in antico milanese dell’“Elucidarium” di Onorio Augustodunense. Padua: Antenore. Dell, Franc¸ois. 1980. Generative Phonology and French Phonology. Cambridge: Cambridge University Press. [revised trans. of Les re`gles et les sons. 1973. Paris: Hermann] Deloffre, Fre´de´ric. [1961] 1999. Agre´ables confe´rences de deux paysans de Saint-Ouen et de Montme´rency sur les affaires du temps (1649–1651). Repr. with additions. Geneva: Slatkine. Densusianu, Ovid. [1901–38] 1975. Histoire de la langue roumaine. 2 vols. [Original edition, Paris: Welter]. Bucharest: Editura Minerva. Desgranges, J.-C.-L.-P. 1821. Petit dictionnaire du peuple a` l’usage des quatre cinquie`mes de la France. Paris: Chaumerot Jeune. Dessau, H. 1892–1914. Inscriptiones latinae selectae. 3 vols. Berlin: Weidmann. Deutschmann, Otto. 1971. Lateinisch und Romanisch. Munich: Hueber. Diehl, E. 1924–31. Inscriptiones Latinae Christianae Veteres. 3 vols. Berlin: Weidmann. Diez, Friedrich. [1836–43] 1856/58/60. Grammatik der romanischen Sprachen. 2nd edn. 3 vols. Bonn: Weber. Dinguirard, J.-C. 1979. “Observations sur le gascon des plus anciennes chartes”. Annales de la Faculte´ des Lettres de Toulouse-Le Mirail 15: 9–46. Doussinet, Raymond. 1971. Grammaire saintongeaise. La Rochelle: Rupella. Dressler, Wolfgang. 1965. “i-Prosthese vor s-impurum in Kleinasien (und in Vulgarlatein)”. Balkanso Ezikoznanie 9: 93–100. Dubois, Jacques. [= Iacopus Sylvius]. 1531. In linguam gallicam isagge, vna cum eiusdem Grammatica Latino-gallica, ex Hebræis, Græcis, & Latinis authoribus. Paris: R. Stephanus. [Repr. Geneva: Slatkine, 1971]. Dubuis, Roger. 1996. Lexique des “Cent nouvelles nouvelles”. Paris: Klincksieck.
260
Bibliography
Ducibella, Joseph W. 1934. The Phonology of the Sicilian Dialects. Washington DC: Catholic University of America. Durand, Jacques. 1990. Generative and Non-linear Phonology. London: Longman. Durand, Marguerite. 1945. “Quelques observations sur un exemple de parisien rural”. Le Franc¸ais Moderne 13: 83–91. Echenique Elizondo, Marı´a Teresa. 2004. “La lengua vasca en la historia lingu¨ı´stica hispa´nica”. In Cano (ed.), 59–80. Elwert, W. Theodor. 1972. Die Mundart des Fassa-Tals. Wiesbaden: F. Steiner. Erasmus, Desiderius. [1528]. In J. Kramer (ed.). Desiderii Erasmi Roterodami. De recta Latini Graecique sermonis pronuntiatione dialogus. 1978. Meisenheim am Glan: Anton Hain. Ernout, Alfred. 1957. “Exsto¯ et les compose´s latins en ex-”. In Alfred Ernout, Philologica.II. Paris: Klincksieck, 198–207. Ernst, Gerhard. 1985. Gesprochenes Franzo¨sisch zu Beginn des 17. Jahrhunderts. ZrP Beih. 204. Tu¨bingen: Niemeyer. Espinosa, Aurelio M. 1925. “Syllabic consonants in New Mexican Spanish”. Language 1: 109–18. Estienne, Henri. [1582] 1999. Hypomneses. Ed. J. Chomarat. Paris: Champion. Everett, Nicholas. 2003. Literacy in Lombard Italy, c. 568–774. Cambridge: Cambridge University Press. Ewen, Colin J. 1982. “The internal structure of complex segments”. In Hulst and Smith (eds), 27–67. Falcone, Giuseppe. 1976. Calabria. Profilo dei dialetti italiani, 18. Pisa: Pacini. Fermin, Maria H. J. 1954. Le vocabulaire de Bifrun dans sa trduction des quatre e´vangiles. Amsterdam: L. J. Veen. Fe´ry, Caroline and Vijver, Ruben van de (eds). 2003. The Syllable in Optimality Theory. Cambridge: Cambridge University Press. Fleischhacker, Heidi. 2001. “Cluster-dependent epenthesis asymmetries”. UCLA Working Papers in Linguistics 7: 71–116. Floricic, Franck. 2004. “A propos de certaines e´penthe`ses en sarde”. Cahiers de Grammaire 29: 59–87. Flutre, Louis-Fernand. 1955. Le parler picard de Mesnil-Martinsart (Somme). Geneva: Droz. —— 1977. Du moyen picard au picard moderne. Amiens: Muse´e de Picardie. Fouche´, Pierre. [1924] 1980a. Phone´tique historique du roussillonnais. Toulouse: Privat. Repr. Geneva: Slatkine. —— [1924] 1980b. Morphologie historique du roussillonnais. Toulouse: Privat. Repr. Geneva: Slatkine. —— 1966. Phone´tique historique du franc¸ais. III. Les consonnes. 2nd edn. Paris: Klincksieck. —— 1969. Phone´tique historique du franc¸ais. II. Les voyelles. 2nd edn. Paris: Klincksieck. Fougeron, Ce´cile. 2001. “Articulatory properties of initial segments in several prosodic constituents in French”. 29: 109–35.
Bibliography
261
Francard, Michel. 1980. Le parler de Tenneville. Introduction a` l’e´tude linguistique des parlers wallo-lorrains. Louvain-la-Neuve: Cabay. —— 1981. “Voyelles instables en wallon: propositions pour une approche globale”. Cahiers de l’Institut de Linguistique de l’Universite´ de Louvain 7: 169–200. Franc¸ois, Denise. 1974. Franc¸ais parle´. 2 vols. Paris: SELAF. Frati, Lodovico. 1900. La vita privata di Bologna dal secolo XIII al XVII. Bologna: Zanichelli. Fudge, Erik. 1969. “Syllables”. JL 5: 253–86. Gaeng, Paul. 1968. An Inquiry into Local Variations in Vulgar Latin as reflected in the Vocalism of Christian Inscriptions. Chapel Hill: University of North Carolina Press. Gaeta, Franco (ed.). 1981. Machiavelli. Lettere. Milann: Feltrinelli. Galme´s de Fuentes, A´lvaro. 1962. Las sibilantes en la Romania. Madrid: Gredos. —— 1983. Dialectologı´a moza´rabe. Madrid: Gredos. Galme´s, Salvador (ed.). 1935. Ramon Llull. Libre de Evast e Blanquerna. Vol. 1. Barcelona: Barcino. Garcı´a Arias, Xose´ Lluis. 1988. Contribucio´n a la grama´tica histo´rica de la lengua asturiana y a la caracterizacio´n etimolo´gica de su le´xico. Oviedo: Biblioteca de Filoloxı´a Asturiana. Garcı´a Rey, Verardo. 1934. Vocabulario del bierzo. Madrid: Centro de Estudios Histo´ricos. Garcı´a Turza, Claudio and Garcı´a Turza, Javier. 1998. “Los glosarios hispa´nicos: el manuscrito 46 de la Real Academia de la Historia”. In Actas del IV Congreso Internacional de Historia de la Lengua Espaola, vol. II: 939–60. Gartner, Theodor. 1883. Raetoromanische Grammatik. Heilbronn: Gebr. Henninger. —— 1892. “Die Mundart von Erto”. ZrP 16: 183–209, 308–71. —— 1912. Das Neue Testament des Jakob Bifrun. Gesellschaft fu¨r romanische Literatur 32, 2. Dresden: Niemeyer. Gasca Queirazza, Giuliano. 1995. “Piemont, Lombardei, Emilia-Romagna / Piemonte, Lombardia, Emilia e Romagna”. LRL II, 2: 98–111. Gaspari, Gianluigi and Tirondola, Giovanna. 1976. “Analisi dell’area vocalica nel linguaggio infantile dai due a quattro anni”. In R. Simone, U. Vignuzzi, and G. Ruggiero (eds). Studi di fonetica e fonologia. Rome: Bulzoni, 117–28. Gaudenzi, Augusto. 1889. I suoni, le forme e le parole dell’odierno dialetto della citta` di Bologna. Turin: Loescher. Gavel, Henri. 1920. E´le´ments de phone´tique basque. Paris-Biarritz: Champion. —— 1936. “Remarques sur les substrats ibe´riques, re´els ou suppose´s, dans la phone´tique du gascon et de l’espagnol”. RLiR 12: 36–43. Ghet¸ie, Ion. 1974. Inceputurile scrisului ˆın limba romaˆnaˇ. Bucharest: Editura Academiei Republicii Socialiste Romaˆnia. Giammarco, Ernesto. 1968–79. Dizionario abruzzese e molisano. 4 vols. Rome: Edizioni dell’Ateneo. Giannelli, Luciano. 2000. Toscana. Profilio dei dialetti italiani 9. Pisa: Pacini. Giese, W. 1965. “Urruma¨nisches anlautendes r- im Aromunischen”. In (no ed.) Omagui lui Alexandru Rosetti la 60 de ani. Bucharest: Editura Academiei Republicii Socialiste Romaˆnia, 299–301.
262
Bibliography
Gil, Juan. 2004. “El latı´n tardı´o y medieval (siglos VI–XIII)”. In Rafael Cano (ed.). Historia de la lengua espaola. Barcelona: Ariel, 149–82. Glatigny, M. 1989. “Norme et usage dans le franc¸ais du XVIe sie`cle”. In P. Swiggers and W. van Hoecke (dir.). La langue franc¸aise au XVIe sie`cle: usage, enseignement et approches descriptives. Louvain: Leuven University Press–Peeters, 7–31. Glessgen, Martin-Dietrich. 1989. Lo Thesaur del hospital de Sant Sperit. Edition eines Marseiller Urkundeninventars (1399–1511). ZrPh Beiheft 226. Tu¨bingen: Niemeyer. —— 1995. “Okzitanische Skriptaformen III. Provence, Dauphinois / Les scriptae occitanes III. Provence, Dauphinois”. LRL II, 2: 425–34. Go¨kc¸en, Adnan M. 1996. I volgari di Bonvesin da la Riva. New York: Lang. Goldsmith, John. 1990. Autosegmental and Metrical Phonology: An Introduction. Oxford: Blackwell. Gonza´lez Palencia, Angel. 1926–30. Los moza´rabes de Toledo en los siglos XII y XIII. 4 vols. Madrid: Instituto de Valencia de Don Juan. Gorra, Egidio. 1890. “Fonetica del dialetto di Piacenza”. ZrP 14: 133–58. —— 1892. “Il dialetto di Parma”. ZrP 16: 372–9. Gossen, Carl T. 1970. Grammaire de l’ancien picard. Paris: Klincksieck. Goudeket, Maurice (ed.). 1964. Lettres de la Princesse Palatine de 1672 a` 1722. Paris: Editions Le Club du Livre. Gouskova, Maria. 2001. “Falling sonority onsets, loanwords, and syllable contact”. CLS 37: 175–85. Grafstro¨m, A˚ke. 1958. E´tude sur la graphie des plus anciennes chartes languedociennes avec un essai d’interpre´tation phone´tique. Uppsala: Almqvist & Wiksell. Grammont, Maurice. 1894. “La loi des trois consonnes”. MSLP 8: 53–90. Granda, Germa´n de. 1966. La estructura sila´bica y su influencia en la evolucio´n fone´tica del dominio ibero-roma´nico. Madrid: CSIC. Grandgent, Charles H. 1907. An Introduction to Vulgar Latin. Boston-London: Heath. [1963. Introduccio´n al latı´n vulgar. Tr. F. B. de Moll. 3rd edn. Madrid: CSIC.] Gratwick, Adrian. 1967. “Ipsithilla: a vulgar name. Catullus, XXXII, 1”. Glotta 44: 174–6. Green, Antony D. 2003. “Extrasyllabic consonants and onset well-formedness”. In Fe´ry and Vijver, 238–53. Greenberg, Joseph H. 1978. “Some generalizations concerning initial and final consonant clusters”. In J. H. Greenberg (ed.). Universals of Human Language 2. Phonology. Stanford: Stanford University Press, 243–79. Gregory, Stewart. 1981. “Le dialecte wallon avant 1165”. TraLiLi 19: 7–51. —— 1982. “Quelques attestations de mots wallons au 12e sie`cle”. RLiR 45: 271–322. —— 1990. The Twelfth-Century Psalter Commentary for Laurette d’Alsace (an Edition of Psalms I-L). 2 vols. London: Modern Humanities Research Association. —— 1994. La traduction en prose franc¸aise du 12e sie`cle des Sermones in Cantica de saint Bernard. Amsterdam: Rodopi. Grignani, Maria Antonietta. 1980. “Testi volgari cremonesi del XV secolo”. SFI 38: 55–70.
Bibliography
263
Grisch, Mena. 1939. Die Mundart von Surmeir (Ober- und Unterhalbstein). Paris, Zurich, and Leipzig: Droz-Niehans. Gro¨ber, Gustav. 1878. “Gli, egli, ogni”. ZrPh 2: 594–600. Guarnerio, Pier Enea. 1906. “L’antico campidanese dei sec. XI-XIII secondo “le antiche carte volgari dell’Archivio arcivescovile di Cagliari””. Studi Romanzi 4: 189–259. —— 1918. Fonologia romanza. [repr. 1978]. Milan: Hoepli. Guiraud, Pierre. 1978. Le franc¸ais populaire. 4th edn. Paris: PUF. Hadlich, Roger L. 1965. The Phonological History of Vegliote. Chapel Hill: University of North Carolina Press. Haiman, John and Beninca`, Paola. 1992. The Rhaeto-Romance Languages. London: Routledge. Hajek, John and Goedemans, Rob W. N. 2003. “Word-initial geminates and stress in Patani Malay”. The Linguistic Review 20: 79–94. Hall, Robert A. 1964. “Initial consonants and syntactic doubling in West Romance”. Language 40: 551–6. Hall, T. A. 1997. The Phonology of Coronals. Amsterdam and Philadelphia: Benjamins. Ham, William. 2001. Phonetic and Phonological Aspects of Geminate Timing. New York and London: Routledge. Hammarstro¨m, Go¨ran. 1953. E´tude de phone´tique auditive sur les parlers de l’Algarve. Uppsala: Almqvist & Wiksell. Hammond, Robert M. 2000. “The phonetic realizations of /rr/ in Spanish—a psychoacoustic analysis”. In H. Campos et al. (eds). Hispanic Linguistics at the Turn of the Millennium. Somerville, MA: Cascadilla Press, 80–100. Harris, James W. 1983. Syllable Structure and Stress in Spanish. A Nonlinear Analysis. Cambridge, MA: MIT Press. Harris, Roy. 2000. Rethinking Writing. London: Athlone. Harris, William V. 1989. Ancient Literacy. Cambridge, MA: Harvard University Press. Hartman, Stephen L. 1986. “Learned word, popular words, and “first offenders””. In O. Jaeggli and C. Silva-Corvala´n (eds). Studies in Romance Linguistics. Dordrecht: Foris, 87–98. Haugen, Einar. [1966] 1972. “Dialect, language, nation”. In J. B. Pride and J. Holmes (eds). Sociolinguistics. Selected Readings. Harmondsworth: Penguin, 97–111. Haust, Jean. 1933. Dictionnaire lie´geois. Lie`ge: Vaillant-Carmanne, Imprimerie de l’Acade´mie. Helfenstein, Ruth, Keller, Deborah, and Kristol, Andres. 1993. “Bethmale: la fin d’une tradition”. In Wu¨est and Kristol, 83–108. Herculano de Carvalho, Jose´. 1958. Fonologia mirandesa. I. Coimbra: s.n. Herman, Jo´zsef. 1990. Du latin aux langues romanes. E´tudes de linguistique historique. Tu¨bingen: Niemeyer. Herzog, Eugen. 1904. Streitfragen der romanischen Philologie. I. Die Lautgesetzfrage zur franzo¨sischen Lautgeschichte. Halle: Niemeyer.
264
Bibliography
Hesseling, D. C. and Pernot, H. 1919. “E´rasme et les origines de la prononciation e´rasmienne”. Revue des E´tudes Grecques 32: 278–301. Hill, Archibald A. 1954. “Juncture and syllable division in Latin”. Language 30: 439–47. Hock, Hans Henrich. 1986. Principles of Historical Linguistics. Berlin: Mouton de Gruyter. —— and Joseph, Brian D. 1996. Language History, Language Change, and Language Relationship. Berlin and New York: Mouton de Gruyter. Hoenigswald, Henry M. 1949. “A note on Latin prosody: initial s impure after short vowel”. Transactions of the American Philological Association 80: 271–80. —— 1964. “Graduality, sporadicity, and the minor sound change processes”. Phonetica 11: 202–15. Holm, Catherine. 1992. “La structure syllabique dans l’histoire du franc¸ais”. In R. Lorenzo (ed.). Actas do XIX Congreso Internacional de Lingu¨´ıstica e Filoloxı´a Roma´nicas. A Corua: Fundacio´n “Pedro Barrie´ de la Maza, Conde de Fenosa”. Vol. 5, 171–82. Holmes, Urban T. 1935. “French words with e for o in unaccented initial syllables”. Language 11: 231–7. Holt, D. Eric (ed.). 2003. Optimality Theory and Language Change. Dordrecht: Kluwer. Holt, D. Eric. 2004. “Optimization of syllable contact in Old Spanish via the sporadic sound change metathesis”. Probus 16: 43–61. Hope, T. Edward. 1971. Lexical Borrowing in the Romance Languages. 2 vols. New York: New York University Press. Horning, Adolf. 1887. Die ostfranzo¨sischen Grenzdialekte zwischen Metz und Belfort. Franzo¨sische Studien 5, no. 4. Heilbronn: Gebr. Henninger. Huguet, E. 1925–67. Dictionnaire de la langue franc¸aise du seizie`me sie`cle. 7 vols. Paris: Didier. Hulst, Harry van der and Smith, N. (eds). 1982. The Structure of Phonological Representations. II. Dordrecht: Foris. —— and Ritter, Nancy (eds). 1999. The Syllable. Views and Facts. Berlin and New York: Mouton de Gruyter. Ive, A. 1886. “L’antico dialetto di Veglia”. AGI 9: 114–87. Jackson, Kenneth. 1953. Language and History in Early Britain. Edinburgh: Edinburgh University Press. Ja¨ggli, Peter. 1959. Die Mundart von Sennori. Zu¨rich: Juris-Verlag. Jennings, Augustus C. 1940. A Linguistic Study of the Cartulario de San Vincente [sic] de Oviedo. New York: Vanni. Jespersen, Otto. 1904. Lehrbuch der Phonetik. Leipzig and Berlin: Teubner. Joos, Martin. 1952. “The medieval sibilants”. Language 28: 222–31. Jungemann, Fredrick H. 1956. La teorı´a del sustrato y los dialectos hispano-romances y gascones. Madrid: Gredos. Kasten, Lloyd A. and Nitti, John J. (eds). 1978. Concordances and Texts of the Royal Scriptorium Manuscripts of Alfonso X, El Sabio. Madison: Hispanic Seminary of Medieval Studies.
Bibliography
265
—— 2002. Diccionario de la prosa castellana del Rey Alfonso X. 3 vols. New York: Hispanic Seminary of Medieval Studies. Kaye, Jonathan. 1992. “Do you believe in magic? The story of s+C sequences”. SOAS Working Papers in Linguistics and Phonetics 2: 292–313. Keating, Patricia, Cho, Taehong, Fougeron, Ce´cile, and Hsu, Chai-Shune. 2003. “Domaininitial articulatory strengthening in four languages”. In John Local, Richard Ogden, and Rosalind Temple (eds). Phonetic interpretation. Papers in Laboratory Phonology 6. Cambridge: Cambridge University Press, 145–63. Keller, Emil. 1934. Die Reimpredigt des Pietro da Barsegape´. 2nd edn. Frauenfeld: Huber. Kelly, Reine Cardaillac. 1973. A Descriptive Analysis of Gascon. The Hague: Mouton. Kenstowicz, Michael. 1994. Phonology in Generative Grammar. Oxford: Blackwell. —— 2003. “The role of perception in loanword phonology”. Studies in African Linguistics 32: 95–112. Kent, Roland G. 1945. The Sounds of Latin: A Descriptive and Historical Phonology. 3rd edn. Baltimore: Publications of the Linguistic Society of America. Kerswill, Paul and Williams, Ann. 2002. ‘“Salience” as an explanatory factor in language change: evidence from dialect levelling in urban England’. In Mari C. Jones and Edith Esch (eds). Language Change: The Interplay of Internal, External, and Extra-Linguistic Factors. Berlin and New York: Mouton de Gruyter, 81–110. King, Robert D. 1969. Historical Linguistics and Generative Grammar. Englewood Cliffs, NJ: Prentice Hall. Kiparsky, Paul. 1968. “How abstract is phonology?”. Indiana University Linguistics Club. Repr. in P. Kiparsky. 1982. Explanation in Phonology. Dordrecht: Foris, 119–63. Kiss, Sa´ndor. 1971. Les transformations de la structure syllabique en latin tardif. Debrecen: Kossuth Lajos Tudoma´nyegyetem. —— 1992. “La portion initiale de la syllabe en pre´roman. E´tude de phonologie diachronique”. Actas do XIX Congreso Internacional de Lingu¨´ıstica e Filoloxı´a Roma´nicas V, 71–8. Corua: Fundacio´n Barrie´ de la Maza. Conde de Fenosa. Kitto, Catherine and De Lacy, Paul. [1999] 2006. “Correspondence and epenthetic quality”. Proceedings of AFLA VI. Toronto: Toronto Working Papers in Linguistics [ROA #337], 181–200. Klob, Otto. 1901. “A Vida de Sancto Amaro. Texte portugais du XIVe sie`cle”. Romania 30: 504–18. Kohler, Klaus. 1967. “Modern English phonology”. Lingua 19: 145–76. Kontzi, Reinhold (ed.). 1982. Substrate und Superstrate in den romanischen Sprachen. Darmstadt : Wissenschaftliche Buchgesellschaft. Kramer, Johannes. 1977. Historische Grammatik des Dolomitenladinischen. Gerbrunn bei Wu¨rzburg: A Lehmann. Kro¨tsch, Monique. 2004. “La formation de groupes de consonnes au de´but des unite´s phoniques en franc¸ais parle´”. In T. Meisenburg and M. Selig (eds). Nouveaux de´parts en phonologie. Tu¨bingen: Gunter Narr, 217–33. Kuen, Heinrich. 1995. “Ladinisch / Le ladin”. LRL II, 2: 61–8.
266
Bibliography
Kurylowicz, Jerzy. 1966. “A problem of German alliteration”. In Mieczyslaw Brahmer, Stanislaw Helsztynski, and Julian Krzyzanowski, (eds). Studies in Language and Literature in Honour of Margaret Schlauch. Warsaw: PWN–Polish Scientific Publishers, 195–201. Labov, William. 1994. Principles of Linguistic Change. Internal Factors. Oxford: Blackwell. —— 2001. Principles of Linguistic Change. II. Social Factors. Oxford: Blackwell. Ladefoged, P. and Maddieson, I. 1996. The Sounds of the World’s Languages. Blackwell: Oxford. Laeufer, Christiane. 1991. “Syllabification and resyllabification in French”. In D. Wanner and D. A. Kibbee (eds). New Analyses in Romance Linguistics. Amsterdam: Benjamins, 19–36. Lalanne, Ludovic. [1880] 1970. Lexique des œuvres de Brantoˆme. Paris. Repr. Geneva: Slatkine. La Noue, Odet de. [1596] 1623. Dictionnaire des rimes franc¸oises. Cologny: M. Berjon. La Scala, Frank J. 1975. The Development of Prosthetic Vowels in Latin and the Romance Languages. Ph.D. dissertation. University of Pennsylvania. [Facsimile University Microfilms International 1981]. Lass, Roger. 1980. On Explaining Language Change. Cambridge: Cambridge University Press. —— 1984. Phonology. Cambridge: Cambridge University Press. Lausberg, Heinrich. 1963. Romanische Sprachwissenschaft. I. Einleitung und Vokalismus. 2nd edn. Berlin: De Gruyter. —— 1967. Romanische Sprachwissenschaft. II. Konsonantismus. 2nd edn. Berlin: De Gruyter. Le Blant, Edmond M. 1892. Nouveau recueil des inscriptions chre´tiennes de la Gaule ante´rieures au VIIIe sie`cle. Paris: Imprimerie nationale. Leite de Vasconcellos, Jose´. 1929a. “O dialecto mirande´s”. In Opu´sculos. IV.2. Filologia. Coimbra: Imprensa da Universidade, 679–722. —— 1929b. “Breve estudo dos falares de Riodonor e Guadramil”. In Opu´sculos. IV.2. Filologia. Coimbra: Imprensa da Universidade, 741–90. —— 1970. Esquisse d’une dialectologie portugaise. [Repr. of 1901 edn.]. Lisbon: Centro de Estudos Filolo´gicos. Leonard, Clifford S. 1978. Umlaut in Romance. Grossen-Linden: Hoffman. Le´on, Pierre R. 1992. Phone´tisme et prononciations du franc¸ais. Paris: Nathan. Leumann, M. 1977. Lateinische Laut- und Formenlehre. I. Munich: C. H. Beck’sche Verlagsbuchhandlung. [Rev. edn of Leumann-Hofmann-Szantyr, Lateinische Grammatik. I. 5th edn. 1926] Lindau, Mona. 1985. “The story of /r/”. In V. A. Fromkin (ed.). Phonetic Linguistics. Essays in Honor of Peter Ladefoged. Orlando: Academic Press, 157–68. Lindley Cintra, M. L. F. 1963a. “Les anciens textes portugais non litte´raires. Classement et bibliographie”. RLiR 27: 40–58.
Bibliography
267
—— 1963b. “Observations sur l’orthographe et la langue de quelques textes non litte´raires galiciens-portugais de la seconde moitie´ du XIIIe sie`cle”. RLiR 27: 59–77. Lindsay, W. M. 1894. The Latin Language. Oxford: Clarendon. Linnell, Per. 2005. The Written Language Bias in Linguistics. London: Routledge. Liver, Ricarda. 1982. Manuel pratique de romanche. Chur: Lia Rumantscha. Lloyd, Paul M. 1971. “A Note on Latin Syllable Structure”. Classical Philology 64: 41–2. Lodge, R. Anthony. 1993. French. From Dialect to Standard. London: Routledge. —— 1996. “Stereotypes of vernacular pronunciation in 17th-18th century Paris” ZrP 112: 205–31. —— 2004. A Sociolinguistic History of Parisian French. Cambridge: Cambridge University Press. Lo¨fstedt, Bengt. 1961. Studien u¨ber die Sprache der langobardischen Gesetze. Stockholm: Acta Universitatis Uppsala. Lombard, Alf. 1976. Le ˆı prosthe´tique du roumain. Acta Universitatis Upsaliensis Nova Series 2:5. Uppsala: Almqvist and Wiksell, 111–18. Lombardi, Lisa. 2003. “Markedness and the typology of epenthetic vowels”. Unpublished MS, University of Maryland. [ROA #578]. Longobardi, Monica. 1994. “Un frammento di ricettario del Trecento”. L’Archiginnasio 89: 249–78. Loporcaro, Michele. 1988. Grammatica storica del dialetto di Altamura. Pisa: Giardini. —— 1996. “On the analysis of geminates in Standard Italian and Italian dialects”. In B. Hurch and R. A. Rhodes (eds). Natural Phonology: The State of the Art. Berlin: Mouton de Gruyter, 153–87. —— 1997a. L’origine del raddoppiamento fonosintattico: saggi di fonologia diacronica romanza. Basle: Francke. —— 1997b. “On vowel epenthesis in Alguer Catalan”. In P. M. Bertinetto (ed.). Certamen Phonologicum III. Turin: Rosenberg & Sellier, 205–27. —— 1998. “Syllable structure and sonority sequencing. Evidence from Emilian”. In A. Schwegler, B. Tranel, and M. Uribe-Etxebarria (eds). Romance Linguistics. Theoretical Perspectives. Amsterdam: Benjamins, 155–70. —— 1999. “On possible onsets”. In Rennison and Ku¨hnhammer (eds), 133–51. —— 2005. “La sillabazione di muta cum liquida dal latino al romanzo”. In S. Kiss, L. Mondin, and G. Salvi (eds). Latin et langues romanes. Tu¨bingen: Niemeyer, 419–30. Lorck, J. E´tienne. 1893. Altbergamaskische Sprachdenkma¨ler (IX.-XV. Jahrhundert). Halle: Niemeyer. Lorenzo, Emilio. 1996. Anglicismos hispa´nicos. Madrid: Gredos. Lorenzo Va´zquez, Ramo´n. 2003. “El gallego en los documentos medievales escritos en latı´n”. In H. Perdiguero Villareal (ed.). Lengua Romance en textos latinos de la Edad Media. Burgos: Universidad de Burgos, 161–92. Loriot, Robert. 1984. Les parlers de l’Oise. Dijon-Amiens: Association bourguignonne de dialectologie et d’onomastique & Socie´te´ de linguistique picarde.
268
Bibliography
Lovera, L., Bettarini, R. and Mazzarello, A. (eds). 1975. Concordanza della Commedia di Dante Alighieri. 3 vols. Milan: Einaudi. Lowenstamm, Jean. 1981. “On the maximal cluster approach to syllable structure”. Linguistic Inquiry 12: 575–603. —— 1996. “Remarks on Mutae cum Liquida¯ and branching onsets”. In J. Durand and B. Laks (eds.). Current Trends in Phonology: Models and Methods. Salford: ESRI, University of Salford, 419–43. —— 1999. “The beginning of the word”. In Rennison and Kuhnhammer (eds), 153–66. Lucas A´lvarez, M. 1986. El tumbo de San Julia´n de Samos (siglos VIII–XII). Santiago de Compostela: Caixa Galicia. Luchaire, Achille. 1877. Les origines linguistiques de l’Aquitaine. Extrait du Bulletin de la Socie´te´ des Sciences, Lettres et Arts de Pau. Pau. —— 1879. E´tudes sur les idiomes pyre´ne´ens de la re´gion franc¸aise. [Repr. 1973. Geneva: Slatkine]. —— 1881. Recueil de textes de l’ancien dialecte gascon. Paris: Maisonneuve. Lu¨dtke, Helmut. 1957. “Sprachliche Beziehungen der apulischen Dialekte zum Roma¨nischen”. Revue des E´tudes Roumaines 3–4: 130–46. Lurati, Ottavio. 1988. “Lombardei und Tessin / Lombardia e Ticino”. LRL IV: 485–516. —— 2002. “La Lombardia”. In Cortelazzo et al., 226–60. Lutta, C. Martin. 1923. Der Dialekt von Bergu¨n. ZrP Beih. 71. Halle: Niemeyer. Lyche, Chantal. 1994. “Government phonology and s+C clusters in French”. In C. Lyche (ed.). French Generative Phonology: Retrospective and Perspectives. University of Salford, Salford: AFLS & ESRI, 259–75. Machado, Jose´ Pedro. 1977. Diciona´rio etimolo´gico da lı´ngua portuguesa. 5 vols. Lisbon: Livros Horizonte. McMahon, April M. S. 1994. Understanding Language Change. Cambridge: Cambridge University Press. —— 2003. “When history doesn’t repeat itself ”. In Holt (ed.), 121–42. Macinghi Strozzi, Alessandra. 1987. See Bianchini 1987. Maffei Bellucci, Patrizia. 1977. Lunigiana. Profilo dei dialetti italiani 9,1. Pisa: Pacini. Maia, Clarinda de Azevedo. 1975–8. “Os falares do Algarve”. RPF 17: 37–205. —— 1977. Os falares fronetiric¸os do concelho de Sabugal e da vizinha regia˜o de Xalma e Alamedilla. Coimbra: Instituto de Estudos Romaˆnicos. —— 1986. Histo´ria do Galego-Portugueˆs. Estudo lingu¨´ıstico da Galiza e do Noroeste de Portugal desde o se´culo XIII ao se´culo XVI. Coimbra: INIC. Maiden, Martin. 1995. A Linguistic History of Italian. London: Longman. Maiden, Martin and Parry, Mair (eds). 1997. The Dialects of Italy. London and New York: Routledge. Maiden, Martin and Robustelli, Cecilia. 2000. A Reference Grammar of Modern Italian. London: Arnold. Mainoldi, Pietro. 1967. Vocabolario del dialetto bolognese. Bologna: Forni.
Bibliography
269
Malagoli, Giuseppe. 1910–13. “Studi sui dialetti reggiani. Fonologia del dialetto di Novellara”. AGI 17: 29–146, 147–97. —— 1930. “Fonologia del dialetto di Lizzano in Belvedere (BO)”. ID 6: 125–96. —— 1934. “Studi sui dialetti reggiani: fonologia del dialetto di Vale`stra”. ID 10: 63–110. —— 1954. “Intorno ai dialetti dell’alta montagna reggiana”. ID 19: 111–42. Malkiel, Yakov. 1975. “Conflicting prosodic inferences from Ascoli’s and Darmesteter’s Laws ?”. RPh 28: 483–520. Maneca, Constant. 1965. “Considerazioni sopra la protesi vocalica in italiano”. Revue roumaine de linguistique 10: 499–507. Maniet, Albert. 1975. La phone´tique historique du latin. Paris: Klincksieck. Maragliano, A. 1976. Dizionario dialettale vogherese. Bologna: Patron. Marchello-Nizia, Christiane. 1995. L’e´volution du franc¸ais. Paris: Armand Colin. Marshall, M. M. 1984. The Dialect of Notre-Dame-de-Sanilhac. Saratoga, CA: ANMA Libri. Marotta, Giovanna. 1999. “The Latin syllable”. In Hulst and Ritter, 285–310. Marrese, Maria (ed.). 1996. Lettere. Marco Parenti. Florence: Olschki. Martinet, Andre´. 1952. “Celtic lenition and western Romance consonants”. Language 28: 192–217. [Cf. Martinet 1955 [1964]: 257–96]. —— [1955] 1964. E´conomie des changements phone´tiques. Traite´ de phonologie diachronique. 2nd edn. Berne: Francke. —— 1956. Description phonologique, avec application au parler franco-provenc¸al d’Hauteville (Savoie). Geneva: Droz. —— 1969. “Qu’est-ce que le “e muet””. In id. Le franc¸ais sans fard. Paris: PUF, 209–19. Mateus, Maria H. and d’Andrade, Ernesto. 2000. The Phonology of Portuguese. Oxford: Oxford University Press. Mayerthaler, Eva. 1982. Unbetonter Vokalismus und Silbenstruktur im Romanischen. Tu¨bingen: Niemeyer. McCarthy, John J. (ed.). 2004. Optimality Theory in Phonology. A Reader. Oxford: Blackwell. McKenzie, Kenneth. 1912. Concordanza delle rime di Francesco Petrarca. Oxford and New Haven: Oxford University Press and Yale University Press. McMahon, April. 2000. Change, Chance, and Optimality. Oxford: Oxford University Press. Meillet, Antoine. 1927. “De la prothe`se vocalique en grec et arme´nien”. BSLP 27: 129–35. Melillo, A. M. 1977. Corsica. Profilo dei dialetti italiani 21. Pisa: Pacini. Mene´ndez Pidal, Ramo´n. 1964a. Orı´genes del espaol. 5th edn. Madrid: Espasa-Calpe. —— 1964b. Cantar de Mio Cid. 3 vols. 4th edn. Madrid: Espasa-Calpe. —— 1966. Manual de grama´tica histo´rica espaola. 12th edn. Madrid: Espasa-Calpe. Merci, Paolo. 1992. Il Condaghe di San Nicola di Trullas. Sassari: Delfino. Merlo, Clemente. 1959. “Contributi alla conoscenza dei dialetti della Liguria odierna”. In Clemente Merlo. Saggi linguistici. Pisa: Pacini-Mariotti, 127–60. Meyer-Lu¨bke, Wilhelm. 1890. Grammaire des langues romanes. I. Phone´tique. Paris: Welter.
270
Bibliography
—— 1929. “Unterschicht und Oberschicht und der Lautwandel”. In Behrens-Festschrift. Supplementheft der Zeitschrift fu¨r franzo¨sische Sprache und Litteratur 13. Jena-Leipzig: Gronau, 16–36. Michel, Franciscus. 1860. Libri Psalmorum. Versio Antiqua Gallica. Oxford: Clarendon. Michel, Louis. 1948. “La vocalisation de l’S dans l’Aude”. RLaR 70: 29–38. —— 1956. E´tude du son “s” en latin et en roman. Montpellier: PUF. Michelena, Luis. 1990. Fone´tica histo´rica vasca. [Repr. of 2nd edn. 1977]. San Sebastia´n: Diputacio´n Foral de Gipuzkoa. Migliorini, Bruno. 1984. The Italian Language. Rev. edn. London: Faber and Faber. [tr. and recast by T. G. Griffith from Storia della lingua italiana. 2nd edn. 1960. Florence: Sansoni]. Mignani, Rigo (ed.). 1979. El Conde Lucanor. Florence: Licosa Ed. Mihaˇescu, Haralambie. 1978. La langue latine dans le sud-est de l’Europe. Bucharest and Paris: Ed. Academiei-Les Belles Lettres. Mihaˇilaˇ, Gheorghe. 1974. Dict¸ionar al limbii romaˆne vechi. Bucharest: Editura enciclopedicaˇ romaˆnaˇ. Milanesi, Gaetano (ed.). 1875. Le lettere di Michelangelo Buonarroti. Florence: Le Monnier. Millardet, Georges. 1909. “Le domaine gascon (jusqu’en 1907)”. Revue de Dialectologie Romane 1: 122–56. —— 1910. E´tude de dialectologie landaise. Toulouse: Privat. Molho, Maurice. 1961. “Les home´lies d’Organya`”. Bulletin Hispanique 63: 186–210. Moll, Francisco B. de. 1952. Grama´tica histo´rica catalana. Madrid: Gredos. Monfrin, Jacques. 2001. “Humanisme et traductions au Moyen-Aˆge”. In Jacques Monfrin. E´tudes de philologie romane. Paris: Droz, 757–85. Montreuil, Jean-Pierre. 1986. “Null segments in Romance”. In O. Jaeggli and C. SilvaCorvalan (eds). Studies in Romance Linguistics. Dordrecht: Foris, 265–81. —— 2000. “Sonority and derived clusters in Raeto-Romance and Gallo-Italic”. In L. Repetti (ed.). Phonological Theory and the Dialects of Italy. Amsterdam and Philadelphia: Benjamins, 211–37. Morelli, Frida. 2003. “The relative harmony of /s+stop/ onsets: obstruent clusters and the Sonority Sequencing Principle”. In Fe´ry and Vijver, 356–71. Morin, Yves Charles. 1979. “La morphophonologie des pronoms clitiques en franc¸ais populaire”. Cahier de Linguistique 9: 1–36. —— 2003a. “Remarks on prenominal liaison consonants in French”. In Ploch (ed.), 385–400. —— 2003b. “Le statut linguistique du chva ornemental dans la poe´sie et la chanson franc¸aises”. In J.-L. Aroui (ed.). Le sens et la mesure: de la pragmatique a` la me´trique. Hommage a` Benoıˆt Cornulier. Paris: Champion, 459–98. Morosi, G. 1888. “L’odierno linguaggio dei valdesi del Piemonte”. AGI 11: 309–415. Mott, Brian. 1989. El habla de Gistaı´n. Huesca: Diputacio´n Provincial. Munthe, A˚ke W. 1887. Anteckningar om folkma˚let i en trakt af vestra Asturien. Uppsala: Almqvist & Wiksell.
Bibliography
271
Mushacke, Wilhelm. 1884. Geschichtliche Entwicklung der Mundart von Montpellier (Languedoc). Franzo¨sische Studien 4, Schlussheft 5. Heilbronn: Gebr. Henninger. Mussafia, Adolf. 1873. Beitrag zur Kunde der norditalienischen Mundarten im XV. Jahrhunderte. Denkschriften der philosophisch-historischen Classe der kaiserlichen Akademie der Wissenschaften, 22. Vienna: Kaiserliche Akademie der Wissenschaften, 103–228. Nandris¸, Octave. 1963. Phone´tique historique du roumain. Paris: Klincksieck. Nauton, Pierre. 1974. Ge´ographie phone´tique de la Haute-Loire. Paris: Belles Lettres. Nebbia, Sergio. 2001. Dizionario Monferrino. Savigliano: Ed. Artistica Piemontese. Nebrija, Antonio de. [1481] 1981. Introductiones Latinae. Ed. E. de Bustos. Salamanca: Universidad de Salamanca. —— [1492] 1946. Gramatica Castellana. Ed. Galindo Romeo and L. Ortiz Muoz. 2 vols [critical edition and facsimile]. Madrid: Ed. de la Junta del Centenario. —— [1517] 1957. Reglas de orthographia en la lengua castellana. Ed. A. de Quilis. Bogota´: Publicaciones del Instituto Caro y Cuervo. Nespor, Marina and Vogel, Irene. 1986. Prosodic Phonology. Dordrecht: Foris. Niedermann, Max. 1954. Recueil Max Niedermann. Neuchaˆtel: Faculte´ des Lettres, Universite´ de Neuchaˆtel. Nieri, Idelfonso. 1902. Vocabolario lucchese. Memorie e documenti per servire alla storia di Lucca, 15. Lucca: Tipografia Giusti. Nigra, Costantino. [1901] 1973. “Il dialetto di Viverone”. In Miscellanea linguistica in onore di Graziadio Ascoli. Turin. Repr. Geneva: Slatkine, 247–62. Nisard, Charles. 1872. E´tude sur le langage populaire en patois de Paris et de sa banlieue. Paris: A. Franck. Noske, Roland. 1982. “Syllabification and syllable changing rules in French”. In Hulst and Smith (eds), 257–310. Nyrop, Kristoffer. [1899] 1935. Grammaire historique de la langue franc¸aise. 4th edn. Copenhagen: Gyldendalske Boghandel. Ohala, John J. 1990. “Alternatives to the sonority hierarchy for explaining segmental sequential constraints”. Chicago Linguistic Society 26, vol. 2: 319–38. Omeltchenko, Stephen W. 1977. A Quantitative and Comparative Study of the Vocalism of the Latin Inscriptions of North Africa, Britain, Dalmatia, and the Balkans. Chapel Hill: University of North Carolina Press. Orazi, Veronica. 1997. El dialecto leone´s antiguo. Madrid: Univ. Europea-CEES Ediciones. Ott, Wilhelm. 1973–85. Metrische Analysen zu Vergil Aeneis. 12 vols. Tu¨bingen: Niemeyer. Palay, Simin. 1971. Dictionnaire du Be´arnais et du Gascon modernes. Paris: CNRS. Palsgrave, John. 1530. Lesclarcissement de la langue francoyse. In F. Genin (ed.). Documents ine´dits sur l’histoire de France. Deuxie`me se´rie. Histoire des lettres et des sciences. Paris: Imprimerie nationale, 1852. [Facsimile and French translation and notes: Susan Baddeley (ed.). 2003. L’e´claircissement de la langue franc¸aise. Paris: Champion]. Paoli, Cesare and Piccolomini, Enea. 1871. Lettere volgari del secolo XIII scritte da Senesi. Bologna: G. Romagnoli.
272
Bibliography
Papahagi, Tache. 1974. Dict¸ionarul dialectului aromaˆn. 2nd edn. Bucharest: Ed. Acad. Republ. Soc. Romaˆnia. Paradis, C. and Prunet, J.-F. (eds). 1991. Phonetics and Phonology. The Special Effect of Coronals: Internal and External Evidence. San Diego: Academic Press. Parenti, Marco. 1996. See Marrese 1996. Pariente, Angel. 1968. “El problema de las formas stlocus, stlis, stlat(t)a y stlatarius”. Emerita 36: 247–69. Parry, Mair. 2005. Sociolinguistica e grammatica del dialetto di Cairo Montenotte. Savona: Societa` Savonese di Storia Patria. Passy, Paul. 1891. “Patois de Sainte-Jamme (Seine-et-Oise)”. Revue des patois gallo-romans 4: 7–16. Peer, Oscar. 1962. Dicziunari rumantsch ladin - tudais-ch. Samedan: Lia Rumantscha. Pei, Mario A. 1932. The Language of the Eighth-Century Texts in Northern France. New York: Carranza & Co. Peixoto de Fonseca, F. V. 1978. “Prothe`se, e´penthe`se et e´pithe`se en ancien portugais”. RLiR 42: 53–5. Peletier, Jacques. 1550. Dialogue de l’ortografe e prononciation francoese. [Facsimile: Geneva: Slatkine 1964]. Poitiers: Ian & Enguilbert de Marnef. Penny, Ralph. 2002. A History of the Spanish Language. 2nd edn. Cambridge: Cambridge University Press. Pensado, Carmen. 1985. “On the interpretation of the non-existent: nonoccurring syllable types in Spanish phonology”. Folia Linguistica 19: 313–20. —— 2006. “Existio´ alguna vez la “variacio´n del romance occidental””. RLiR 70: 5–19. Petrocchi, Policarpo. 1966–7. La Divina Commedia. 2 vols. Societa` Dantesca Italiana. Milan: Mondadori. Petrovici, Emil. 1930. De la nasalite´ en roumain. Recherches expe´rimentales. Cluj: Institutul de arte graficaˇ “Ardealul”. —— 1957. Kann das Phonemsystem einer Sprache durch fremden Einfluß umgestellt werden ?. The Hague: Mouton. Piccitto, Giorgio. 1977–2002. Vocabolario siciliano. 5 vols. Catania: Centro di Studi Filologici e Linguistici Siciliani. Pieri, Silvio. 1890–2. “Fonetica del dialetto lucchese”. AGI 12: 107–34. Pieros, Carlos-Eduardo. 2005. “Syllabic-consonant formation in Traditional New Mexican Spanish”. Probus 17: 253–301. Pirandello, Luigi. [1891] 1973. Laute und Lautentwickelung der Mundart von Girgenti. Halle: Buchdruckerei des Waisenhauses. Repr. Pisa: Edizioni Marlin. Pittau, Massimo. 1972. Grammatica del sardo-nuorese. Bologna: Patron. Ple´nat, Marc. 1993. “Observations sur le mot minimal franc¸ais. L’oralisation des sigles”. In B. Laks and M. Ple´nat (eds). De Natura Sonorum. Essais de phonologie. Paris: Presses Universitaires de Vincennes, 143–72. Ploch, Stefan (ed.). 2003. Living on the Edge. 28 Papers in Honour of Jonathan Kaye. Berlin and New York: Mouton de Gruyter.
Bibliography
273
Plomteux, Henri. [1975] 1981. I dialetti delle Liguria orientale odierna, la val Graveglia. Bologna: Patron. Politzer, Robert L. 1949. “A study of the language of the eighth-century Lombardic documents”. Ph.D. dissertation. New York: Columbia University. —— 1959. “A note on the distribution of prothesis in Late Latin”. Modern Language Notes 74: 31–7. —— 1961. “The interpretation of correctness in Late Latin texts”. Language 37: 209–14. —— 1967. Beitrag zur Phonologie der Nonsberger Mundart. Romanica Aenipontana 6. Innsbruck: Institut fu¨r Romanische Philologie der Leopold-Franzens-Universita¨t. Politzer, Frieda N. and Politzer, Robert L. 1953. Romance Trends in 7th and 8th-Century Documents. Chapel Hill: University of North Carolina Press. Pop, Sever. 1966. “Romania e Sardegna: rapporti linguistici”. Acta Philologica 4: 547–85. Pope, Mildred K. 1952. From Latin to Modern French. 2nd edn. Manchester: Manchester University Press. Posner, Rebecca. 1961. Consonantal Dissimilation in the Romance Languages. Oxford: Blackwell. —— 1996. The Romance Languages. Cambridge: Cambridge University Press. Price, P. J. 1980. “Sonority and syllabicity: acoustic correlates of perception”. Phonetica 37: 327–43. Prince, Alan and Smolensky, Paul. [1993] 2004. “Optimality Theory: Constraint interaction in Generative Grammar”. In McCarthy, 2004, 3–71. Prinz, Otto. 1938. “Zur Entstehung der Prothese vor s-impurum im Lateinischen”. Glotta 26: 97–115. Prou, Maurice. 1892. Les monnaies me´rovingiennes. Paris: Rollin & Feuardent. Pulgram, Ernst. 1958. The Tongues of Italy. Cambridge, MA: Harvard University Press. Pult, Gaspar. 1897. Le parler de Sent (Basse-Engadine). Lausanne: Pache. Rafel i Fontanals, J. 1980. “Sobre el benasque`s”. In J. Bruguera (ed.). Actes del cinque col. loqui internacional de llengua i literatura catalanes. Montserrat: Publicacions de l’Abadia de Montserrat. Reichenkron, Gu¨nter. 1965. Historische Latein-altromanische Grammatik. I. Einleitung: das sogenannte Vulga¨rlatein und das Wesen der Romanisierung. Wiesbaden: Harassowitz. Reighard, John. 1985. “La ve´larisation de l’
274
Bibliography
Restaut, Pierre. [1730] 1773. Principes ge´ne´raux et raisonne´s de la grammaire franc¸oise. 10th edn. Paris: Lottin. Re´zeau, Pierre. 1976. Un patois de Vende´e. Le parler rural de Vouvant. Paris: Klincksieck. Richter, Elise. 1934. Beitra¨ge zur Geschichte der Romanismen. I: Chronologische Phonetik des Franzo¨sischen bis zum Ende des 8. Jahrhunderts. [ZrP Beih. 82]. Halle: Niemeyer. Rickard, Peter. 1976. Chrestomathie de la langue franc¸aise au quinzie`me sie`cle. Cambridge: Cambridge University Press. Rohlfs, Gerhard. 1966/68/69. Grammatica storica della lingua italiana e dei suoi dialetti. 3 vols. I. Fonetica, II. Morfologia, III. Sintassi e formazione delle parole. Turin: Einaudi. —— 1970. Le gascon. E´tudes de philologie pyre´ne´enne. ZrP Beih. 85. 2nd edn. Tu¨bingen: Niemeyer. —— 1977. Nuovo dizionario dialettale della Calabria. 2nd edn. Ravenna: Longo. Ronjat, Jules. 1930–41. Grammaire istorique des parlers provenc¸aux modernes. 4 vols. I. Fone´tique (Vowels) 1930, II. Fone´tique (Consonants) 1932, III. IV. Montpellier: Socie´te´ des Langues Romanes. Ro¨nsch, Hermann. [1868] 1965. Itala und Vulgata. Munich: Max Hueber. Rosati, Valeria (ed.). 1977. Le lettere di Margherita Datini a Francesco (1384–1410). Biblioteca dell’Archivio storico pratese, 2. Prato: Cassa di Risparmi e Depositi. Rose, Yvan and Demuth, Katherine. 2006. “Vowel epenthesis in loanword adaptation: representational and phonetic considerations”. Lingua 116: 1112–39. Rosetti, Alexandru. 1978. Istoria limbii romaˆne. 2nd edn. Bucharest: Editura s¸tiint¸ificaˇ s¸i enciclopedicaˇ Rousselot, abbe´ Pierre. 1893. “Les modifications phone´tiques du langage e´tudie´es dans le patois d’une famille de Cellefrouin (Charente)”. Revue des patois gallo-romans 5: 209–386. (One of four parts, the others being 4: 65–208, 6: 9–48, 65–208). Rubach, Jerzy and Booij, Geert E. 1990. “Edge of constituent effects in Polish”. Natural Language and Linguistic Theory 8: 427–63. Rupp, Theodor. 1963. Lautlehre der Mundarten von Domat, Trin und Flem. Chur: Sulser. Russell-Gebbett, Paul. 1965. Medieval Catalan Linguistic Texts. Oxford: Dolphin. Sala, Marius. 1976. Contributions a` la phone´tique historique du roumain. Paris: Klincksieck. Salvador Galme´s, Mn. 1935. Libre de Evast e Blanquerna. Barcelona: Ed. Barcino. Salvioni, Carlo. 1886. “Saggi intorno ai dialetti di alcune vallate all’estremita` settentrionale del Lago Maggiore”. AGI 9: 188–234, 249–59. Sampson, Rodney. 1980. Early Romance Texts. An Anthology. Cambridge: Cambridge University Press. —— 1985. “The pattern of evolution of Proto-Romanian //”. Revue roumaine de linguistique 30: 327–59. —— 1999. Nasal Vowel Evolution in Romance. Oxford: Oxford University Press. —— 2004a. “Henri Estienne and vowel prosthesis: a problem in the phonetic adaptation of 16th-century Italianisms”. French Studies 58: 327–41. —— 2004b. “Tendances et contretendances dans la structuration de la syllabe en protoroman”. Aemilianense 1: 481–500.
Bibliography
275
—— 2005. “Vowel prosthesis and its maintenance in Spanish: a comparative perspective”. In Roger Wright and Peter Ricketts (eds). Studies on Ibero-Romance Linguistics Dedicated to Ralph Penny. Newark, DE: Juan de la Cuesta, 241–58. —— 2006. “L’e´volution de la voyelle accentue´e des formes tinto, pinta, punto, unto, etc. en castillan”. RLiR 70: 21–40. Sanga, Glauco. 1995. “Italienische Koine / La koine` italiana”. LRL II, 2: 81–98. Santos Silva, Maria Helena. 1961. “Caracterı´sticas fone´ticas do falar minhoto”. BF 20: 309–21. Saralegui, Carmen. 1977. El dialecto navarro en los documentos des monasterio de Irache (958–1397). Pamplona: Diputacio´n Foral de Navarra & CSIC. Sarrieu, B. 1902–6 “Le parler de Bagne`res-de-Luchon et de sa valle´e”. RLaR 5: 385–446; 6: 317–98; 7: 97–153, 481–534; 9: 5–48, 465–94. Saussure, Ferdinand de. [1916] 1974. Cours de linguistique ge´ne´rale. E´dition critique pre´pare´e par Tullio de Mauro. Geneva: Payot. Scha¨del, Bernd. 1908. “La frontie`re entre le gascon et le catalan”. Romania 37: 140–56. Schane, Sanford A. 1968. French Phonology and Morphology. Cambridge, MA: MIT Press. Schiaffini, Alfedo. 1926. Testi fiorentini del Dugento e dei primi del Trecento. Florence: Sansoni. —— 1928. “Influssi dei dialetti centro-meridionali sul toscano e sulla lingua letteraria”. Italia Dialettale 4: 77–129. —— 1961. I mille anni della lingua italiana. Milan: All’Insegna del Pesce d’Oro. Schiaparelli, Luigi. 1929–33. Codice diplomatico lombardo. 2 vols. Rome: Tipografia del Senato. Schizzerotto, Giancarlo. 1985. Sette secoli di volgare e di dialetto mantovano. Mantua: Publi-Paolini Ed. Schlo¨sser, Rainer. 1985. Historische Lautlehre des Aromunischen von Metsovon. Hamburg: Buske. Schmitt, Christian. 1984. “Varie´te´ et de´veloppement linguistiques. Sur les tendances e´volutives en franc¸ais moderne et en espagnol”. RLiR 48: 397–437. Schneegans, Heinrich. 1888. Laute und Lautentwickelung des sicilianischen Dialectes. Strasbourg: Tru¨bner. Scho¨nthaler, Willy. 1937. Die Mundart des Bethmale-Tales. Tu¨bingen: Go¨bel. Schorta, Andrea. 1938. Lautlehre der Mundart von Mu¨stair. Paris, Zurich, and Leipzig: Droz-Niehans. Schrijver, P. 1995. Studies in British Celtic Historical Phonology. Amsterdam and Atlanta, GA: Rodopi. Schneegans, Heinrich. 1888. Laute und Lautentwickelung der sicilianischen Dialectes. Strassburg: Tru¨bner. Schortz, Miche`le. 1998. Le parler de Senneville-sur-Fe´camp. Uppsala: Uppsala University Press. Schuchardt, Hugo. 1866/67/68. Der Vokalismus des Vulga¨rlateins. 3 vols. Leipzig: Teubner. Schu¨rr, Friedrich. 1918–19. Romagnolische Dialektstudien. 2 vols. Vienna: Ho¨lder.
276
Bibliography
Schwarze, Christoph. 1963. Der altprovenzalische “Boeci”. Mu¨nster: Aschendorffsche Verlagsbuchhandlung. Seelmann, Emil. 1885. Die Aussprache des Latein nach physiologisch-historische Grundsa¨tzen. Heilbronn: Henninger. Se´guy, Jean. [1950] 1978. Le franc¸ais parle´ a` Toulouse. Toulouse: Privat. Seklaoui, Diana R. 1989. Change and Compensation. Parallel Weakening of [s] in Italian, French and Spanish. New York, Berne, Frankfurt, and Paris: Lang. Selkirk, Elizabeth O. 1982. “The syllable”. In Hulst and Smith (eds), 337–83. Serianni, Luca. 1977. Il dialetto pratese nel Medioevo. Prato: Cassa di Risparmi e Depositi. —— and Castelvecchi, Alberto. 1988. Grammatica Italiana. Italiano comune e lingua letteraria. Turin: UTET Sgrilli, Paola (ed.). 1983. Il “Libro di Sidrac” salentino. Pisa: Pacini. S¸iadbei, T. 1958. “La prothe`se vocalique dans les langues romanes”. Revue de Linguistique 3: 153–63. Sievers, E. [1881]1901. Grundzu¨ge der Phonetik. 5th edn. Leipzig: von Breitkopf & Ha¨rtel. Simon, Hans Joachim. 1967. Beobachtungen an Mundarten Piemonts. Heidelberg: Winter. Singh, Rajendra. 1985. “Prosodic adaptation in interphonology”. Lingua 67: 269–82. Skytte, Gunver. 1975. Italiensk Fonetik. Odense: Odense Universitetsforlag. Smith, Colin. 1983. “Vulgar Latin in Roman Britain: epigraphic and other evidence”. In H. Temporini and W. Haase (eds). Aufstieg und Niedergang der ro¨mischen Welt. II, 2. Berlin and New York: de Gruyter, 893–948. Sole´, Maria-Josep. 2002a. “Aerodynamic characteristics of trills and phonological patterning”. Journal of Phonetics 30: 655–88. —— 2002b. “Assimilatory processes and aerodynamic factors”. In C. Gussenhoven and N. Warner (eds). Laboratory Phonology 7. Berlin and New York: Mouton de Gruyter, 351–86. Solmi, Arrigo. 1905. “Le carte volgari dell’Archivo Arcivescovile di Cagliari”. Archivio Storico Italiano 35: 273–330. Spence, Nicol. 1990. ““Sporadic” changes in Jersey French”. In J. N. Green and W. AyresBennett (eds). Variation and Change in French. London: Routledge, 210–25. Spoerri, Teofilo. 1918. “Il dialetto della Valsesia”. Rendiconti dell’Istituto Lombardo. Serie II. 51: esp. 391–409, 683–98. Staaff, Erik. 1907. E´tude de l’ancien dialecte le´onais d’apre`s les chartes du XIIIe sie`cle. Uppsala: Almqvist & Wiksell. Stella, Angelo. 1969. “Tre lettere bolognesi del secolo XIV”. LN 30: 51–2. Steriade, Donca. 1988. “Gemination and the Proto-Romance syllable shift”. In D. Birdsong and J.-P. Montreuil (eds). Advances in Romance Linguistics. Dordrecht: Foris, 371–409. —— 1994. “Complex onsets as single segments: the Mazateco pattern”. In J. Cole and C. Kisseberth (eds). Perspectives in Phonology. Stanford, CA: CSLI, 202–91. Stotz, Peter. 1996–2004. Handbuch zur lateinischen Sprache des Mittelalters. 5 vols. Munich: C. H. Beck’sche Verlagsbuchhandlung. [Esp. vol. 3. Lautlehre. 1996]. Straka, Georges. 1979. Les sons et les mots: choix d’e´tudes de phone´tique et de linguistique. Paris: Klincksieck.
Bibliography
277
Stussi, Alfredo. 1965. Testi veneziani del Duecento e dei primi del Trecento. Pisa: NistriLischi. —— 1982. “Antichi testi salentini in volgare”. In Alfredo Stussi. Studi e documenti di storia della lingua e dei dialetti italiani. Bologna: Mulino, 155–81. Svenson, Lars-Owe. 1959. Les parlers du Marais vende´en. 2 vols. Gothenburg: Almqvist & Wiksell. Sylvius, Iacobus. (= Jacques Dubois). 1531. In linguam gallicam isagge…Paris: R. Stephanus. [Repr. Geneva: Slatkine, 1971]. Taboada, Manuel. 1979. El habla del Valle de Verı´n. Verba, Anejo 15. Compostela: Universidad de Santiago de Compostela. Tabourot, Etienne. 1588. Les bigarrures du Seigneurs des Accords. [Repr. Geneva: Slatkine, 1969]. Tardif, Jules. 1866. Fac-simile´ de chartes et diploˆmes me´rovingiens et carlovingiens. Paris: J. Claye. Taylor, R. 1965. “Les ne´ologismes chez Nicole Oresme traducteur du XIVe sie`cle”. Actes du Xe Congre`s International de Linguistique et Philologie Romanes vol. 2. Paris: Klincksieck, 727–36. Tekavcˇic´, Pavao. 1972. Grammatica storica dell’italiano. I: Fonematica. Bologna: Il Mulino. Telmon, Tullio. 1975. “La prosthe`se vocalique dans les parlers du Pie´mont”. RLiR 39: 122–71. Ternes, Elmar. 1986. “A grammatical hierarchy of joining”. In Andersen (ed.), 11–21. Teyssier, Paul. 1980. Histoire de la langue portugaise. Paris: Presses Universitaires de France. Thomason, Sarah G. 2001. Language Contact. An Introduction. Edinburgh: Edinburgh University Press. —— and Kaufman, Terrence. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley, Los Angeles, and Oxford: University of California Press. Thurot, Charles. 1881. De la prononciation franc¸aise depuis le commencement du XVIe sie`cle. 2 vols. Paris: Welter. [Repr. Geneva: Slatkine, 1966]. Tja¨der, Jan-Olof. 1955. Die nichtliterarischen Papyri Italiens aus der Zeit 445–700, I. Lund: Gleerup. Tomasini, Giulio. 1951. “Minime di grammatica dialettale. La vocale prostetica nel Trentino”. Studi Trentini di Scienze Storiche 30: 115–17. Tomasoni, Piera. 1985. “L’antica lingua non letteraria a Bergamo. Un formulario notarile inedito del secolo XV”. In G. Vitali and G. O. Bravi (eds). Lingue e culture locali. Le ricerche di Antonio Tiraboschi. Bergamo: Lubrina, 229–61. Topintzi, Nina. 2006. “Moraic onsets”. Unpub. Ph.D. dissertation. University College London. Toppino, Giuseppe. 1902–5. “Il dialetto di Castellinaldo”. AGI 16: 517–48. Traina, Alfonso. 1973. L’alfabeto e la pronunzia del latino. 4th edn. Bologna: Patron. Trask, R. Larry. 1996. Historical Linguistics. London and New York: Arnold. —— 1997. The History of Basque. London: Routledge. Tschirch, Fritz. 1966–9. Geschichte des deutschen Sprache. 2 vols. Berlin: Erich Schmidt.
278
Bibliography
Turchi, Laura and Bertinetto, Pier Marco. 2000. “La durata vocalica di fronte ai nessi /sC/: un’indagine su soggetti pisani”. Studi Italiani di Linguistica Teorica e Applicata 29: 389–421. Uffmann, Christian. 2006. “Epenthetic vowel quality in loanwords: empirical and formal issues”. Lingua 116: 1079–111. Ulrich, Jacob. 1885–6. “Sacra rappresentazione del secolo XVII, testo ladino, varieta` di Bravugn”. AGI 8: 263–303, 9: 107–14. Va¨a¨na¨nen, Veikko. 1966. Le latin vulgaire des inscriptions pompe´iennes. 3rd edn. Berlin: Akademie-Verlag. Valde´s, Juan de. [1535] 1967. Dia´logo de la lengua. Ed. C. Barbolani de Garcı´a. MessinaFirenze: G. D’Anna. Valkhoff, Marius. 1931. E´tude sur les mots franc¸ais d’origine ne´erlandaise. Amersfoort: Valkhoff and Co. Vanelli, Laura. 1984. “Pronomi e fenomeni di prostesi vocalica nei dialetti italiani settentrionali”. RLiR 48: 281–95. —— 1987. “I pronomi soggetto nei dialetti italiani settentrionali dal Medioevo a oggi”. Medioevo Romanzo 13: 173–211. Varvaro, Alberto. 1984. “La situazione linguistica della Sicilia nel basso Medioevo”. In Alberto Varvaro. La parola nel tempo. Lingua, societa` e storia. Bologna: Il Mulino, 145–74. Vasconcellos, Jose´ Leite de. See Leite de Vasconcellos, Jose´. Vela´zquez, Isabel. 2000. Documentos de ´epoca visigoda escritos en pizarra (siglos VI–VIII). 2 vols. Turnhout: Brepols. —— 2003. Latine dicitur, vulgo vocant. Aspectos de la lengua escrita y hablada en las obras gramaticales de Isidoro de Sevilla. Logroo: Fundacio´n San Milla´n de la Cogolla. Vennemann, Theo. 1988. Preference Laws for Syllable Structure. Berlin, New York, and Amsterdam: Mouton-De Gruyter. Veny, Joan. 1987. Els parlars catalans. 7th edn. Palma de Mallorca: Ed. Moll. Verner, Karl. 1877. “Eine Ausnahme der ersten Lautverschiebung”. Zeitschrift fu¨r vergleichende Sprachforschung 23: 97–130. [Eng. tr. “An Exception to Grimm’s Law”. In P. Baldi and R. N. Werth (eds). 1978. Readings in Historical Phonology. University Park and London: Pennsylvania State University Press, 32–63]. Vielliard, Jeanne. 1927. Le latin des diploˆmes royaux et chartes prive´es de l’e´poque me´rovingienne. Paris: Champion. Vineis, E. 1998. “Latin”. In A. G. and P. Ramat (eds). The Indo-European Languages. London and New York: Routledge, 261–321. Virdis, Maurizio. 1978. Fonetica del dialetto sardo campidanese. Cagliari: Edizioni della Torre. Vitale, Maurizio. 1992. Studi di storia della lingua italiana. Milan: LED. Vives, Jose´. 1969. Inscripciones cristianas de la Espaa romana y visigoda. 2nd edn. Barcelona: CSIC. —— 1971–2. Inscripciones latinas de la Espaa romana. 2 vols. Barcelona: CSIC.
Bibliography
279
Wagner, Max L. 1936. Restos de latinidad en el norte de Africa. Coimbra: Biblioteca da Universidade. —— 1941. Historische Lautlehre des Sardischen. ZrPh Beiheft 93. Halle: Niemeyer. —— 1951. La lingua sarda. Storia, spirito e forma. Berne: Francke. Walberg, Emmanuel. 1907. Saggio sulla fonetica del parlare di Celerina-Cresta (Alta Engadina). Lund: Gleerup. Walde, A. and Hofmann, J. B. 1938–56. Etymologisches Wo¨rterbuch der lateinischen Sprache. 3rd edn. 2 vols. Heidelberg: C. Winter. Walter, Henriette. 1990. “Une voyelle qui ne veut pas mourir”. In J. N. Green and W. AyresBennett (eds). Variation and Change in French. London and New York: Routledge, 27–36. Waquet, Franc¸oise. 2001. Latin or the Empire of a Sign. [tr. of J. Howe. Latin ou l’empire d’un signe. 1998. Paris: Michel]. London and New York: Verso. Warnant, Le´on. 1956. Constitution phonique du mot en wallon. Paris: Socie´te´ d’e´dition “Les Belles Lettres”. Wartburg, Walther von. 1936. “Die Ausgliederung der romanischen Sprachra¨ume”. ZrP 56: 1–48. —— 1950. Die Ausgliederung der romanischen Sprachra¨ume. Berne: Francke. [Spanish tr. 1952, French tr. 1967]. Weinrich, Harald. 1969. Phonologische Studien zur romanischen Sprachgeschichte. 2nd edn. Mu¨nster: Aschendorffsche Verlagsbuchhandlung. Weinreich, Uriel. 1953. Languages in Contact. The Hague: Mouton. Wheeler, Max. 1979. Phonology of Catalan. Publications of the Philological Society 28. Oxford: Blackwell. —— 2005. The Phonology of Catalan. Oxford: Oxford University Press. Wiese, Richard. 1996. The Phonology of German. Oxford: Clarendon. Woledge, Brian and Clive, H. P. 1964. Re´pertoire des plus anciens textes en prose franc¸aise depuis 842 jusqu’aux premie`res anne´es du XIIIe sie`cle. Geneva: Droz. Wright, Roger. 1982. Late Latin and Early Romance. Liverpool: Francis Cairns. —— 2000. El Tratado de Cabreros (1206): estudio sociofilolo´gico de una reforma ortogra´fica. Papers of the Medieval Hispanic Research Seminar 19. London: Department of Hispanic Studies, Queen Mary and Westfield College. Wu¨est, Jakob. 1979. La dialectalisation de la Gallo-Romania. Berne: Francke. —— 1995. “Dauphinois”. LRL II, 2: 434–40. —— and Kristol, Andreas M. (eds). 1993. Aqueras montanhas. E´tudes de linguistique occitane: Le Couserans (Gascogne pyre´ne´enne). Tu¨bingen and Basle: Francke. Wunderli, Peter. 1969. Die okzitanischen Bibelu¨bersetzungen des Mittelalters. Frankfurt: Klostermann. Wyatt, William F. 1972. The Greek Prothetic Vowel. Philological Monographs of the American Philological Association 31. Cleveland, OH: Case Western Reserve University. Yip, Moira. 1991. “Coronals, coronal clusters, and the coda condition”. In Paradis and Prunet (eds), 61–78.
280
Bibliography
Young, Christopher and Gloning, Thomas. 2004. A History of the German Language through Texts. London: Routledge. Zamboni, Alberto, Cortelazzo, Manlio, Pellegrini, Giovanni Battista, Beninca`, Paola, Vanelli Renzi, Laura and Francescato, Giuseppe (eds). 1984–7. Dizionario etimologico storico friulano. 2 vols. [up to ezzita only]. Udine: Casamassima. Zamora Vicente, Alonso. 1967. Dialectologı´a espaola. 2nd edn. Madrid: Gredos. Zink, Gaston. 1986. Phone´tique historique du franc¸ais. Paris: Presses Universitaires de France. Zirin, Ronald A. 1970. The Phonological Basis of Latin Prosody. The Hague: Mouton. Zo¨rner, Lotte. 1989. Die Dialekte von Travo und Groppallo. Vienna: Verlag der ¨ sterreichischen Akademie der Wissenschaften. O Zufferey, Franc¸ois. 1987. Recherches linguistiques sur les chansonniers provenc¸aux. Paris: Droz.
Index
A-prosthesis actualization 169–71 causation 171–80 chronology 154–9 contact approach to 178–80 enhancement 190–3 vs etymological initial r- 148–9 examples 146 geography 150–1 identification 147–50 and initial rhotic 159–63 maintained 189–90 origins 158–9 overview 37 phonetic approach 175–7 phonological approach 177–8 vs prefixation 148 regression 182–8 vs syncope of pre-tonic vowel 149 vowel quality 164–8 abandonment of I-prosthesis 78–80 in Castilian 106 in Corsica 99 in French 119–26 in Gallo-Romance 129 in Italian 94 in Italo-Romance 143–4 in Portuguese 110–11 in Rheto-Romance 136 in Tuscan 86 in Venetian dialect 141–2 see also regression of A-prosthesis abandonment of schwa in French 125–6 Abruzzo and Molise 153 acoustic similarity of [s] and [i] 63–5 actualization 19 of I-prosthesis 65–7 of A-prosthesis 169–71 of U-prosthesis 202–24
additive processes 1 Agrigento, A-prosthesis 152 Al-Andalus 102–3 Albigeois text 131 Alguere`s, regression of A-prosthesis 187–8 Alsace 126 alternation 10–13 in A-prosthesis 169–70 Ancona, sonorant-initial words and U-prosthesis 218 aphaeresis 1, 4, 56–8, 128, 143–4 in Balkan-Romance 78 Occitan 132–3 in Sardinian 98–9 southern Italian 79–80 apocope 1 Appendix Probi 59 Arabic script 102–3 Aragonese 102, 155 Aromanian 26, 146 and A-prosthesis 158, 169, 191–3 initial rhotic 162 Arquint, J.C. 183 assimilation of vowels 164–5 Asturian dialects 102, 111 Asturo-Leonese area 109, 112 attitude to prosthesis in French 134 Auger, J. 211–12 Badia, A. 150n, 222 Baehrens, W.A. 59n Bagemihl, B. 17n Balkan-Romance 61, 175 and A-prosthesis 158 and I-prosthesis 76–9 initial rhotic 162 Banniard, M. 32 Barsegape`, P. (poet) 142 Bare`ges dialect 159
282
Index
Basque 178–80 Bastogne dialect 130 Bec, P. 151n, 156, 170, 185–7 Belcalzer, V. 144 Bembo, P. 92 Bergu¨n dialect 210, 223n Bezzola, R. and To¨njachen, R. 183 Bible translations 136, 157 and U-prosthesis 204, 229–30 Bifrun, J. 136, 157, 204, 225, 229 Blasco Ferrer, E. 222 bleeding SPIV 200, 201 Blevins, J. 21–2 Blevins, J. and Garrett, A. 218 Boccaccio 85, 92 Boeci 131 Bolognese 144 and SPIV 203, 205–6 and U-prosthesis 194 Bolognesi, R. 189–90 Bonvesin (da la Riva) 93n, 143 borrowing scale 179–80 boundary markers, prosthetic vowels as 26 Brantoˆme 119 Brescia 143 Broselow, E. 51n Brun, A. 134 Busachi dialect, vowel copying 164 Cabre´, T. 12 Cairo Montenotte dialect (Piedmontese) 140 Calabrian, A-prosthesis 152–3 Caltanissetta, A-prosthesis 152 Campidanese 39 maintenance of A-prosthesis 189–90 vowel copying 164 vowel quality 168 see also Sardinian Cancioneiros 108 Capidan, T. 191–2 Carnin and U-prosthesis vowel quality 227 Carolingian reforms 113–14 Castilian, I-prosthesis 104–6 Catalan 12, 97, 102, 107 and A-prosthesis 150n, 187–8
proclitics vs lexical forms 221–2 and U-prosthesis 195 vowel quality 16 causation of A-prosthesis 171–80 of I-prosthesis 67–73 in language change 19 lexical alignment 27 morpholexical factors 27 morphophonological factors 25–6 phonological factors 20–5 sociolinguistic factors 28–33 of U-prosthesis 228 Celerina dialect 236 and SPIV 197–8, 201, 203 and U-prosthesis 194, 230 Cesena, complex onsets 203–4 Cevio variety and SPIV 197 Chanson de Sainte Foi d’Agen 131 charters 115, 158 Chiampel, D. 231 Christian inscriptions 60 Christian Spain 102 chronology of U-prosthesis 204–8 Cisalpine Gaul 137 classification of prosthesis 40 clitic phrases 25 Clivio, G.P. 34, 214n, 221, 225, 231 Clivio, G.P. and Danesi, M. 139 Cochet, E. 226 Coco, F. 194 coins, inscriptions in 113 complex onsets in Latin 45–6 and SPIV 203 and U-prosthesis 209, 225 Consonantal Strength 22n consonants Classical Latin 41–2 complexity 23 contact approach to A-prosthesis 178–80 Corneille, T. 126n Coromines, J. 187 Correa Rodrı´guez 179n Corsican 99 and I-prosthesis 95 vowel quality 165–7
Index Coulmas, F. 31 Coupier, J. 135 Cremonese 143 Dacia 61–2 Daco-Romanian and A-prosthesis 193 and I-prosthesis 77 initial rhotic 162 Dalbera, J.-P. 133 Dalmatian, I-prosthesis 77–8 D’Ambra, R. 153 Dante, A. 84–5 De Mauro, T. 94 De Sathana cum Virgine (Bonvesin) 143 deletion of [n] as explanation of I-prosthesis 70 Desgranges, J.-C.-L.-P. 122 dictionaries as evidence of A-prosthesis 153 Engadinish-German 183 Dinguirard, J.-C. 156 diphthongization of word-initial vowels 5–6 direct indicators of vowel prosthesis 3–7 Divine Comedy (Dante) 84–5 Dolomitish, regression of A-prosthesis 184 Donzac dialect 134 Dordogne 135 d’Orle´ans, C. 116 Dressler, W. 56, 67 Dubois, J. 118–19 Ducibella, J.W. 152 Dutch, Middle, loanwords 115 Elba, I-prosthesis 95 Elizondo, E. 179n Elwert, W.T. 184 Emilian-Romagnolo 142, 144–5, 214n and SPIV 199, 202–3 and U-prosthesis 209, 231 Engadinish 136, 206 and A-prosthesis 169, 183–4 and SPIV 200 and U-prosthesis
developments 229–31 and falling sonority 215 vowel quality 225 environments for I-prosthesis epenthesis 1, 71–2 vs I-prosthesis 126–7 vs prosthesis 51 vowel quality 166 in Walloon 129–30 Erasmus, D. 123 Estienne, H. 119, 120n, 134 etymological initial rand A-prosthesis 155, 159–63 and vowel insertion 148–9 etymological prefixal vowels 73 evidence 34–5 excrescent vowels 17n extant inscriptions 60 Falcone, G. 152 falling sonority onsets and U-prosthesis 214–16 Farserotic 191n Fassan, regression of A-prosthesis 184 feeding SPIV 200–1 Fleischhacker, H. 51 Florentine writers 85 Flutre, L.-F. 210–11, 218 Fonni dialect, vowel copying 164 Fouche´, P. 125, 188n, 200n Fougeron, C. 171, 171n Francard, M. 130 Franc¸ois, D. 226 Frankish, loanwords 115 French 4, 9, 25 abandonment of I-prosthesis 123–6 alternations 13 influence on Gascon 187 initial rhotic 160–1 loanwords in 115, 125 from Italian 119–20 Old 39 published grammar 118–19 resemblance to Sardinian 100n SPIV 200–1
283
284
Index
French (cont.) standardization 30n, 120 effect on langue d’oc 134 word-initial syllables and U-prosthesis 217 frequency of [s] and [i] 63 Friulian 136 and A-prosthesis 157–8, 184–5 Gaeng, P. 60 Galician dialect 102, 108, 109–10 vowel deletion 111 Gallo-Romance and I-prosthesis 112–35 northern varieties 133 and U-prosthesis 196, 204, 210–12, 228 development 232 vowel quality 225–6 and word-initial syllables 217 Gallurese dialect 99 see also Sardinian Gartner, T. 183–4 Gascon dialect 121, 146, 156–7, 234 and A-prosthesis 147–51 actualization 170–1 implementation 169 regression of 185–7 and I-prosthesis 66 initial rhotic 160, 163 SPIV 198 vowel quality 168 word-initial morphemes 4 Gavel, H. 179n geminate consonants 42 geminate rhotic 172, 175, 177–8 gemination and ign- forms 181 in southern Italian 190 geography and I-prosthesis 74–5 Germanic influences 129, 136–7 and SPIV 207–8 Giammarco, A-prosthesis 153 Giannelli, L. 95 Giese, W. 191n, 192
Gorra, E. 145 Gouskova, M. 17 Government Phonology (GP) 24, 177–8 grammatical factors in vowel prosthesis 39 grammatical vowel addition 4 Greek 67 geminated rhotic 175 influence on Latin syllabification 48 loanwords 43, 72 Gregory of Tours 113 Grisch, M. 196 Grisons 136 Grizzando Morandi variety 145 Grizzanese falling sonority and U-prosthesis 215–16 proclitics vs lexical forms 220–1 Hall, R.A. 173 Hall, T.A. 44n Hall and Cravens, T.D. 172 Hall and Weinrich, H. 173 Hammond, R.M. 176 Harris, J.W. 11 Herman, J. 47n heterosyllabic onsets 50–1 hierarchy, phonological 20 Hindi, sibilants and stop segments 51 Hindret, J. 120–1n Holmes, U.T. 200 Homelies de Organya` 107 Huguet, E. 119–20n hypercorrection 56–8, 101, 113 I-prosthesis actualization 65–7 and Balkan-Romance 76–9 causation 67–73 environments 54–5 and Gallo-Romance 112–35 and Ibero-Romance 100–12 and La Spezia-Rimini line 74–5 and langue d’oc 130–5 and langue d’oı¨l 114–23 and northern Italo-Romance 137–45 origins 56–60
Index overview 36–7 and Rheto-Romance 135–7 and Sardinian 96–100 and southern Italian 79–80 spread during Roman times 60–2 and Tuscan 80–96 vowel quality 62–5 Iberian 178–9 Ibero-Romance and A-prosthesis 155 and I-prosthesis 100–12 initial rhotic 160 and SPIV 199 ign- forms and A-prosthesis 180–2 indirect indicators of vowel prosthesis 7–8 initial rhotic and A-prosthesis 159–63, 191 articulation problems 176–7 strengthening 171–5 internal prosthesis 54–5 interplay between word edges 25–6 intervocalic -n- in Gascon 156 Italian 5, 9, 10–11 and A-prosthesis 157, 169 dialects, falling sonority and U-prosthesis 215–16 loanwords in French 119–20 northern and U-prosthesis 205 sandhi 26 southern 79–80, 190–1 Italo-Romance and A-prosthesis 151–4 initial rhotic 161–2 northern and I-prosthesis 137–45 rhotic strengthening 175 and U-prosthesis 195, 225 Jungemann, F.H. 163n, 176 Keating, P. et al. 171 Kenstowicz, M. 16–17 Kiss, S. 44n Kramer, J. 184 Kurolywicz, J. 48n, 50
285
La Noue, O. 161 La Scala, F.J. 34 La Spezia-Rimini line and I-prosthesis 74–5 Ladefoged, P. and Maddieson, I. 176 Ladin area 136 Langobards (Lombards) 81 language contact 28–9, 187 langue d’oc (Occitan) and I-prosthesis 130–5 langue d’oı¨l and I-prosthesis 114–23 initial rhotic 161 and U-prosthesis 228 Languedocian 135 Latin 25, 235 Classical, syllable structure in 41–9 to French, change in formal settings 116–17 geminate rhotic 172 influences on 67 loanwords in 115 rising-sonority onsets 213 syllabicity of [s] 68–9 syllables 49–52, 70 variation 29 written 31–3 Latinisms, Castilian 105–6 Lausberg, H. 67 Leben, W. 202n legal charters 138 lenition and SPIV 206 Leopardi, verse of 93 letters Bolognese 205 Tuscan 87–91 lexical alignment in language change 27 lexical forms vs proclitics 220–2 Libre de Evast e Blanquerna (Llull) 107, 221–2 Libro di Sidrac 80 Ligurian 133 linearization of syllabic consonants 6–7 literary language, Tuscan 84 Llull, R. 107, 221–2
286
Index
loanwords 10–11, 115 in French 119–20, 125 in Gascon from French 163, 186 from Greek 43, 72 in Italian 94 in Italo-Romance 154 in pre-Roman languages 179 Lodge, R.A. 30n, 87n, 121 Logudorese 96–7, 99 see also Sardinian Lombard, A. 138 Lombards (Langobards) 81 Lombardy 142–4 Lopocaro, M. 100n, 153, 153n, 214n Lorraine 126, 128 Lowenstamm, J. 24n Lucchese dialects evidence of Tuscan I-prosthesis 81 vowel quality 165–6, 167–8 Luchaire, A. 187
Migliorini, B. 93 Miha˘ila˘, G. 158 Millardet, G. 175 minimal saliency 16–17, 63 and U-prosthesis 224 Moll, F.B. 150n Monferrato dialect, U-prosthesis 209 moraic thory, geminates in 177 Morelli, F. 50–1 Morosi, G. 133n morphemes, word-initial 4 morpholexical factors in language change 27 morphological basis to A-prosthesis 157 morphological boundaries, reinterpretation 4–5 morphophonological factors in language change 25–6 Mozarabic 102–3 Mushacke, W. 132
Machado, J.P. 155n Mainoldi, P. 194 Malagoli dialect 194, 199, 231 and U-prosthesis 209 Malkiel, Y. 57n Maneca, C. 34 Mantua 143 Manzoni, A. 93 Marchello-Nizia, C. 39n Marotta, G. 44n, 70 Marseille 134 Martinet, A. 174n Mateus, M.H. and d’Andrade, E. 12–13 maximal formal identity 72 McKenzie, K. 85 McMahon, A. 10n medial vowels 9 medieval period 76 Merovingian period 113 Mesnil-Martinsart dialect 220 and U-prosthesis 210–11 Messina dialect 5–6 metathesis 1, 71n Meyer-Lu¨bke, W. 34, 150n
[n] deletion as explanation of I-prosthesis 70 Nandris¸, O. 78 Nauton, P. 133 Navarrese, A-prosthesis 155 Nebrija, A. 106 neologisms 10–12 Neapolitan, A-prosthesis 153 newly appearing vowels 16–18 Nieri, I. 94, 157, 167–8 Nigra, C. 209, 213n Nisard, C. 122 Norse, loanwords 115 Noske, R. 44n Notre-Dame-de-Sanhilac dialect 135 Novellarese dialect and SPIV 197 and U-prosthesis 194, 209 Nuorese 99 obligatory contour principle (OCP) 202 Occitan (langue d’oc) and I-prosthesis 130–5 Old French 39
Index Omeltchenko, S.W. 60–1 on-glides 5 A-prosthesis as 177 Onset Theorem 23–4n onsets heterosyllabic 50–1 word-initial change 49–50 complexity 23–5 Latin 42–5 and U-prosthesis 212–20 word-medial, Latin 45 Optimality Theory (OT) 23–4, 35 Osprandus (Tuscan scribe) 83 Oxford Psalter 115 pagan inscriptions 60 palatal nasal in word-initial position and A-prosthesis 180–2 palatalization and SPIV 206–7 Palsgrave, J. 118 Papahagi, T. 146n, 192 papyri 137–8 paragoge 1 Pariente, A. 44 Parisian and U-prosthesis vowel quality 226 Peer, O. 183, 230 Pei, M.A. 114 Peletier, J. 126 Pensado, C. 173 Petrarch 85, 92 phonetic approach to A-prosthesis 175–7 phonetic realization of pre-consonantal <s> 118 phonological approach to A-prosthesis 177–8 phonological bond between [s] and voiceless plosive 72 phonological factors in language change 20–5 phrase medial vowels 9 Piacenza 145 Picard varieties and U-prosthesis 226–7, 229 Piccitto, G. 153 Pidal, M. 155n
287
Piedmontese 139–41 and U-prosthesis 209 developments 231 falling sonority 216 sonority levels 213–14 vowel quality 225 Pieri, S. 165 Pirandello, L. 152 Pisa, I-prosthesis in 95 Pittau, M. 99 Pliny the Elder 57n poetry see verse Politzer, R.L. 33 Pontremoli dialect, U-prosthesis 209, 217 Portomarin 108 Portuguese 12–13, 109, 112 and A-prosthesis 155n standard 110 and U-prosthesis 195 pre-consonantal <s>, phonetic realization 118 prefixal vowels 73 prefixation 27 vs A-prosthesis 148 prosthesis in word-medial position 11 prepositional phrases, lexicalized 40 pre-Roman linguistic influence 67, 178–80 prestigious varieties 30 principle of minimal saliency see minimal saliency Prinz, O. 60, 62–3, 66–7, 79, 137 private letters Bolognese 205 Tuscan 87–91 probabilistic approach to describing causes of change 19 proclitic forms Tuscan 86–7 and U-prosthesis 220–4 prosodic domains of U-prosthesis 208–12 prosthesis common properties 234–7 definition 1–2 vs epenthesis 51 Provenc¸al 135
288
Index
Psalter Commentary 127 Puglia, A-prosthesis 152 Pulon Matt 205 Pult, G. 231 quality of vowels see vowel quality rafforzamento fonosintattico (RF) 172–3 Ranrupt dialect 129 reductive processes 1 regional differences in northern Italy 138–9 regression of A-prosthesis 182–8 see also abandonment of I-prosthesis Reighard, J. 161 Renaissance period 106 restructuring of syllabic consonants 6–7 restructuring of syllable as explanation of I-prosthesis 71 resyllabification 48 as explanation of I-prosthesis 69 RF (rafforzamento fonosintattico) 172–3 Rheto-Romance 93n and A-prosthesis 150, 157 regression of 183–5 and I-prosthesis 135–7 initial rhotic 163 and U-prosthesis 195–6, 204, 210 and proclitics 223 rhoticity 218–19 rhotics and A-prosthesis 159–63, 191 articulation problems 176–7 strengthening 171–5 Richter, E. 63 Rickard, P. 117n rising-sonority onsets and U-prosthesis 213 Rohlfs, G. 95, 153–4, 174n, 179n, 180, 218 Roine´ 139–40 Romania continua 29–30 and A-prosthesis 150 Romanian 6 evidence of I-prosthesis 78 and U-prosthesis 195, 222–3
Ronjat, J. 132n Rose, Y. and Demuth, K. 64n, 166 Rosetti, A. 162, 191n, 192–3 Rossellone`s, U-prosthesis 222 rural vs urban development of U-prosthesis 231 Russell-Gebbett, P. 102 [s] and voiceless plosive, phonological bond 72 s impure 53–5, 67–8 s lı´quida 68 s-palatalization and SPIV 206–7 Sainte-Jamme variety and U-prosthesis vowel quality 225 San Nicola di Trullas 97 San Pietro di Silki 96–7 sandhi 26 Sanga, G. 92n Sardinian 146 and A-prosthesis 151, 169 and I-prosthesis 96–100 vowel copying 164–5 Sarrieu, B. 66 Sassarese 99–100 see also Sardinian Scha¨del, B. 187 Schiaparelli, L. 81 Schlo¨sser, R. 192 Schneegans, H. 151–2 Scho¨nthaler, W. 187 Schortz, M. 123 Schuchardt, H. 33, 62–3, 67 schwa deletion 125–6, 211 in newly appearing prosthesis 17–18 strengthening 129–30 and U-prosthesis 224 scribes, Tuscan 82–3 segmental phonology and language change 20 Se´guy, J. 134 Sennori dialect 100 Sent variety and U-prosthesis developments 231
Index Sermone 142 Sermoni subalpine 139 S¸iadbei, T. 33 sibilants and stop segments 50–1 Sicilian, A-prosthesis 151–2 signatures 137–8 Slavic influence on Balkan-Romance 79 sociolinguistic factors in language change 28–33 Sole´, M.-J. 176 sonorant-initial onsets and U-prosthesis 217–20 sonority hierarchy 21–2 sonority levels in onsets and U-prosthesis 212–16 Sonority Sequencing Generalization (SSG) see SSG sources of data 34–5 Spanish 27 alternation 11–12 Latin American, trill production 176 speech communities, variation in 19 spelling pronunciation 31 SPIV (syncope of pre-tonic initial vowels) 196–202 vs A-prosthesis 149 dating 205–8 surrounding consonants 202–4 Spoerri, T. 194 spoken language and I-prosthesis 144 sporadic changes 14–15 SSG (Sonority Sequencing Generalization) 21–2, 50, 235–6 St Isidore 59 staged view of I-prosthesis 65–6 Standard French, effect on langue d’oc 134 strengthening of initial rhotic 159–63, 171–5 stress 2, 5 structuralist view of A-prosthesis 176 of rhotic strengthening 171–5 substratum languages 28 suffricates 50 superstratum languages 28–9
289
Surmeiran proclitics and U-prosthesis 223 Surselvan variety 5, 199 syllabic change, Latin 49–52 syllabic rhotics 218–19 syllabicity of [s] 68–9 syllabification and U-prosthesis 228 of word boundaries 47–9 syllable simplification 70 syllable structure 20–3 Classical Latin 41–9 as explanation of I-prosthesis 71 synchronicity of prosthesis 8–14 syncope 1 syncope of pre-tonic initial vowels see SPIV Tabourot, E´. 124–5n Tardif, J. 114 Tekavcˇic´, P. 49n, 70 Telmon, T. 34, 213n, 225 Temes 108 tensing 68 Terentianus, C. 57 Thesaur del hospital de Saint Sperit 132 Thomason, S.G. 179–80 Thomason, S.G. and Kaufman, T. 179–80 Thurot, C. 31n Tja¨der, J.-O. 137 Toulouse 134 Touraine dialects and U-prosthesis vowel quality 226–7 Travo 145 Trento 144 trill, phonetic problems 176 Turinese, proclitics vs lexical forms 220–1 Tuscan 138 and I-prosthesis 80–96 initial rhotic 162 medieval, ign-forms 180–2 vowel quality 165, 167 U-prosthesis causation 228
290
Index
U-prosthesis (cont.) chronology 204–8 developments 229–32 geography 195–6 overview 38 and proclitic forms 220–4 prosodic domains 208–12 structural preconditions 196–204 and vowel quality 224–7 and word-initial onset 212–20 unstressed vowels subsystem 15 Upper-Engadinish, falling sonority and U-prosthesis 215 urban vs rural development of U-prosthesis 231 uvularization of rhotic 161 Valsesia dialect (Piedmontese) 141 and SPIV 197, 202 and U-prosthesis 194 see also Piedmontese Vanelli, L. 34, 221, 229 variable frequency of I-prosthesis in Tuscan letters 90 Venetian texts 141–2 Vennemann, T. 22n, 48n Vermandois variety and U-prosthesis development 232 vernacular prose, Tuscan 85 vernacular usage of Ibero-Romance 104 Verner, K. 14n verse 108 in langue d’oc 131 late medieval 116 Lombardy 143 Villacidro dialect 99 Villette variety, SPIV 197 Villon, F. 116 Vimeu dialect and U-prosthesis 211 Visigothic Spain 101 Viverone dialect (Piedmontese) 140–1 and SPIV 197 and U-prosthesis 209, 216 vowel copying 164–5
vowel deletion see abandonment of I-prosthesis; abandonment of schwa in French vowel quality 15–18 and A-prosthesis 164–8 factors determining 17–18 in Ibero-Romance 107 in Piedmontese 139–40 and SPIV 198–201 and U-prosthesis 224–7 Wagner, M.L. 98–9 Walberg, E. 183, 194, 198, 230 Walloon dialects 112, 126–30 and U-prosthesis vowel quality 227 weakening of prosthetic vowels 111 to schwa 129–30 Weinreich, U. 136n Wheeler, M. 12, 150n Wiese, R. 50 word boundaries interplay between 25–6 syllabification 47–9 word-final consonants in Rheto-Romance 135–6 word-initial morphemes 4 word-initial onsets change 49–50 complexity 23–5 Latin 42–5 and U-prosthesis 212–20 word-initial palatal nasal and A-prosthesis 180–2 word-initial syllables and U-prosthesis 217–20 word-medial onsets, Latin 45 written language 31–3 Wu¨est, J. 156 Wunderli, P. 132n Zamboni, A. et al. 185 Zirin, R.A. 48n Zufferey, F. 132n