, comes after a sonorant instead of , the sonorant is in fact followed by a punctuation sign.
6. Outlook This survey of potential clues to prosodic information in Old High German manuscripts is probably far from complete, and I have not been able to treat all the features exhaustively or in the required depth. Still, I hope to have shown that accents, word separation and other paleographical pheno-
mena potentially provide an interesting resource for prosodic research. The difficulties in interpreting such paleographic features and the danger of circular argumentation must be emphasized, however. So far very little systematic work exploring the paleographical phenomena discussed here and their potential value for prosodic studies has been carried out. Further research in this area, based on careful analyses of the manuscripts, could lead to interesting new insights. Acknowledgments The problems and ideas discussed in this paper first captivated me in December 2003, when I was invited by the project Die Rolle der Informationsstruktur bei der Herausbildung von Wortstellungsregularitäten im Germanischen (part of the Sonderforschungsbereich 632: Informationsstruktur: Die sprachlichen Mittel der Gliederung von Äußerung, Satz und Text), then in its initial phase, to discuss some of the methodological issues facing the project. Since then, the same questions have kept reappearing and have been discussed intensively on many occasions. For help and suggestions regarding the present paper I am grateful to Brigitte Bulitta, Oliver Ernst, Nikolaus Henkel, Roland Hinterhölzl, Astrid Kraehenmann, Thomas Krisch, Andreas Nievergelt, Svetlana Petrova, Oliver Schallert, Eva Schlachter, and Michael Solf. Needless to say, all remaining errors are my responsibility. Finally, many thanks are due to Mark Pennay for smoothing out my English and to the Stiftsbibliothek St. Gallen for allowing me to reproduce the illustrations. My work on Old High German syntax was financed by a Fellowship for Advanced Researchers granted by the Swiss National Science Foundation (PA001--105063), which allowed me to spend a very fruitful time at the Humboldt-Universität zu Berlin; it is herewith gratefully acknowledged. Notes 1.
Part of this passage (namely, amplius enim intellectus instruitur quando uox legentis quiescit) is quoted and translated by Parkes (1993: 21); the translation of this passage is taken from there. This might have been different in the western, Romance-speaking parts of the empire, where Latin, according to McKitterick (1989: 21), “was in no sense a second or foreign language”, but “continued to be understood.” (see also Saenger 1997: 101).
Prosodic information in Old High German manuscripts 3.
Due to the apparent lack of work relating to accent usage in Carolingian manuscripts I will sometimes refer to the works relating to the Beneventana. Since especially the monastery of Montecassino, one of the leading scriptories using this script, influenced the whole monastic world of its time (Bischoff 1986: 151), this can to some extent be justified, although it must be stressed that most of the evidence quoted is younger than the Old High German records with which we are dealing. Still, some parallels can be observed, which makes comparison worthwhile, its anachronism and geographic distance notwithstanding. 4. Due to space limitations I cannot discuss the (early Middle High German) accentuation system found in the various Williram manuscripts, which seems to be an offspring of Notker’s system (see Kruse 1913: 8, Gärtner 1991: 47, Schneider 2007: 19). For a discussion of this system see Kruse (1913) and Gärtner (1991: 45–55). Most interestingly from a sentence-prosodic perspective, it seems to be clear for this system that monosyllabic function words displaying an accent bear sentence stress (see Gärtner 1991: 49–50). 5. In this passage Notker mentions only articulos, but in his actual usage it is not only articles that do not bear accents. One might wonder whether Notker’s Medieval Latin usage of articulos encompasses other parts of speech in addition to articles. Unfortunately King (1986) and Hellgardt (1979) do not discuss this question; in his New High German translation, Sonderegger (1970: 83) renders articulos as “Artikel”. 6. According to Gabriel (1969: 64–65) Notker in his diphthong accentuations éi, éu, óu and íu as opposed to ûo, îe, îa, and îo also intended to render pitch differences. This view is rejected by King (1986: XLI–XLII). 7. In this article texts are quoted from reliable editions (Otfrid: Kleiber 2004a) or from manuscript sources (Tatian: Cod. Sang. 56; Notker, Boethius: Cod. Sang. 825); in the latter case the page and line of the respective codices is quoted. This information easily allows to find the respective references in the editions by Masser (1994) and King (1986), which are diplomatic; they are therefore not quoted additionally. 8. Since the examples quoted are usually quite long, no gloss indicating the grammatical categories is provided; the only abbreviations used are NEG = “negative particle” and PREF = “verbal prefix”. The meaning of the examples is clarified by more idiomatic translations. If an Old High German example is phrased in close accordance to a biblical passage, the idiomatic translation is taken from the King James Version. 9. The illustrations of this article are all details from manuscripts hosted at St. Gall’s Abbey library (Stiftsbibliothek St. Gallen); facsimiles can be accessed via: http://www.cesg.unifr.ch/virt_bib/manuscripts.htm 10. In the youngest Otfrid manuscript the different accents are not distinguished any longer as far as their paleographical shape is concerned (see Sievers 1909: 8). As a matter of fact, in this manuscript the scribe put (rhythmic) ac-
Jürg Fleischer cents nearly randomly, obviously not knowing what he was doing (see Pivernetz 2000: 121). Some systematization can be seen in his use of accents on íá, íó, and íú, however (see Pivernetz 2000: 122); thus, the principles of phonetic accentuation seem to have been followed to some extent. The different accents are distinguished in print in Kleiber’s (2004a) edition by use of two different acute accents symbols. In the present article this cannot be reproduced. In order to distinguish Otfrid’s different types of accents, the “rhythmic accents” (along with the letters above which they are placed) are printed in bold. Incidentally, there is a third type of accent of somewhat unclear function in the Viennese manuscript, disregarded here. It occurs only in a small portion of the text, while the phonetic and rhythmic accents are used throughout the codex (see Kleiber 2004b: 120). For instance, it seems that in direct speech, mentioned by Masser (1997: 60) with respect to scribe ȗ of the Tatian manuscript, more accents are encountered as compared to other passages. The same phenomenon seems to be observed in the Beneventana (see Newton 1999: 181). This example is taken from Loew (1914: 277, note 1), where the same is said with respect to the Beneventana. According to Saenger (1997: 56), this disambiguating function of the accent became especially prominent in the late tenth century; Schneider (1999: 93, 2007: 21), referring to German manuscripts, describes this function for the thirteenth century, too. Incidentally, the accent on ós ‘mouth’ might also be used to distinguish ǀs ‘mouth’ from ǂs ‘bone’; this function of the accent is noticeable in the writings of Isidore (Keller 1908: 103; see also Keller 1908: 106, Gabriel 1969: 49). According to Saenger (1997: 98), who refers to European vernacular languages in general, but particularly also mentions “Old German”, acute accents were employed “to supplement the deficiencies of the Roman alphabet and to avoid homographs for words distinguished by vowel length.” The use of an acutelike symbol (termed apex originally, but eventually lumped together with the acute sign) to avoid confusion with a homograph displaying a short vowel seems to be very old (see Keller 1908: 103–104 and Saenger 1997: 54, who quote a passage by Quintilian, c. 40–c. 118). Newton (1999: 181) mentions a similar function with respect to the Beneventana. Illustrations of text from the Tatian manuscript (Cod. Sang. 56) which contain both Latin and Old High German text (Illustrations 2, 4 and 5) are reproduced with the Latin text before the Old High German text. This differs from the layout in the manuscript, but allows the relevant paleographic details to be shown in a greater magnification. Apart from scribe ȕ this usage can only be found three times in the portion of scribe Ȗ in the St. Gall Tatian manuscript (see Sievers 1909: 15). By “separation” I am referring to separation by spaces only. An alternative and/or supplementary indication of word or syllable separation by points usually called prosodiae (see Saenger 1990: 55–56, 1997: 26, 53–58) is at-
Prosodic information in Old High German manuscripts
tested in some Old High German records (see Krotz 2002: 119, Voetz 2006: 57). 18. It seems that in glosses and interlinear versions spaces might perform a rather different function, namely, to indicate the morphological structure of the Latin words that were glossed (see Henkel 1996: 59–60). 19. This even holds for the otherwise very faithful edition of the Viennese Otfrid manuscript by Kleiber (2004a; see Voetz 2006: 60). 20. An equivalent to punctuation by means of punctuation symbols is an arrangement principle known as per cola et commata that goes back to Jerome (see Müller 1964: 28, 70, Parkes 1993: 35). It is frequently encountered in Vulgate manuscripts. In this arrangement, every constituent to be separated by pauses from other constituents is written on a new line; the needed information to recite a text with the appropriate pauses is thus easily provided (see Müller 1964: 70–71). Among Old High German records the Tatian manuscript displays this kind of arrangement (see Bischoff 1986: 48).
References Bästlein, Ulf Christian 1991 Gliederungsinitialen in frühmittelalterlichen Epenhandschriften: Studie zur Problematik ihres Auftretens, ihrer Entwicklung und Funktion in lateinischen und volkssprachlichen Texten der Karolinger- und Ottonenzeit. Frankfurt am Main etc.: Peter Lang. Bischoff, Bernhard 2004 Reprint. Paläographie des römischen Altertums und das abendländischen Mittelalters. 2., überarbeitete Auflage. Berlin: Erich Schmidt. Original edition, Berlin: Erich Schmidt, 1986. Braune, Wilhelm and Ingo Reiffenstein 2004 Althochdeutsche Grammatik I: Laut- und Formenlehre. 15. Auflage bearbeitet von Ingo Reiffenstein. Tübingen: Niemeyer. Bruckner, Albert 1980 Akzente. Lexikon des Mittelalters. Vol. 1, 259–260. München/Zürich: Artemis. Fleischer, Oskar 1882 Das accentuationssystem Notkers in seinem Boethius. Zeitschrift für deutsche Philologie 14: 129–172, 285–300. [also partially published as diss. 1882 Halle: Waisenhaus.] Frey, Evelyn 1988 Wortteilung und Silbenstruktur im Althochdeutschen. Mit einem Anhang zur mittelhochdeutschen „Speculum ecclesiae“-Handschrift. Diss. München: Kopier- und Druckzentrale.
Gabriel, Eugen 1969 Die Entwicklung der althochdeutschen Vokalquantitäten in den oberdeutschen Mundarten. Wien: Österreichische Akademie der Wissenschaften. Gärtner, Kurt 1991 Die Williram-Überlieferung als Quellengrundlage für eine neue Grammatik des Mittelhochdeutschen. Zeitschrift für deutsche Philologie 101, Sonderheft: Mittelhochdeutsche Grammatik als Aufgabe, ed. Klaus-Peter Wegerea, 23–55. Green, D. H. 1994 Medieval Listening and Reading: The Primary Reception of German Literature 800–1300. Cambridge: Cambridge University Press. Hellgardt, Ernst 1979 Notkers des Deutschen Brief an Bischof Hugo von Sitten. In Befund und Deutung. Zum Verhältnis von Empirie und Interpretation in Sprach- und Literaturwissenschaft, eds. Klaus Grubmüller, Ernst Hellgardt, Heinrich Jellissen and Marga Reis, 169–192. Tübingen: Niemeyer. Henkel, Nikolaus 1996 Die althochdeutschen Interlinearversionen: zum sprach- und literaturhistorischen Zeugniswert einer Quellengruppe. In Übersetzen im Mittelalter. Cambridger Kolloquium 1994, eds. Joachim Heinzle, L. Peter Johnson and Gisela Vollmann-Profe, 46–77. Berlin: Erich Schmidt. Illich, Ivan 1993a In the Vineyard of the Text: A Commentary to Hugh’s Didascalicon. Chicago and London: The University of Chicago Press. 1993b Lectio Divina. In Schriftlichkeit im frühen Mittelalter, ed. Ursula Schaefer, 19–35. Tübingen: Narr. Keller, Wolfgang 1908 Über die Akzente in den angelsächsischen Handschriften. In Untersuchungen und Quellen zur germanischen und romanischen Philologie. Johann von Kelle dargebracht von seinen Kollegen und Schülern. Erster Teil, 97–120. Prag: Bellmann. King, James C. 1986 Die Akzente bei Notker und in der Fachliteratur, besonders bei Sehrt und Starck. In Notker der Deutsche, Boethius, »De consolatione Philosophiæ«, ed. Petrus W. Tax, XXXVII–XLIII. Tübingen: Niemeyer. Klaper, Michael 2004 Musikhistorische Interpretation. In Kleiber 2004b: 148–153.
Prosodic information in Old High German manuscripts
Kleiber, Wolfgang 1971 Otfrid von Weißenburg: Untersuchungen zur handschriftlichen Überlieferung und Studien zum Aufbau des Evangelienbuches. Bern/München: Francke. 2004a Otfrid von Weißenburg: Evangelienbuch. Band I: Edition nach dem Wiener Codex 2687. Herausgegeben und bearbeitet von Wolfgang Kleiber unter Mitarbeit von Rita Heuser. Teil 1: Text. Tübingen: Niemeyer. 2004b Otfrid von Weißenburg: Evangelienbuch. Band I: Edition nach dem Wiener Codex 2687. Herausgegeben und bearbeitet von Wolfgang Kleiber unter Mitarbeit von Rita Heuser. Teil 2: Einleitung und Apparat. Mit Beiträgen von Wolfgang Haubrichs, Norbert Kössinger, Otto Mazal, Norbert H. Ott und Michael Klaper. Tübingen: Niemeyer. Krotz, Elke 2002 Auf den Spuren des althochdeutschen Isidor: Studien zur Pariser Handschrift, den Monseer Fragmenten und zum Codex Junius 25. Mit einer Neuedition des Glossars Jc. Heidelberg: Winter. Kruse, Heinrich 1913 Die Accente in den Handschriften von Willirams Übersetzung und Auslegung des hohen Liedes. Diss. Greifswald: Adler. Lahiri, Aditi and Astrid Kraehenmann 2004 On maintaining and extending contrasts: Notker’s Anlautgesetz. Transactions of the Philological Society 102: 1–55. Loew, Elias A. 1980 Reprint. The Beneventan Script. A History of the South Italian Minuscule. Second edition prepard and enlarged by Virginia Brown. Volume I: Text. Roma: Storia e Letteratura. Original edition, Oxford: Clarendon, 1914. Masser, Achim 1994 Die lateinisch-althochdeutsche Tatianbilingue Stiftsbibliothek St. Gallen Cod. 56. Göttingen: Vandenhoeck & Ruprecht. 1997 Wege zu gesprochenem Althochdeutsch. In Grammatica Ianua Artium. Festschrift für Rolf Bergman zum 60. Geburtstag, eds. Elvira Glaser and Michael Schlaefer, 49–70. Heidelberg: Winter. McKitterick, Rosamond 1989 The Carolingians and the Written Word. Cambridge: Cambridge University Press. Müller, Rudolf Wolfgang 1964 Rhetorische und syntaktische Interpunktion. Untersuchung zur Pausenbezeichnung im antiken Latein. Diss. Tübingen.
Newton, Francis 1999 The Scriptorium and Library at Monte Cassino, 1058–1105. Cambridge: Cambridge University Press. Nübling, Damaris 1992 Klitika im Deutschen: Schriftsprache, Umgangssprache, alemannische Dialekte. Tübingen: Narr. Parkes, Malcolm B. 1993 Pause and Effect. An Introduction to the History of Punctuation in the West. Berkeley/Los Angeles: University of California Press. Piper, Paul 1882 Zu Otfrid: 1. Otfrids accente. Beiträge zur Geschichte der deutschen Sprache und Litteratur 8: 225–244. Pivernetz, Karin 2000 Otfrid von Weißenburg. Das ‘Evangelienbuch’ in der Überlieferung der Freisinger Handschrift (Bayerische Staatsbibliothek München, cgm. 14). Edition und Untersuchungen. II: Untersuchungen. Göppingen: Kümmerle. Saenger, Paul 1990 The separation of words and the order of words: the genesis of medieval reading. Scrittura e Civilità 14: 49–74. 1997 Space between Word: The Origins of Silent Reading. Stanford: Stanford University Press. Scholz, Manfred Günter 1980 Hören und Lesen. Studien zur primären Rezeption der Literatur im 12. und 13. Jahrhundert. Wiesbaden: Steiner. Schneider, Karin 1999 Paläographie und Handschriftenkunde für Germanisten: eine Einführung. Tübingen: Niemeyer. 2007 Akzentuierung in mittelalterlichen deutschsprachigen Handschriften. In Edition und Sprachgeschichte. Baseler Fachtagung 2.–4. März 2005, eds. Michael Stolz, Robert Schöller and Gabriel Viehhauser, 17–24. Tübingen: Niemeyer. Sievers, Paul 1909 Die Accente in althochdeutschen und altsächsischen Handschriften. Berlin: Mayer & Müller. Simmler, Franz 1997 Interpungierungsmittel und ihre Funktionen in der Lorscher Beichte und im Weißenburger Katechismus des 9. Jahrhunderts. In Grammatica Ianua Artium. Festschrift für Rolf Bergman zum 60. Geburtstag, eds. Elvira Glaser and Michael Schlaefer, 93–114. Heidelberg: Winter.
Prosodic information in Old High German manuscripts 1998
Makrostrukturen in der lateinisch-althochdeutschen Tatianbilingue. In Deutsche Grammatik – Thema in Variationen. Festschrift für HansWerner Eroms zum 60. Geburtstag, eds. Karin Donhauser and Ludwig M. Eichinger, 299–335. Heidelberg: Winter. 2003 Geschichte der Interpunktionssysteme im Deutschen. In Sprachgeschichte. Ein Handbuch zur Geschichte der deutschen Sprache und ihrer Erforschung. 2., vollständig neu bearbeitete und erweiterte Auflage, 3. Teilband, eds. Werner Besch, Anne Betten, Oskar Reichmann and Stefan Sonderegger, 2472–2504. Berlin/New York: Walter de Gruyter. Sobel, Naphtali 1882 Die Accente in Otfrids Evangelienbuch: eine metrische Untersuchung. Strassburg: Trübner. Sonderegger, Stefan 1970 Althochdeutsch in St. Gallen: Ergebnisse und Probleme der althochdeutschen Sprachüberlieferung in St. Gallen vom 8. bis ins 12. Jahrhundert. St. Gallen: Ostschweiz, Sigmaringen: Thorbecke. Stotz, Peter 2002 Handbuch zur lateinischen Sprache des Mittelalters. Erster Band: Einleitung, lexikologische Praxis, Wörter und Sachen, Lehnwortgut. München: Beck. Tax, Petrus W. ed. 1986 Notker der Deutsche: Boethius, “De consolatione Philosophiæ” Buch I/II. Tübingen: Niemeyer. 1996 Notkers Brief. In Notker der Deutsche: Die kleineren Schriften, herausgegeben von James C. King und Petrus W. Tax, 347–349. Tübingen: Niemeyer. Voetz, Lothar 2006 Einige Beobachtungen zur Getrennt- und Zusammenschreibung im Althochdeutschen. In Neue Perspektiven der Sprachgeschichte: Internationales Kolloquium des Zentrums für Mittelalterstudien der Otto-Friedrich-Universität Bamberg 11. und 12. Februar 2005, eds. Ursula Götz and Stefanie Stricker, 51–64. Heidelberg: Winter. Zürcher, Josef 1978 Graphetik – Graphemik – Graphematik unter besonderer Berücksichtigung von Notkers Marcianus Capella. Diss. Zürich: akuFotodruck.
On the “syntax of silence” in Proto-Indo-European Thomas Krisch
Abstract Using material from the ancient Indo-European languages Latin, Greek, Sanskrit and Hittite and insights of modern linguistic theory, this paper discusses two phenomena of ellipsis: gapping and object ellipsis. Both kinds of ellipsis are shown to be operative in these languages and can be reconstructed for Proto-Indo-European. Gapping appears in the two variants of forward and backward gapping in the languages discussed. In accordance with observations in literature using generative theory, which tell us that backward gapping is only possible in SOV languages, we conclude that backward gapping is a further piece of evidence for an underlying SOV structure of these languages and of Proto-Indo-European. The fact that there exists forward gapping in all of the languages discussed is interpreted as a reflex of a V-to-C movement. Overt V-to-C-movement is only attested in Latin, Greek and Sanskrit and not in Hittite, though. This fact is interpreted as a Hittite innovation, forward gapping being a remnant of Proto-Indo-European V-to-C-movement in this language. Object ellipsis operates in forward direction and depends on factors of functional sentence perspective. 1. Introduction1 This paper deals with some aspects of the “syntax of silence” (ellipsis) in Indo-European. There are not many investigations into this phenomenon for ancient Indo-European languages and also for modern languages there is still much work to be done. During the last decades, Generative Grammar, though, has made some substantial contributions to our understanding of ellipsis where it is syntactically conditioned. The moderate aims of the present paper are to discuss some theoretical issues, to describe some phenomena appearing in ancient Indo-European languages (AIELs) and to reconstruct some elliptical constructions for Proto- Indo-European (PIE).
In this paper, I leave aside the ubiquitous ellipsis of the verb “to be” in AIELs and refer to Karl Praust’s recent illuminating paper (Praust 2003), where he reconstructs a PIE (phonologically) silent injunctive of the verb “to be”, which he sees reflected in ancient IE verbless nominal sentences. I here also leave aside the question of an ellipsis of the verb in case of verbs of movement with preverbs (cf. e.g. Krisch 1984: 93-94) in AIELs. This paper also does not deal with the well known type of ellipsis in more or less lexicalized NPs of AIELs where the head (the N) is not expressed and an originally attributively used adjective is nominalized.2
2. Ellipses and implicatures Ellipses are a performance factor of natural languages. Speakers and (to a little lesser degree) also writers of texts often use them. Speaker and hearer have the tacit knowledge that certain parts of utterances may remain unexpressed. Modern relevance theory claims that speakers/writers do not express truth values in a direct way but rather suggest truth values (“schematic logical forms” cf. Blakemore 2002: 77) to the hearer/reader who has to complete the utterance by deriving inferences if necessary.3 Speaker and hearer follow the principle of maximal relevance of the utterance in the context and the speaker as well as the hearer expect implicatures. Implicatures are involved even in seemingly complete utterances as in (1): (1)
Mein Schwager wohnt in München, aber seine Tante ist Ärztin in Wien. “My brother-in-law lives in Munich but his aunt is a medical doctor in Vienna”
The meaning of this sentence contains the truth values of both sentences that form part of it. Thus, the sentence is only true iff the speaker’s brotherin-law lives in Munich and iff his aunt is a medical doctor in Vienna. But in addition to that, the hearer/reader can draw the implicature (activated by the conjunctive particle aber “but”) that there is some conflict between the utterance Mein Schwager wohnt in München “my brother-in-law lives in Munich” and seine Tante ist Ärztin in Wien “his aunt is a medical doctor in Vienna”, and the reader/hearer will worry about the reason of this conflict.4 This type of implicature triggered by the semantics of certain elements of speech has been called “conventional implicature” for the last 30 years.5 The other type of implicature (“conversational implicature”) is triggered by the context and by encyclopaedic knowledge. Cf. the much cited sentence in (2):
On the “syntax of silence” in Proto-Indo-European 193 (2)
Herr M. beherrscht seine Muttersprache und hat meine Lehrveranstaltungen regelmäßig besucht. “Mr. M. has a command of his mother tongue and regularly attended my lectures”.
If you utter this sentence in the context of an expert opinion to a scholarship this is a negative statement, namely Mr. M. does not have any other merits. If one sees the notions of conventional implicature vs. conversational implicature as poles of a continuum, our paper looks at examples which are nearer to the pole of conventional implicature.
3. Remarks on the history of generative research in ellipsis The most influential researcher in this field has been John Robert (‘Haj’) Ross (e.g. Ross 1970). Among other things he introduced the notion of “gapping” to denote ellipsis of the verb caused by coordinate constructions. I adopt Richards’ definition of gapping here: “Gapping involves ellipsis of a portion of the verb phrase, including the verb but excluding one or more VP-internal constituents” (Richards 1998: 158). The direction of gapping depends on the syntactic type. Backward gapping is only allowed in SOV languages (cf. (3c)), whereas forward gapping may appear in SVO (cf. the English examples in (3a)), SOV (cf. German subordinate clauses in (3b)) or VSO6 languages (cf. the Irish example in (3d)). Only SOV languages allow gapping in both directions (cf. (3b) vs. (3c)). In our examples the element(s) that can be gapped is/are put in deleted capital letters. (3)
a. English (SVO) John likes fish and Peter LIKES meat. Max seemed to be trying to begin to make love to Harriet and Fred SEEMED TO BE TRYING TO BEGIN TO MAKE LOVE to Sue. b. German (SOV) Ich weiß, dass die Kinder Fisch lieben und I know that the children fish love and die Eltern Fleisch LIEBEN the parents meat LOVE “I know that the children love fish and the parents (love) meat.” c. Ich weiß, dass die Kinder Fisch LIEBEN und and I know that the children fish LOVE die Eltern Fleisch lieben. the parents meat love “I know that the children (love) fish and the parents love meat.”
Thomas Krisch d. Modern Irish (VSO) (cf. Steedman 2000:177) Chonaic Eoghan Siobhán agus CHONAIC Saw Eoghan Siobhán and SAW Ciarán Ciarán “Eoghan saw Siobhán and Eoghnaí (saw) Ciarán”
Gapping thus seems to correlate with the directionality of verbal government. In languages with government to the right (SVO, VSO) only forward gapping is allowed, whereas SOV languages with government to the left show backward gapping. A subclass of SOV languages like German allow a verb before the object in some clause types (in German especially in main declarative clauses), cf. (4): (4)
Kinder lieben children love
In generative grammar, this characteristic has been explained by a movement of the verb from original position after its object inside the verb phrase (VP) into a position called “C(omplementizer) position” (cf. also 4.3). The result may be a structure like (4), which superficially looks like an SVO structure.7 By syntactic analogy (abduction8), then, the same gapping pattern as in SVO languages, namely forward gapping can be applied to these SOV languages if the verb in the first sentence of the conjoined sentences appears in final position (cf. (3b)).
4. Gapping as a phenomenon of Proto-Indo-European syntax 4.1. Examples from ancient Indo-European Languages Gapping is, indeed, a type of ellipsis that can be traced back to the PIE language. Let us first look at some examples from AIELs. A sentence with backward gapping that reoccurs in the Old Hittite9 ritual corpus a number of times is (5a). There likewise exists forward gapping in Hittite (cf. (5b)):
On the “syntax of silence” in Proto-Indo-European 195 (5)
Hittite a. (backward ellipsis) StBoT 12 (Neu 1970; Hittite thunderstorm ritual), Rs. III, 21´ G]IŠ DINNANA.GAL.GAL LÚ.MEŠIal-li-ri-e[ i-]Ia-mi-an-zi “The big Itar-instrument [probably a lyra, TK] (resounds), the priest-singers resound” LÚ.MEŠ GIŠ DINNANA GAL.GAL IŠ)AMAI Iallirie Itar-instrument big.big RESOUNDS priest-singer.NOM.PL iIamianzi resound.PRS.3PL b. (forward ellipsis) StBoT 12 (Neu 1970; Hittite thunderstorm ritual), Rs. III, 38´: LUGAL-uš Iu-u-up-pa-ri ši-pa-an-ti MUNUS.LUGAL-ša n[a-at-]ta “The king libates into a tureen, and the queen (does) n[o]t (libate).” LUGAL-uš IVppari šipanti MUNUS.LUGAL-ša king.NOM.SG tureen.LOC.SG libate.3SG.PRS queen-and[CLT.] n[at]ta ŠIPANTI not LIBATES
In Latin, too, one can observe backward and forward ellipses, cf. the examples (6a) and (6b): (6)
Latin a. (backward ellipsis) (cf. Gaeta and Luraghi 2001: 95) Caes. Bell.Gall. 1,40,13 (AcI in indirect speech, a speech of Julius Cesar in a war assembly) suam innocentiam perpetua vita, felicitatem Helvetiorum bello esse perspectam. “(that) his [scil. Ceasar’s TK] selflessness (was to be seen) in (his) whole life, (his) success was to be seen in the war against the Helvetians” suam innocentiam perpetua selflessness.ACC.SG.F. continuous.ABL.SG his.ACC.SG.F.REFL felicitatem vita ESSE PERSPECTAM success.ACC.SG.F life.ABL.SG. TO BE SEEN Helvetiorum bello esse war.ABL.SG be.INF Helvetians.GEN.PL perspectam seen.PTCP.PRF.PASS.ACC.SG.F
Thomas Krisch b. (forward ellipsis) Plaut. Mil. 990 Viden tu illam oculis venaturam facere atque aucupium auribus? “Do you see her making a hunt with her eyes and (making) a bird-hunt with her ears?” Vide-n tu illam oculis venaturam facere atque eye.ABL.PL hunt.ACC.SG make.INF and See.2SG-Q you her auribus aucupium FACERE ear.ABL.PL bird-catching.ACC.SG. MAKE
Also Ancient Greek provides attestations for both, backward and forward gapping, as the following examples illustrate: (7)
Ancient Greek a. (backward ellipsis) Hdt. 2,56,1 (cf. Gaeta and Luraghi 2001: 105)
Ei alFth‚Ps hoi Phoínikes exÇgagon t…s hir…s gynaĩkas kaì t¼n mèn autéPn es LibýFn, t¼n dè es t¼n [email protected] apédonto, … “If the Phoenicians did in fact carry away the sacred women and (sell) one of them in Libya and sell one in Hellas, …”
Ei alFth‚Ps hoi Phoínikes If
the Phoenicians they.carried.away
t…s hir…s the sacred
and this.ACC.SG.F EMPH of.them [GEN.PL] to APÉDONTO t¼n dè es t¼n Libya.ACC SOLD this.ACC.SG.F but to ART.ACC.SG.F
Greece.ACC.SG.F sold.3PL.AOR.MID b. (forward ellipsis) Hdt. 2,180,2 (cf. Gaeta and Luraghi 2001: 100)
… Ámasis mèn gár sphi édPke chília styptFríFs tálanta, hoi dè en AigýptPi oikéontes HéllFnes eíkosi mnéas … for Amasis gave them a thousand talents’ weight of alum, but the Greek settlers in Egypt (gave them) twenty minae (of silver).” Ámasis mèn gár sphi édPke Amasis.NOM.SG EMPH for them.DAT gave.3SG.AOR tálanta hoi dè chília styptFríFs thousand alum.GEN.SG talent.ACC.PL ART.NOM.PL.M but oikéontes HéllFnes en AigýptPi in Egypt.DAT live.PTCP.PRS.NOM.PL.M Greek.NOM.PL.M mnéas SPHI ÉD0SAN eíkosi THEM GAVE twenty minae.ACC.PL.F
On the “syntax of silence” in Proto-Indo-European 197
In (7b), one may observe two types of elliptical expressions gapping of the verb édPsan including one of its arguments and also the usual ellipsis of “silver” with the unit of measurement “Mnéa”. Vedic Sanskrit, too, shows gapping in both directions, cf. (8a) and (8b): (8)
Vedic Sanskrit a. (backward ellipsis) (cf. Gaeta and Luraghi 2001: 96) RV 6,75,2 [email protected]à gx [email protected]¡ê jayema “with the bow (we want to win) the cows with the bow we want to win the battle.”
bow.INS.SG cow.ACC.PL.F WE.WANT.TO.WIN bow.INS.SG
battle.ACC.SG win.1PS.PL.OPT.PRS b. (forward ellipsis) RV 10,42,10
g¢bhiù ñaremxmatiê dur‚vàê [email protected] kù£dham puruhåta v¡÷vàm “Through cows we want to overcome malignant helplessness, through corn (we want to overcome) all (kinds of) hunger, o much invoked one!”
malignant.ACC.SG.F corn.INS.SG WE.WANT.TO.OVERCOME
4.2. Sloppy identity in ancient Indo-European languages The Hittite example (5a) as well the Greek example (7b) show a phenomenon well attested in modern languages: the elliptical form does not need to be completely identical to its model. In (5a), the overt verbal form iIamianzi is a third person plural form but one has to recover a third person singular (išIamai) by conventional and conversational implicature. In (7b), the overt verbal form édPke by (conventional and conversational) implicature triggers a third plural form édPsan to be recovered. This phenomenon of only partial identity of the overt and the recovered element has been termed “sloppy identity” in the literature. (5a) and (7b) show that this property of ellipsis also existed in the AIELs.
4.3. Backward and forward gapping as a PIE phenomenon and SOV The examples given above all belong to the four ancient Indo-European language families that are the most important ones for the reconstruction of PIE syntax10 (Anatolian, Italic, Greek and Indo-Aryan). They all show the same phenomena: backward and forward gapping. It is therefore safe to reconstruct both possibilities for the Proto-language. The question why PIE and the daughter languages allowed both directions of gapping may be answered by the predominantly SOV character of PIE (cf. also section 3. above)11 and its daughter languages Hittite, Latin, Greek and Vedic Sanskrit and by the fact that there exists V to C movement in all of these languages (as in the German example (4) above) except Hittite, cf. the examples (7b) and (8b) above and the examples in (9) (the moved verb is put in bold letters): (9)12
a. Latin Ter. Ph. 594–596 vixdum dimidium dixeram, intellexerat / gaudebat, me laudabat. quaerebat senem, / dis gratias agebat ... “I scarcely had said half (of it) he had understood. He was glad, praised me, looked for the old man, thanked the gods ...” me laudabat, quaerebat senem look.for.3SG.IPF old.man.ACC.SG me.ACC.SG praise.3SG.IPF b. Greek Il. 18, 476–477 (cf. also Krisch 2001: 169-171) [... th©ken en akmothétᯉ mégan ákmona, génto dè cheirì / raist©ra krater¼n,] hetérēphi dè génto pyrágrēn “[he [scil. Hephaistos, TK] set on the anvil-block a great anvil, he seized a massive hammer with one hand,] but with the other hand he seized the fire-tongs” dè génto pyrágrēn hetérēphi with.the.other but seize.3SG.MID.AOR fire.tongs.ACC.SG.F c. Vedic Sanskrit RV 1,85,7
v¡ùõur [email protected] dhvad vÆùaõam madacy£taü [email protected] [email protected] sãdann @dhi barh¡ùi priy‚ “when Viùõu supported the bull reeling with excitement [scil. god Indra, TK], they [scil. the Marut, the storm-gods, TK] sat down like birds on the dear Barhis [i.e. the sacrificial grass, TK].”
On the “syntax of silence” in Proto-Indo-European 199
This point becomes clearer when one uses a generative model of description that I have applied to PIE several times in the last years.13 There are two main sentence structures (Figure 1) and (Figure 2) in the SOV-language PIE. If the verb is fronted to C-position (cf. (7b), (8b)) or if it remains in its original sentence final position in the first part of the sentence (cf. examples (5b), (6b)), then forward gapping can be applied in the second part of the coordinate construction. As pointed out in section 3. of this paper, I attribute this Janus-like behaviour of V in sentence final position to analogy (abduction). If the verb remains overt in sentence final position in the second part of the sentence, then only backward gapping can be triggered in coordinate constructions (examples (5a), (6a), (7a), (8a). In Table 1 and Table 2 the arrows indicate the direction(s) of the possible ellipses. CP (=S’) qp XP C’ TOPIC qp W1-W2 C IP (=S) ru XP I’ VP 6
Figure 1. Structure of the PIE sentence with arrows indicating possible directions of ellipses CP (=S’)
CP (=S’) W1 XP TOPIC1 qp XP C’ qp TOPIC2 W2 C IP (=S) ru XP I’ (I)
VP 6 ...........
Figure 2. Structure of the PIE sentence with Chomsky adjunction and arrows indicating possible directions of ellipses
4.4. Reconstructing the Proto-Indo-European situation I propose to reconstruct the situation of Latin, Greek and Vedic Sanskrit as the Proto-Indo-European state: a) PIE was an SOV language which allowed V-to-C-movement. b) PIE thus allowed not only for backward gapping (predicted by the direction of government in an SOV-language) but it allowed also for forward gapping. This peculiarity is reflected in Latin, Greek and Vedic Sanskrit. How does Hittite fit into this picture? I do not know of any good case in Old Hittite which would force us to assume V-to-C movement for this language. Perhaps the possibility of a forward ellipsis (5b) in Hittite represents a last piece of evidence in Hittite that V-to-C-movement once existed also in this branch of Indo-European. If this interpretation is correct, it would provide us with an instance where Hittite, the oldest attested AIEL, exhibits syntactic innovation.14
5. Ellipsis and information structure Apart from structural points of view (cf. sections 3. and 4. above), there exist a number of other approaches to deal with ellipsis in modern linguistic theory. I restrict myself here to approaches that interpret ellipsis as an epiphenomenon of the TOPIC-FOCUS distinction.15 This type of functional explanation in my view does not contradict structural views but supplements them. Kuno emphasizes the importance of TOPIC-FOCUS (/theme-rheme) for ellipsis in his famous “pecking order of deletion” (Kuno 1980:132) “Pecking order of deletion Delete order [sic! correct “older” TK] (less important) information first, and newer (more important) information last”. Klein’s formulation is a bit more precise: (10)
Klein (1993: 791) “Genau jene lexikalischen Einheiten, die eine beibehaltene Topik ausdrücken, können p-reduziert werden” (those and only those elements that express maintained topic/thematic information may be preduced).16
Merchant (2001) starts from a different approach but goes in a direction similar to Klein’s (1993) formulation. Merchant's focus condition on ellipsis (Merchant 2001: 38) reads “A constituent α can be deleted only if α is e-GIVEN” whereby e-GIVEN means (in a slightly simplified way) that
On the “syntax of silence” in Proto-Indo-European 201
there is an antecedent for α which entails the deleted constituent by an existentially bound variable and vice versa. If you take the sentence in (11), (11)
you can abstract an existentially bound variable of the form in (12) (12)
∃x sing (x)
In a sentence like (13) (with VP-ellipsis), (13)
Abby sang because Ben did SING.
one can recover the verb SING because an existentially bound variable like the one in (12) may be abstracted (“∃-type shifting”). This view makes it a bit easier to understand the possibility of “sloppy identity” (as exemplified by the Greek example (7b)). The verbal form édPke “he gave” (3Sg.Aor) appears in the first part of the coordinate construction and thus represents thematic material for the second part of the coordination. An interpretation of édPke analogous to the one in (12) would be something like “there exists an x who gives”. This abstraction of a variable evidently is strong enough to make it possible to interpret the elided form as a plural (perhaps by thinking about the variable x as a collective expression). With an example taken from German, Klein (1993: 774) claims that for forward ellipses only the “lexical content” is important and not morphological marking. For backward ellipsis, Klein (1993: 797) formulates the rule in (14): (14)
“Identisches Endstück in parallelen Konjunkten kann beim ersten Vorkommen p-reduziert werden” (the first instance of an identical piece of the final part in parallel coordinated constructions can be p-reduced).
In other words, Klein claims that there must be complete formal identity of the elided part and the overt part in backward ellipsis (cf. Klein 1993: 773). The ungrammatical German example (15) (Klein 1993: 774) confirms this claim:17 (15)
*weil ich because I trinkst drink.2Sg.Prs
und du and you
Our Vedic example [(8a), repeated as (16)], too, meets Klein's requirement in (14): the verbal form jayema “we want to win” which appears in the second part of the coordinate construction is exactly the form that is missing in the first part of the coordinate construction. (16)
Vedic Sanskrit RV 6,75,2
[email protected]à gx [email protected]¡ê jayema “with the bow (we want to win) the cows with the bow we want to win the battle.” [email protected]à
bow.INS.SG cow.ACC.PL.F WE.WANT.TO.WIN
But our Hittite example for backward ellipsis [(5a), repeated as (17)] shows sloppy identity in backward gapping: (17)
Hittite StBoT 12 (Neu 1970; Hittite thunderstorm ritual), Rs. III, 21´ G]IŠ DINNANA.GAL.GAL LÚ.MEŠIal-li-ri-e[ i-]Ia-mi-an-zi “The big Itarinstrument [probably a lyra, TK] (resounds), the priest-singers resound” LÚ.MEŠ GIŠ DINNANA GAL.GAL IŠ)AMAI Iallirie Itar-instrument big.big RESOUNDS priest-singer.NOM.PL iIamianzi resound.PRS.3PL
This means that Klein’s claim in (14) cannot be a universal one. Languages seem to differ in this respect. This can also be shown by the following Russian examples from Philippa Cook, taken from the description of a project of Lang (2006: 4): (18)
(Russian) a a. Ja naunuju sta'tju !ITAJU I scientific article READ.1SG but itaet. read.3SG vodu i b. Ja *PIL / PILA I *DRANK.SG.M / F water and vodku vodka
detektiv detective story
On the “syntax of silence” in Proto-Indo-European 203
Russian tolerates differences in person in backward ellipses (18a) but not differences in gender (18b). (18b) is only grammatical if ja “I” refers to a female person. The exact conditions for the appearance of backward gapping in AIELs and in PIE await further investigation. A thorough analysis of a large corpus of several AIELs will be necessary for such a project.
6. Object deletion as a phenomenon of Proto-Indo-European syntax This section deals with deleted objects in AIELs and with the possibility to reconstruct this type of ellipsis for the proto-language.
6.1. Latin data Consider first the Latin example in (19), taken from “Amphitruo”, a comedy by the greatest Roman comic playwright, Titus Maccius Plautus (254184 B.C.): (19)
Plaut. Amph. 387-392 (God Mercurius, having taken the shape of Sosia, a servant of Amphitruo, beats up Sosia, because Sosiaa (rightly) claims that hea is Sosia. Mercurius wants Sosiaa to deny that hea is Sosia ) Merc. Ego sum Sosia ille quem tu dudum esse aiebas mihi. Sos. Obsecro ut per pacem liceat te alloqui, ut ne vapulem. Merc. Immo indutiae parumper fiant, si quid vis loqui. Sos. Non loquar nisi pace facta, quando pugnis plus vales. Merc. Dic si quid vis, non nocebo. Sos. Tuae fide credo? Merc. Meae. Sos. Quid si falles? Merc. Tum Mercurius Sosiae iratus siet. “Merc. I am this Sosia about whom you just told me that you are him. Sos. I implore you that you allow me to speak to you in a friendly way, that you do not beat me up. Merc. Well, there shall be armistice for a short time if you want to say anything. Sos. I shall not speak unless peace has been made because you are the stronger one with your fists. Merc. Say SOMETHING, if you want TO SAY anything, I shall not harm YOU. Sos. May I believe in your honesty? Merc. Yes [lit. mine] Sos. What if you will deceive ME? Merc. Then Mercurius will be angry with Sosia.”
It is evident from this example (cf. the deleted material in the translation) that the immediate context gives us hints to interpret the elliptical elements. Let us first take a closer look at line 391: (20)
Plaut. Amph. 391 si quid vis Dic (ALIQUID) Say.IMP.2SG (SOMETHING) if anything.ACC.SG want.2SG.PRS LOQUI, non TIBI nocebo YOU.DAT.SG harm.1SG.FUT SAY.INF.PRS not
In (20), the imperative dic is construed without an object, an “ellipsis” provided for by the Latin lexicon (actually changing the semantics of dic to “speak”18), and with the sentential adjunct (a conditional clause) si quid vis. In verse 389 the verbal form loqui (which has to be supplemented in verse 391) appears in the phrase si quid vis loqui uttered by the same speaker (Mercurius). The verb loqui in verse 389 thus represents thematic, already known material and therefore neatly meets Klein’s criterion cited in (10). This type of verbal ellipsis resembles forward gapping in a sense but goes beyond it since the distance between the antecedent and the gap is bigger, there is no coordinate construction involved and the elliptic element is not the finite auxiliary vis but the infinite part of the verbphrase (loqui). The Latin verb nocFre “harm” is normally construed with a direct object in the dative case19 “to harm somebody”. An “unspecific” reading (see also note 15) without the object in the dative case is achieved by using the figurae etymologicae noxam nocFre or noxiam nocFre, which mean something like “to harm a harm”.20 Since this construction is not present in our case (20) we have to assume an ellipsis of the dative object TIBI here. This ellipsis is also based on given thematic material. All of the sentences uttered by Mercurius before our example (20) contain a pronoun of the second person singular either explicitly (cf. (19) verse 387) or implicitely [as “pro” in the verbal form vis in verse 389 and in the verbal forms dic and vis in the immediate context before nocebo, cf. also (19)]. In Latin, the dative object with the verb nocēre behaves like a normal direct object. As a good piece of evidence for that one can show that, in passive constructions, this dative may be changed into a nominative case: (21) Vitr. 2,9,14 Larix ... ab suci vehementi amaritate ab carie aut tinea non nocetur ... “The larch ... is not harmed by decay or by a caterpillar because of the enormous bitterness of its sap ...” Larix .... non nocetur not harm.PRS.PASS.3SG larch.NOM.SG
On the “syntax of silence” in Proto-Indo-European 205
In verse 392 of (19), repeated as (22), the object of the verb falles (which normally governs an accusative case) is missing: (22)
Plaut. Amph. 392 Quid si (ME) What if (I.ACC.SG )
Also this object can be easily retrieved by the context of a dialogue situation.21 The omission of direct objects is such a common phenomenon in Latin that quite complex constructions can be found, cf. (23). Here the verbs with missing objects are put in bold letters. The normal valency frames of these verbs are listed in (24). (23)
Tac. Ann. 12,46,18 – 12,47,1 (cf. Luraghi 1997: 250) qua necessitate Mithridates diem locumque foederi accepit castelloque egreditur. (47) Ac primo Radamistus in amplexus eius effusus simulare obsequium, socerum ac parentem appellare; adicit ius iurandum, non ferro, non veneno vim adlaturum; simul in lucum propinquum trahit ... “Under this compulsion, Mithridates accepted a day and a place for the completion of a contract and quitted [lit. “quits”, TK] the fortress. (47) And first, Radamistus threw himself into his embraces, simulated [historical infinitive, TK] respect, called [historical Infinitive, TK] HIM father-in law and father, swore [lit. “swears”, TK] an oath too that he would do no violence TO HIM, neither by sword nor by poison. At the same time he drew [lit. “draws”, TK] HIM into a neighbouring grove ...” appellare... socerum ac parentem EUM father-in-law.ACC.SG and father.ACC.SG HE.ACC.SG.M call.INF vim non ferro non veneno EI not sword.ABL.SG not poison.ABL.SG HE.DAT.SG.M violence.ACC.SG adlaturum ESSE22 bring.ACC.SG.M. PTCP.FUT BE.INF ... in lucum propinquum simul EUM at the same time HE.ACC.SG in grove.ACC.SG nearby.ACC.SG trahit draw.3SG.PRS
a. appellare “call, refer to as” (+ acc (someone) + acc (as something/ someone)) Tac. Ann. 4,52,19 Afer primoribus oratorum additus, divulgato ingenio et secuta asseveratione Caesaris qua suo iure disertum eum appellavit. “Afer was ranked among the foremost orators, through his publicly known ability and through the subsequent assertion of the emperor by which he called him ‘eloquent by his own right’”. suo iure disertum eum appellavit own.ABL.SG right.ABL.SG eloquent.ACC.SG him.ACC.SG he.called b. vim afferre “do violence to” ([lit. vim (violence.ACC) afferre (carry.to)] (with dat.)23 Liv. 39,54,6 (an accusativus cum infinitivo construction) oppidum quoque aedificare coepisse, quod indicium esset nec agro nec urbi ulli vim adlaturos venisse. “(they [scil. the Gauls, TK ] said that) they also had started to build a town, and that would be an indication for the fact that they had not come in order to do violence to the country or to any town.” nec agro nec urbi ulli and.not country.DAT.SG and.not town.DAT.SG any.DAT.SG vim adlaturos venisse carry.to.PTCP.FUT.ACC.PL come.INF.PERF violence.ACC.SG c. trahere “draw” + acc. (someone / something) + prepositional phrase (denoting the GOAL) Tac. 4,21,4 Et spreta potentia Augustae trahere in ius Urgulaniam domoque principis excire ausus erat. “He also dared to hand over Urgulania to the courts neglecting the power of Augusta and to summon (her) from the palace of the emperor.” trahere in ius Urgulaniam Urgularia.ACC.SG.F draw into law.ACC.SG.N
Thus, in the case of appellare in (23) a pronoun (3SG.M) in the accusative is missing, in the case of vim afferre in (23) there is an ellipsis of a pronoun (3SG.M) in the dative and in the case of trahit in (23), a pronoun of the accusative (3SG.M) is missing, all of which refer to the same person addressed by the form eius “his” (lit. “of him”, GEN.SG.M of the anaphoric personal pronoun of the third person singular is, ea, id) in the immediate context before. Eius, again, (carrying thematic material, as an “anaphoric” pronoun) refers back to Mithridates in the sentence before. Thus, we have a neat “chain” of thematic material here, first producing an anaphoric element (eius) and then a series of null elements.
On the “syntax of silence” in Proto-Indo-European 207
6.2. Greek data The same type of object ellipsis is ubiquitous in Homeric Greek. Consider example (25). The verbs with missing objects are put in bold letters. (25)
Hom. Il 5, 22-24 (Dares, a priest of the god Hephaistos, has two sons, Phegeus and Idaios. In the Troian war, they fight against Diomedes. Diomedes kills Phegeus and would have killed Idaios as well, but Hephaistos rescues Idaios)
oudè gàr oudè ken autòs hypékphyge kËra mélainan, / all’ HÇphaistos éryto, sáPse dè nyktì kalýpsas, / hPs dÇ hoi m¼ págchy gérPn akachÇmenos eíē. “Nay, he [Idaios, TK] would himself not have escaped the black goddess of death, but Hephaistos guarded HIM, rescued HIM, enfolding HIM in darkness so that the aged one [= Dares TK] might not be utterly fordone with grief ”24 all' MIN HÇphaistos éryto, sáPse but HE.3SG.ACC Hephaistos guard.3SG.IPF.MID rescue.3SG.AOR nyktì MIN kalýpsas dè MIN night.LOC.SG HIM hide.PTCP.AOR.NOM.SG.M and HIM
Here, again, we are dealing with a thematic “chain”. The name Idaios is the explicit subject of the sentence before (Hom. Il. 5,20). The anaphoric pronoun autós refers back to this person. The following missing direct objects to the verbal forms éryto, saPse, and kalýpsas all refer back to the same person. The normal valency frames of the verbs with elliptic objects in (25) are exemplified in (26): (26)
a. erý(e)sthai “guard” + acc. (someone) Hom. Il. 13,554–555 péri gár ra PoseidáPn enosíchthPn / Néstoros hyiòn éryto “for mightly did Poseidon, the Shaker of Earth, guard Nestor’s son” péri gár ra PoseidáPn all.around for indeed Poseidon.NOM.SG enosíchtPn Néstoros hyiòn Shaker.of.Earth.NOM.SG Nestor.GEN.SG son.ACC.SG éryto guard.3SG.IPF.MID
Thomas Krisch b. sᮼzein “rescue”+ acc. (someone)
óphra kaì Automédonta saÔseton ek polémoio “that you two [scil. two horses TK] will also rescue Automedon out of the war” óphra kaì Automédonta saÔseton that also Automedon.ACC.SG.M rescue.2DU.FUT ek polémoio out.of war.GEN.SG c. kalýptein “hide” + acc. (someone/something) Hom. Il. 4,461 tòn dè skótos ósse kalypse “and darkness enfolded his eyes” (lit. and darkness enfolded him, the eyes”25) tòn dè skótos ósse kalýpse he.ACC.SG.M and darkness.NOM.SG eye.ACC.DU hide.3SG.AOR
The null object construction is not allowed in the languages continuing Latin and Ancient Greek, viz. the Romance languages and Modern Greek.26 On the other hand, the attestation of the “null object” type of ellipsis in Ancient Greek and in Latin make this construction a candidate for PIE reconstruction.
6.3. Vedic data The data from Vedic Sanskrit fit in with what was just said about Latin (6.1.) and Greek (6.2.), cf. (27). The verbs with missing objects are put in bold letters. (27)
RV 2,35,1 ( cf. van der Wurff 1997: 345)
£pem asçkùi vàjay£r vacasyxê [email protected] dadhãta nàdy¢ g¡ro me | apxê [email protected]àd à÷uh‚mà kuv¡t [email protected] sup‚÷asas karati j¢ùiùad dh¡ || “Eager for reward I have poured out (my) eloquence. The child of the river may accept my songs [gíro] with favour. Will he, the rapidly rushing Apàm Napàt [name of a god in the waters, lit. “offspring of the waters” TK], make THEM [scil. the songs, TK] well adorned? Will he enjoy THEM [scil. the songs, TK]?” 27
water.GEN.PL offspring.NOM.SG.M rapidly.rushing.NOM.SG.M
make 3SG.AOR.SBJV THEY.ACC.PL.F enjoy.3SG.AOR.SBJV EMPH
On the “syntax of silence” in Proto-Indo-European 209
There is thematic continuity of the elided null objects with the word for “songs” (gíro) in the first line of our example (27). The thematic “chain” is not as “dense” as it was in the Latin and the Greek examples (23) and (25). There is no “intermediate” stage with an overt pronoun in (27). The normal valency frames of the verbs with elliptic objects in (27) are exemplified in (28): (28)
a. kar- “make” + acc. (someone/something) + secondary predicate (adjective) RV 10,18,6 [email protected] [email protected]ùñà [email protected]à saj¢ùà dã[email protected] xyuþ karati jã[email protected] vaþ “May [email protected]ùñç [Indian god of creation, TK], giver of good birth, being gracious, make long the life-span for your life here.”
[email protected] [email protected]ùñà
here [email protected]ùñç.NOM.SG.M
life.DAT.SG. you.GEN./DAT.PL.CLIT make 3SG.AOR.SBJV b. joù - “enjoy” + acc. (something)29 RV 1,25,18 etx juùata me g¡raþ “He shall enjoy these songs from me”
I.GEN / DAT. SG (CLT)
6.4. Hittite data In Hittite, there are not many examples for null objects. The specialist for Hittite syntax, Silvia Luraghi (2005: 244) remarks “Transitive verbs ... only very infrequently occur with N[ull] O[bject]s”. She brings an example from the relatively young (Neo-Hittite) Annals of Mursilis for this phenomenon (Luraghi 2005: 244) and mentions the Old Hittite law texts (Luraghi 2005: 242), where null objects occur. In addition to that I have found several attestations for this construction in the Old Hittite thunderstorm ritual. Three examples are cited in (29). The verbs with missing objects are marked with bold letters.
(29) StBot 12 (Neu 1970, Hittite thunderstorm ritual) RS. III, 44'–46' LÚ SÌLA.ŠU.DU8 NINDAIar-ši-in EM-A GE6 / LUGAL-i pa-a-i ta [email protected]ši-K[a LÚ ] SÌLA.ŠU.DU8 e-ip-zi ta LÚ GIŠBANŠUR pa-a-i .... “The cupbearer gives sour dark bread to the king and he cuts IT. The cupbearer takes IT and he gives IT to the table servant ...” LÚ SÌLA.ŠU.DU8 NINDAIarši-n EMA GE6 30 Cupbearer bread-ACC.SG.COMM sour.ACC black AN LUGAL-i pài ta king-DAT.SG give.3SG.PRS and IT-ACC.COMM. (CLT) LÚ SÌLA.ŠU.DU8 AN paršiKa cut.3SG.PRS.MID cupbearer IT.ACC.SG.COMM. (CLT) ēpzi ta (AŠ)31 take.3SG.PRS and (HE.NOM.SG.COMM. (CLT)) LÚ GIŠBANŠUR pài AN give.3SG.PRS IT.ACC.SG.COMM. (CLT) table.servant32
The first sentence of (29) shows the full valency of the verb “to give”, piKe-IIi (appearing in the form pài).33 The last part of the cited text exhibits ellipsis of the accusative object with the same verb. In between these two sentences there are two further examples of null objects (elliptical accusative objects). In all of these cases there is thematic continuity with the word for “bread” in the first line of the cited text. Like the ègvedic example (27), the Hittite example in (29) does not show “intermediate” pronominalization. In (30) one can find examples for the normal valencies of the verbs paršKe-a “cut” and ēpp/ app- “take, seize” in Old Hittite: (30) a. paršKe-a “cut” + acc. (something) StBoT 12, (Neu 1970; Hittite thunderstorm ritual) RS IV 25' NINDA Iar-ši-in-na pár-ši-Ka “and he cuts bread” NINDA Iaršinn-a páršiKa bread-ACC-and cut.3SG.PRS.MID b. ēpp/ app- “take, seize” StBoT 8 (Otten / Souek 1969; Old Hittite ritual for the royal couple) Vs. II 19 ma-a-an MUŠENIa-a-ra-na-an Iu-š[(u-Wa-an-da-an ap-pa-an-zi)] ...“if one seizes [lit. “they seize”, TK] a living eagle ...” màn MUŠENIàrana-n IušuWanda-n app-anzi if eagle-ACC.SG.COMM living-ACC.SG.COMM seize-3PL.PRS
On the “syntax of silence” in Proto-Indo-European 211
6.5. Germanic data (?) Like the modern Romance languages (cf. section 6.2. above), the modern Germanic languages are said to lack clear instances of object ellipsis, whereas this phenomenon is claimed to exist in older Germanic dialects.34 But there are many questions open here. We can cite one instance from Old High German which suggests, in our view erroneously, that this older state of German could have been more permissive in this respect than modern German, namely the Straßburg oaths: (31)
Old High German Straßburg oaths (Steinmeyer 1916=1971: 82) Oba Karl then eid, then er sinemo bruodher Ludhuuige gesuor, gileistit, indi Ludhuuuig, min herro, then er imo gesuor, forbrihchit, ob ih inan es iruuenden nemag ... “If Charles keeps the oath which he swore to his brother Ludwig, and Ludwig, my lord, (on his part) breaks THE OATH / IT which he swore to him, if I cannot prevent him from it, ....”
In Modern German the corresponding sentence with ellipsis would be ungrammatical: (32)
Modern German *... und (wenn) Ludwig, mein Herr, DEN EID/den, den er ihm schwor, bricht,
A grammatical rendering of (32) in Modern German would be .... und (wenn) Ludwig, mein Herr, den (Eid), den er ihm schwor, bricht (with an overt correlative pronoun (den) as head of the relative clause). The Romance text of the oath shows an overt pronoun in the accusative (lo) in the corresponding construction.35 In my opinion, the lack of a (correlative) pronoun as head of a relative construction in the Old High German text (32) [Ludhuuuig, min herro, then (relative pronoun) er imo gesuor] could have been immediately influenced by Latin which showed such constructions.36
6.6. Reconstruction The examples taken from AIELs and discussed in sections 6.1–6.4. showed deletion of the object. This leads us to the conclusion that PIE had a rule that allowed deleting objects which refer back to already mentioned “thematic” material. This “null object” type of ellipsis thus is operating in a
forward direction and meets Klein’s criterion cited in (10). Modern IndoEuropean languages like the Romance languages, the Germanic languages (also partly in their old attestations) and Modern Greek do not permit null objects any more (cf. sections 6.2. and 6.5).
7. Conclusions a)
Gapping was possible in the ancient Indo-European languages and can be reconstructed for Proto-Indo-European. It operates backwards and forwards in these languages and in the protolanguage and it is determined by syntactic (basic word order) factors and by factors of functional sentence perspective. Object ellipsis was possible in the ancient Indo-European languages and in Proto-Indo-European. It operates in forward direction in these languages and in the protolanguage and it is determined by factors of functional sentence perspective.
Appendix: Abbreviations AIEL=Ancient Indo-European Language; Amph.=Amphitruo; Ann.=Annales; AOR=Aorist; Ar.=Aristophanes; Bell. Gall.=De Bello Gallico; C=position of the “Complementizers”; Caes.=Caesar; CLT=clitic; COMM=genus commune; CP=Complementizer Phrase; EMPH=emphasizing particle; Hdt.=Herodotos; Hom.=Homeros; Il.=Ilias; INJ=injunctive; INTERR=interrogative particle; IP=Inflection Phrase; IPF=imperfect; Liv.= Livius; MID=middle voice; OPT=optative. Mil.=Miles Gloriosus; PERF=perfect; Ph.=Phormio; PIE=Proto-Indo-European; Plaut.=Plautus; preduced=phonologically reduced; Ra.=Ranae; RV=ègveda; StBoT=Studien zu den Bo÷azköy–Texten; Tac.=Tacitus; Ter.=Terentius; TK=Thomas Krisch; TOPIC=Topic or contrastive focus position; TOPIC1=topic position; TOPIC2=focus-position; Vitr.=Vitruvius; VP=verbal phrase; W=Wackernagel particle (W1= enclitical sentence connectors; W2= other enclitics); XP=any phrase
On the “syntax of silence” in Proto-Indo-European 213
Interlinear glosses of examples follow the Leipzig glossing rules to be found at the following URL: http://www.eva.mpg.de/lingua/resources/glossingrules.php (April 2, 2008). Additional abbreviations used in this article are listed as an appendix before the notes. Many thanks go to Jürg Fleischer for numerous suggestions, to Thomas Lindner for helping the author with the Straßburg oaths, to Hermann Bieder for confirming the Russian data in example (18), to an anonymous reviewer of this article for some useful hints and to Christina Katsikadeli and Stefan Niederreiter for proofreading the text. All remaining shortcomings remain in the responsibility of the author, of course. Cf. e.g. the following examples for the “right hand” which show nominalization of the word for “right” and do not express the (feminine) word for “hand” (put in parentheses) lat. dext(e)ra f. (scil. manus), germ. die Rechte f. (scil. Hand), gr. dexia/ f. (scil. cheír); ved. [email protected]ùiõa- m. (scil. hásta-). A similar case is the ellipsis of the head noun in the following ancient Greek syntagm és Haidou “to the (houses of) Hades” (e.g. Ar. Ra. 69). Blakemore (2002: 71): “Pragmatics does not simply enter when linguistic decoding fails; on the contrary, the linguistic system is subservient to pragmatic inference in the sense that it functions as an aid to the inferential system.” Of course, I am aware of the fact that the exact semantic interpretation of English but and German aber is more complex, as is clear from the analyses of Abraham (1979: e.g. 92) and Lang (1977: e.g. 168). As is well known, the term “implicature” has been made popular in linguistics by Grice (1975). The concept of conventional implicature is viewed upon in a critical way by Bach (2006; in press), especially p. 14 (cited after http//userwww.sfsu.edu/~kbach/TopTen.pdf (seen April 2, 2008)). Bach wants to get rid of the notion of “conventional implicature” which is “counterintuitive” for him because to his opinion the conjunction but states something rather than implicates something. I cannot follow this way of argumentation, though. By hearing aber “but” the hearer starts worrying and expects additional information. Cf. also Maling (1972: 103). Of course, not every case of V-to-C movement looks like an SVO structure. Consider the following German coordinate sentences where the first position in the sentences is occupied by an adverb: Gestern liebten die Kinder Fisch, aber heute lieben sie Yesterday loved the children fish but today love they Gemüse. vegetables “Yesterday the children loved fish but today they love vegetables”. Cf. e.g. Krisch (1984: 44–47); Krisch (1992: 157) with reference to Andersen (1973).
12. 13. 14. 15. 16.
Thomas Krisch The Hittite texts are first given in their cuneiform spelling using the traditional transcription system. Sumerograms are given in capital letters, as determinatives they are written in upper case. Akkadograms appear in capital italicized letters and Hittite words appear in syllabic writing with a hyphen between the syllables (broad transcription). In glossed words the hyphens of syllable writing are removed (“narrow transcription”) and if hyphens are used there, they indicate morpheme boundaries. The importance of these daughter-languages for the syntactic reconstruction of PIE lies in the fact that they are the oldest attested AIELs and that we have a great amount of texts at our disposal from all of them. Since backward gapping is only attested in SOV languages (cf. section 3.) this may in turn count as quite safe diagnostics for the status of these languages as SOV. Cf. also Krisch (2001: 169–170); Krisch (1997: 305); McCone (1979: 224). Krisch (1998, 2002, 2004). Another case of syntactic innovation in Hittite is proposed in Krisch 1990: 72. Cf. e.g. Kuno (1982); Klein (1993); Merchant (2001). By p(honologically)-reduced Klein means that ellipsis takes place in the phonological component of grammar and that the elided material is still present in syntax. Forward ellipsis, on the other hand allows sloppy identity of person in German. The sentence “weil ich Bier trinke und du Wein TRINKST” is grammatical (cf. Klein 1993: 774). This type of ellipsis in the lexicon creates an “unspecific” reading of the verb and this type of ellipsis of an object normally is not counted as “null object” (cf. e.g. Luraghi 2005: 235). Therefore, in example (20), I have put the elided object (SOMETHING) into parentheses. We shall mainly deal with “referential null objects” (Luraghi 2005: 235) here, but, admittedly, it is not always clear in the texts whether one has to do with an “unspecific reading” (a possibility provided for by the lexicon) or with a “referential null object”. This dative object originally was an adjunct with “malefactive” meaning. It is a causative formation to the PIE root *ne,- (“to get lost”, “to vanish”, “to die”) with the original meaning “bring death to someone”. Cf. LIV (2001: 452) with footnote 9. Cf. e.g. tab. XII, 2a (Law of the Twelve Tables) Si servus furtum faxsit noxiamve noxit ... lit. “If a slave has committed theft or has harmed a harm ...” The interpretation of fallere as an optionally intransitive verb with the meaning “afflicting harm” (an “unspecific reading”) already provided for by the lexicon cannot be ruled out completely here. Therefore, in (22), I put the elided object (ME) into parentheses. Cf. also footnote 18. This is an AcI-construction dependent from adicit ius iurandum “he adds an oath”. Adlaturum esse is the periphrastic future infinitive of afferre in Latin.
On the “syntax of silence” in Proto-Indo-European 215 23. Afferre and the form adlaturos (PART. FUT) are suppletive members of a paradigm. Our example (24b), taken from Livy exhibits this suppletive future participle like the Tacitus-example (23). 24. I do not translate hoi (Dat.Sg.M) “him” here in order to avoid confusion. Hoi refers to the old man. 25. The schema kat' holon kai meros (“Construction of the whole and the part”) is a typical stylistically marked construction in Ancient Greek and in other AIELs. cf. e.g. Smyth (1963=1956=1920: 267): “... a verb may take two objects, one denoting the person, the other the part especially affected by the action.” 26. For a more detailed analysis cf. e.g. Luraghi (2005: 246). 27. Geldner (2003=1951: 321) interprets both of the interrogative clauses of our translation as emphatic assertions. Hettrich (1988: 154) translates only one of our interrogative clauses as interrogatrive and the second one as emphatic. 28. The Sanskrit word gír- “song” is a feminine noun. 29. There is also an example with instrumental case (RV 10,6,4) and one example with mixed cases (accusative and instrumental RV 5,39,4). The forms of this verb have to be analysed all as forms originally belonging to a middle paradigm, cf. Joachim 1978: 82. 30. Sumerograms (Logograms) often are not characterized morphologically and can serve several functions. In this case, the Sumerogram functions as a nominative case. 31. As a rule in Hittite “no subjects of transitve verbs are ever cliticized” (Luraghi 1990: 41). Therefore I put the deleted nominative clitic (AŠ) in parentheses. 32. In this case, the Sumerogram functions as a dative case (cf. also note 30). 33. The Hittite verbal forms are cited in the form in which they are cited in Oettinger 1979. 34. Cf. Luraghi (2005: 247–248) for Old English and Old Norse examples. 35. Si Lodhuuigs sagrament, quĊ son fradre Karlo iurat, conservat, et Karlus meos sendra de suo part ñ lostanit, ... . Having consulted the facsimile [in Becker 1972: 28–29 and the facsimile to be found at the URL http://www.fh-augsburg.de/~harsch/germanica/Chronologie/09Jh/Strassburger Eide/eid_text.html (seen 3 April 2008) (Enneccerus 1897: 24–26, 34–36 non vidi)] I agree with the interpretation of the much discussed (cf. e.g. Elcock 1960: 335–336, especially footnote 1 on these pages) ñ lostanit in the cited passage as ñ lo franit [in the late Carolingan minuscle writing ductus the group fr may look very much like st; franit can be a writing variant for fraint “breaks” (Thomas Lindner, Salzburg, personal communication; cf. also Elcock 1960: 339)], whereby ñ (probably an abbreviation for the dative of nomen. “name”) is a fill-in for a proper noun, in this case Lodhuuige (in the dative case). The Romance text can be translated as follows: “If Ludwig keeps the oath that he swore to his brother Charles, and Charles, my lord, breaks (franit) it (lo) to Ludwig (ñ), as far as he is concerned... ”. This text shows an accusative pronoun (lo) that refers back to sagrament “oath”. There is no relative clause in the Romance text.
36. The construction without the pronominal head was possible in Latin, cf. Plaut. Mil. 367–368 PHIL(OCOMASIUM) Tun me vidisti? SCE(LEDRUS) Atque his quidem hercle oculis. PHIL Carebis OCULIS, credo, qui plus vident quam ID quod vident “PHIL You saw me? SCE And even, alas, with these eyes PHIL You will have to do without THE EYES which see more than THAT what they see” (cf. also Hoffmann / Szantyr (1965: 555–556). No Latin text is attested for the Straßburg oaths but one may assume that it existed, because the speech of Ludwig to the troops is handed down in Latin only (cf. also de Boor 1971: 48). Elcock (1960: 337–338) even goes so far as to reconstruct the Latin original of the text. If this assumption (not representing the current commnis opinio, cf. Schmidt-Wiegand 1995: 379) is correct, then the use of the ellipsis in the Old High German text (without a correlative demonstrative pronoun) could have been immediately influenced by the Latin original.
References Abraham, Werner 1979 But. Studia Linguistica 23: 89–119. Andersen, Henning 1973 Abductive and Deductive Change. Language 49: 765–793. Bach, Kent 2006 The Top 10 Misconceptions about Implicature. In Drawing the boundaries of Meaning: Neo-Gricean Studies in Pragmatics and Semantics in Honor of Laurence R. Horn, ed. Betty Birner and Gregory Ward. Amsterdam/Philadelphia: John Benjamins. Cited after http://userwww.sfsu.edu/~kbach/TopTen.pdf (seen 2 April 2008). Becker, Siegfried 1972 Untersuchungen zur Redaktion der Straßburger Eide. Frankfurt a.M.: Peter Lang. Blakemore, Diane 2002 Relevance and Linguistic Meaning: The semantics and pragmatics of discourse markers. Cambridge: Cambridge University Press. de Boor, Helmut 1971 Die deutsche Literatur von Karl dem Großen bis zum Beginn der höfischen Literatur. Mit einem bibliographischen Anhang von Dieter Haacke. 8. Auflage. München: C.H. Beck‘sche Verlagsbuchhandlung. Elcock, W. D. 1960 The Romance Languages. London: Faber & Faber Limited. Enneccerus, Magda 1897 Die ältesten deutschen Sprach-Denkmäler in Lichtdrucken. Frankfurt a.M.: F. Ennecerus.
On the “syntax of silence” in Proto-Indo-European 217 Gaeta, Livio and Silvia Luraghi 2001 Gapping in Classical Greek Prose. Studies in Language 25: 89–113. Geldner, Karl Friedrich 2003 Reprint. Der Rig-Veda aus dem Sanskrit ins Deutsche übersetzt und mit einem laufenden Kommentar versehen. Cambridge, Massachusetts and London, England: Harvard University Press. Original edition, Cambridge, Massachusetts: Harvard University Press, 1959. Grice, H. Paul 1975 Logic and Conversation. In Syntax and Semantics 3: Speech Acts, eds. Peter Cole and Jerry L. Morgan, 41–58. New York/San Francisco/London: Academic Press. Hettrich, Heinrich 1988 Untersuchungen zur Hypotaxe im Vedischen. Berlin/New York: Walter de Gruyter. Hoffmann, J. B. and Anton Szantyr 1965 Lateinische Syntax und Stilistik. München: C.H. Beck‘sche Verlagsbuchhandlung. Joachim, Ulrike 1978 Mehrfachpräsentien im ègveda. Frankfurt a.M.: Peter Lang. Klein, Wolfgang 1993 Ellipis. In Syntax: Ein internationales Handbuch zeitgenössischer Forschung, eds. Joachim Jacobs, Armin von Stechow, Wolfgang Sternefeld and Theo Vennemann. 1. Halbband, 763–799. Berlin/New York: Walter de Gruyter. Krisch, Thomas 1984 Konstruktionsmuster und Bedeutungswandel indogermanischer Verben: Anwendungsversuche von Valenztheorie und Kasusgrammatik auf Diachronie und Rekonstruktion. Frankfurt a.M.: Peter Lang. 1990 Das Wackernagelsche Gesetz aus heutiger Sicht. In Sprachwissenschaft und Philologie: Jacob Wackernagel und die Indogermanistik heute, eds. Heiner Eichner and Helmut Rix, 64–81. Wiesbaden: Ludwig Reichert Verlag. 1992 Analogische Prozesse in der lateinischen Sprachgeschichte. In Latein und Indogermanisch: Akten des Kolloquiums der Indogermanischen Gesellschaft, Salzburg, 23.–26. September 1986, eds. Oswald Panagl and Thomas Krisch, 155–181. Innsbruck: Institut für Sprachwissenschaft. 1997 Delbrücks Arbeiten zur Wortstellung aus heutiger Sicht. In Berthold Delbrück y la sintaxis indoeuropea hoy: Actas del Coloquio de la Indogermanische Gesellschaft Madrid, 21–24 de septiembre de 1994, eds. Emilio Crespo and José Luis García Ramón, 283–309. Madrid: Ediciones de la UAM and Wiesbaden: Ludwig Reichert Verlag.
Zum Hyperbaton in altindogermanischen Sprachen. In Sprache und Kultur der Indogermanen: Akten der X. Fachtagung der indogermanischen Gesellschaft Innsbruck 22.–28. September 1986, ed. Wolfgang Meid, 351–384. Innsbruck: Institut für Sprachwissenschaft. 2001 „Man kann sich ein Klavier ja auch um den Bauch binden“. In Fremd und Eigen: Untersuchungen zu Grammatik und Wortschatz des Uralischen und Indogermanischen in memoriam Hartmut Katz, eds. Heiner Eichner, Peter-Arnold Mumm, Oswald Panagl, Eberhard Winkler with collaboration by Roland Hemmauer, Susanne Knopp and Velizar Sadovski, 155–174. Wien: Edition Praesens. 2002 Indogermanische Wortstellung. In Indogermanische Syntax: Fragen und Perspektiven, ed. Heinrich Hettrich with collaboration by JeongSoo Kim, 249–261. Wiesbaden: Ludwig Reichert Verlag. 2004 Some aspects of word order and sentence type From Indo-European to New High German. In ANALECTA HOMINI UNIVERSALI DICATA. Arbeiten zur Indogermanistik, Linguistik, Philologie, Politik, Musik und Dichtung: Festschrift für Oswald Panagl zum 65. Geburtstag, eds. Thomas Krisch, Thomas Lindner and Ulrich Müller with editorial collaboration by Michael Crombach, Stefan Niederreiter, Helga Panagl and Ursula Pavii. Vol I, 106–129. Stuttgart: Akademischer Verlag Hans-Dieter Heinz. Kuno, Susumo 1980 Functional Syntax. In Syntax and Semantics 13: Current Approaches to Syntax, eds. Edith Moravcsik and Jessica R. Wirth, 117–135. New York/London/Toronto/Sydney/San Francisco: Academic Press. Lang, Ewald 1977 Semantik der koordinativen Verknüpfung. Berlin: Akademie Verlag. 2006 Parallelismus in der Grammatik: Reichweite, Status, Herkunft. Förderungsantrag (Antrag auf Gewährung einer Sachbeihilfe) P 6. Berlin: Humboldt-Universität. Cited after http://www.zas.gwz-berlin.de/ research/projects/p6.pdf (seen April 3, 2008). LIV 2001 Lexikon der indogermanischen Verben: Die Wurzeln und ihre Primärstammbildungen, ed. Helmut Rix with collaboration by Martin Kümmel, Thomas Zehnder, Reiner Lipp and Brigitte Schirmer. 2d. revised and extended ed. by Martin Kümmel and Helmut Rix. Wiesbaden: Ludwig Reichert Verlag. Luraghi, Silvia 1990 Old Hittite Sentence Structure. London/New York: Routledge. 1997 Omission of the Direct Object in Latin. Indogermanische Forschungen 102: 239–257.
On the “syntax of silence” in Proto-Indo-European 219 2005
Null Objects in Latin and Greek and the Relevance of Linguistic Typology for Language Reconstruction. In Proceedings of the Fifteenth Annual UCLA Indo-European Conference, eds. Karlene Jones-Bley, Martin E. Huld, Angela Della Volpe and Miriam Robbins Dexter, 234–256. Wahington D.C.: Institute for the Study of Man. Maling, Joan M. 1972 On gapping and the order of constituents. Linguistic Inquiry 3: 101– 108. McCone, Kim Robert 1979 Aspects of Indo-European Sentence Patterns and Their Role in the Constitution of the Old Irish Verbal System. Oxford Phil. Diss. Merchant, Jason 2001 The syntax of silence: Sluicing, Islands and the Theory of Ellipsis. Oxford: Oxford University Press. Neu, Erich 1970 Ein althethitisches Gewitterritual. Wiesbaden: Otto Harrassowitz. Oettinger, Norbert 1979 Die Stammbildung des hethitischen Verbums. Nürnberg: Verlag Hans Carl. Otten, Heinrich and Vladimir Souek 1969 Ein althethitisches Ritual für das Königspaar. Wiesbaden: Otto Harrassowitz. Praust, Karl 2003 A Missing Link of PIE Reconstruction: The Injunctive of *H1es- ‘to be’. In Proceedings of the Fourteenth Annual UCLA Indo-European Conference, eds. Karlene Jones-Bley, Martin E. Huld, Angela Della Volpe and Miriam Robbins Dexter, 112–144. Washington D.C.: Institute for the Study of Man. Richards, Norvin 1998 The Full Pursuit of the Unspeakable. In NELS 28 (2): 153–168. Ross, John Robert 1970 Gapping and the order of constituents. In Progress in Linguistics: A Collection of Papers, eds. Manfred Bierwisch and Karl Erich Heidolph, 249–259. The Hague/Paris: Mouton. Schmidt-Wiegand, Ruth 1995 ‚Straßburger Eide‘. In Die deutsche Literatur des Mittelalters: Verfasserlexikon, ed. Burghart Wachinger, Vol. IX, 377–380. Berlin/New York: Walter de Gruyter. Smyth, Herbert Weir 1963 Reprint. Greek Grammar. 2d. ed., revised by Gordon M. Messing. Cambridge: Harvard University Press. Original edition, Cambridge: Harvard University Press, 1956.
Steedman, Mark 2000 The Syntactic Process. Cambridge, Massachusetts: MIT Press. Steinmeyer, Elias von 1971 Reprint. Die kleineren althochdeutschen Sprachdenkmäler. Dublin/Zürich: Weidmann. Original edition, Berlin: Weidmann, 1916. van der Wurff, Wim 1997 Syntactic reconstruction and reconstructability. Proto-Indo-European and the typology of null objects. In Linguistic Reconstruction and Typology, ed. Jacek Fisiak, 337–355. Berlin/New York: Mouton de Gruyter.
Information-structural categories in the main texts of early German inheritance
Word order variation and information structure in Old High German: An analysis of subordinate dhazs-clauses in Isidor Eva Schlachter
Abstract One of the earliest texts written in Old High German exhibits an extraordinarily high percentage of verb-early position in subordinated clauses introduced by dhazs. This paper shows that copula and auxiliary verbs end up more frequently in a higher clause position than main verbs. It demonstrates that these verb types occur with different information-structural values and concludes that they relate to different sentence structures. At the same time it is a contribution to the philological debate about differences between the so-called quotation syntax and the rest of the text.1
1. Introduction Until recently it was considered to be a well-established fact that Old High German already displayed the same verbal syntax as Modern German (Fourquet 1974, Lenerz 1984): in main clauses the finite verb shows up in second position (Verb-second) while in subordinate clauses it occurs in final position (Verb-end). This complementary distribution led to the bynow classic generative analysis of den Besten (1977): in main clauses the finite verb ends up in the Complementizer-position (linke Satzklammer ‘left sentence brace’), while in subordinate clauses it remains in the final INFLposition (rechte Satzklammer) since the Comp-position is blocked by a lexical complementizer. Old English, which shows a similar distribution, received an analogous analysis in Kemenade (1987). However, more recent studies on Old and Middle English (Pintzuk 1993, 1999; Kiparsky 1996; Fuß and Trips 2002) support the view that there are more positions than the ones standardly assumed. There is a high number of embedded clauses showing a non-final verb position, which is identified with an INFL-medial grammar.2 Since
there are at the same time many cases of subordinations with the finite verb in final position, it is assumed that INFL-final and INFL-medial grammars coexisted side-by-side in the same speaker (Kroch 1989). The ultimate change, then, is seen as the result of a competition process between these two grammars.3 For Old High German, however, contradictory data have been known for more than 100 years. Diels (1906: 164) considered early verb placement (or Verb-second) in subordinate clauses as a quite common fact, triggered by the intonation of the first “word” (1906: 168). Rannow, who investigated exclusively the syntax of Isidor in comparison to the Latin original, also regarded intonation to be responsible for non-canonical ordering, but considered it only as a relevant factor in explaining the occurrence of the postverbal parts of the clause (1888: 119). Finally, when later work pointed out the existence of early verb placement in subordinate clauses, it was mainly analysed as a type of extraposition in support to maintain the Verb-end analysis (Lenerz 1984: 129, Borter 1982: 198). In contrast, Tomaselli (1995) and Fuß (1998) have argued for an INFL-medial position in Old High German. Schlachter (2004), adopting this analysis, has shown that the postverbal constituents in embedded Verbsecond clauses are to be interpreted as focussed. Robinson (1997) has investigated all embedded clauses of the Old High German Isidor translation on the basis of their semantic and syntactic function in relation to the main clause, the active/passive status, and the kind of selecting verb. His data seems to show that Verb-end is more common in adverbial and relative clauses than in indirect questions and complement dhazs-clauses. Nevertheless, he notes that in adverbial dhazs-clauses in which the postverbal constituent is not a sentential constituent or a predicate the verb occurs in 50% of all cases in a higher position and speculates that such clauses seem to be less subordinated, in “that their meaning is often the main content of the whole sentence in which they appear, and almost always asserted information” (1997: 83). The main problem with all these analyses is that they only focus on the type of clauses that fit the given explanation. For instance, Robinson’s group of adverbial dhazs-clauses is a very small one indeed. But what is the reason for early placement in the other cases? Tomaselli bases her analysis on a few Verb-third main clauses, which are very rare in Old High German, and Schlachter mainly argues with copular constructions. Thus, the results cannot be generalized.
Subordinate dhazs-clauses in Isidor 225
This paper therefore wants to narrow the gap between theoretical explanation and empirical findings. It aims at a classification of the different causes of early verb placement in subordinate clauses, thus creating a secure starting point for further theoretical work. We limit the empirical base of this study to the dhazs-clauses, because the syntactic status of dhazs as formal marker of subordination is not in doubt in the majority of cases. In addition, the dhazs-clauses have played an important role in the traditional view of how subordinate structures developed (cf. Müller and Frings 1959). As a data base we chose the earliest prose text, the so-called Old High German Isidor from the end of the 8th century. The paper is organized as follows. After a short presentation and evaluation of the Old High German text we present a word class distinction of the finite verb in early position and a classification of the postverbal constituents. In the next section it is shown that these constituents have different values of focus. The results and some of the consequences are discussed in the final section. 2. The significance of the Isidorian treatise The texts of the so-called Isidor group vary a great deal with respect to their length, which is certainly one of the reasons why they get different degrees of scholarly attention. The text most known and studied is a translation of a theological treatise written in Latin by Bishop Isidor of Sevilla (560–636) De fide catholica ex veteri et novo testamento contra Iudaeos ‘About the catholic faith from the Old and New Testament against the Jews’. Together with the Latin text it has been handed down in a codex BN, Ms lat. 2326 in Paris named (P). Until folia 22 the Old High German text is written side by side with the Latin text, then the Old High German part remains empty up to folia 34, where there is no more space left for the translation (Krotz 2002: 20). It has been edited by Hench (1893) and more recently by Eggers 4 (1964), the latter of which will be the basis for the quotations in this paper. Since the beginning of the treatise is not recorded in (P), Eggers restores parts of it from a codex from the monastery in Mondsee (M) which is nowadays kept in the Austrian National Library of Vienna. This manuscript consists of five small texts: beside the translation of the Isidorian treatise, which is a fragment of only five pages, it contains parts of the Gospel of St. Matthew, fragments of a further treatise and two more homilies. All the texts were transcribed from non-Bavarian texts into the Bavarian language (Matzel 1970: 45).
The Isidorian translation was written in the last decades of the 8th century (Eggers 1964: VIII) in the southwest Franconian dialect and originates from the area of Lorraine (Sonderegger 2003: 129). The basic motivation for its origin is the theological debate about the nature of Christ: is he the real son of God or was he ‘as human being’ adopted by God the Father? The dispute between adherents of the Trinity doctrine and followers of Arianism had undergone a revival at the end of the 8th century, at which time Arianism was known as Adoptianismus, a movement represented by Felix von Urgel. This heresy was condemned by church councils in Regensburg in 792 and in Frankfurt in 794. The text defends the doctrine of the Trinity, arguing with quotations from the Old Testament which show “that Christ, God, and the Holy Spirit are, contrary to the proclamations of the followers of Adoptianism, for example, one equally in essence and in substance” (Nordmeyer 1958: 33). Nordmeyer supposes that the text “was intended to be read aloud, perhaps before a mixed audience of clerics and lay persons assembled in council to judge the Felician heresy” in order to “account with some plausibility for the translator’s rhetorical and exegetical additions and enlargements of his text” (1958: 33). If Nordmeyer is right, the huge number of deviations from the Latin text would find a functional explanation: the translator tried to make the content of the treatise as explicit as possible so that even a theologically uneducated lay public would be able to understand it. Because of its relative independence from the Latin original, the Old High German Isidor version is considered an excellent piece of translation (Lippert 1974: 28). Nevertheless, it‘s exactly because of this outstanding quality that it hasn’t always counted as a good object for syntactic investigation. Gering, as quoted in Robinson (1996: 3) points out that the Isidor “is so free a translation that we cannot be quite sure whether a deviation from the Latin is required by the grammar of German, or simply better style. Such a deviation in Tatian is so infrequent, however, that when one occurs it m u s t be because of German grammar“. We will argue, however, that the Isidor is a well-suited text for the study of information structure for the following reasons. First, stylistic effects are grounded in the grammatical system. Scrambling in Modern German, for instance, which refers to the non-canonical sequencing of constituents in the middle field, is certainly governed by grammatical rules that, at the same time, create stylistic or informationstructural effects.
Subordinate dhazs-clauses in Isidor 227
Second, the frequency of a certain structure in the corpus is indicative of its being a core property of grammar. In anticipation of one of the results, we found 50% of early verb positions in subordinate dhazs-clauses. Since a huge number of these occurs independently of the Latin original and since these clauses are part of a text which has been evaluated as an outstandingly good translation, we have to take them as part of the Old High German syntax. Third, since we conclude that both verb positions (the early and the final one) are part of the Old High German syntax, we can have a closer look at the ordering of the other constituents. Deviations from the Latin text can lead to interesting clues about different structures of German and Latin with respect to information flow. The influence of the Latin original is a factor which has to be controlled for although it is not as important as, for instance, in the Tatian translation (see Petrova and Solf, this volume). On the contrary, it seems that the Latin original of the Isidorian treatise was revised in order to make it more “Germanic”. Ostberg (1979), who investigated all Latin versions of the text, points ot that as “the [Latin] version represented by [Paris-Latin] and [Mondsee-Latin] is consistently closer to the German text than any of the other [manuscripts] examined, it would appear that the translation effort was closely connected with the drafting of the [Latin] ‘Vorlage’ itself” (Ostberg 1979: 205). He furthermore concludes that „most of the alternations apparent in the translation, especially those involving clarification and emphasis, must – in the absence of evidence to the contrary – continue to be regarded as the translator’s own contribution.“ (Ostberg 1979: 205) Furthermore, there is another factor besides the Latin influence which has to be controlled for: the so-called quotation syntax, where the author quotes parts from the Old Testament to strengthen his arguments. Matzel (1970: 357) therefore differentiates between two techniques of translation: the argumental parts should be considered as free paraphrasing which mainly aims to explain the difficult theological reasoning, while the quotation parts are much closer to the Latin original using more archaic means of expression. However, regarding the status of the quotation syntax, a systematic investigation is still lacking about the differences (and similarities) between the two systems, differences which, for example, refer to the attraction of the relative pronoun or the collocation of the definite article (see Matzel 1970: 357, note 855). An example of the different evaluation of syntactic phenomena with regard to quotation syntax is the view on Verbfirst sentences. Robinson (1994) concludes that Verb-first sentences, which
mainly occur in the quotation syntax, aren’t an Old High German pattern at all, but a foreign one. However, this interpretation can’t be correct as Verbfirst patterns are well established in all Germanic languages. Hinterhölzl and Petrova (2005) show for the Old High German Tatian that verb-initial placement can be analysed as encoding a coordinating discourse relation with mainly two functions: first it coordinates pieces of discourse at the same level of text structure (typically of narration), and second, it serves to return from a “subordinating”, elaborating discourse part to the main line of discourse structure (Petrova 2006: 162). Although the precise status of the quotation syntax is still unclear, its potential influence should nevertheless be taken into account. In sum, the treatise was intented to be as clear as possible in order to convince the hearers with its theological content. We therefore expect it to use the tools of information structure in a consistent manner. The results, however, should be checked against other Old High German texts to make sure that they are not simply idiosyncratic properties.
3. Syntactic subgroups of dhazs-clauses 3.1. Methodology and first results As already mentioned, we have built up a corpus of subordinate dhazsclauses of the treatise. Dhazs can occur in adverbial clauses, complement clauses and relative clauses. Relative dhazs-constructions have not been considered since their status as main or subordinate clauses is often unclear. Compare (1a), where dhazs is analysed as a demonstrative pronoun evoking a Verb-second configuration,5 and (1b), where it is interpreted as a relative pronoun of a relative clause with an extraposed subject: (1)
Dhazs ni saget apostolus noh forasago that not says apostle nor prophet ni bifant not found-out a. ‘that neither the apostle says nor the prophet found [it] out’ b. ‘neither what the apostle says nor the prophet found-out’ ... nec apostolus dicit nec propheta conperit. (II.3, 103)
Subordinate dhazs-clauses in Isidor 229
Even if the Latin original points to a certain translation, we don’t mechanically transfer it to the Old High German text when a different interpretation is possible. Consider, for instance, the following clause, where the Latin text suggests a verb-final interpretation of the first dhazs-clause. At the same time it could be interpreted as a relative pronoun or as a correlative 6 demonstrative pronoun of a main Verb-second clause. (2)
Dhazs ir chundida dhazs dher ‘that he announced that the got. God’ Ut eundem spiritum ostenderet esse deum.
gheist ist spirit is
(III.10, 248) Without considering such unclear cases, we found 56 clauses introduced by dhazs in the whole manuscript. More than half of these, 30, display an early verb position. On the basis of the standard analysis of Old High German as a Verb-end language, this is a spectacular result. Before we start to analyse these clauses we have to make sure that early verb placement is not simply conditioned by the dominance of the Latin syntax or the quotation style, whatever it might be. Table 1. The influence of Latin and quotation syntax
V-early position parallel to Latin 11 deviant from Latin 19 quotation syntax 9 part of argumentation 21
58% 51% 82% 47%
V-end position 8 18 2 24
Total 42% 49% 18% 53%
19 37 11 45
A first look at the two subsystems ‘parallelisms’ and ‘deviations’ leads to the conclusion that verb position is not primarily conditioned by the Latin. 58% of the clauses with Verb-early position are parallel, which is only slightly higher than the 51% which are deviant. Within the deviant clauses Verb-early and Verb-end position are nearly equally represented. These results point to the fact that the translator selected the verb position very deliberately and independently of the Latin original. Therefore, the Latin influence on verb placement is not a significant factor.
With respect to quotation syntax we come to a different result: while in the theological argumention parts Verb-early and Verb-final placement are equally represented (48% : 52%), the 11 dhazs-clauses occurring in quotation show a different distribution. With 9:2 cases, Verb-early in quotation syntax is much more common than Verb-end. This extraordinarily high proportion of Verb-early placement points to the fact that quotation syntax indeed behaves differently. We might hypothesize that these Verb-early positions are in fact due to a Latin influence. But a comparison reveals that only 3 cases are totally parallel to the Latin original, while 4 sentences differ with respect to verb placement and in one case only the nominal constituents are changed. Thus, quotation syntax indeed seems to be more than just a copy of the Latin text although it might have been influenced by it. Table 2. Position of the dhazs-clause and clausal function
Dhazs-clause extraposed preposed adverbial clause complement clause
Verb-early position 30 0 8 22
Verb-end position 23 3 5 21
The position of the subordinated clause doesn’t seem to be decisive: nearly all of the sentences are extraposed; the only three instances of preposing out of the group with Verb-end cannot be regarded as significant. Thirteen of the total of 56 clauses are adverbial ones; more than half of these, 8, show an early verb position. However, 22 of the remaining 43 complement clauses also display an early verb position, which is a little bit more than half of all cases. Thus, we conclude that the degree of syntactic integration cannot be the decisive factor. Rather it seems to be the distribution of the informational content between main and subordinated clauses.
3.2. Types of finite verbs At this point we understand ‘early verb position’ as a purely descriptive term for the position of the finite verb. It could thus be a INFL–medial position as proposed by Tomaselli (1995) and Schlachter (2004), and it has recently also been interpreted as a TP selecting a ȣP by Weiß (to appear). It could also be a Verb-end position as suggested by Axel (2007), leading to
Subordinate dhazs-clauses in Isidor 231
the consequence that all postverbal material has to be interpreted as extraposed. Before interpreting the structural position of the finite verb we want to determine its status as an auxiliary, copula or main verb in order to see if a more finely differentiated word class distinction helps to identify preferences for certain positions. Robinson observes that “the verb uuesan ‘be’ is different from all other verbs. In subordinate clauses it frequently ends up before predicate nominals or adjectives, or, if they precede, before the subject” (1997: 149) and he states that “this is clearly not just because the specific form of the verb found is a ‘light element’” (1997: 149).7 Apparently he is refering to the copula status of the verb. However, as table 3 shows, copulas and auxiliaries show the same syntactic behaviour: Table 3. Position and type of the finite verb
auxiliary verb main verb copula verb
Verb-early 12 9 9
Verb-end 8 12 6
Copulas and auxiliaries prefer early placement with 60:40% (if such a small number of examples allows a statistically relevant result at all). Main verbs, on the other hand, seem to prefer the end position with nearly the same percentage. If we look at the clauses with early auxiliary placement with respect to their lexical fillings, we find the following distribution: Table 4. Position and type of the auxiliary
ist sii uuari uuerde uuard uurdi mahti scolda
Verb-early 5 2 1 1 2
The table shows a nearly complementary distribution: the auxiliary forms based on the verb be mainly appear in an early verbal position, whereas the forms with uuerdan tend to appear in final position. Thus Robinson’s statement about the significance of the verb uuesan can be strengthened and modified: uuesan ends up in early position with an extraordinarily high frequency, differently than other verbs. Since its behaviour is the same as a copula and as a auxiliary verb and since the other auxiliary verbs prefer final position, one of the reasons for early placement could be its light phonological and semantic weight. On the other hand, the forms of uuerdhan which prefer the final position behave rather as main verbs, which could be due to the fact that they are still at the very beginning of the grammaticali8 zation path from main verb to auxiliary. 9 The 9 copula constructions are mainly selected by verbs of saying; consider for instance: (3)
dhoh ir in cyres nemin quadhi, dhazs ir he although he in Kyros name saidSUBJ that ioh druhtin ist got is God and Lord ‘Although he used Kyros’ name when he said that he is God and the Lord’ deum et dominum ita esse testatur dicens (III.2, 151)
Nevertheless, the selecting main verb doesn’t seem to be decisive for early verb placement since Verb-end clauses occur with the same type of verb: (4)
Hear quidit umbi dhazs christus got here says about that Christ God ‘Here is said that Christ is God and the Lord’ Quia christus deus et dominus est.
druhtin ist. Lord is.
(III heading, 130) Fourquet (1939: 139–140) already observed that in clauses with only one nominal element (as he called it) the final verb could occur either in final or in early position. But in clauses with two nominal elements as in (5) the finite verb behaves quite regularily, usually occuring between these two elements. We will return to this type of clause in section 4.
Subordinate dhazs-clauses in Isidor 233
dhazs ir gote uuas that he GodDAT was ‘that he was equal to God’ …esse se equalem deo
3.3. Syntactic functions of postverbal elements Conforming to the three verbal classes (main verb, copula and auxiliary) we distinguish between the following kinds of postverbal elements. Main verbs typically select a DP or a PP, copulas select a predicative nominal or adjective and auxiliaries either are followed by a DP and a non-finite verbal form or they introduce the verbal complex followed by a prepositional phrase. Accusative or dative objects never occur after the non-finite verb; they always follow the finite verb. Only prepositional phrases tend to be placed at the end of the clause either after the finite or after the non-finite verb. We found the following distribution: Table 5. Syntactic functions of the postverbal constituents
Vmain + DP/PP
Vcop + DP/AP predicative
Vaux + XP + Vnon-finite
Vaux + Vnon-finite + PP
The group [Vmain + DP/PP] is an important one for the claim of an INFLmedial position since examples with postverbal accusative or dative objects or with pronominal forms are usually not analysed as extraposed. We found 10 the following examples: (6)
“dhazs ih fora sinem anthlutte that I in front of his face dheodun,..” imu him peoplePL ‘That I subjugate his peoples in front of him’ ut subiciam ante faciem eius gentes
“dhazs dhu firstandes heilac chiruni“ that you understandSUBJ holy secret ‘that you understand the holy secret’ ...et archana secretorum… (III.2, 159)
“dhazs uuerodheoda druhtin sendida that army Lord sent dhir“ you ‘that the Lord of the armies sent me to you’ quia dominus exercitum misit me ad te
(III.9, 236) (9)
dhazs imu arsterbandemu siin fleisc ni flesh NEG that heDAT dieParttPres/DAT his enigan unuuillun. any decomposition ‘That when he died his flesh did not decompose’ quia moriens caro eius non uidit corruptionem ...
(IX.12, 719) (10)
dhazs sie ni eigun eouuihd, that they NEG own anything, dhar uuidar setzan. there against oppose ‘that they have nothing to object’ dum non habeant quod proponant
huuazs sie what they
(V.5, 430) (11)
Endi dhazs mittingart firleizssi diubilo and that earth leaveSUBJ/PRET devilsGEN endi auur aruuegodi zi sines drugidha his idols and again returnSUBJ/PRET to scheffidhes huldin. mercy CreatorGEN ‘and that the world leaves the deceptions of the devils and again returns to its Creator’s mercy’ Omissisque mundus dĊmonum simulacris reconciliaretur gratiĊ conditoris. (V.10, 507)
Subordinate dhazs-clauses in Isidor 235
dhazs ir dhoh in dheru chihuurfi zi that he after all in thisFEM/DAT turnsSUBJ/PRET to rehtnissa uuerchum. gotes minniu endi zi to justice’s work GodGEN love and ‘that after all he turns by the virtue of this to God’s love and to the work of justice’ ut uel per ipsam reuerteretur ad amorem dei et operationem iustitiĊ (V.10, 501)
dhinem fordhrom“ “dhazs dhu faris zi that you go to your ancestors ‘that you go to your ancestors’ ut uadas ad patres tuos (IX.2, 622)
The first three examples with postponed accusative object or postponed pronouns are part of quotation syntax. Examples (9) and (10) display accusative objects independently of quotation syntax. Both occur in negated clauses with postverbal indefinite pronouns or adjectives which have to be interpeted as focussed. Only (11) exhibits no specially marked accusative object, neither influenced by Latin nor quotation syntax. It could, however, be interpreted as a main clause. Interestingly (12) and (13) are constructed exactly as the Latin original. Nevertheless the clause final PPs in (12) and (13) have to be regarded as features of a Germanic syntax and are typically analysed as extrapositions. We conclude that the evidence for a structural early verb position within this group is based on just a few examples, which are mainly part of quotation syntax. The group [Vcop + DP/AP predicative], in contrast, doesn’t seem to be influenced either by quotation syntax or by the Latin text. We only find deviations from the Latin original as in (14) or (15), the only example which seems to imitate at least the ordering of the last constituents being (16): (14)
got dhazs ir selbo Christ ist chiuuisso that he himself Christ is certainly God druhtin ioh and Lord ‘that he himself, Christ, is certainly God and the Lord’ Quia idem deus et dominus est (III.1, 135)
“dhazs ir gote uuas that he GodDAT was ‘that he was equal to God’ Esse se equalem deo
dhazs dher selbo gheist ist got. that the same spirit is God ‘that the same spirit is God’ Ut eundem spiritum ostenderet esse deum.
(III.10, 248) One possible analysis of the types [Vaux + XP + Vnon-finite] is to regard them as cases of verb projection raising with the finite verb occurring in final position and the verbal projection with the non-finite verb extraposed to the right (cf. Axel 2007: 98-104) Between the finite and the non-finite verbal form only little material is included: out of a total of 8 cases the subject appears 5 times, as in (17) and (18), and only once does an object (19), an adverb (20) or PP (21). (17)
dhazs fona dhemu almahtigin fater dhurah inan father through him that from theDAT almighty uuordan ist al is all created ‘that all is created by the almighty father and by means of him’ quando a patre per illum cuncta creata esse noscuntur (II.3, 99)
dhazs dhar ist christ chizeihnit that there is Christ meant ‘that there is meant Christ’ intellege christum (III.2, 148)
“dhazs druhtin dhir ist huus zimbrendi“ house building that Lord youDAT is ‘that the Lord is building a house for you’ quod Ċdificaturus sit domum tibi dominus (IX.2, 620)
Subordinate dhazs-clauses in Isidor 237
dhazs (...) christ iu ist that (...) Christ yet is ‘that Christ came long ago’ (...) christus olim uenisse cognoscitur
lange long ago
(V.6, 455) (21)
dhazs ir selbo gotes sunu uuard in was in that he self GodGEN son chiboran born ‘that God’s son himself was born as a human’ eundem filium dei natum in carne monstremus
(V.1, 381) As Weiß points out (to appear: 7), clauses like (17) with an extraposed pronominal are not grammatical in modern dialects which allow verb projection raising. He therefore assumes that such constructions with a right adjoined VP are also not possible in Old High German and proposes an analysis with leftward movement of the finite verb. However, independently of the proposed analysis all of these clauses have to be regarded as examples of a typical Old High German syntax. The group [Vaux + Vinfinit + DP/PP] only consists of four examples. They all belong to the argumentative part of the treatise.11 (22)
unseres druhtin dhazs dhiz ist chiquedan in that this is said in our LordGEN nemin name ‘that this said in the name of our Lord’ Quod in persona specialiter christi domini nostri accipitur. (III.3, 174)
dhazs ir sih auur dhurah hreuun that he himself again through remorse chinisti. mahti chigarauuan zi could prepare to salvation ‘that he could prepare himself for salvation by remorse’ ut per penitentiam reparari possit ad ueniam (V.10, 498)
dhazs ir iesus uuardh chinemnit in that he Jesus was named in iesuses bauhnungum dhes chiuuarin JesusGEN naming theGEN trueGEN ‘that he was called Jesus in naming the true Jesus’ ut iesus nominaretur ad significandum illum uerum iesum. (IV.2, 545)
These examples have to be evaluated like the ones in (12) and (13): although they apparently copy the Latin original, they are instances of a genuine Germanic syntax which is still used in present-day German. In the next section we try to find out if all the postverbal positions are motivated by the same information-structural purpose.
4. The Influence of information-structural factors We start our investigation with the hypothesis that the last or the postverbal parts of the clause are reserved for focussed information. This assumption is a very old one (cf. Rannow 1888: 119; Erdmann 1886: 190). Lacking the notion of focus it was understood as a specially stressed position. Since access to pronunciation in historical texts is not available, one method to find the focussed parts of the clause is text interpretation. Admittedly this isn’t a very transparent instrument either since discourse often allows different interpretations. Only in some cases do we find lexical items which we can take to be reliable indicators for focus interpretation. Before interpreting the information-structural values of the dhazs-clauses with early verb position, we have to define the concepts to be used. As a starting point we adopt Molnár’s (1993) approach to information structure, where Focus-Background, Topic-Comment and Theme-Rheme are located on different layers (for a detailed discussion, see Petrova and Solf in this volume). Thus, Topic and Focus, which are often interpreted as complementary categories, may overlap under certain circumstances. The TopicComment layer will not be treated here since topics, which are the better studied concept of this dichotomy, typically occur at the beginning of the clause and not at the end. The notions of Theme and Rheme refer to the status of discourse referents and are classified by the features [new] and [given]. The former means “newly introduced into discourse”, while the latter covers previously mentioned or accessible entities. In the theological argumentative Isidor treatise the given items are, for instance, God, Christ, the Holy Spirit,
Subordinate dhazs-clauses in Isidor 239
the prophet, and so on, which are under discussion throughout. A first approximation of the concept of Focus might use the notions of highlighting or marking of the relevant part of the utterance (cf. Petrova and Solf, this volume) Thus, Focus can be used to highlight old or new information. If it overlaps with new information, it is to be characterized as “new information Focus”, but this overlap is not obligatory. It also may coincide with old information. Other definitions of Focus recognize the potential overlap of Focus with old information and try to save the notion of “newness” by defining it as, for instance, a “new state of information in the addressee’s mind” (Lambrecht 1994: 210, quoted by Petrova and Solf, this volume). Nevertheless, such a definition cannot explain the rhetorical relations of the argumentative theological treatise we are dealing with in a satisfying way. The often repeated or derived claim that Jesus is God can never be considered as new information; it is always a known proposition wether it is accepted or not. We therefore adopt the semantic definition of Focus proposed by Rooth (1985, 1992) and developed by Krifka (2007: 6). This says that “Focus indicates the presence of alternatives that are relevant for the interpretation of linguistic expressions.” In our example the excluded alternative in interpretation is the heretic view that Jesus should be regarded as a human being and not the son of God. This special pragmatic use could also be described as a means to “correct or confirm information” (Krifka 2007: 11), which can be considered as a case of contrastive Focus (Krifka 2007: 21). In any case, these functions of Focus, including new information Focus and contrastive Focus, should also be subsumed under the general definition of Focus as exclusion of alternatives. Using this definition we try to determine typical interpretations of the postverbal material and to find a motivation for its position, be it extraposed or not.
4.1. DPs or PPs after the finite main verb We cannot confirm the traditional view that the constituents after the finite main verb represent focussed information if we understand focus in the sense of exclusion of alternatives. Sometimes the reading of exclusion is only possible if we extend the focus phrase and include the predicate. Restriction to the postverbal NP leads to an interpretation which can better be described as “new information” or as rhematic material. Consider, for example, (7), repeated here for convenience as (25).
“dhazs dhu firstandes heilac chiruni“ that you understandSUBJ holy secret ‘that you understand the holy secret’ ... et archana secretorum... (III.2,159)
The ‘holy secret’ is not contrasted with something else; its reference is vague as it is newly introduced within the quotation part. The focus rather includes the predicate answering the question: “What do you want me to do?” In contrast, the postverbal personal pronouns in (8), repeated here as (26), which typically indicate givenness, seem to contradict the observation that the postverbal material is new. As already mentioned, however, these sentences are quotations from the Old Testament. Being imported into the text of the Isidor treatise the pronouns cannot refer to known material. They count, therefore, as unknown material. (26)
“dhazs uuerodheoda druhtin sendida that army Lord sent dhir“ you ‘that the Lord of the armies sent me to you’ quia dominus exercitum misit me ad te.
(III.9, 236) Only the continuation of the text in (27) explicitly suggests an interpretation for the referent of the pronoun mih in (26): (27)
Huuelih ist auur nu dhese druhtin which one is however now this Lord fona uuerodheoda drohtine chisendit, nibu from army Lord sent, if not auur dher selbo druhtin nerrendeo christ again this same Lord Redeemer Christ ‘Which one, however, is this Lord sent by the Lord of the armies, if not just the same Lord Redeemer Christ? (III.9, 238)
The same holds for the postverbal preposition phrase in (13), here (28).
Subordinate dhazs-clauses in Isidor 241
“dhazs dhu faris zi dhinem fordhrom“ ‘that you go to your ancestors’ ut uadas ad patres tuos. (IX.2, 622)
In fact, we do not know who this person addressed by dhu is, nor who his ancestors are. Thus, if our interpretation of the postverbal parts in quotation syntax is on the right track, we have found one of their characteristics: they typically introduce new material. Nevertheless, in the argumentative part there are some cases where the postverbal constituent is focussed in the sense that other alternatives are excluded. They typically involve negations with a focussed indefinite adjective or pronoun where the excluded alternative consists of the positive counterpart. Thus, the postverbal constituents only provide focussed information if they include a focus-inducing element: (29)
dhazs imu arsterbandemu siin fleisc ni his flesh NEG that heDAT diePartPresDAT enigan unuuillun. any decomposition ‘that when he died his flesh did not decompose’ quia moriens caro eius non uidit corruptionem ...
(IX.12, 719) (30)
dhazs sie ni eigun eouuihd, that they NEG own anything, dhar uuidar setzan. PART oppose ‘that they have nothing to object’ dum non habeant quod proponant
huuazs sie what they
(V.5, 430) 4.2. Constituents within or after the verbal complex Our findings that material after the main verb is only interpreted as new information unless it is accompanied by a focus-inducing pronoun or adjective does not seem to be true if we look at the examples (22) and (24): both consist of a complex DP where the Genitive and the adjective can be interpreted as contrastively focussed. We conclude that in addition syntactic
complexity can also be a motivation for postverbal Focus. This seems to be true if we look at the simple DP zi chinisti ‘to salvation’ in (23), which cannot be evaluated as focussed. The picture is different for constituents between the finite and the nonfinite verb. Except for (33), they all have to be interpreted as contrastively focussed. (31) (32) (33) (34) (35)
ist al uuordan (II.3) ist christ chizeihnit (III.2) ist huus zimbrendi (IX.2) ist lange quhoman (V.6) uuard in lihhe chiboran (V.1)
We thus conclude that focussed (contrasted) material is not inserted in relation to finiteness, but in relation to the main verb and that it prefers the preverbal position. Postverbal focussed material apparently needs a special lexical or syntactic marking to be licensed. However, in establishing this generalization a methodological problem may arise: having no access to pronunciation we have no means to decide wether a constituent which is not lexically marked can be interpreted as focussed or not. I am aware of this problem, but the only way to deal with this issue is to rely on the weaker method of textual interpretion.
4.3. Identificational and predicative constructions Typical focussed information occurs in copular constructions that are analysed as identificational clauses,12 where the second NP serves to fix the identity of the first NP: (36)
dhazs ir selbo Christ ist that he self Christ is druhtin Lord quia idem deus et dominus est (...)
Subordinate dhazs-clauses in Isidor 243
dhazs ir that he (III.2,151)
dhazs iesus ist that Jesus is dominum esse iesum
druhtin Lord (VI.2, 549)
The context is always the same: after a quotation or argumentational part the author confirms that the previously mentioned Christ has to be God and Lord, where we interpret the last NPs as being focussed. None of the constituents delivers new information. Nevertheless, other interpretations are possible: it could be that the first part of the equation is the prominent one (if ir selbo or iesus were stressed). It is even possible that the copula itself is stressed, which leads to an interpretation in which the truth value of the assertion is focussed (cf. Höhle’s Verum-Fokus 1992). Nevertheless, we assume that it is the postverbal constituent which is focussed, the sentence adverb chiuuisso in (36) can be seen as an indication of this interpretation. In this respect copulas are identical to auxiliaries, which also appear before the focussed material. We thus get the following ordering: (39)
Background / Verbaux/cop / Focuscontr / Verbmain/
New information Focus (lex. induced) Focus (syntact. compl.)
5. Results We have argued that only by distinguishing different types of dhazs-clauses can we arrive at insightful explanations of the interaction between sentence structure and information structure. The widely held belief that the preferred position for focussed constituents is the clause-final one must be modified in the sense that not all constituents at the end are focussed in the same manner or are focussed at all. Prepositional phrases often simply provide new material, and have to be interpreted as rhematic. Sometimes they provide given material as in (23). The majority of Verb-early clauses headed by main verbs also lack postverbal contrastive focus, the postverbal material has to be classified as new rather than contrastive. In this case the new information focus com-
prises the entire VP. At the same time it has been shown that these clauses mainly occur in the quotation parts of the text. The situation is different for auxiliaries and copulas in early position, which are usually followed by a contrastive constituent. A typical case is the identificational copular clause, where the postverbal nominal element is contrasted with alternative interpretations. The majority of these structures are independent of the Latin or quotation syntax. Furthermore, the observation that the verb uuesan (to be) plays an outstanding role in early verb placement can be strengthened and modified. Not only as a copula verb but also as an auxiliary uuesan prefers an earlier position in the clause. We therefore assume a structural Verb-early position for auxiliaries and copula verbs, which, following Fuß and Trips (2002), we identify with ȣP. This genuine Old High German structure is then used for structures with early main verbs as they typically occur in quotation syntax. They have the common feature of a special information-structural pattern and point to a different stylistic system, which could be described either as archaic or as contact induced. At the same time there are many Verb-end structures with copulas and auxiliaries (and, of course, main verbs). These belong to the subsystem of subordinated clauses which already existed in Old High German and became generally established in the later development; consider (40) and (41):13 (40)
dhazs christ (…) Ċr allem uueraldim fona that Christ before all worlds by fater uuard chiboran. (the) father was born ‘that Christ before the beginning of the worlds was born by the father’ …ante omnia secula filius a patre genitus esse declaratur. (II.3, 97-98)
dhazs christus got
here says about that Christ God ‘here is said that Christ is God and the Lord’ Quia Christus deus et dominus est.
(III. heading, 130) The later unification of the syntactic position for finite verbs has the disadvantage of an information-structural ambiguity common to subordinated clauses with Verb-end. If there is more than one constituent before the verbal complex, as for example in (40), it is not necessarily clear whether the preverbal constituent is focussed or if the whole subordinate clause is to be
Subordinate dhazs-clauses in Isidor 245
interpreted as backgrounded. The contrastive reading means that ‘Christ is the son of the divine father and not of a human being’. In contrast, the background reading just summarizes the knowledge of Christ’s birth as a starting point for further argumentation. This interpretation gains support by the fact that Verb-end clauses typically occur at the beginning or at the end of a paragraph. The Verb-end analysis also accounts for clauses where main verbs are followed by constituents which are marked by a focus-inducing element or by syntactic complexity. These are analysed as extrapositions triggered by a special prosodic pattern. Finally, we don’t regard these different verb positions as parts of different grammars but as different means of the same grammar with special discourse functions. It is possible that information structure allows different sentence structures as in the example quoted by Fourquet (mentioned in section 3.2 as (5), here (42)). Fourquet obviously did not see any regularity in the position of the predicative adjective with respect to the dative object except for the fact that the verb apparently has to occur in the penultimate position. In (42) the predicative adjective shows up after the copula and in (43) it is preverbal. In both positions they have to be interpreted as focussed as revealed by the context given in brackets [ ]. (42)
“dhazs ir gote uuas ebenchiliih.” that he GodDAT was equal ‘that he was equal to God [and not just in the same form] esse se equalem deo. (V.3, 406)
dhazs ir chihoric uuari gote that he obedient wasSUBJ GodDAT that he should be obedient to God [and no longer disobedient] ut esset deo subiectus (V.9, 491)
Following our analysis of copular structures, uuas in (42) occurs in ȣ°. The focussed predicative adjective corresponds to the established informationstructural pattern. Nevertheless, the Verb-end analysis is not excluded since the extraposed constituent is lexically marked: the adjective ebenchiliih ‘equal’ is a composition of eban and chiliih, both meaning ‘equal’. Its specific lexical (and prosodic) structure determines the final position in the clause. In contrast, chihoric ‘obedient’ in (43) has no special lexical focus
device and makes use of the usual preverbal focus placement of the ‘regular’ Verb-end construction. The extraposition of gote guarantees that there is no information-structural ambiguity in the sense that a background reading is not available. We are aware of the fact that our results are based on a very small number of examples and might be better described as hypotheses. In future research we hope to test these findings against all the subordinated clauses of the Isidor text.
Notes 1. 2.
5. 6. 7. 8. 9. 10.
I want to thank Svetlana Petrova and Roland Hinterhölzl for interesting discussions and comments on an earlier version of this paper. At this point in the discussion, we don’t distinguish between the different proposals for the landing-site of the finite verb be it VIP as in Kiparsky (1996), ȣP as in Fuß and Trips (2002) or TP and ȣP as in Weiß (to appear). But see Hinterhölzl (2004) for a critical evaluation of the double base hypothesis. We opted for Egger’s edition because it is accessible to a greater public. Since we not only quote the line, but also the chapter and the paragraph, it will be possible to find the quoted clauses in Hench’s edition as well. The Verb-second analysis is based on Jäger’s (2005) account of negation, where the negative particle is clitised to the finite verb. See Fleischer (2006) for methodological problems in general. This fact, already noted by Fourquet (1974: 318–319) has not yet received a satisfactory explanation. This point was stressed by Svetlana Petrova in a personal communication. Seven times with the pattern uuesan/sin + NP/AP, two times with the pattern uuerdan + NP/AP. We don’t consider Verb-first constructions. Many of them are parallel to the Latin text as in the following example: Dhazs ni bilibun ano herrun iudaeoliudi fona iudases chunne ‘that the Jews of the family of Judas are without masters’ Non defuisse principes iudeorum populi ex genere iuda (VIII.1, 581). Again, we won’t consider a Verb-first construction from the quotation part: „dhazs chiendot uuerdhe dhiu aboha ubarhlaupnissi…” That finished beSUBJ this bad offence. Ut consummetur praeuaricatio (V.6, 448). For a recent discussion of copular clauses, see Geist (2006). We take examples with a final verbal complex as cases of Verb-end even if the finite verb occurs in the penultimate position.
Subordinate dhazs-clauses in Isidor 247
References Axel, Katrin 2007 Studies on Old High German Syntax. Left sentence periphery, verb placement and verb-second. Amsterdam/Philadelphia: John Benjamins. Besten, Hans den 1977 On the Interaction of Root Transformation and Lexical Deletive Rules. Ms. University of Amsterdam. Borter, Alfred 1982 Syntaktische Klammerbildung in Notkers Psalter. Berlin/New York: Walter de Gruyter. Diels, Paul 1906 Die Stellung des Verbums in der älteren althochdeutschen Prosa. Berlin: Mayer & Müller. Eggers, Hans 1964 Der althochdeutsche Isidor. Nach der Pariser Handschrift und den Monseer Fragmenten. Tübingen: Niemeyer. Erdmann, Oskar 1886 Grundzüge der deutschen Syntax nach ihrer geschichtlichen Entwicklung. Stuttgart: Cotta. Fleischer, Jürg 2006 Zur Methodologie althochdeutscher Syntaxforschung. Beiträge zur Geschichte der deutschen Sprache und Literatur 128 (1): 25−69. Fourquet, Jean 1939 L’ ordre des élément de la phrase en germanique ancien. Paris: Les Belles Lettres. 1974 Genetische Betrachtungen über den deutschen Satzbau. In Festschrift für Hugo Moser zum 65. Geburtstag. Studien zur deutschen Literatur und Sprache des Mittelalters, eds. Werner Besch, Günther Jungbluth, Gerhard Meissburger and Eberhard Nellmann, 314−323. Berlin: Erich Schmidt. Fuß, Eric 1998 Zur Diachronie von Verbzweit. Die Entwicklung von Verbstellungsvarianten im Deutschen und Englischen. M.A. thesis, Universität Frankfurt/Main. Fuß, Eric and Carola Trips 2002 Variation and change in Old and Middle English: On the validity of the Double Base Hypothesis. Journal of Comparative Germanic Linguistics 4: 171−224. Geist, Ludmilla 2006 Die Kopula und ihre Komplemente. Zur Kompositionalität in Kopulasätzen. Tübingen: Max Niemeyer.
Hench, George A. 1893 Der althochdeutsche Isidor. Facsimile-Ausgabe des Pariser Codex nebst critischem Texte der Pariser und Monseer Bruchstücke. Straßburg: Karl J. Trübner. Hinterhölzl, Roland 2004 Language change versus grammar change: What diachronic data reveal about the distinction between core grammar and periphery. In Diachronic Clues to Synchronic Grammar, eds. Eric Fuß and Carola Trips, 131−160. Amsterdam/Philadelphia: John Benjamins. Hinterhölzl, Roland and Svetlana Petrova 2005 Rhetorical relations and verb placement in Early Germanic languages: Evidence from the Old High German Tatian translation (9th century). In Salience in Discourse. Multidisciplinary Approaches to Discourse, eds. Manfred Stede, Christian Chiarcos, Michael Grabski and Luuk Lagerwert, 71–79. Münster: Stichting/Nodus. Höhle, Tilman 1992 Über Verum-Fokus im Deutschen. In Informationsstruktur und Grammatik, ed. Joachim Jacobs, 112−141. (Linguistische Berichte, Sonderheft 4.) Opladen: Westdeutscher Verlag. Jäger, Agnes 2005 Negation in Old High German. Zeitschrift für Sprachwissenschaft 24 (2): 227−262. Kemenade, Ans van 1987 Syntactic Case and Morphological Case in the History of English. Dordrecht: Foris. Kiparsky, Paul 1996 The shift to head-initial VP in Germanic. In Studies in comparative Germanic Syntax II, eds. Höskuldur Thráinsson, Samuel D. Epstein and Steve Peter, 140−179. Dordrecht: Kluwer. Krifka, Manfred 2007 Basic notions of information structure. Interdisciplinary Studies on Information Structure (ISIS) 6: 13–56. Kroch, Anthony S. 1989 Reflexes of grammar in patterns of language change. Journal of Language Variation and Change 1: 199−244. Krotz, Elke 2002 Auf den Spuren des althochdeutschen Isidor. Studien zur Pariser Handschrift, den Mondseer Fragmenten und zum Codex Junius 25. Mit einer Neuedition des Glossars Jc. Heidelberg: C. Winter. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: Cambridge University Press.
Subordinate dhazs-clauses in Isidor 249 Lenerz, Jürgen 1984 Syntaktischer Wandel und Grammatiktheorie. Eine Untersuchung an Beispielen aus der Sprachgeschichte des Deutschen. Tübingen: Niemeyer. Lippert, Jörg 1974 Beiträge zu Technik und Syntax althochdeutscher Übersetzungen unter besonderer Berücksichtigung der Isidorgruppe und des althochdeutschen Tatian. München: Wilhelm Fink. Matzel, Klaus 1970 Untersuchungen zu Verfasserschaft, Sprache und Herkunft der althochdeutschen Isidor-Sippe. (Rheinisches Archiv 75.) Bonn: Ludwig Röhrscheid. Molnár, Valéria 1993 Zu Pragmatik und Grammatik des TOPIK-Begriffes. In Wortstellung und Informationsstruktur, ed. Marga Reis, 155−202. Tübingen: Niemeyer. Müller, Gertraud and Theodor Frings 1959 Die Entstehung der deutschen daß-Sätze. Berlin: Akademie Verlag. Nordmeyer, George 1958 On the OHG Isidor and its significance for Early German prose writings. Publications of the Modern Language Association 73: 23−35. Ostberg, Kurt 1979 The Old High German Isidor in Its Relationship to the Extant Manuscripts (Eighth to Twelfth Century) of Isidorus ‘De Fide Catholica’. Göppingen: Kümmerle. Petrova, Svetlana 2006 A discourse-based spproach to verb placement in Early West-Germanic. Interdisciplinary Studies on Information Structure (ISIS) 5: 153−185. Petrova, Svetlana and Michael Solf this vol. On the methods of the information-structural analysis of historical texts: A case study on Old High German. Pintzuk, Susann 1993 Verb seconding in Old English: Verb movement to Infl. The Linguistic Review 10, 5−35. 1999 Phrase Structures in Competition: Variation and Change in Old English Word Order. New York: Garland. Rannow, Max 1888 Der Satzbau des althochdeutschen Isidor im Verhältnis zur lateinischen Vorlage. Berlin: Weidmannsche Buchhandlung. Robinson, Orrin, W. 1994 Verb first position in the Old High German Isidor translation. Journal of English and Germanic Philology 93: 356−373.
Clause subordination and verb placement in the Old High German Isidor translation. Heidelberg: Carl Winter.
Rooth, Mats 1985 Association with focus. Ph.D. dissertation, University of Massachusetts, Amherst. 1992 A theory of focus interpretation. Natural Language Semantics 1: 75−116. Schlachter, Eva 2004 Satzstruktur im Althochdeutschen. Eine Skizze zur Position des Verbs im Isidor-Traktat des 8. Jahrhunderts. In Beiträge zu Sprache und Sprachen 4. Vorträge der Bochumer Linguistik-Tage, ed. Karin Pittner, Robert J. Pittner and Jan C. Schütte, 179−188. München: Lincolm. Sonderegger, Stefan 2003 Althochdeutsche Sprache und Literatur. Berlin/NewYork: Walter de Gruyter. Tomaselli, Alessandra 1995 Cases of verb third in Old High German. In Clause Structure and Language Change, eds. Adrian Battye and Ian Roberts, 345−369. New York/Oxford: Oxford University Press. Weiß, Helmut (to appear) Die rechte Peripherie im Althochdeutschen. Zur Verbstellung in dass-Sätzen. To appear in: Akten der Arbeitstagung der Indogermanischen Gesellschaft, Jena 2006.
Information structure and word order variation in the Old High German Tatian Svetlana Petrova
1. Introduction Word order variation in the right periphery of subordinate clauses is one of the most striking properties of Old High German (OHG) syntax. It belongs to the central and most vividly discussed topics in the historical treatment of sentence structure in German, namely to the description of the rules and principles that determine the position of the finite Verb (Vfin) in the earlier stages of the German language. The same issue has been intensively discussed with respect to the remaining early Germanic languages as well. In recent generative work on Old English (OE), two different accounts have been put forward. The first one launched by Kemenade (1987)1 attributes surface orders with postverbal constituents in clauses with overt complementizers to extraposition from a uniform SOV base. This kind of operation applies in modern SOV langages as well, especially with PPs and CP-complements which are regulary extraposed to the right of the selecting verb. Additionally, some restructuring operations in verb clusters leading to orders with a tensed auxiliary before the untensed main verb in subordinate clauses (verb raising and verb projection raising) were originally analysed as instances of rightward movement of the VP as well (Haegeman and van Riemdijk 1986). In line with the theoretical discussion on properties of asymmetric SOV languages, structural variation in the right periphery of subordinate clauses in OE was explained as the result of rightward movement while Vfin always remains in its basic position in the end of the clause (see also Tomaselli 1995, 350–351). However, the idea that OE has a uniform SOV structure in the base has been challanged by Pintzuk (1991) who discovered evidence for postverbal phrases, e.g. pronouns and light adverbs, which are excluded form extraposition in modern SOV langauges. To explain structural variation in the data, Pintzuk claimed that OE displays variation in the head-complement parameter in both I(nfl)P and VP. In line with this model, non-Vend orders
in OE are explained partly as a result of leftward movement of Vfin to a clause-medial I(nfl)P, and partly as instances of VO in the base. The basic points in the discussion on word order variation in OE have been also applied to the interpretation of the OHG data. Weiß (2006) discusses word order in complement clauses introduced by dass ‘that’ in the so-called ‘Minor texts’ of the OHG tradition. He is able to derive a great part of the non-Vend orders form a basic SOV order, although he is forced to assume, apart from extraposition of PPs and heavy complements, a series of leftward movement operations according to which Vfin targets two different functional projections (TP and ȞP) below CP. A different approach is pursued by Schallert (2006) who discusses evidence for mixed word OV/VO order in OHG claiming that the early Germanic languages were unspecified with respect to the head-complement parameter in the VP. The present study addresses this complex discussion from the perspective proposed by Hinterhölzl (2004) who relates word order variation to properties of the information-structural organization of the utterance. A first and by now unique empirical investigation on variation in the right periphery of subordinate clauses in OHG is provided by Schlachter (2004). Her analysis reveals that the different placements of Vfin in complement clauses in the OHG Isidor correlates with the iconic separation of the domains of focus and background in the clause. These findings are in line with a long tradition in the descriptive literature which related the principles of verb placement in dependent clauses to stylistic effects and properties of theme-rheme (see the summary in Ebert 1978, 39–43). The foregoing observations suggest that pragmatic considerations play an important role in the explanation of word order variation in the earlier stages of German. Therefore, the aim of the present paper will be to explore in more detail the extent to which information-structural principles are responsible for the different placements of Vfin in subordinate clauses in OHG. Special attention will be placed on the correlation between pragmatic properties of constituents like givenness/novelty, contrast, emphasis and the like, and their positional realization with respect to Vfin.
2. Properties of the database This study analyses data from the OHG Tatian translation, which is the largest prose text of the classical period of OHG. In order to base the observations on genuine OHG structures, we will examine only sentences in which the word order differs from that of the corresponding Latin original.2
Information structure in the Old High German Tatian
Project B4 of SFB 632 “Information Structure” has provided a data collection of all clauses displaying differences in word order between OHG and Latin in the text parts assigned to three different scribes (Į, ȕ, and ε). For the purpose of the present study, we will analyse the dependent clauses found in this data collection. First, let us look at the quantitative distribution of Vend vs. non-Vend orders in all conjunctional and relative clauses found in the database. The figures provided in Table 1 clearly show that non-Vend orders are highly frequent in subordinate clauses in OHG. Causal clauses are listed separately because they are ambiguous between coordinate root conjuncts with an extra-clausal connective comparable to modern German denn ‘because’ and subordinate clauses with a lexically filled complementizer. This ambiguity may explain the high number of non-Vend orders among the causal clauses. However, the ratio of non-Vend among the unambiguously subordinate clauses is only slightly below 50 per cent. This means that in nearly half of the conjunctional clauses the scribe decided to depart from the structure of the Latin original but ended up in a structure which is not Vend in OHG3: Table 1. Relative frequency of Vend vs. non-Vend orders in subordinate clauses in the OHG Tatian in the database of the study. clause type
There is crucial evidence suggesting that non-Vend is an authentic native pattern in subordinate clauses in OHG. First, we find cases where Vend order is given in the Latin original but suspended in the OHG translation, see (1a). Second, non-Vend is attested in clauses whose Latin equivalent lacks a finite verb, see (1b–c). In such cases, we can assume that the insertion of Vfin is ruled by native OHG grammar. As the examples show, the scribes disregard the oportunity to preserve or create Vend patterns in subordinate clauses:
a. thaz [...] thie dar gisehent daz sie sin blinte (T 224, 4–6) that […] those PRT see that they are-SUBJ blind-PL ‘[I have come,] so that […] those who see may be made blind’ lat. ut […] qui uideant caeci fiant b. salige sint thiethar sint sibbisame (T 60, 16) blessed are who-PRT are peaceful ‘Blessed are the peacemakers’ lat. Beati pacifici c. In thie burg/ galileĊ thero namo ist nazar&h (T 28, 4–5) in that city Galilee whose name is Nazareth ‘into that city of Galilee whose name is Nazareth’ lat. In ciuitatem/ galileae cui nomen nazar&h
Moreover, the postverbal domain in OHG hosts types of constituents which do not undergo movement to the right in modern SOV languages, e.g. single (unmodified) NPs (2a), predicative nouns (2b) and adjectives (2c):4 (2)
a. Inti thie thár hab&un diuual (T 59, 1) and who PRT had devil ‘and those who were possessed by the devil’ lat. & qui demonia habebant b. thaz sie hiezzin boanerges (T 59, 22) that they were called Boanerges ‘that they be called Boanerges’ lat. boanerges c. oba thin ouga uuirdit luttar (T 69, 22) if you eye becomes bright ‘if your eye becomes bright’ lat. si fuerit occulus tuus simpex
The number of postverbal single elements after Vfin in subordinate clauses is raised by non-finite forms of main verbs in complex predicates, e.g. in constructions with modal verbs (3a–b), or in the combination of sîn ‘to be’ and uuerdhan ‘to become’ with the past participle (3c), a construction which is formally identical with the passive in modern German:
Information structure in the Old High German Tatian
a. thaz sie Inan mohtin ruogen (T 199, 2) that they Him might-SUBJ accuse-INF ‘that they had something of which to accuse Him’ lat. ut possent accusare eum b. Inti thiethár uuolle mit thír uuehslon (T 65, 12) and who-PRT want-SUBJ from you-DAT borrow-INF ‘and whoever wants to borrow from you’ lat. & uolenti mutuare a té c. nibi ir uuerdet giuuentite/ inti gifremite soso theser luzilo (T 151, 12) NEG-if you become converted-PL and formed-PL like this young [boy] ‘if you do not convert and become like this young boy’ lat. nisi conuersi fueritis/ & efficiamini sicut paruuli
Orders in which the tensed auxiliary precedes the untensed main verb in subordinate clauses have been related to phenomena like verb raising and verb projection raising typical for verb clusters in modern SOV languages as well (Fuß and Trips 2002). However, the degree of grammaticalization of periphrastic forms is questioned in OHG, especially with respect to the formal equivalent of the modern passive construction. According to the common view, this expression is still undergoing a process of grammaticalization from a copular construction to a periphrastic form (Valentin 1987 among others). The presence of inflectional endings on the participle, which agrees in number, gender and case with the corresponding subject constituent as in (3c), is a strong indication of the copular status of the construction. Copular construction, however, are not among the clusters discussed in relation with verb raising or verb projection raising in the literature (Wurmbrand 2004). These observations prompt the view that variation in verb placement in subordinate clauses is a genuine syntactic property of OHG which calls for an alternative description. For this reason, a more detailed analysis of the grammatical and pragmatic properties of constituents in different syntactic patterns in subordinate clauses from the OHG period is needed. The present study is based on the analysis of 100 clauses with Vend order and 100 clauses with non-Vend order which were selected from the main corpus. Representatives of all types of subordinate clauses established above, i.e. conjunctional, relative and causal clauses, are included. For reasons of unambiguous classification, sentences containing only one constituent apart from Vfin have been left aside, as a serialisation of the
type Conjunction – XP – Vfin may be viewed both as V2 and Vend. Furthermore, the influence of the translation technique known for the OHG Tatian has been also reflected in the selection of the database: as the requirement not to shift material across the lines of the manuscript is the main translation principle stated for this text (Masser 1997), examples where the transposition of constituents appears to be blocked by the line break also remained unconsidered.
3. Analysis 3.1. Clauses with Vend order 3.1.1. The placement of discourse-anaphoric material Among the total of 100 instances with Vend order, there are 34 cases in which the only syntactic difference in clause structure results form the insertion of the subject pronoun in the translation while the rest of the sentence remains unchanged. See (4a–b) where the subject pronouns her ‘he’ and ír ‘you-2pl.nom’ have no equivalent in the Latin clause but the order of the PP mit imo ‘with him’ and the direct object zeichan inti uuvntar ‘signs and miracles’ with respect to Vfin is identical in both OHG and Latin: (4)
a. thaz her mit imo uuari (T 88, 26) that he with Him was-SUBJ ‘that he was with Him’ lat. ut cum eo ess& b. nibi ír zeichan inti uuvntar giseh& (T 90, 18) NEG-if you signs and miracles see ‘unless you see signs and miracles’ lat. nisi signa & prodigia uideritis
These examples are revealing as to where subject pronouns are usually placed in OHG. In fact, our group of Vend sentences shows a 100 percent consistency with respect to the placement of subject pronouns, as in all cases investigated here the position chosen for the pronoun in the OHG clause is always the one immediately after the subordinating conjunction, i.e. the so called Wackernagel position (see also Tomaselli 1995, 349). In 28 additional cases, the same syntactic positon is targeted by other pronominal elements in the OHG text, see the direct object thiu ‘these
Information structure in the Old High German Tatian
things-3pl.acc.neutr’ in (5a), the indirect object thir ‘you-2sg.dat’ in (5b), and the prepositional object in (5c): (5)
a. tho siu thiu gisah (T 28, 12) when she this-PL saw ‘as she saw these things’ lat. quae cum uidiss& b. unzan ih thir quede (T 40, 28) until I you-DAT tell ‘until I tell you’ lat. usquedum dicam tibi c. soso zi In gisprochan uuas (T 37, 5) as to them-DAT said was ‘as has been said to them’ lat. sicut dictum est ad illum
It can be also shown that pronominal arguments appear to the left of adverbials which according to the standard syntactic assumptions mark the left edge of the VP, see thara and thar ‘there’ in (6a–b): (6)
a. Inti thô her thara quam (T 42, 03) and when He there came ‘and when he came there’ lat. & ueniens b. thô sie thar uuarun (T 35, 22) when they there were ‘as they got there’ lat. cum essent ibi
These facts about OHG sentence structure have not passed unnoticed in the literature. According to Behaghel (1932, 4–6, §1426), word order in early Germanic is subject to an intricate interplay of two basic sets of principles, the first one concerning the informational relevance of sentence constituents and the second one concerning their “physical” properties (ibid. 5) in terms of relative length and phonological heaviness. According to the first set of rules, less relevant information tends to precede more relevant one in the clause. Additionally, the principle of growing constituents requires shorter constituents to precede longer ones in the clause. From this perspective, structures like (4)–(6) show a perfect interplay of the mentioned requirements. It is obvious that pronouns representing previously men-
tioned information can be analysed as less relevant than the rest of the utterance conveying the new information in the discourse. Additionally, they represent short items which tend to be de-accented and cliticised to other tonic elements and thus being typical cases of phonologically light elements which tend to be placed before heavy, full lexical material. This evidence raises the question whether anaphoricity is the factor leading to the identical positional distribution of pronominal arguments in OHG. In order to examine this issue, we will turn our attention to the syntactic realization of full lexical arguments with anaphoric properties in OHG. According to Dittmer and Dittmer (1998, 21), full phrases tend to be less often shifted across Vfin in contrast to pronominal elements in the OHG Tatian. However, our data base yields 8 instances in which full constituents with anaphoric properties are shifted from the postverbal position in the Latin sentence to the preverbal domain in OHG. The reverse transposition is not found in the corpus. Consider (7a–b), in which the discourse-given object DP thén buoh is shifted across the verb against the Latin original in two subsequent clauses: (7)
a. so hér thén buoh int&a (T 53, 21) when He this book opened ‘as he opened the book’ lat. & ut reuoluit librum b. inti mit thiu hér thén buoh bit&a (T 53, 32) and when He this book closed ‘and as he closed the book’ lat. & cum plicuiss& librum
The fact that, just like pronouns, full lexical DPs are also regularly shifted to the preverbal domain when they are discourse-given, strongly supports Behaghel’s rule of relevance. But what about the principle of growing constituents and the role of phonological heaviness, given the fact that we deal with full lexical categories bearing an overt determiner and therefore comprising several syllables? In a historical corpus, we have no opportunity to judge about the prosodic realization of sentence constituents. However, we can draw parallels to the situation in some contemporary intonational languages. Lakoff (1976, 288) presents some well-known facts for English. He shows that full DP-expressions regularly give up accent to the verb when they refer back to a previously mentioned antecedent, see (8a), Lakoff’s (91). At the same time, main accent on a DP blocks its interpretation as an anaphor to a pre-established referent, see (8b), Lakoff’s (92)5:
Information structure in the Old High German Tatian
a. When Harryi entered the room, Mary KISsed the presidenti. b. When Harryi entered the room, Mary kissed the PREsidentj/ *PREsidenti.
So anaphoric reading correlates with de-accentuation, and vice versa. From this we can assume that in OHG too, full DPs were de-accented when used in anaphoric relation to an antecedent in the previous context. Consequently, they do not count as heavy constituents but share the prosodic behaviour of pronouns and light adverbs. Again, Behaghel’s principles cooccur: anaphoric, i.e. informationally less relevant material, as well as phonologically de-accented, i.e. light material, appears early in the clause.
3.1.2. Non-anaphoric information The observations made on the positional realization of discourse-given material are confirmed by the fact that Vend order is found in clauses containing familiar information only, e.g. resuming a pre-established fact or conveying an expected, inferable event. In (9), the entire information in the purpose clause, namely that the first-born son shall be presented to God, is inferrable from the common knowledge of the customs of the Jewish people explicitly refered to in the context (after moyseses euuu ‘according to the Law of Moses’): (9)
Inti after thiu gifulta uuarun taga /[…] brahtun sie Inan thô In and after filled were [the] days/ […] brought they Him-ACC then to hierusalem/ thaz sie Inan gote giantuuvrtitin (T 37, 11–14) Jerusalem/ that they Him-ACC God-DAT presented-SUBJ ‘after the days of her purification they brought Him to Jerusalem to present Him to the Lord’ lat. & postquam Impeti sunt dies/ […] tullerunt illum In herusalem/ ut sisterent eum domino
However, unlike the examples discussed so far, there are subordinate clauses with Vend order in which the preverbal domain does not convey anaphoric or inferable information. Instead, the expressions preceding Vfin can be seen to achieve special prominence over the rest of the utterance for different pragmatic or contextual reasons. Mainly two types of preverbal phrases can be distinguished here: parts of idioms and narrowly focused expressions.
In the first group, the preverbal constituent forms a complex semantic unit with Vfin which immediately follows it, i.e. the pre-verbal phrase forms a kind of an idiom with the following verb. Consider thurft sîn ‘to need’ in (10a) as well as heim uuverban ‘to return home’ in (10b) in which the non-finite part provides the semantic core of the complex predicate: (10) a. uueiz íuuar fater/ uues íu thurft ist (T 67, 29) knows your Father/ what-GEN you-DAT need is ‘your Father knows the things you have need of’ lat. scit enim pater uester/ quibus opus sit uobis b. mit thiu sie heim uuvrbun (T 42, 17) when they home returend ‘when they returned home’ lat. cum redirent Non-verbal elements of complex semantic units cannot be pronominalized or referred back to by any means of anaphoric reference. In other words, parts of idioms fail to display properties which according to Karttunen (1976) are distinctive for referential expressions, i.e. they lack referential status and are not subject to the given/new-distinction. In modern German, non-verbal parts of complex predicates, like e.g. the NPs in Ball spielen ‘to play ball’, Schlange stehen ‘to stand in line’etc., are known to stay in a close relation to the verbal head not only with respect to semantics but to syntax as well. So in basic order, the nominal part has to be left adjacent to the verb: it is not subject to scrambling and does not allow insertion of adverbials or negation elements between itself and its verbal head (Pittner 1998), see (11a–b): (11) a. *dass die Kinder Ball oft/ nicht spielen that the children ball often/ NEG play b. dass die Kinder oft/ nicht Ball spielen that the children often/ NEG Ball play By contrast, ordinary arguments of verbs show no restrictions in this respect, see (12a–b): (12) a. dass die Kinder das Geld oft/ nicht ausgeben that the children the money often/ NEG spend b. dass die Kinder oft/ nicht das Geld ausgeben that the children often / NEG the money spend
Information structure in the Old High German Tatian
The incorporation of non-verbal parts of idioms to form a unitary whole with the verbal head is best represented in the case of separable verb prefixes which historically go back to directional or locative adverbials closely related with the verb (Di Meola 2000, 129). In the second group, the position before Vfin is occupied by material revealing properties of narrow, e.g. contrastive or operator-bound focus, though the focus operator is phonologically empty in most of the cases. Consider (13) in which the preverbal position in the embedded question hosts the DP thin zesuua ‘your right hand’ as an alternatives to the DP thin uuinistra ‘your left hand’ mentioned in the preceding main clause: (13) niuuizze íz thin uuinistra/ uuaz thin zesuua tuo (T 67, 4–5) NEG-know-SUBJ it your left hand what your right hand does ‘your left hand should not know what your right hand is doing’ lat. nesciat sinistra tua/ quid faciat dextra tua Furthermore, we find examples in which the preverbal phrase bears properties of exhaustiveness similar to those described for preverbal focus in Hungarian (Kiss 1998). The only contextually adequate reading of (14) is that Jesus spoke about no one else than about the Pharisees. As we can see, the constituent conveying exhaustiveness occupies the preverbal position, all remaining background material precedes: (14)
Inti pharisei […]/ furstuontun thaz her Iz fon In quad (T 204, 22) and the Pharisees […]/ understood that He it about them said ‘and the Parisees realized that he spoke of them [and of no one else]’ lat. & phairsei […]/ cognouerunt quod de ipsis dicer&
Similar effects are given in the relative clauses in (15a–b) which are uttered to exclude any alternative to the divine origin of Jesus. The ability to hear or speak the words of God is restricted to a referent with a special property only, namely to the one sent by God. In this interpretation, the pre-verbal phases fon gote ‘from God’ and got ‘God’ act as focus expressions bound by a phonologically empty focus operator triggering an exhaustive effect. In both cases, the focus expressions are shifted from the postverbal domain in the Latin text to the position immediately before the verb in the OHG translation:
(15) a. ther fon gote ist ther horit gotes uúort (T 219, 1) who from God-DAT is DEM hears God-GEN words ‘who came from God can hear God’s words’ lat. qui est ex deo uerba die audit b. then got santa ther sprihhit gotes uúort (T 57, 26) who-ACC God sent DEM speaks God-GEN words ‘who was sent by God may speak God’s words lat. quem enim missit deus uerba dei loquitur There are also examples in which exhaustive interpretation applies to adjuncts or modifiers, see êrist ‘as the first one’ in (16): (16) íogiuuelih gommanbarn/ thaz uuamba êrist Intuot (T 37, 16–17) every male child who womb first opens ‘every male who opens the womb [shall be called holy to the Lord]’ lat. omne masculum/ adaperiens uuluam The constituent êrist reveals exactly that part of the clause which is crucial to the proper understanding of the utterance: the presentation of a newborn child in the temple according to the Law of Moses applies to the firstborn son only, not to the others. Remarkably, êrist has no proper lexical equivalent in the Latin original but is included in the semantics of the present participle adaperiens ‘opening’. In the OHG sentence, the participle construction is transformed into a relative clause, and the semantics of the Latin participle is split into the focused modifier êrist and a finite verb Intuot ‘open-3sg.pres.ind’. The focus phrase is placed immediately before the finite verb while the object DP uuamba ‘the womb’, which is inferable in this context, is shifted to the position immediately after the conjunction. To sum up, the preverbal parts in the two kinds of patterns considered in (10)–(16), i.e. parts of idioms and narrowly focussed phrases share some important common features with respect to Behaghel’s classification. First, they both represent relevant information, the one with respect to the overall semantics of the complex predicate, and the other with respect to the context. According to this, both types of preverbal phrases can be assumed to carry main stress and consequently to meet the condition of prosodic heaviness. Main stress on the preverbal constituents in (11)–(16) is evident from their focal status. As far as non-finite parts of idioms are concerned, their prosodic behaviour in modern German is rather suggestive. As they carry word stress in the complex unit, they also take the functions of the focus exponent in the clause if there is no other argument suitable to carry
Information structure in the Old High German Tatian
main stress, e.g. because it is de-accented in anaphoric use. Exactly this condition applies in the sentences quoted above, where all the rest of the information in the sentences is anaphoric, i.e. de-accented. Therefore, informational relevance and accentuation are two features which link together narrow focus and nominal parts of idioms placed before Vfin.
3.1.3. Interim conclusion Taken together, the clauses with Vend order analysed above prompt the assumption that the domain between CP and Vfin in OHG is organized according to information-structural principles. First, two basic informationstructural domains can be identified here, one preserved for background material, and another one fixed for lexically or contextually relevant, i.e. focused material in the clause. Second, it is obvious that preverbal focus is subject to a further specification. From the examples viewed above, it is clear that it involves single phrases with a contrastive or operator-bound reading, and does not include wide focus or cases of focus projection. Furthermore, there is strong evidence that there are firm principles governing the serialisation of these information-structural domains in the clause: elements belonging to the background are associated with the position immediately following the subordinating conjunction or the relative pronoun while focused XPs has to be left-adjacent to Vfin. In the entire sample of clauses with Vend order, this principle is violated only once: (17) trisiuuet íu treso in himile/ thar noh rost noh miliuua íz nifurmelit (T 69, 15–16) deposit you-DAT treasure in Heaven where neither rust nor moth it NEG-destroys ‘But lay up for yourselves treasures in heaven, where neither rust nor moth destroys them’ lat. Thesaurizate autem uobis/ thesaurus in cello/ ubi neque erugo neque tinea demolitur In (17), the constituent placed left-adjacent to Vfin against the Latin original is the object pronoun iz ‘it-3sg.acc.neutr’. It takes up the previously metnioned antecedent treso ‘treasure’ and therefore represents given information, and the relevant context does not provide any indications leading to the contrastive interpretation of this pronoun. However, this remains
a single occurrence which cannot be taken as representative for the OHG situation in general.6 The results concerning the distribution of the information-structural domains in OHG gained from the analysis of clauses with Vend order may be therefore summarized as follows: (18) CP – [XP….]BGR … – [XP]FOC – Vfin Two relevant questions with respect to the anaylsis of clauses with nonVend order arise from these observations. First, do sentences displaying non-Vend order also show the same distribution of information-structural categories in the preverbal domain, and second, which positional distribution may be provided for wide, e.g. VP-focus in the OHG Tatian.
3.2. Clauses with non-Vend order 3.2.1. The placement of anaphoric material Within the group of non-Vend subordinate clauses, the proportion of cases deviating form the Latin only with respect to the transposition or insertion of the subject pronoun is higher than within the group of Vend clauses. Here, it applies to 52 of all 100 cases; a typical example is given in (19) where the subject pronoun is inserted in the OHG clause but the order of the remaining constituents adheres to that in the original: (19) thaz sie gihórtin gotes uuort (T 55, 2) that they heared-SUBJ God-GEN words ‘in order to hear God’s words’ lat. ut audierent uerbum dei The tendency to place pronominal elements in the domain immediately after the conjunction is confirmed by the syntactic behaviour of nonsubjects in further 19 cases, see the indirect object imo ‘him-3sg.dat.masc’ for lat. ei in (20): (20) só imo gibot truhtines engil (T 35, 2) as him-DAT commanded Lord-GEN angle ‘as the angel of the Lord commanded him’ lat. sicut precepit ei angelus domini
Information structure in the Old High German Tatian
There are only 5 cases in which a pronominal argument remains in postverbal position, each time in accordance with the Latin structure, see e.g. inan ‘him-3sg.masc.acc’ for lat. eum in (21):7 (21) thaz sie fiengin inan (T 119, 9) that they arrest-SUBJ Him ‘in order to arrest Him’ lat. ut raperent eum This evidence confirms the view that background elements have their usual position immediately below C°. However, the DP truhtines engil ‘God’s angel’ in (20), which is discourse-anaphoric as well, is not preposed into this domain. But in the group of non-Vend clauses, we nevertheless find 6 instances in which a full anaphoric DP is shifted from a postverbal position in the Latin to a preverbal position in OHG; the reverse transposition, i.e. to shift anaphoric material after the verb against the Latin original, does not occur. See (22) where the entity uueralt ‘the world’ established in the governing root clause is resumed in the following embedded clauses and placed both times before Vfin against the Latin order:8 (22) nisanta got sínan sun/ In uueralt thaz her uueralt tuome/ NEG-sent God His son/ to world that He world condemned-SUBJ uzouh thaz uuerolt si giheilit thuruh inan (T 197, 30–32) but that world is-SUBJ healed though Him ‘God didn’t send his son into the world to condemn the world, but that the world might be saved through Him’ lat. non enim missit deus filium suum/ In mundum ut l[sic!]udic& mundum/ sed ut salute&ur mundus per ipsum In the database, we find examples which provide an interesting minimal pair given with respect to the placement of given vs. new material in OHG. In (23), the DP thin elimosina ‘your charity’ which is mentioned for the first time in the discourse is retained in postverbal position in accord with the original. But in (24), where it represents given material, the same phrase is shifted to the preverbal domain: (23) thanne thú tuos elimosinam (T 66, 29) then you do charity ‘when you give charity’ lat. Cum ergo facies elimosinam
(24) thaz thin elimosina sí in tougalnesse (T 67, 6) that you charity is-SUBJ in secret ‘in order that your charity be done in secret’ lat. ut sit elimosina tua in abscondito Thus, we can conclude that anaphoric material, be it light or heavy lexical material, is placed as a rule adjacent to C°, while material that represents new information is placed after Vfin. The question is whether this is an accidental distribution resulting from the shift of anaphoric material before the verb, or part of a regular tendency applying in OHG independently of the Latin. In order to check this, we shall turn to the analysis of examples in which the OHG sentence contains postverbal material in contrast to the Latin original.
3.2.2. Properties of postverbal constituents in OHG The following examples contain postverbal material only in the OHG version of the text. First, we shall look at instances in which preverbal material of the Latin structure is realized postverbally in OHG. In our data base, this occurs 7 times. In 3 of the examples, the shifted material is a predicative adjective or participle in a copular construction; see (25a)9. In the remaining 4 cases, the shifted constituent is the direct object of the finite verb, see (25b)10. In all cases, the postverbal information is new and therefore part of the domain of new-information (i.e. presentational) focus: (25) a. giueh& uúarlihho/ thaz íuuere namon sint giscribane/ in himile (T 103, 26–28) be happy PRT/ that your names are written-PL in Heaven ‘Be happy that your names are written in heaven’ lat. gaud&e autem/ quod nomina uestra scripta sunt/ in caelis b. thaz in mir habet sibba (T 290, 8) that in Me have peace ‘that in Me you may have peace’ lat. ut In me pacem habeatis Furthermore, there is a group of sentences which are formed independently of the original to translate a nominal group or a participial construction of the Latin text. As the placement of the verb relative to the remaining con-
Information structure in the Old High German Tatian
stituents is not influenced by the original, we can assume that the linear order attested here obeys the principles of native OHG syntax. Among these rather valuable sentences, we find 7 instances displaying postverbal material only in the OHG text. Two of them are given in (26):11 (26) a. soso thie lihhazara sint gitruobte (T 68, 23) as the hypoctires are sad-PL ‘like the hypocrites are with a sad countenance’ lat. sicut hypocrite tristes b. thes namo uuas giheizzan simeon (T 37, 24) whose name was called Simeon ‘whose name was calles Simeon’ lat. cui nomen simeon All cases involve copular constructions in which the finite copula (sîn ‘be’, heizzan ‘be called’) precedes the nominal part of the predicate. However, exactly the latter carries the new or relevant lexical information in the particular context while the copula is semantically empty and only represents grammatical features like tense and agreement. Quite interestingly, ‘to be’ and ‘to call’ make up the majority of the cases in the group of examples in (25). Following this, we can assume that predicative adjectives and nouns as parts of copular constructions form a stable class of constituents being generally realized in the postverbal domain in OHG in evident contrast to the Latin structure. This is further supported by the fact that the postverbal realization of the predicative part of copular constructions is also typical for Old English (27) as well as for late-OHG texts, see (28) from the Physiologus (mid-11th century). At the same time, postverbal placement of nominal parts of predicates is ungrammatical in modern German subordinate structures of any kind: (27) On hiera dagum Hengest 7 Horsa […] gesohton Bretene on þam in their days Henges and Horsa […] sought Britain on that staþe þe is genemned Ypwinesfleot (ASChr 449) shore that is called Ebbsfleet ‘In their days [in the days of the reign of Mauritius and Valentiunus] Hengest and Horsa arrived in Britain on the shore which is called Ebbsfleet’
(28) dinen schephare, der dir ist ganemmet oriens (Ph 142)12 ‘your creator who is called Oriens’ These observations prompt the assumption that at the right periphery of the sentence in early Germanic, next to the very well known position of extraposed heavy constituents (PPs, heavy NPs and CP-complements), there was also a position occupied by close arguments of the verb providing new information in the discourse. This assumption shall be elaborated further in two more steps. First, constituents maintained in the postverbal position already given in the Latin structure shall be examined with respect to novelty and focus, and then, postverbal focus shall be compared with preverbal focus argued for in section 3.1.2 above.
3.2.3. The nature of postverbal focus We shall look back at the examples in which apart from the placement of background material, no further syntactic differences between the Latin and OHG structure occur. The task will be to find clues for the retention of postverbal material related to novelty or focus on the constituents. Quite certainly, this cannot be assumed for all examples of the kind. See e.g. the PP in (29) which clearly provides given information, as at that particular point in the story, it is known that Zacharias is still in the temple. However, the postverbal PP in (30) conveys a new direction and thus is associated with new-information focus: (29) Inti uuvntorotun thaz her lazz&a in templo (T 27, 23) and marveled that he stayed in [the] temple ‘and marveled that he lingered so long in the temple’ lat. & mirabantur quod tardar& ipse in templo (30) mitthiu her quam ubar thén giozon / in lantscaf gerasenorum (T 86, 31-32) when he came over the river to [the] country Gergesenes ‘when he had come over the river, to the country of the Gergesenes’ lat. Et cum ueniss& trans fr&um/ In regione geraseorum While examples of the type in (29) will not be interpreted further but left aside as equivalents of today’s cases of PP-extraposition, those like in (30) may be associated with focus in OHG. Among the non-Vend sentences,
Information structure in the Old High German Tatian
there are 32 cases in which the phrase retained in postverbal position represents new information. As the examples show, there is no restriction as to the grammatical type of the postverbal phrase: it can be a participle (31a), a PP (31b) as well as an object (31c): (31) a. Quamun thô thie firnfollon man/ thaz sie uuvrdin gitoufit (T 46, 24-25) came PRT the sinful men/ that they became-SUBJ baptized ‘The sinful men also came in order to be baptized’ lat. Uenerunt autem & publicani/ ut baptizarentur b. mit thiu her tho arsteig in skef (T 88, 23) when He PRT went into [a] boat ‘when He got into a boat’ lat. Cumque ascender& nauem c. thaz her giuuente herzun fatero In kind (T 27, 23) that He turned-SUBJ [the] hearts of fathers towards [the] children ‘that He turned the hearts of the fathers towards their children’ lat. ut conuertat corda partium In filios Looking at these examples more carefully, we discover that the new information is not only provided by the postverbal constituents alone but rather comprises the entire VP. In (31a), it covers not only the participle but also the finite verb used to translate the Latin synthetic passive. In (31b), the focus domain includes the verb as referring to a new action together with the discourse-new directional phrase in scef ‘into a boat’. The same applies to the VP in (31c) which assigns new information to the discouse-given referent John the Baptist. In a further group of examples, wide VP-focus spreads over the Vfin and an argument that is being re-activated at that particular point in the discourse. In (32a), a previous event, namely the birth of Christ, is now being related to the king, therefore, the new information is that King Herod also heard about the birth of Christ. In (32b), respectively, the previous action, namely the unexpected catch of fish, is now presented from the point to view of one participant of a group of referents established before. What is achieved here is a kind of change of perspective, or a topic shift with respect to the continuation of the narration. So in these cases, the verb provides together with the postverbal constituent the new-information focus domain, while background information, e.g. the preceding action resumed in the anaphor thaz ‘this’ is placed preverbally:
(32) a. thô thaz gihorta herodes ther cuning (T 39, 17) when that heard Herod the kind ‘when Herod the king heard this’ lat. audiens autem herodes rex b. mit thiu thaz tho gisah simon petrus (T 55, 29) when this PRT saw Simon Peter ‘when Simon Peter saw this’ lat. Quod cum uider& simon petrus These examples confirm the view that apart from cases of PP-extraposition (see (29) above), which is common in modern German as well, the postverbal field in OHG can also host constituents carrying new information, or more precisely, being part of a wide new-information focus domain opened by the finite verb. Therefore, the right periphery of such clauses is structured according to the following scheme: (33)
CP – [XP…]BGR … – VP[Vfin…]FOC
But how does this statement relate to the existence of a preverbal focus position claimed above? The intuitive answer to this question relates to the idea that wide, i.e. new-information focus, and narrow, operator-bound focus, are realized in two distinct syntactic positions distinguished by the placement of the finite verb in the clause. The following section shall provide more empirical support in favour of this view. 3.2.4. Multiple foci in OHG Crucial evidence supporting the existence of two distinct positions for focus material in OHG comes from sentences with multiple foci. Consider (34)–(35) which contain both a discourse-given but contrastively focussed constituent and additional new information after the verb: (34) [b&onte nicur& filu sprehan/ sósó thie heidanon mán = ‘And when you pray, do not use vain repetitions as the heathen do’] sie uuanen thaz sie in iro filusprahhi / sín gihórte (T 67, 23–26) they think that they in their many words/ are-SUBJ heard ‘They think that they will be heard for their many words’ lat. orantes autem. nolite multum loqui/ sicut &hnici.’/ putant enim quia in multiloquio/ exaudiantur
Information structure in the Old High German Tatian
(35) [thanne thu fastes/ salbo thin houbit/ Inti thin annuzi thuah = ‘when you fast, anoint your head and wash your face’] thaz thu mannon nisís gisehan/ fastenti. úzouh thinemo fater (T 68, 29–32) that you men-DAT NEG-are-SUBJ seen/ fasting but your-DAT father‘ ‘so that you do not appear to men to be fasting but to your Father’ lat. tu autem cum ieiunas/ unge caput tuum/ & faciem tuam laua/ ne uideatis hominibus/ ieiunans. sed patri tuo Let us now look at the postitional distribution of the different focus types in these examples. Narrow focus on constituents left adjacent to Vfin is evident in both cases. In (34), the PP in iro filusprahhi ‘for their many words’ acts as a focus exponent bound by an empty focus operator yielding the interpretation that only many words guarantee the fulfilment of the prayers. In (35), narrow focus on mannon ‘men-dat.pl’ results from the fact that it forms a constrastive pair with the explicitly mentioned constituent thineno fater ‘your father’. In both cases, the structures may be viewed to be chosen deliberately by the scribe. So in (34), the equivalent of the Latin synthetic passive could also be constructed in the order ‘participle – Vfin’. Instead, the scribe opted for the reverse order, namely ‘Vfin – participle’, which allows to retain the narrowly focused material before Vfin and to place the new information after it. Similarly, the deponens lat. uideatis is dissolved into a periphrastic construction involving the order ‘Vfin – participle’, while the narrow focus is shifted across Vfin against its original position in the Latin sentence. The sentence in (35) is notable in one more respect. According to the standard view, full PPs like in iro filusprahhi are a typical candidate for extraposition in modern German. It is striking, however, that exactly when pragmatic conditions apply yielding a narrow focus on the PP, it is put in preverbal position like any other type of phrase acting as operator-bound focus in the utterance. It is interesting what causes this particular distribution of focus material in OHG. In order to arrive at a plausible explanation to this question, we shall consider two more examples from the OHG Tatian:
(36) nimág ther man Iouuiht intphahén/ noba imo íz gigeban uuerde fon himile (T 57, 6–7) NEG-can the man anything receive/ NEG-if him-DAT it given became-SUBJ from Heaven ‘a man can receive nothing unless it has been given to him from Heaven’ lat. Non potest homo quicquam accipere/ nisi ei fuerit datum a caelo (37) [thisu sprahih íu = ‘these things I have spoken to you’] thaz in mir habet sibba/ In theru uueralti habet ir thrucnessi (T 290, 7–9) that in Me-DAT have peace/ in the world have you tribulation ‘that in Me you may have peace; in the world, you will have tribulation.’ lat. Haec locutus sum uobis/ ut in me pacem habeatis/ In mundo presuram habebitis In each of these sentences, two different constituents receive focus interpretation. One of them is set in an explicit contrastive relation to another entity in the utterance. In (36) this is the participle gigeban ‘given’ which refers to the only way to obtain spiritual power, namely by being given it, not by acquiring it oneself. In (37) the PP in me ‘in me’ is set into contrast to the expression In therru uueralti ‘in this world’. Additionally, there is also material supplying new information to the context. In (36) this is the source of the spiritual power, namely Heaven, and in (37) it is the direct objects sibba ‘peace’ which is also contrasted to the expression thrucnessi ‘pressure’ in the following conjunct. In the Latin version, in both cases the different types of foci are placed on the same side of the verb, after the verb in (36) and before the verb in (37). In the OHG text, however, these two types of foci are spaced in such a way that the contrastive or narrowly focused information is immediately before Vfin while the new one follows it. This invites the assumption that Vfin in OHG is used to avoid the stacking of two different types of focus in one and the same structural domain in the sentence.13 This enables the distinction of the focus types and the disambiguation of focus interpretations in the sentence.
Information structure in the Old High German Tatian
4. Results: different positional distribution of focus types in OHG The present paper investigated the role of pragmatic factors for the explanation of syntactic variation in the right periphery of subordinate clauses in OHG. It started with the claim that the standard account on word order in subordinate clauses in the early Germanic languages cannot be maintained without any modifications for a number of reasons. If in accord with the previous literature, we assume that OHG displays a basic SOV order maintained in clauses introduced by an overt complementizer, while exceptions to Vend are due to the extraposition of PPs, heavy NPs or CP-complements as well as to verb raising and verb projection raising in verb clusters, we are in need for an explanation of postverbal material like single NP or nominal parts of copular constructions. Therefore, we approached variation in the right periphery from a different perspective which subscribes to the view that verb placement in early Germanic is a grammatical correlate of pragmatic, discourse-based principles. More precisely, our analysis shows that in early Germanic, there is a tight correlation between the informationstructural properties of sentence constituents and their realization with respect to Vfin. This approach was applied to a sample of 100 Vend and 100 non-Vend sentences from the OHG Tatian which deviate from the structure of their Latin counterparts. First, it was shown that background material regularly appears in the domain immediately below C°, while focus material is found in two different structural positions adjacent to Vfin in the clause. Second, it was possible to account for principles governing this kind of distribution of focus material in the clause. On the one hand, narrowly focused information, e.g. contrastive focus on a single XP as well as operator-bound focus, tends to be placed left adjacent to Vfin. On the other hand, the domain of wide, new-information focus is opened by Vfin while the remaining elements of the focus projection follow it. The different positional distribution of focus types gained crucial support by the realization of multiple foci in OHG which occupy distinct syntactic positions with respect to Vfin. The picture derived for OHG evokes clear parallels to the situation in Yiddish as described by Diesing (1997, 390-396). According to her, the different syntactic realization of object NPs with respect to the selecting main verb triggers three different types of semantic interpretation. As Hinterhölzl (2004, 154) observes, these interpretations correspond to different categories of information structure. Leftward movement of an NP outside the VP associates with definitness and specificity, i.e. with background, while postverbal (in-situ) placement yields an existential reading
of discourse-new indefinite NPs as instances of new-information (presentational) focus. Additionally, both definite and indefinite objects in the position left-adjacent to the verb gain a special, marked status only possible when contrastive or corrective emphasis is put on them, i.e. when they are contrastively focussed. From this Hinterhölzl (2004) concludes that OHG, similarly to Yiddish, establishes two different syntactic positions for contrastive vs. non-contrastive, i.e. presentational focus. One significant difference to the situation described for OHG, however, remains: the position of narrow (contrastive and operator-bound) focus is left adjacent not to the main selecting verb but to Vfin in the clause. This, in turn, fits with the observations of Sapp (2006) on verbal clusters in Early New High German. He reports that contrastive interpretation on the immediately preceding XP is among the most influential factors leading to orders in which the Vfin is placed before all non-finite verbs in verb clusters. The question is why these two focus positions were distinguished in the system of OHG. Two hypotheses can be put forward to explaining this issue. The first one relates to aspects of the prosodic realization of focus especially in cases of multiple foci: as focus is prototypically associated with main stress, the placement of the finite verb between two different types of focus was a means of avoiding a clash of two heavily stressed phrases in one and the same structural domain of the clause. This scenario, however, does not account for the regular association of narrow vs. wide focus with a special position in the clause. This feature is reflected in the second hypothesis claiming that the different positional realization of focus types allows the unambiguous interpretation of the pragmatic value of the constituents involved. It guarantees that preverbal focus is interpreted as narrow XP-focus only, excluding the option of focus projection. In this way, OHG avoids a phenomenon known as ‘focus ambiguities’ in modern German (as well as in a number of other non-related languages). It is wellknown that in modern German, main accent on the rightmost XP in basic order yields both VP- or XP-focus while in scrambled order, the rightmost surface constituent receives an unambiguous contrastive interpretation (Abraham 1992). By contrast, in the system reconstructed for OHG, phrases belonging to the domains of new-information focus surface in postverbal position while preverbal focus only triggeres the option of XP-focus with additional effects of contrast, emphasis, and exclusion of alternatives.
Information structure in the Old High German Tatian
Kemenade (1987) proposes a model according to which OE displays properties of an asymmetric SOV language like modern German or Dutch. She assumes a base-generated SOV order maintained in subordinate clauses introduced by an overt complementizer in the head of a functional projection CP. In root clauses, the empty position of the complementizer is filled by Vfin while an optional movement of another constituent to SpecCP yields V2 in the surface. A similar approach has been proposed by Lenerz (1984) for OHG as well. Some basic differences between OE and OHG consist in the obligatoriness of V-to-C movement in main clauses. For Old English (OE), residual V2 in clauses with syntactic operators in SpecCP (wh- and negation words, sentence adverbials like þa/þonne, or a silent imperative-mood operator) is assumed, while in all other contexts Vfin targets another projection below CP (van Kemenade 1987 and 1997, Eythórsson 1996). By contrast, fronting of Vfin to C in cases with non-operators in SpecC is said to apply regularly in OHG (Axel 2007 and in this volume). Basic methodological considerations in favour of this view were put forward by Dittmer and Dittmer (1998). Schlachter’s statistics (in this volume) comprising all dhazs-sentences counting as unambiguous cases of subordination in the OHG Isidor confirms this picture. As Schlachter also shows, Latin influence has to be definitely excluded as a factor leading to this situation in the OHG Isidor. Pronouns are also excluded form extraposition in modern SOV languages. However, all instnaces involving a postverbal pronoun in our database may be explained as imitations of the original, which also involves a postverbal pronominal object, see (i): (i) thaz sie úz uuvrphin sie (T 76, 2) that they PRT threw-SUBJ them ‘that they threw tham away’ lat. ut eicerent eos Additional instances with postverbal pronouns are found in T 50, 21, T 119, 9, T 122, 15 and T 220, 10. But note that Dittmer and Dittmer (1998) provide examples in which pronouns are placed after Vfin against the Latin original, see (ii): (ii) thiedar giotmotigot sih (T 195, 19) who-PRT humiliates ReflPr ‘who humiliates himself’ lat. qui se humiliate (Dittmer and Dittmer 1998, 148). Uhmann (1991, 200 and 217) provides similar facts for modern German, too. In 2 additional cases involving narrow focus on a single phrase, the pattern Background–XP-Focus–Vfin is obviously blocked by the line break. Consider (i) and (ii) where the phrases thin zesuuua ouga ‘your right eye’ and thin ze-
7. 8. 9. 10. 11. 12.
Svetlana Petrova suuúa hant ‘your right hand’ represent a pair of alternatives and therefore receive an interpretation as narrow, contrastive focus: (i) oba thin zesuuua ouga / thih bisuihhe (T 63, 24–25) if your right eye/ you-ACC troubles ‘if your right eye causes you to sin’ lat. quodsi oculus tuus dexter/ scandalizet te (ii) Inti oba thin zesuuúa hant / thih bisuihhe (T 63, 31–32) and if your right hand/ you-ACC troubles ‘and if your right hand causes you to sin’ lat. & si dextra manus tua / scandalizat té Here, the pronoun thih ‘you-2sg.acc’ belonging to the background intervenes between the focus phrase and Vfin. However, it is clear that placing the pronoun in the Wackernagel position and above the focus phrase would violate the line principle. But see Note 4 above. The remaining instances are: T 30, 19–20, T 69, 22–24 and T 84, 10–11. The remaining instances are: T 151, 12, T 224, 4–6. The remaining instances are: T 46, 2–4, T 59, 1, T 89, 26–28. The remaining instances are: T 35, 14–16, T 59, 22, T 60, 3, T 60, 12, T 60, 14, T 60, 16. I owe this example to Richard Schrodt (University of Vienna) who discussed evidence for non-Vend orders in late OHG in a talk “An den Rändern des Satzes. Kommunikative Dynamik im Althochdeutschen” on 17th Feb 2006 at Humboldt University Berlin. In a similar way, Speyer (2008) argues that focus spacing in double-focus constructions is responsible for topicalisation in Old and Middle English.
Primary texts [ASChr]
The Anglo-Saxon Chronicle. A collaborative edition. Volume 3. MS A. ed. Janet M. Bately, 1986. A semi-diplomatic edition with introduction and indices. Cambridge: D.S. Brewer. Die lateinisch-althochdeutsche Tatianbilingue Stiftsbibliothek St. Gallen Cod. 56. hg. Masser, Achim. 1994. Göttingen: Vandenhoeck & Ruprecht.
Information structure in the Old High German Tatian
References Abraham, Werner 1992 Clausal focus versus discourse rhema in Geman: a programmatic view. In Language and cognition 2, ed. Dicky Gilbers and Sietze Looyenga, 1–18. Groningen: Universiteitsdrukkerij. Axel, Katrin 2007 Studies in Old High German Syntax: left sentence periphery, verb placement and verb second. Amsterdam/Philadelphia: John Benjamins Publishing Company. this vol. The verb-second property in Old High German: different ways of filling the prefield. In New Approaches to Word Order Variation and Change in the Germanic Languages, eds. Roland Hinterhölzl and Svetlana Petrova. Berlin: Mouton de Gruyter. Behaghel, Otto 1932 Deutsche Syntax. Eine geschichtliche Darstellung. Band IV. Wortstellung. Periodenbau. Heidelberg: Carl Winter. Diesing, Molly 1997 Yiddish VP order and the typology of object movement in Germanic. Natural Language and Linguistic Theory 15.2: 369–427. Di Meola, Claudio 2000 Die Grammatikalisierung der deutschen Präpositionen. Tübingen: Stauffenburg Verlag. Dittmer, Arne and Dittmer, Ernst 1998 Studien zur Wortstellung - Satzgliedstellung in der althochdeutschen Tatianübersetzung. Göttingen: Vandenhoeck & Ruprecht. Ebert, Robert Peter Historische Syntax des Deutschen. Stuttgart: J. B. Metzlersche Verlagsbuchhandlung. Eythórsson, Thórhallur 1996 Functional Categories, Cliticization, and Verb Movement in the Early Germanic Languages. In Studies in Comparative Germanic Syntax, eds. Höskuldur Thráinsson, Samuel David Epstein and Steve Peter, 109–139. Dordrecht/Boston/London: Kluwer Academic Publishers. Fuß, Eric and Trips, Carola 2002 Variation and change in Old and Middle English – on the validity of the Double Base Hypothesis. Journal of Comparative Germanic Linguistics 4: 171–224. Haegeman, Liliane, and Riemsdijk, Henk van 1986 Verb Projection Raising, Scope and the Typology of Rules Affecting Verbs. Linguistic Inquiry 17: 417–466.
Hinterhölzl, Roland 2004 Language Change versus Grammar Change: What diachronic data reveal about the distinction between core grammar and periphery. In Diachronic Clues to Synchronic Grammar, eds. Carola Trips and Eric Fuß, 131–160. Amsterdam/Philadelphia: John Benjamins Publishing Company. Karttunen, Lauri 1976 Discourse Referents. In Syntax and Semantics 7: Notes from the Linguistic Underground, ed. James McCawley, 363–385. New York/ San Francisco/London: Academic Press. Kemenade, Ans van 1987 Syntactic Case and Morphological Case in the History of English. Dordrecht: Foris. 1997 V2 and embedded topicalization in Old and Middle English. In Parameters of morpho-syntactic change, eds. Ans van Kemenade and Nigel Vincent, 326–352. Cambridge: Cambridge University Press. Kiss, Katalin. E. 1998 Identificational focus versus information focus. Language. Journal of the Linguistic Society of America 74: 245–273. Lakoff, George 1976 Pronouns and Reference. In Syntax and Semantics 7: Notes from the Linguistic Underground, ed. James McCawley, 275–335. New York/ San Francisco/London: Academic Press. Lenerz, Jürgen 1984 Syntaktischer Wandel und Grammatiktheorie. Eine Untersuchung an Beispielen aus der Sprachgeschichte des Deutschen. Tübingen: Max Niemeyer. Masser, Achim 1997 Syntaxprobleme im althochdeutschen Tatian. In: Semantik der syntaktischen Beziehungen. Akten des Pariser Kolloquiums zur Erforschung des Althochdeutschen 1994, ed. Yvon Desportes, 123–140. Heidelberg: Carl Winter. Pintzuk, Susan 1991 Phrase structures in competition: variation and change in Old English word order. Doctoral dissertation, University of Pennsylvania. Pittner, Karin 1998 Radfahren vs. mit dem Rad fahren: Trennbare Verben und parallele syntaktische Strukturen In Zwischen Grammatik und Lexik, eds. Irmhild Barz und Günther Öhlschläger, 103–112. Tübingen: Max Niemeyer. Sapp, Christopher D. 2006 Verb Order in Subordinate Clauses. From Early New High German to Modern German. Doctoral Dissertation, Indiana University.
Information structure in the Old High German Tatian
Schallert, Oliver 2006 Hybride OV / VO-Systeme und syntaktischer Wandel zu OV und VO in den germanischen Sprachen. Diplomarbeit, Universität Salzburg. Schlachter, Eva 2004 Satzstruktur im Althochdeutschen. Eine Skizze zur Position des Verbs im Isidor-Traktat des 8. Jahrhunderts. In Beiträge zu Sprache & Sprachen 4. Vorträge der Bochumer Linguistik-Tage, eds. Karin Pittner et al. 179–188. München: LINCOM. this vol. Word Order Variation and Information Structure in Old High German. An Analysis of Subordinate dhazs-Clauses in Isidor. In New Approaches to Word Order Variation and Change in the Germanic Languages, eds. Roland Hinterhölzl and Svetlana Petrova. Berlin: Mouton de Gruyter. Speyer, Augustin 2008 Topicalization and Clash Avoidance. On the interaction of Prosody and Syntax in the History of English with a Few Glimpses at German. PhD Thesis, University of Pennsylvania. Tomaselli, Alessandra 1995 Cases of Verb Third in Old High German. In Clause Structure and Language Change, eds. Adrian Battye and Ian Roberts, 345–369. New York/Oxford: Oxford University Press. Uhmann, Susanne 1991 Fokusphonologie: eine Analyse deutscher Intonationskonturen im Rahmen der nicht-linearen Phonologie. Tübingen: Max Niemeyer. Valentin, Paul 1987 Zur Geschichte des deutschen Passivs. In Das Passiv im Deutschen, ed. Centre de Recherche en Linguistique Germanique (Nice), 3–15. Tübingen: Max Niemeyer. Weiß, Helmut 2006 Die rechte Peripherie im Althochdeutschen: Zur Verbstellung in dassSätzen. Ms. Wurmbrand, Susi 2004 West Gemanic Verb Clusters: The Empirical Domain. In Verb Clusters: A Study of Hungarian, German and Dutch, eds. Kathalin É. Kiss and Henk van Riemsdijk, 43–85. Amsterdam/Philadelphia: John Benjamins Publishing Company.
Verb placement and information structure in the OHG Gospel Harmony by Otfrid von Weissenburg Andreas Lötscher
Abstract This paper gives an overview on some central rules and factors that determine word order in the OHG Gospel Harmony by Otfrid von Weissenburg. An outline of the elementary principles for the verb position first shows that Otfrid has the same differentiation between main clauses and subordinate clauses as later stages of German. In addition, the rule of the early position of the finite verb and the late position of infinite verbal elements creates a “frame and field”-structure for sentences as is known for later periods of German. Otfrid, however, has more freedom in placing elements in front and after the respective elements at the left and the right border of the verbal frame structure. It seems useful to consider the V1-pattern as the basic order in main clauses; from this, V2-order and V-late order in main clauses can be explained by the influence of information-structural factors. In detail, Otfrid’s language seems to represent an intermediate stage of language development, showing a combination of older and newer patterns. On the whole, however, Otfrid’s language shows more regularity and consistency with regard to word order than is usually assumed.
1. Preliminaries 1.1. Purpose of this contribution The purpose of this paper is to investigate in more detail rules and principles that possibly govern word order in Otfrid von Weissenburg’s “Evangelienharmonie” (Gospel Harmony) and thereby to contribute to a better knowledge of the situation in Old High German in this respect. Within the limits of this paper, I will not deliver a full-fledged grammatical description in all details (which would presuppose a general grammatical descrip-
tion of the overall system), but restrict myself to collecting the basic observations that seem relevant to the above question.
1.2. The text Otfrid von Weissenburg’s “Evangelienharmonie” seems to be a prime source for the grammatical description of OHG. It is the most voluminous and an outstanding vernacular work in the OHG period before Notker. It was written in a period (a.d. 863–871), in which no other work of comparable importance and richness had been produced. In addition to its value as a corpus, it could be a valuable counterpart to Notker, who produced his work about 200 years later. In spite of such facts, Otfrid’s work is usually not taken into consideration with regard to questions of word order. The main reason is the fact that it is a poetical work, written in end-rhyming verses. It is usually assumed that authors of poetical works tend to stretch or neglect rules of word order, in order to satisfy the rigid rules of versification and rhyming. In Otfrid’s work, word order is apparently more variable than in later Old High German works such as those of Notker of St. Gallen. Such deviations of the normal use are explained as irrelevant because they are supposed to be caused by rhyme.1 In favour of Otfrid, we can assume that even if Otfrid does not apply the usual word order principles of the core grammar of his language his use is not irregular, but simply follows special rules, be it of an older tradition, be it of poetical language as a subsystem of the overall system. Moreover, to dismiss as irrelevant any empirical facts in Otfrid’s language that deviate from the use in other authors or from the rules that are expected amounts to a petitio principii: After all Otfrid has decided to chose the formulations he uses as a competent writer of his language, and it should be proved from other, independent facts that he not only deviates from the usage of other authors, but has violated basic rules of his own language.2 It has been noted, of course, that Otfrid often violates grammatical rules under “Reimzwang” (cf. Nemitz 1962). But these observations relate almost exclusively to morphosyntactic problems such as congruency. Above all, to be able to classify a particular formulation as a violation of a grammatical rule, we first have to know the grammatical rules as a whole. In historical texts, where the grammatical system has to be inferred from the regularity of the use, a particular formulation can be classified as a violation if it is a rare case, compared with the usual construction, if it cannot be
Verb placement and information structure in the Gospel Harmony by Otfrid
explained by specific additional grammatical factors, and if the violation can be attributed to external factors such as “Reimzwang” or problems of performance. None of these cautionary considerations have been made with respect to the word order in Otfrid. Thus, before we can classify a particular word order as ungrammatical, we have to show that it is an exceptional use that does not make grammatical sense. Be this as it may, as long as we do not analyse Otfrid’s language in more detail we cannot assess the justification of any judgment about Otfrid’s language. Thus, I will here take the position that an analysis of the language of Otfrid with respect to the word order might be fruitful and that anyway an unprejudiced description might give valuable insights into word order rules of OHG.
1.3. Some theoretical and methodological considerations One of the key questions in investigating word order is the relation among grammatical and information-structural conditions for word order, in the terminology of Hinterhölzl and Donhauser (2003: 174). If it seems reasonable to base an analysis on this distinction, the question has to be asked what the theoretical premises and implications of such a distinction could be. Any systematic distinction between grammatical rules and informationstructural rules, so it seems to me, presupposes rules on essentially different linguistic levels. Grammatical word order rules by definition function on a purely structural level, whereas information-structural ordering rules are based on pragmatic principles, and probably on cognitive principles of a more general nature. In the following, I will proceed on such assumptions. Whereas grammatical word order rules, which are guided by structural properties of a sentence, seem comparatively easy to identify and define, this is not so for pragmatic principles for the sequencing of constituents. In order to be able to formulate a hypothesis on the relation between pragmatic conditions and word order and to avoid a circular argumentation, we have to determine the pragmatic interpretation of a sentence independent of the grammatical structure. Notoriously, interpretations of the pragmatic properties of an utterance are not easy to make evident, as they imply complex procedures of text understanding, based on conjectures about alleged background knowledge and the purpose of an utterance.
Often, a three-layered approach to this problem is proposed, such as described in Petrova/Solf (in this volume). According to this approach, three different levels of information-structure are to be distinguished: – – –
Given – New Topic – Comment Focus – Background
There are many problems connected with these categories. First, there are definitional problems with all of these pairs of terms. There are many differing definitions, some of them rather vague, others overlapping. E.g. there is no agreement whether “given” should mean ‘shared knowledge’, ‘activated knowledge’, ‘recoverability’, ‘textual accessibility’. Moreover it is not clear what the logical or grammatical category of content the property of ‘given’ or ‘new’ has to attributed to: to an individual as a semantic entity, to an expression denoting an individual or to a constituent within a clause. The same is true for the other pairs (for a discussion of the problems see Petrova/Solf, in this volume). Generally, definitions mostly are made on the basis of rather simple patterns; as soon, as we procede to more complicated utterances in a complex text, the application of the definitions becomes difficult. Second, it is not quite clear how these dimensions correlate. Undoubtedly, givenness has a strong correlation to topicality. Typically, one can make an individual entity a topic only if this individual is known to both of the communication partners; a new individual most often is introduced by way of mentioning it as a comment. However, the reverse is not true: one can treat something as a comment even though it is given and has been introduced earlier in the context: In No, I saw YOUR FATHER yesterday, the expression your father almost certainly is given, but a comment at the same time. The same weak correlation exists between “background” and “givenness”: Only entities that are textually accessible can be mentioned as background, but such entities can be focussed under certain contextual conditions, e.g. if they are disputed with respect to an assertion. It depends on the choice of the topic and the comment of a sentence whether something appears to be part of the background of an utterance. Thus, the distinction between “focus” and “background” possibly is not an independent one, but is dependent on the topic-comment-structure of an utterance. Third, when faced whith a task like that of the present contribution, it is difficult to see how these three layers interact with respect to the surface structure of a sentence. The basic question to be asked here is: How is the
Verb placement and information structure in the Gospel Harmony by Otfrid
word order in the text of Otfrid to be explained by grammatical and/or information-structural factors. For information-structural factors, the question is one of correlating form with content. Word order is linear, i.e. it is one-dimensional. How is a three-dimensional content translated in a onedimensional form? Or how can we detect in a one-dimensional form a three-dimensional content? One important methodological corrolary of these considerations is that one should give definitions for the respective pairs strictly in terms of content and avoid taking recourse to surface structure features. Otherwise, form and content are not distinguishable; moreover, circular definitions and/or descriptions result. Focus usually is defined with respect to stress, topic as the first element in a clause, i.e. with respect to word order. In my view, this is not an appropriate approach. Word order and stress are the main formal indications of the information-structure of an utterance. But first, there is no guarantee that there exists a one-to-one correlation between formal means of expression and a particular pragmatic content. E.g. both comment and certain types of topics can bear strong accent; comments can be fronted under certain conditions in New High German. Second, languages can differ. Evidently, all languages do not function the same way in the translation of informations-structural content into surface structures. We do not want to know how a language fits our predefined correlations of form and content, but to find out the way content factors govern the variancy of forms in a particular language. Third, there are possibly more factors influencing word order than just information-strutural ones, and we must be prepared to find other factors and rules govering word order than we expected.3 For the purposes of the present paper, I will restrict here myself to introducing a few basic concepts that will be necessary for the following analyses, omitting further questions and details.4 I will base my analyses on the assumption that the topic-comment distinction is the basic one, and that the distinctions of givenness-newness and focus-background are corrolaries or conditions of it. My main assumption will be that the topic-commentstructure has to be explicated in pragmatic terms, more specifically in terms of the function of an utterance within a communicative context.5 The function of an assertion is to introduce new assumptions, change existing common assumptions or confirm existing common assumptions (Sperber and Wilson 1986: 121). An utterance cannot establish completely new knowledge, but has to relate to some piece of existing common knowledge. We proceed from common knowledge (“given” knowledge in the broadest sense) with the intention to change or confirm certain pieces of knowledge
that seem to us controversial or worth being disputed.6 Making something a “topic” means declaring that this particular part within a common assumption is that part about whose properties one perceives a dispute and wants to introduce new assumptions, change assumptions or confirm assumptions. The corresponding elements introducing new assumptions, changing or confirming the relevant part of the assumptions are called “comment”. I take that which is usually called “background” as that portion of common assumptions connected with the utterance and possibly verbalized that the speaker neither wants to topicalize nor to dispute, that is irrelevant to his communicative intentions, either because he agrees with his audience or because a possible disagreement is not important to him in the given situation. Topics are specially marked if new background assumptions are introduced, specified or a change to other topics takes place. In many situations, in the course of a communication, a topic, once established, can become part of the background. Topics and changes of common assumptions within a communicative process are the result of continuous processes of changing, narrowing down and contrasting pieces of “background” and assumptions disputed.7 This has important implications for the semantic and pragmatic nature of topics and focus: When starting a communication, only elements with a referential counterpart in a context may be disputed; in the course of a communicative process, however, it usually happens that only parts of utterances are discussed, among such that have no referential counterpart or have only grammatical status, e.g. verbal elements, adjectives (predicative, attributive), quantifiers.8 In the following example, in the utterance of B, GOETHE is narrowing down the topic of the discourse (assuming that everybody knows in the given context that among the books there is a volume by Goethe), UNTER is a correction of auf in the preceding sentence; the utterance thus is intended to change a common assumption (‘Where are the books’) with regard to a specific book. In the utterance of C, the “background” for which a change of assumptions will be attempted by the speaker C is narrowed down to UNTER dem Tisch as opposed to AUF dem Tisch (C thus indicates, e.g., that she does not want to dispute the books on the table); she changes the common assumptions about what is under the table in a very general way to ‘many other books’ (Bücher bearing no contrastive value to the accepted background assumptions).
Verb placement and information structure in the Gospel Harmony by Otfrid
A: Ich habe die Bücher auf den Tisch gelegt. B: Der GOETHE liegt aber UNTER dem Tisch C: UNTER dem Tisch sind noch VIELE ANDERE Bücher. ‘A: I have put the books on the table. B: But the book of Goethe is under the table. C: Under the table, there are a lot more other books.’
Whereas the overall problem under dispute may well be covered by the term “topic” in the sense of “expression about whose referent the sentence is” (although this might be only a prototypical case), for other elements that make up only such details of a proposition under dispute, such a definition seems rather odd. The methodological consequences are: We cannot dismiss an element as a topic because it has no referential status; and possibly there are several layers of topics of different status present in an utterance, such as topical elements taken over from previous discourse (“continued topic”) and topics newly introduced or topicalized again. How can we detect the informational structure of a historical written text whose grammar we want to find out? As I said earlier, directly arguing from word order is not suitable, because it begs the question. Stress is not accessible directly in historical written texts, either. In the case of Otfrid, though, we can interprete the accent structure of sentences from the given verse structure; moreover Otfrid has noted accents carefully in his manuscript, although he does not differentiate primary from secondary stress. Basically, we have to interpret the information structure by interpreting the communicative function of an utterance within its context. On the basis of the connectivity of the text (recurrence of expressions referring to entities) and morphosyntactic indications (articles etc.), we can make inferences about the givenness of an entity and consequently about its topicality. But in important respects, our interpretation of the function of an utterance relies on implicatures about its communicative relevance within a given context. Such implicatures are based on assumptions about what a reasonable text should be. Thus, in important respects, the attribution of a particular information-structural value to an expression in a given sentence is the result of a text interpretation based both on universal premises and individual expectations.
2. General observations on the position of verbal elements As mentioned above, the position of verbal elements in Otfrid’s work shows great variety. In a first step I will attempt to establish a general grouping of the different types of order on a more observational level in order to have a basis for more detailed analyses of the particular problems.9
2.1. The position of the finite verb It is important to differentiate between the position of verbs in main clauses and in subordinate clauses, although in the particular sentence the differences are not always so evident.
2.1.1. Main clauses In main clauses, the finite verb (V-fin), can be in the first position (V-1), in the second position (V-2), in a position later than V-2 (V-late) or even at the end of the clause (V-last). V-fin=V-1: (2) Spráchun sie tho zimo sár: \ "meistar, zéllen wir thir wár, wir woltun wízan in giwís, \ war thu émmizigen bíruwis." (II, 7, 17–18) ‘They told him at once: „Master, we tell you in truth we wanted to know surely where you live usually “’ V-fin=V-2: (3) In búachon ist nu fúntan: \ thaz wort theist mán wortan, iz ward héra in worolt fúns \ joh nu búit in úns; Wir sáhun sinaz ríchi \ joh sina gúallichi, thaz was scóni al so frám, \ so sélben gotes súne zam. (II, 2, 31–34) ‘In the books it is found: the word has become man, it was here in the world at his own will and lives among us now. We have seen his realm and his goodness, that was so good, as it befits to the son of god.’
Verb placement and information structure in the Gospel Harmony by Otfrid
V-fin=V-late: (4) óuh man thara ládota \ thie júngoron thier tho hábeta. (II, 8, 8) ‘They had also invited the disciples he then had’. V-fin=V-last: (5) Er fíngar sinan thénita, \ then júngoron sar tho zélita, joh sár in tho giságeta \ thia sálida in thar gáganta. (II, 7, 9–10) ‘He pointed with his finger, he told his disciples and explained them at once that salvation was approaching them.’
2.1.2. Subordinate clauses In subordinate clauses, the possibilities of the positions of the finite verb are more restricted, compared to that in main clauses: the usual position either is final (V-last) or later than second (“late position”, V-late): V-fin = V-last: (6) Sie lóbont inan hárto \ frónisgero wórto, joh thánkont es mit wórte \ Kriste themo wírte; Want ér unsih fréwita, \ then gúaton win uns spárota, ther fúrdir uns ni wénkit, \ joh géistlicho drénkit; Thaz únsih es gilúste, \ thera fréwida ni bréste, joh wír zi themo gúate \ io wesen fástmuate. (II, 10, 17–22) ‘They praise him with magnificent words and thank it to Christ the host with words, because he delighted us and saved the good wine for us. that in future does not falter and gives us a spiritual potion, so that we rejoice it and do not lack joy and always be eager to do the good.’ V-fin = V-late: (7) Maht lésan wio iz wúrti \ zi theru drúhtines gibúrti, thaz éngil mit giwúrtin \ iz kundta sar then hírtin; Joh theiz ni wás ouh bóralang \ thaz hériskaf mit ímo sang; wio éngilo ménigi \ fúar thar al ingégini. (II, 3, 11–14) ‘You can read how it happened with the birth of the Lord that the angel with joy announced it to the shepherds: and it was within a short time that the hosts sang with him how the crowd of angels approached them.’
In this classification, some types of positions in main clauses and subordinate clauses are grouped to a simple “V-late”-position. We could differentiate in more detail between, say, Verb-third, Verb-fourth for main clauses and V-late (or “Verb-second-to-last” etc.) for subordinate clauses, as does e.g. Wunder (1965: 482–483). In most cases, such a differentiation does not yield much information about the underlying grammatical facts. On the contrary, on a purely observational level, a more differentiated description suggests more of a grammatical description than can be accounted for. In a clause with, say, four elements and the finite verb placed in the middle of this sequence, it is rather arbitrary or it even begs the question to classify this sequence either as a V3-structure or a V-late-structure.
2.2. The relative position of finite and infinite verbal elements Things become more transparent if we consider the position of nonfinite verbal elements relative to that of the finite verb. Infinite elements appear in Otfrid in the perfect tense, (participle + auxiliary), passive or passivelike constructions (participle + sîn/werdan), modal constructions (infinitive + modal verb) and other infinitive constructions, (e.g. infinitive + biginnan).
2.2.1. Main clauses For the relative positions of a finite verb and a nonfinite element, we have to differentiate between main clauses and subordinate clauses. In main clauses the finite verb always precedes the nonfinite element (apart of some special cases to be discussed later). The position of the nonfinite verbal element relative to other nonverbal elements again is variable: it can be in the ultimate, penultimate or antepenultimate position, and there can be one or more constituents between the finite and the nonfinite verbal element. This means, among other things, that e.g. a direct object of a nonfinite element can be placed before or after this nonfinite element. Taken together, this yields quite a number of combinations of different distributions of the two verbal elements in a clause. Nevertheless, the principle that the nonfinite element has to follow the finite one implies that the nonfinite element statistically is very often in final position.
Verb placement and information structure in the Gospel Harmony by Otfrid
V-fin = V-1, V-infin = V-last: (8) wólt er sar mit wíllen \ thaz sin gibót irfullen. (II, 9, 42) ‘He eagerly wanted to fulfil his commandment.’ V-fin = V1, V-infin = V-late: (9) Liaz ínan waltan álles \ thes wúnnisamen féldes; (II, 6, 11) ‘He let him cultivate all the beautiful land.’ V-fin = V-2, V-infin = V-last: (10) Ni wolt ér fon níawihti \ (thoh er so dúan mohti, ob ér thes wolti thénken) \ then selbon wín wirken; (II, 10, 1–2) ‘He did not want at all (although he could have done so, if he had intended) to create this wine (i.e. from the water).’ V-fin = V-2, Vinfin = V-late: (11) Wir scúlun uns zi gúate \ nu kéren thaz zi múate, mit wiu ther díufal so frám \ bisueih then ériston man; Wir sculun dráhton bi tház, \ thaz wir giwárten uns thiu báz. (II,5, 1-3) ‘We shall to our benefit pay attention to that, how the devil deceived very much the first human being; we shall strive thereby to beware of that.’
2.2.2. Subordinate clauses In subordinate clauses both the position of the whole group of verbal elements and the position of the finite element with respect to the infinite element differ from that of main clauses. First of all, in the overwhelming majority of the instances (e.g. in book II, in 90 % of all cases, i.e. in 28 of 31 instances), one verbal element or the whole group has clause final position. As a rule, the whole group is clause final: (12) Scúld bilaz uns állen, \ so wír ouh duan wóllen, (II, 20, 35) ‘Release all of us from our debts, as we will do as well.’ In a few cases the verbal group is followed by another element: (13) Ther evangélio thar quit, \ theiz móhti wesan séxta zit; (II, 14, 9) ‘The gospel says that it could have been the sixth hour.’
In most cases, the two verbal elements follow each other directly. It is possible, however, that they are separated by another nonverbal element: (14) Er tháhta odowila tház, \ thaz er ther dúriwart wás, er íngang therera wórolti \ bisperrit sélbo habeti; (II, 4, 8f.) ‘He thought perhaps that he was the doorman, that he himself had locked the entry to this world.’ As an important difference to main clauses, in subordination clauses, the relative positions of finite and nonfinite elements is variable and does not seem to be fixed by a grammatical principle. The sequence V-non-fin – V-fin is used as often as the reverse sequence V-fin – V-non-fin. The order of verbal elements does not seem to be influenced by their grammatical nature; whether the finite verb is an auxiliary or a modal verb does not influence the frequency of the order. (15) a. Ni wolt ér fon níawihti \ (thoh er so dúan mohti, ob ér thes wolti thénken) \ then selbon wín wirken; (II, 10, 1–2) ‘He did not want at all (although he could have done so, if he had intended) to create this wine (i.e. from the water).’ b. Bigínnu ih hiar nu rédinon, \ wio ér bigonda brédigon joh méistera ther uns ónda, \ sámanon bigónda. (II, 7, 1f.) ‘I begin here to tell, how he began to preach and to gather the teachers he gave us.’ An exception from this principle is the well-known case of conditional clauses, in which the finite verb has clause initial position, as in all German varieties up to present day Standard German: (16) Wil thu iz kléinor reken, \ in wíne gisméken: fon Kríste scalt thu iz zéllen, \ gisteist thu tház irwéllen. (II, 9, 69–70) ‘If you want to interpret it more subtly, taste it as wine, you have to relate it to Christ, if you want to do so.’
2.3. Conclusions From this, we may conclude that the distinction between main clauses and subordinate clauses by means of verb placement is a fundamental gram-
Verb placement and information structure in the Gospel Harmony by Otfrid
matical principle in Otfrid. In this respect Otfrid follows the same fundamental rules that are characteristic for all later German variants, with some differences in detail: –
In main clauses the finite verb has to precede a nonfinite element obligatorily; the finite verb is in an “early” position, the infinite element in “late” position. This creates a “frame-and-field structure” comparable to that of all later historical variants of German. In subordinate clauses both elements of the verb group are in “late” or clause final position, but without a strict ordering among them. Otfrid differs from later stages of German such as Notker (cf. e.g. Näf 1979) or Middle High German insofar as in main clauses, the finite verb often appears in the first as well as in a late or the last position.
All this implies that the verb position in Otfrid at least partially is governed by grammatical rules. However, it is not possible to give rules for an absolute position of the respective verbal elements, such as “verb second” or “verb final” position. To speak of “verb first”, “verb second”, “late position” is only a superficial observational characterization that cannot be understood as a grammatical description. In many respects this situation is comparable to the situation in Middle High German. The basic approach to describe the greater freedom of word order in Otfrid, as in Middel High German, to describe the variable position of non-verbal elements with regard to verbal elements as the result of the influence of pragmatic factors such as information-structural conditions, among others.
3. Verb-last position in asyndetic subordinate clauses In Otfrid, clauses are not infrequent that seem to be subordinate from their apparent logical or grammatical connection with the context and or from the mood (subjunctive), although they are only asyndetically linked to it. In (17a), this sentence is a complement to haben, in (17b), the clause is an attributive clause to salida, the relevant clauses in (17c) and (17d) are purpose clauses:
(17) a. Thaz kínd thaz druag thaz wítu mit, \ joh er iz hábeta furi niwíht, er fon thes fáter henti \ tho thar dót wurti. (II, 9, 43–44) ‘The child carried the wood with him, and he considered it as nothing, that he should die from the hands of his father.’ b. Tho uns ward thiu sálida so frám, \ er sélbo in thesa wórolt quam, thaz thiu sin géginwerti \ zi sálidon uns wúrti: (II, 10, 7–8) ‘When the bliss was given to us, that he himself came into this world, so that his presence became our salvation …’ c. Iz ist giscríban fona thír, \ thaz faren éngila mit thír, sie thih biscírmen állan \ joh thíh ni lazen fállan; sie thín giwaro wárten \ jóh thih harto hálten, thaz thin fúaz iowánne \ in stéine ni firspúrne. (II, 4, 57–60) ‘It is written of you that angels escort you that they protect you entirely and do not let you fall. So that they keep you in safety and hold you fast that your feet never hit at a stone.’ d. Bi thiu ílemes io gigáhon \ zi then drúhtines ginádon, er unse wéga irwente \ fon themo fíante; Er únsih ni bisoufe \ áfter themo dóufe. (II, 3, 63–65) ‘Because of this we hurry to come to the grace of the Lord that he turn our ways away from the foe, so that he does not drown us after the baptism.’ In our context, the main relevance of such constructions lays in the fact that the subordination is marked almost exclusively by the late or final position of the verb; this can be interpreted as one more indication that the verb-last position in fact is a grammatical feature of subordinate clauses.
4. Wackernagel position and field structure As a consequence of the variability of the verb positions, in many cases it is not decidable from the superficial word order what the appropriate grammatical description of the verb position or the position of non-verbal elements relative to a verbal element should be. In the following examples, there are at least as many constituents following the verb as are preceding it; sometimes the verb even occupies the position just after the conjunction,
Verb placement and information structure in the Gospel Harmony by Otfrid
although, within the given grammatical context, we would classify the clause as a subordinate clause and thus predict a final or at least a late position of the verb: (18) a. Ni mag thaz mán duan nihéin, \ thaz thiono héreren zuein, thaz er irfúlle io follon \ bédero willon. (II, 22, 1–2) ‘Nobody is able to achieve it that he serves two masters, that he fulfils the will of both of them.’ b. Firnim in álawari, \ thaz got ther fáter wari, joh thaz kínd eino \ Krístan bizéino; Then er zi tóde salta \ bi únsih, sos er wólta, noh themo éinigen ni leip, \ io so Páulus giscréip; Wio er sélbo druag thaz krúzi, \ tho er thúlta thaz wízi, joh irstarp tháre \ in thes cruces áltare; (II, 9, 75–80) ‘Learn in truth that god was the father and the only child means Christ, whom he gave away for death for us, as he wanted to, and did not spare it, as Paul has written, how he himself carried the cross, as he suffered the punishment, and there died on the altar of the cross.’ Such examples demonstrate that a purely numerical classification of the verb positions is not a good grammatical description; such a description has to be of a more abstract nature. A passage such as thaz thiono héreren zuein (II, 22, 1), showing a null-subject construction, can superficially be either characterized as a V-1 or as a V-late pattern, or, assuming a late deletion of a subject pronoun, even as an underlying V-2 construction. The appropriate description can only be given by correlating the surface order to a corresponding grammatical derivation, accounting for the whole range of variations. One indication that there is really a field structure is the position of enclitic weakly stressed elements, mostly pronouns. In New High German their regular position is at the left side of the middle field or, in other words, directly at the left bracket of the middle field, the so-called „Wackernagel position”. In main clauses, this left bracket is constituted by the finite verb in V-2-postion, in subordinate clauses by the conjunction or the relative pronoun. The differences between main clauses and subordinate clauses with regard to the position of the finite verb create differences in the position of enclitic elements: in main clauses they follow the finite verb, in subordi-
nate clauses they precede it. A pronoun with the function of an accusative object thus precedes the subject: (19) a. Vielleicht ADV
dich dein Bruder angerufen. OBJakk SUBJ Vinf ↑ Wackernagel position ‘Perhaps your brother has called you up’. b. dass dich vielleicht dein Bruder angerufen hat. CONJ OBJakk ADV SUBJ Vinf Vfin ↑ Wackernagel position ‘that perhaps your brother has called you up’
These regularities apply for Otfrid’s text as well. Weakly stressed pronominal and other weak elements regularly have their enclitic place in the Wackernagel position. This is clearly borne out in the above examples, cf. e.g. (15a), where pronominal elements follow or precede verbal elements, depending from the type of clause: (15) a. Ni wolt ér fon níawihti \ (thoh er so dúan mohti, ob ér thes wolti thénken) \ then selbon wín wirken; (II, 10, 1–2) ‘He did not want at all (although he could have done so, if he had intended) to create this wine (i.e. from the water).’ The principle of the Wackernagel position for weak elements, incidentally, applies also to V-1-structures: (20) a. Gab er mo ántwurti \ mit súazeru giwúrti (II, 7, 57) ‘He answered him with sweet joy.’ b. Fúart er sar tho thárasun \ then sélbon sinan drútsun; wólt er sar mit wíllen \ thaz sin gibót irfullen. (II, 9, 41–42) ‘He led there his dear son; he wanted wilfully fulfil his order. ’ These observations have several implications. First, the position of enclitic elements indicates that in subordinate clauses in fact verb late position applies. Second, this can be taken as one more indication that both for V-1structures and V-2-structures Otfrid uses a field structure comparable to the field model as we know it from later German periods. A sensible generalization will be that this is the result of a general grammatical rule applicable
Verb placement and information structure in the Gospel Harmony by Otfrid
to all clauses; in other words, we have to interprete infinite verbal element as being in verb late position even when there no other element between the finite and the infinite verbal element and even when there is possibly another element following the infinite element, such as in (18a) (thaz thiono héreren zuein). On the basis of these generalizations, we can decide for particular clauses whether it is a main clause or a subordinate clause. When a weakly stressed pronominal element precedes a finite verb as in (18a), we have an instance of a V-late position, even though the finite verb may have the second position in a clause and is followed by other nominal elements. This may be especially relevant for clauses introduced by wanta, which are notoriously ambiguous between main clauses and subordinate clauses; e.g., in the following passage the wanta-clause has to be interpreted as a subordinate clause: (20) c. Sálig sint zi gúate \ thie rózegemo múate, wanta in firtílot thaz sér \ dróst filu mánager; (II, 16, 9–10) ‘Blessed are those with a sorrowful mood, because much consolation removes their sorrow.’
5. Verb-late and verb-final position As a result of the above considerations, the verb in late or final position of any type of clause has to be interpreted as the right bracket of the middle field. Usually, in situations as this, the final position is taken to be the basic position of the verb (finite or infinite), and the position of other elements right of the verb is attributed to extraposition. Apparently, this rule of extraposition works differently in the different stages of German. As to Otfrid, postposition relative to a “late” verb is only possible for certain types of constituents: Complements to a verb only occur in postverbal position if they are “heavy” constituents in a double sense: they are composed of fully lexical material and they carry strong accent, being comment. Otherwise these constituents have to be in the middle field (or Wackernagel position): (21) a. Ther evangélio thar quit, \ theiz móhti wesan séxta zit; (II, 14, 9) ‘The Gospel says that it might have been the sixth hour.’
b. Sie kúndtun thar then líutin, \ thóh si es tho ni rúahtin, thaz ín was queman hérasun \ ther gotes éinigo sun. (II, 3, 25–26) ‘They announced to the people, although they did not realize it at that time, that the only son of God had arrived.’ c. Nu léru ih íuih hárto \ kúrzero wórto, wio ír giduet fóllon \ then drúhtines wíllon. (II, 23, 1–2) ‘Now I teach you in very short words, how you fulfil completely the will of God.’ This restriction does not apply to adverbial constituents: They can follow a final verb even when they are weakly stressed short constituents (cf. also the position of hérasun in example (21b) above): (22) a. Ih mag iz wóla midan, \ mag hiar nídarstigan; ziu scal ih íowanne \ gótes koron thánne. (II, 4, 77–78) ‘I can easily avoid this and stay down here, why should I tempt God then ?’ b. Iz ist giscríban fona thír, \ thaz faren éngila mit thír. (II, 4, 57) ‘It has been written of you that your are escorted by angels’ A particularly interesting case is (22b) in our context. Two phrases are following the finite verb faren, which has to be assumed in “late position” from our former generalizations. That éngila is following the verb, can be accounted for by the extraposition possibility just mentioned: it can safely be assumed that éngila has the IS-function of a comment, being the relevant new information within the given context. Furthermore, it can be assumed with some certainty that mit thír is a weakly accentuated verbal phrase. Thus, an weakly accentuated adverbial element follows an heavily accentuated complement, being a comment, when both are extraposed. Whether this is a general rule or just a possibility remains to be investigated more closely and I leave it open as a question. We can describe this situation in the following grammatical terms: –
We can assume that verb final position is the basic position for the respective cases (nonfinite verbal elements in main clauses, both finite and nonfinite verbal elements in subordinate clauses). This verbal final rule can be overruled by extraposition of complements and adverbial elements, with clear differences among the different types of elements:
Verb placement and information structure in the Gospel Harmony by Otfrid
Extraposition of complements can apply under restricted pragmatic conditions. Only a complement with heavy accent being a comment can be extraposed. The general semiotic principle underlying this rule is the tendency for comments to be placed at the end of a clause. Other constituents such as adverbs are free in their position with regard to the clause final verb. Without going further in details we can hypothesize that pragmatic factors rule the position of such elements, too. As far as I can see, extraposed adverbial elements tend to mention contextual parameters already fixed within the context such as time, place or addressee etc.; in other words, they are “background elements”. Of course, other major factors in Otfrid’s case are rhyming and rhythm. But we may assume that the extraposition of adverbial elements is of another nature as that of complements. Semiotically, it has more the character of “Nachtrag” (postscript), grammatically, the extraposition of adverbials places them in a position outside the clause.
6. Verb-first and verb-second position As mentioned earlier, Otfrid displays a much wider variability for the position of the finite verb in the main clause than later German periods. A central question is how this variability is to be accounted for. I will first try to give a description of the use of V-1 and V-2 positions.
6.1. Indications for a grammatical V-2-rule There are some indications that under certain grammatical conditions, a verb-second rule is valid: Some grammatical facts seem only to be explicable in a straightforward way if we assume a verb-second rule as a more general grammatical principle. The usual verb-second rule implies a defined prefield that can only be filled with one constituent. There are certain well-defined cases in Otfrid’s work that allow exactly one constituent in the prefield. If there were no such verb-second rule it should be possible that also in these cases other constituents accompany such constituents, which does not seem to be the case.
6.1.1. Question words One such case are question words in direct questions, which always take the place of the prefield and are not followed by any other constituent: (23) a. wio mág thaz sin firlóugnit, \ thaz hímil theru wórolti ougit? (II, 3, 20) ‘How can this be denied that heaven shows itself to this world?’ b. Waz mag ih zéllen thir ouh mér? \ ther púzz ist filu díofer; (II, 14, 29) ‘What can I tell you more? The well is very deep.’ Without a verb second rule it should be possible that other constituents follow such constituents, a situation that as far as I can see never arises.
6.1.2. Resumptive elements Another case of apparent V-2-structures are resumptive constructions of different types, either with noun phrases or with subordinate clauses, where a left dislocated structure is resumed by an anaphoric expression, which appears in the prefield position. The left dislocated structure must be classified as being outside the field structure (for New High German in the “linkes Außenfeld”, cf. Zifonun et al. 1999: 1577), grammatically outside the clause structure. The nominal type is usually known as “left dislocation” (“Linksversetzung”): (24) a. Unsere áltfordoron \ thie bétotun hiar in bérgon; (II, 14, 57) ‘Our ancestors prayed here in the mountains.’ b. Ther géist ther ist drúhtin \ mit fílu hohen máhtin; (II, 14, 71) ‘The holy ghost is the lord with very high power.’ Another type of such resumptive constructions are subordinate clauses resumed by an adverb or a resumptive pronoun: (25) a. Ther fon ther érdu hinana íst, \ ther scal spréchan thanana er íst. (II, 13, 19) ‘Whoever is from this earth, he shall speak from where he is.’ b. Wio íh iu hiar gibíete, \ thaz hóret ío zi gúate; (II, 19, 13) ‘How I give you commands, listen that to your benefit.’
Verb placement and information structure in the Gospel Harmony by Otfrid
Resumptive elements, like question words, are always alone in filling the position in front of the verb, which can again be interpreted as an argument in favour of a V-2-rule in Otfrid. Note, however, that a left dislocated element is not necessarily resumed by a resumptive element. If no such element occurs, the following clause can be organized differently, according to the general rules for sentences without precedent subordinate clause, either with the usual V-2-structure or even with a V-1-structure: (26) a. So sie tho thára quamun, \ thaz héimingi gisáhun, sie núzzun thera héimwisti \ then dág tho mit gilústi. (II, 7, 21–22) ‘When they came there and saw his home, they enjoyed the house that day with delight.’ b. Ob ér sih thoh biknáti, \ jáhi sos er dáti, zaltiz állaz ufan síh: \ ni wúrtiz alles so égislih; (II, 6, 43–44) ‘If he had repented it and admitted it when he had done it, and had taken it upon himself, it would not have become so terrible.’ c. Thie ínan thoh irkántun \ joh múates sih biwántun, giéret er se in thén sind, \ tház sie warin gótes kind. (II, 2, 27–28 ‘Those who recognized him and changed their mind, he honoured them so that they became the children of god.’ Thus, the argument in favour of a V2-rule holds only in case resumptive elements are used. Note by the way that the direct connection of a main clause with a verb after a preceding subordinate clause such as in (26b) and (26c) has to be interpreted as an instance of a V-1-structure and not as a V-2-structure with a subordinate clause filling the prefield slot (“integrated front position”). In later medieval German, subordinate clauses of any type are not allowed to fill the prefield position, as in New High German, but have to be preposed and resumed by a resumptive particle (cf. Lötscher 2005).
6.1.3. Obligatory verb-first constructions There are constructions that apparently demand verb first position obligatorily, namely direct questions and conditional clauses. Verb first position in these cases seems to be a grammatical rule. Such a rule can only have
structural sense if these constructions are opposed to grammatically defined verb second constructions. (27) Lási thu io thia rédina, wio drúhtin threwit thánana? (V, 19, 31) ‘Have you ever read the passage how the Lord appears threatening from there?’ Taken together, such observations seem to indicate that Otfrid obeys a grammatical rule of V-2 comparable to later variants of German.
6.2. Information-structural factors for the choice between verb-first and verb-second constructions Besides V-2-structures, we find V-1-sentences in declarative clauses as well, and they are fairly frequent, in fact. Thus, we cannot assume a V-2rule as an obligatory general principle. This leads to the question what factors could decide for a choice between V-1- and V-2-structures, especially for the cases of V-1- and V-2-structures that are not grammatically specifiable as those mentioned in the previous section.
6.2.1. Topics Hinterhölzl and Donhauser (2003) propose an information-structural account for the alternatives of V-1-sentences and V-2-sentences. They state for Old High German in general that V-2-sentence are used when a new topic is established or a topic is changed, whereas V-1-sentences are used for “all-focus” utterances (utterances without an overt topic) or sentences where an established topic is taken over from the context. Basically, this hypothesis is born out for Otfrid as well. This can be seen in a text sequence such as the following one: (28) 15 “Heil mágad zieri, \ thíarna so scóni, állero wíbo \ gote zéizosto! Ni brútti thih múates, \ noh thines ánluzzes fárawa ni wenti; \ fol bistu gótes ensti! Fórosagon súngun \ fon thir sáligun, 20 wárun se allo wórolti \ zi thir zéigonti. Gímma thiu wíza, \ magad scínenta,
Verb placement and information structure in the Gospel Harmony by Otfrid
múater thiu díura \ scált thu wesan éina. Thú scalt beran éinan \ alawáltendan érdun joh hímiles \ int alles líphaftes, Scépheri wórolti \ (theist min árunti), fátere gibóranan \ ebanéwigan. Got gíbit imo wíha \ joh éra filu hóha (drof ni zuívolo thu thés), \ Davídes sez thes kúninges. Er ríchisot githíuto \ kúning therero líuto; (thaz steit in gótes henti) \ ána theheinig énti. Állera wórolti \ ist er líb gebenti, tház er ouh inspérre \ hímilrichi mánne.” Thiu thíarna filu scóno \ sprah zi bóten frono, gab si imo ántwurti \ mit súazera giwurti: … (I, 5, 15–34) ‘ “Hail, beautiful maiden, young lady so beautiful, the dearest of god of all women! Do not be frightened and do not loose the colour of your face: you are full of god’s grace. The prophets have sung of you, the blessed one and they have prophesied you to all the centuries. Brilliant gem, shining girl, you shall be the illustrious mother, you alone. You shall bear the one who alone rules heaven and earth and all living beings. The creator of world, (this is my message), the son who is as eternal as the father. God gives him glory and high honour, (do not doubt about this), and the seat of David the king. He will rule gloriously as a king of the people, (this is in the hand of God), without ending. He will give life to all the world. so that he will unlock heaven for mankind. The beautiful girl spoke to the messenger of the Lord, answered him with sweet joy: …” ’
Passages that confirm the observations of Hinterhölzl and Donhauser (2003) are e.g. the verses 19 and 20, 28, 29 and 33–34: Forasagun in v. 19 is a newly introduced topic, that is resumed and continued by se in v. 20; in v. 19, the expression is placed in the prefield, in v. 20 enclitically after the verb (in the Wackernagel position), the prefield is left empty. A similar sequencing of topic introduction and continuation can be found in verse 33,
where thiu thiarna is introduced as a new topic, (implicitly opposed to the angel speaking, mentioned before in v. 15), which is again continued in pronominal form si in enclitic position in verse 34. Similarly in V. 28 Got is introduced as a new topic and posited in the prefield. A slightly different case is represented in v. 29, however, where the seemingly anaphoric pronoun er (as a continued topic, connected to the complement imu in the preceding sentence) appears in the prefield. On a closer look, er in v. 29 is not a continued topic, but it’s mentioning brings about a topic change from got to er (= the son of God) as well. As hypothesized by Hinterhölzl and Donhauser (2003) for Old High German in general, V-1-sentences occur in Otfrid’s work for statements without a constituent identifiable as a new topic in a textually relevant sense. (29a) and (29b) are characteristic in that a V-1-sentence is opening a text sequence with a new thought or a new textual topic, not having a indivual referent as a single topic, however, about which the sentence is a statement. II, 7, 1 in (29a) is a typical opening of a chapter; ih as the speaker of the sentence cannot be considered as a topic in a precise sense, rather, it is a background element. Similarly, in II, 7, 5, the sentence as a whole is a holistic desription of a situation where Johánnes is not treated as the topic about which the utterence is. II, 6 47 in (29b) is a kind of a gnomic statement; a generic ‚everybody’ without precise referent cannot be a real topic in such a statement either. (29) a. Bigínnu ih hiar nu rédinon, \ wio ér bigonda brédigon joh méistera ther uns ónda, \ sámanon bigónda; Mit zúhtin sier mo húldta \ joh wísduames irfúlta, sant er thíe tho in allahánt, \ so himil thékit thaz lant. Stuant Johánnes gomono éin \ mit sinen júngoron zuein, gisáh er gangan thárasun \ then selben drúhtines sun; (II, 7, 1–6) ‘I begin here to tell how he began to preach and to gather the teachers that he gave us. He won them for himself by teaching them and he filled them with wisdom. Johannes, one of the men, stood there with his disciples, he saw going there God’s son.’ b. Ságe mir nu, friunt mín, \ wio dati só bi then win, thih sus es nu inthábetos, \ so lángo nan gispáratos? Gíbit giwelih mánno, \ ther fríunta frewit gérno, (ih weiz, thu es ínnana bist) \ then fúriston io sar zi érist. (II, 6, 45–48)
Verb placement and information structure in the Gospel Harmony by Otfrid
‘Tell me, dear friend, why did you proceed with the wine in this way, that you have hold it back, that you have saved it for such a long time? Everybody who wants to please his friends (I know that this is well known to you) serves the best wine first.’ In other cases, referents are mentioned pronominally that have been introduced earlier in the context. They could perhaps be interpreted as “continued topic”. As a matter of fact, however, such elements cannot considered as textually relevant topics; they rather give background information: (30) a. Thiu muater hórta thaz tho thár; \ si wéssa thoh in álawar, thaz íru thiu sin gúati \ nirzígi thes siu báti. Gibót si then sar gáhun \ then thes lídes sahun, so wás so er in giquáti, \ iz íagiliher dati. (II, 6, 23–26) ‘His mother heared this, but she knew for sure. that his goodness would not refuse what she begged for. She then asked those who were responsable for the beverages to do whatever he would order.’ b. er kérta sih sar widar zín, \ quad: “gúate man, waz skel iz sín?” Spráchun sie tho zimo sár: \ “meistar, zéllen wir thir wár, wir woltun wízan in giwís, \ war thu émmizigen bíruwis.” (II, 7, 16–18) ‘He turned to them and said: “Good men, what do you want?” They anwered him: “Master, we tell you honestly, we wanted to know for sure, where you live everyday.” ’ These principles apparently are not in any case followed consistently. In some cases, pronominal elements resuming a topic referent of a previous clause (“established topics”) sometimes do not appear in the middlefield in the Wackernagel position, but in the prefield; it seems that this is restricted to subjects: (31) a. Thie ínan thoh irkántun \ joh múates sih biwántun – giéreta er se in thén sind, \ tház sie warin gótes kind. Ni quámun sie fon blúate \ noh fon fléislichemo múate; sie wárun er firlórane, \ nu sint fon góte erbórane In búachon ist nu fúntan: \ thaz wort theist mán wortan, iz ward héra in worolt fúns \ joh nu búit in úns; Wir sáhun sinaz ríchi \ joh sina gúallichi, (II, 2, 27–33)
‘Those who recognized him and changed their mind, he led them to the path to become children of God. There origin was not the blood nor carnal will. Earlier, they were lost, now they are reborn from God. In the books it is found that the word has become a man, it came in this world willingly, and now it lives among us. We have seen his kingdom and his goodness.’ b. So er thaz suért thenita, \ ther éngil imo háreta, er híaz inan irwíntan; (II, 9, 51–52) ‘When he lifted the word, the angel hared to him, he commanded him to hold on.’ Especially the pronouns ih ‘I’ and wir ‘we’, which are normally weakly accentuated, can be found before or after the verb, without apparent difference in the informational function. Rather, the reasons are usually rhythmical. Within a theory according to which in sentences with an anaphoric background element, V-1 has to be chosen, such patterns should not be possible, as such an element should be located in the Wackernagel position. One factor that can be made responsible for the fronting of such elements is the fact that subjects in a neutral ordering have to precede all other constituents within the same information-structural level and that this could be a particular reason overriding the Wackernagel position. Apart from this, it has to be taken into account that writers in many cases have some freedom as to how treat given elements such as subject pronouns within the informational structure of an utterance. The context of an utterance does not force automatically the functional interpretation to be given a pronoun with respect to its informational status.
6.2.2. “Frame topics” There are cases of constituent fronting in Otfrid that are not predicted by Hinterhölzl and Donhauser (2003), however. In particular the fronting applies not only for “aboutness topics”, i.e. topics that have as referent the object of a statement and denotate the item an utterance is about, but other elements such as temporal, local or causal adverbs with a deictic function, that function as text connectors to the preceding text or as delimitations of the contextual frame of a statement, i.e. elements that could be classified as cases of “frame-setting elements” (Jacobs 2001):
Verb placement and information structure in the Gospel Harmony by Otfrid
(32) a. Fúar er thuruh Samáriam, \ zi einera burg er thar tho quám, in themo ágileize \ zi éinemo gisáze. Tho gisaz er múader, \ so wir gizáltun hiar nu ér, (II, 14, 5–7) ‘He wandered through Samaria. There he came to a borough in this toil, to a resting place. Then he sat down tired, as we told earlier.’ b. Úaptun thar thie líuti \ eino brútloufti themo wírte joh theru brúti \ in sáligeru zíti. […] Thar was Kríst guater \ joh sélba ouh thiu sin múater, (II, 8, 3–4, 7) ‘There was a wedding a happy time for the host and his bride. […] The good Christ was there as well as his mother.’ c. Bi thiu ílemes io gigáhon \ zi then drúhtines ginádon, (II, 3, 63) ‘Therefore we strive to go to the graces of the Lord.’ d. Sie íltun tho bi mánne \ fon theru búrg alle, íltun al bi gáhin, \ tház sie nan gisáhin Innan thés batun thár \ thie júngoron then méistar, tház er thar gisázi \ zi dágamuase inti ázi. (II, 14, 93–96) ‘All people of the village hurried then they hurried to see him. Meanwhile the disciples asked the master
to sit down and to eat the meal.’ 6.2.3. Focussed comments Moreover, not only topics, but also comments may be placed in the prefield, as long as there is no marked topic (new topic or topic change with heavy stress) in the clause simultaneously: (33) a. Druhtin Kríst sar zi imo sprah, \ so er nan érist gisah: “Symon bistu, muates línd, \ joh bistu ouh dúbun kind;” (II, 7, 35f.) ‘Christ, the Lord, told him, when he first saw him: “You are Simon, of peace-loving mind, and you are the son of a dove.” ’ b. Wil thu iz kléinor reken, \ in wíne gisméken: fon Kríste scalt thu iz zéllen, \ gisteist thu tház irwéllen. (II,9,69–70) ‘If you want to interpret it more subtly, taste it as wine, you have to relate it to Christ, if you want to do so.’
c. Tharana maht thu irthénken, \ mit brúnnen thih gidrénken, gifréwen ouh thie thíne \ mit géistlichemo wíne Ih zéllu thir in alawár: \ luzil dránk ih es thar. (II, 9, 23–25) ‘You might think about this and quench your thirst with wather and enjoy your people with spiritual wine. I tell you honestly: I drank little of it here.’ d. Sie sint úbil thrato \ wérko joh thero dáto; míhil ist ir úbili \ thuruh thaz hérza frávili. (II, 12, 89–90) ‘They are very evil in works and in deeds; their evilness is great through their sinful heart.’ e. Zéllu ih thir ouh hiar tház \ bi thiu stéininun fáz: hérza iz sint gidígano \ thero gotes drútthegano. (II, 9, 11–12) ‘I also give you an explanation of the stoneware vessels: they signify the pure harts of the disciples of god.’ Within this group, there are some special cases. One type concerns modal adverbs, often to verba dicendi: (34) a. Tho drúhtin themo mán luag, \ thes ih hiar óbana giwúag, óbaz theih hiar fóra quad, \ thaz er mo hárto firspráh: Hárto sageta er imo tház, \ thaz er mo bórgeti thiu baz; (II, 6, 3–5) ‘When the Lord prohibited to the man, as I told earlier, the fruit I mentioned before and he banned, he told him severely that he should refrain from it.’ b. Scono zált er imo tház \ (so drúhtin io giwón was). (II, 12, 51) ‘He explained it to him in good words, (as the Lord used to do).’ Semantically, the adverbs belong to the verb as a modal adverbial. As to the information structure, undoubtedly, the adverbial and the finite verb together make up part of the comment of the main clause. A second group concerns sentence-initial, prefield position of nonfinite verbal elements. At first sight, such cases seem to contradict the rules stated earlier that nonfinite verbal elements in main clauses follow the finite verbal element and are placed be in V-late position. The examples show that such nonfinite verbal elements can also be fronted like other constituents:
Verb placement and information structure in the Gospel Harmony by Otfrid
(35) a. Thó sprah Kríst zi imo sár: \ “giscríban ist in álawar, thaz mánnilih giwéreti, \ selb drúhtines ni kóroti.” (II, 4, 75) ‘Christ said to im then: „It is written truly, That everybody shall fight not to tempt the Lord.” ’ b. Giléitit ward tho druhtin Kríst \ thar ein éinoti ist, (II, 4, 1) ‘Christ was lead into a desert.’ c. Spúan er io zi nóti \ jénan zi úbarmuati, zi gíri ioh zi rúame, \ zi suaremo ríchiduame. Níazan sah er inan tház, \ thaz ímo ju gisuás was; (II, 5, 7–9) ‘He lead him by force to arrogance, to greed and He saw him make use of that was agreeable to him.’ Such verbs undoubtedly are part of comment, too, often together with an end-placed subordinate clause: The verb is not a concept introduced earlier in the text, it is not in contrast to another predicative expression that possibly could have the function of a topic, and they are nevertheless in a position with heavy accent. The motivations for preposing of a comment element instead of leaving it in the usual comment position at the end of the middlefield are often difficult to determine. One factor seems to be that preposing a comment element helps solving other ordering problems in a sentence. In the verses cited in (34), in the usual word order, the adverb as comment would have to be located at the end of the clause, which could make it appear as the sole comment element of the clause. In other cases, preposing an comment element makes it possible to keep another element at the end of the main clause that otherwise would land in a position considered as infelicitous. Often a pronoun belonging to the background part of the utterance would have to be placed in the first – topic – positioning front of the finite verb. If in (35c) the infinitive would be in its grammatically normal final position there could arise rhythmical problems for the demonstrative pronoun “thaz”. Furthermore, the placement of a heavily accented infinite verb usually facilitates an appropriate versification of a sentence; these verbs as a rule occupy the first accent position of the verse. Another possible factor is special emphasis. Such special emphasis can be seen in II, 9, 26 (ex. (33c)): (33) c. Tharana maht thu irthénken, \ mit brúnnen thih gidrénken, gifréwen ouh thie thíne \ mit géistlichemo wíne Ih zéllu thir in alawár: \ luzil dránk ih es thar, (II, 9, 23–25)
‘You might think about this and quench your thirst with wather and enjoy your people with spiritual wine. I tell you honestly: I drank little of it here.’ An element preposed that according to the usual rules is expected in the the end of a sentence is much more marked when it is fronted.10 Both the case of preposed modal adverbs and the case of nonfinite verbal elements raise the additional question why preposing or fronting is necessary or possible for these elements when they form a common comment together with the finite verb in the V-2-positon. Structurally, they should form a single constituent whose internal ordering should not be affected by information-structural factors. The explanation lies in the assumption that in fact, the finite verb as a surface constituent forms a constituent of its own, structurally separated from the other parts of the other elements of the verbal group. Whether the finite verb is placed the V1position or in the V2-position, this position is grammatically fixed and cannot be changed by information-structural factors. In contrast to this, the other elements of the verbal group can be placed independently according to information-structural needs, i.e. they can be placed in front of the clause-initial finite verb. Independently of the grammatical explanation of these facts, they in itself are one more indication that the usual German field-and-frame structure is valid for Otfrid as well.
6.3. Discourse markers Another apparently problematic type of ordering is found in adverbials that represent illocutive specifications of a speech act. They can appear in the prefield position as well, without having a particular contrastive value either as a topic or as a focus: (36) a. giwisso zéllu ih thir nú: \ finfi hábotost thu jú. (II, 14, 52) ‘I tell you for certain, that you had five (husbands).’ b. Thísu selba rédina \ theih zálta nu hiar óbana, bréitit siu sih hárto \ géistlichero wórto; Thoh wíll ih es mit wíllen \ hiar étheswaz irzéllen. (II, 9, 1–3) ‘Although this events that I telled earlier about are open to many different spiritual interpretations nevertheless I will explain some of them.’
Verb placement and information structure in the Gospel Harmony by Otfrid
From a grammatical point of view, elements as giwisso in (36a) are adverbials to the verb. Thoh in (36b) on the other hand is a concessive particle. In the usual word order, in the verses cited in (36), these elements would normally have to be located at the end of the clause. This could make them appear as the sole comment element of the clause. This could motivate preposing the element the point of view of the information structure of such sentences, the adverbial and the finite verb form the comment of the main clause. Semantically and pragmatically, however, these elements are not adverbials in the strict sense, but are discourse markers, expressing a perlocutive attitude of the speaker or the perlocutive connection of an utterance to the context. As such, these expressions are not part of the propositional content of the utterance and cannot be integrated into its informational structure. It is quite usual that such elements are placed in front of a sentence. Thus, such sentence patterns do not contradict the earlier findings; rather, they show that besides the information-structural factors, there are more factors influencing the order of elements in a sentence
6.4. Conclusions In section 6.2., it was attempted to explain the choice between V-1- and V2-structures along the pragmatic lines proposed by Hinterhölzl and Donhauser (2003). It was observed that their theory can be corroborated partially; but there are more constituents that may be in prefield position than predicted by their hypothesis, namely constituents that are not topics in a strict sense, but rather frame delimitations or textual connectors on the one side and comment elements on the other hand. This seems to be a rather heterogeneous collection of elements, which makes it difficult to derive a simple information-structural rule for explaining the choice between V-1and V-2-structure. Looking from a different angle, however, a rather simple principle guiding the distribution can be formulated. Obviously, there are different factors that may induce the fronting of an element: Elements may be fronted if they are topics, if they serve as textual connectors, if they are discourse markers, if they are given special emphasis as comment, or in some special cases, to circumvent ordering problems that otherwise might impair a correct pragmatic interpretation. The particular factor inducing the fronting or the particular pragmatic function of a constituent has no particular relevance for the position in the prefield, however; any element that takes the first position among the non-verbal constituents in a particular utterance for particular pragmatic or textual reasons is fronted left to the
finite verb, except when it is a weakly accentuated element to be located in the Wackernagel position. The place in front of the verb thus does not have a specific quality marking the element taking that place as having a particular pragmatic function in the sentence. This principle has some resemblance to the V-2-rule in present day Standard German. The main difference is that unlike in Standard Present Day German, the prefield is not to be filled obligatorily, but only when there exists an element that is marked pragmatically or stylistically. Looking at the ordering rules from the point of view of the verb, we may conclude that in the case where there is no particular reason for any element to be fronted, the position of the finite verb is the leftmost position in the sentence and that this is the basic, i.e. the unmarked case. V-1 thus appears to be the basic order. These considerations seem to contradict the findings in the first subsection, namely that there must be a V-2-rule for grammatical reasons. In the light of the previous analyses, these findings can be integrated in the following way: Grammatically definable elements such as question words or grammaticalised elements such as resumptive pronouns and particles form a separate group of preposing, i.e. preposing for strictly grammatical reasons, as opposed to preposing for information-structural reasons. A common restriction in this system seems to be that exactly one constituent can be fronted for grammatical reasons. For the fronting rule for pragmatical reasons, the system is more variable, as will be shown in section 7. One problem remains, namely the rule that yes-no-questions are marked by V-1-position. If this is a grammatical rule, the question arises how this grammatical rule is to match with a basic V-1-structure for the “unmarked” case. Superficially, simple V-1-structures in assertive sentences and in questions cannot be distinguished. There are two possible answers to this question: Either the V-1-order represents the same basic order as that for assertions; questions just do not have a topic usually. Or there may be a structural difference between assertions and yes-no-questions: For questions there is no place open in front of the verb to be filled optionally. I must leave this question open to further discussion.
7. Verb-late/verb-last position of the finite verb in main clauses Besides Verb-first and Verb-second structures, Otfrid uses Verb-late or Verb-last position for the finite verb in main clauses as well. The pattern is by no means rare and not exceptional in Otfrid’s work.
Verb placement and information structure in the Gospel Harmony by Otfrid
(37) a. Sin wórt iz al gimeinta, \ sus mánagfalto déilta al io in thésa wisun \ thuruh sinan éinegan sun (II, 1, 33f.) ‘His word decided it all, divided it up in diversity, all in this manner through his only son.’ b. So er nan zi ímo brahta, \ Kríst inan irknáta, (II, 7, 53) ‘When he brougth him to him, Chris recognized him.’ c. Druhtin Kríst sar zi imo sprah, \ so er nan érist gisah: “Symon bistu, muates línd, \ joh bistu ouh dúbun kind”(II, 7, 35f.) ‘Lord Christ said to him, when he saw him the first time: “You are Simon, of a gentle mind, and you are the son of a dove.” ’ d. “Séhet”, quad er, “hérasun, \ war geit ther drúhtines sun;[…]” Thiu wórt sie sar intfíangun \ joh after ímo giangun (II, 7, 11, 15) ‘ “Behold”, he said, “where God’s son is going […]” They accepted his words at once and followed him.’ e. María thaz bihúgita, \ joh Kríste si iz giságeta. (II, 8, 12) ‘Maria noticed it and told it Christ.’ f. Ein scaf er stántan gisah, \ thaz was zem ópphere gimah(II, 9, 59) ‘He saw a sheep standing there that was suitable for a sacrifice’ g. Zít ward tho giréisot, \ thaz er gíangi furi gót; ópphoron er scólta \ bi thie síno súnta, (I, 4, 11–12) ‘The time had arrived that he should appear with got He should make a sacrifice for his sins.’ The question is how this pattern is to be interpreted within Otfrid’s system so far described. One can think of the following two possibilities: –
“Verb-last-pattern”: We may hypothesize that Otfrid uses the grammatical verb-last pattern similar to that of asyndetic subordinate clauses. This would amount to the hypothesis that Otfrid stretches or even violates the grammatical system at disposition using a pattern reserved for subordinate clauses in main clauses, simply in order to fulfil the necessities of verse structure. “Expanded fronting”: A second possible description consists in the assumption that the Verb-late pattern in main clauses arises from a more general application of the fronting rule of the system described in the previous chapter, together with a rule placing weak elements in a Wackernagel position after the first element. The hypothesis would implicate that in fact such V-late-patterns are the result of the same or at least similar rules as that derived from V-1/V2-structures.
Certain aspects speak in favour of the second approach. The overwhelming number of main clause V-late structures show a specific well-defined distribution of constituents: The first constituent is an element with heavy accent; this element is followed by one or more weak constituents, such as anaphoric pronominal constituents (“continued topic” or background elements). The elements with a heavy accent are fronted under the same conditions as that discussed with V-2-patterns: They are heavily stressed topics (contrastive topics or new topics), as in (38a) – (38e), or comments, as in (38f) and (38g): (38) a. Sin wórt iz al gimeinta, \ sus mánagfalto déilta al io in thésa wisun \ thuruh sinan éinegan sun (II, 1, 33f.) ‘His word decided it all, divided it up in diversity, all in this manner through his only son.’ b. So er nan zi ímo brahta, \ Kríst inan irknáta, (II, 7, 53.) ‘When he brougth him to him, Chris recognized him.’ c. Druhtin Kríst sar zi imo sprah, \ so er nan érist gisah: “Symon bistu, muates línd, \ joh bistu ouh dúbun kind;” (II, 7, 35f.) ‘Lord Christ said to him, when he saw him the first time: “You are Simon, of a gentle mind, and you are the son of a dove.” ’ d. “Séhet”, quad er, “hérasun, \ war geit ther drúhtines sun […]” Thiu wórt sie sar intfíangun \ joh after ímo giangun;(II, 7, 11,15) ‘ “Behold”, he said, “where God’s son is going […]” They accepted his word at once and followed him.‘ e. María thaz bihúgita, \ joh Kríste si iz giságeta. (II, 8, 12) ‘Maria noticed it and told it Christ.’ f. Ein scaf er stántan gisah, \ thaz was zem ópphere gimah (II,9,59) ‘He saw a sheep standing there that was suitable for a sacrifice’ g. Zít ward tho giréisot, \ thaz er gíangi furi gót; ópphoron er scólta \ bi thie síno súnta, (I, 4, 11–12.) ‘The time had arrived that he should appear with got He should make a sacrifice for his sins.’ The weak constituents, on the other hand, are placed in enclitic attachment to this heavy constituent in front position. If the underlying structure followed the verb-final clause structure the ordering among weak and heavy elements should be the reverse one. This position of weak elements enclitically after the first heavy element of the sentence can be seen as the typical Wackernagel position in the original Indo-European version.
Verb placement and information structure in the Gospel Harmony by Otfrid
Such a sequence of constituents thus can be seen as a combination of fronting heavily stressed topic or comment constituents under the same conditions as described earlier and the enclitic attachment of weakly accentuated constituents directly to this first constituent, instead of being attached to the right of the verb. The difference between V-2-structures, as described above, and V-late-structures consists in the fact that for V-late-structures, we have a different version of the Wackernagel position and the absence of the restriction of the fronting to a single constituent.11 As this concerns only fronting for pragmatical reasons and not fronting for grammatical reasons, this could be an argument in favor of the hypothesis that fronting for grammatical reasons is a rule different from that of fronting for pragmatical reasons. The question remains, why the V-2-structures are so more frequent than V-late-structures. For one thing, the information-structural value of the verb might have an influence. One important question is, however, how in Otfrid’s system the Wackernagel position is to be defined. In the present account, we have to assume two different versions. Depending on which version of the Wackernagel rule is applied, the result is either a V-2structure or a V-late-structure. The choice of the particular version of the Wackernagel rule may depend on extragrammatical factors, such as rhythm, rhyming necessities etc. But eventually, we can speculate that we are confronted with two versions of the constituent defining the Wackernagel position belonging to two different diachronic periods. The older one, of course, is the definition of the Wackernagel position by the first constituent with heavy accent. In the newer one, the Wackernagel position is defined by the verb in V-2-positon. In this view, Otfrid is using two patterns from two diachronically different systems, a phenomenon that is typical for intermediate stages in language change. We can speculate with some plaubility that the older stage has been carried on in poetic works as the new type was developed. Thus, the V-late pattern in main clauses can be interpreted as typical for poetic language. Apart from the specific properties of V-late-patterns mentioned, one more argument in favour of the second approach of a description can be found in the fact that V-late ordering appears also in other early sources of OHG, such as the Isidor-translation Contra Iudaeos.12 In sentences as the following ones the sequence of elements and their informational characteristics is similar to the patterns described above – a heavy constituent as a first element is followed by a weak pronominal constituent: (39) a. erino portun ih firchnussu. (Isidor 157) ‘I shatter iron doors.’
b. Dhes martyrunga endi dodh uuir findemes mit urchundin dhes heilegin chiscribes. (Isidor 516) ‘We demonstrate his martyrdom and his death with evidence from the Holy Writings.’ Sentences as these have been the object of discussions on several occasions, such as in Lenerz (1984), Tomaselli (1995) and Ramers (2005), with varying grammatical interpretation, which take directions comparable to the proposals put forth above, although the relatively small number of examples makes it even more difficult to draw general conclusions from the facts. Independently from the grammatical interpretation, such sentences seem to show that the possibility of V-late structures existed in early Old High German outside of Otfrid’s work. Furthermore, Tomaselli (1995) points to the fact that in Old English similar structures can be found even more frequently. Thus it is not unfounded to assume that Otfrid did not invent such a word order rule by forcing grammar but that he used a traditionally given pattern. It has to be added, though, that V-late structures can be found that do not fit into this explanation and apparently can only be explained as strict verb-final patterns. (40) a. So was er io mit ímo sar, \ mit imo wóraht er iz thar; so wás ses io gidátun, \ sie iz allaz sáman rietun. (II, 1, 15f.) ‘He thus was with him all the time, he did it together with him; whatever they did, they planned it together.’ b. Stuant Johánnes gomono éin \ mit sinen júngoron zuein, gisáh er gangan thárasun \ then selben drúhtines sun; […] Er fíngar sinan thénita, \ then júngoron sar tho zélita, joh sár in tho giságeta \ thia sálida in thar gáganta. (II, 7, 5–6,9–10) ‘John, one of the men, stood there with two of his disciples, he saw passing there god’s own son. […] He (= John!) pointed with his finger, he told his disciples and explained them at once that salvation was approaching them.’ In sentences as these, the weak pronoun is in the first position and a heavy constituent is following it.13 This cannot be explained by the factors developed above. These sentences show the characteristics of strict verb-final structures, used e.g. in asyndetic subordinate clauses. Such patterns are rare exceptions, though, and they cannot put to question the explication given for the more frequent cases.
Verb placement and information structure in the Gospel Harmony by Otfrid
8. Conclusions We may summarize the results of the above observations and considerations as follows. Although the order of elements in Otfrid’s work, especially with regard to verb position, is more variable than e.g. in Notker, his use is by no means irregular; on the contrary, it is possible to find relatively clear patterns and principles: –
The position of verbal elements clearly follows grammatical rules: There are distinct grammatical differences between main clauses and subordinate clauses, especially with respect to the relative position of finite and nonfinite verbal elements. These rules indicate that Otfrid applies a system of “frame-and-field-structure” comparable to that of later variants of German. Moreover, for yes-noquestions and conditional clauses, the verb position is clearly defined as V-1. This implies that Otfrid has a system with grammatically defined verb positions. Several factors seem to blur the rules of “Rahmenbildung” and verb position: There are no absolute rules for the position of the verbal elements at the left side and at the right side of the clause. This variability can be attributed to pragmatic influences, especially information-structural factors. Pragmatic factors take a more important role in the ordering of elements than in later stages of German; in other words, grammatical rules are more restricted in their range, and ordering is subject to pragmatic factors more frequently. One major difference of Otfrid to later systems is the variability of the position of finite verb in the main clause. It can be placed in the first, the second and even in a late position. The choice can be attributed to specific pragmatic factors: − V2-structures and V-late arise from preposing elements under specific conditions. Most often, these factors are of information-structural nature, such as topicalization and focussing comment. Other elements that tend to be preposed are textual connectors and discourse markers. As a consequence, it is not possible to attribute a specific pragmatic value to the position in front of the finite verb, restricted only to informationstructure. Rather, there are several different factors that play a role in preposing, and this place is open to any element that is to be preposed for any reason whatsoever.
The V-1-structure in this view appears as a structure that is used in absence of such specific factors; in a certain sense, V1 thus appears to be the basic order. − V-late-structures may be captured by the same principles, applied in a more general way, if we assume a diachronically older definition of the Wackernagel position. This implies that Otfrid is applying ordering rules originating in different diachronic stages of Germanic languages. Looking at the Old High German Isidor translation, we notice that Otfrid is not an isolated case. In one way or other, V-1 and V-late patterns can be found in the Isidor translation as well, although the material is not abundant, which makes an interpretation for Isidor more difficult. Moreover, Robinson (1996) correlates some of the more marginal patterns to special conditions; e.g. he characterizes the V-1 pattern as a “distinctively foreign pattern, which [...] led the Isidor translator to use it as a flag to mark Biblical quotations [...]“ (Robinson 1996: 25f.) In the light of Otfrid, this statement could be revised to some extent Otfrid shows some preferences to V-2-structures in main clauses when fronting specific elements. This might show a tendency to restrict the number of elements in front of the verb to a single constituent, although this cannot be an absolute rule as in later stages of German. It may well be that with his combination of freedom of word order left to the finite verb and his preference for and partial grammaticalization of V-2-structures, Otfrid combines newer developments with the use of older pattern that are vanishing in his own period. This would be no unusual trait for a poetical text such as his Gospel Harmony.
There certainly remain many problems and apparent exceptions to be considered, and at a closer look, we can certainly find more descriptional problems. Nevertheless, I hope to have made plausible that, what concerns word order, Otfrid’s language is not as irregular as is often assumed.
Notes 1. 2.
“Die Späterstellungen bei Otfrid sind reimbedingt und damit keine aussagekräftigen Belege”. (Schrodt 2004: 196). As P. Piper notes in a commentary to O II, 4, 97: “Der Reim kann wohl den Dichter zu weniger gebräuchlichen Wendungen und Constructionen, zur Ver-
Verb placement and information structure in the Gospel Harmony by Otfrid
7. 8. 9. 10.
wendung von Flickwörtern u. dgl. verleiten, nie aber dazu, sprachlichen Unsinn zu schreiben.” (Otfrid 1882, 1: 161) Cf. Jacobs (2001). Jacobs lists varying semantic attributes of different kinds of “topics”. His list seems to be incomplete, though; moreover it remains open to questioning whether any expression that shows the linear and accentual properties of a topic on the surface structure level must be classified as a topic semantically. It is impossible, of course, to give here an overall outline of the theoretical questions in this domain. E.g., I leave out the discussion of different kinds of topics such as « aboutness-topics », « familiarity topic », « contrast topic » etc. (cf. Frascarelli/Hinterhölzl 2007), as these distinctions play no role for the following discussion. Other terms are explained in its context. In Lötscher 2006, this dimension is called “informational relevance”.We do not only discuss knowledge, but also hypotheses and other concepts in nonreal worlds, sometimes within ad hoc modalities. Cf. the notion ‘disputable’ in Büring (1997: 700). I assume that it is part of a communicative act of a speaker to put something specific to dispute in a communicative act (granted that it must be disputable for some reason). But things are not simply disputed because they are disputable; most assumptions are disputable, but we do not discuss them anyway. This dynamic aspect of topic treatment wthin a text gives place to the different types of topics mentioned in fn. 4. The topic-comment interpretation of adjectives and quantifiers is discussed in more detail in Büring (1997). The examples are usually taken from book II. The fronting of comment elements to the prefield has been a common possibility in all periods of the history of German, although it apparently has a not the same stylistic and rhetorical functions at all times. Tomaselli (1995) treats similar patterns in the Old High German translation of Isidor as V-3-patterns. Whether this is correct or not, could only be decided on a broader set of examples than are attested in Isidor. In any case, the situation in Otfrid shows that a purely numerical definition (“V-2”, “V-3” etc.) on the basis of a few examples is not sufficient for a general grammatical description. More examples are given in Schrodt (2004: 202f.) Actually, the pragmatic function of sie in II, 1, 16 in example (40a) is not altogether clear; we could assume as well that the pronoun has a special topic function within the context, which would make this example an instance in favour of the “fronting”-hypothesis.
References Text sources Isidor 1964
Der althochdeutsche Isidor. Edited by Hans Eggers. Tübingen: Niemeyer.
Otfrid 1882/87 Otfrids Evangelienbuch. Edited by Paul Piper. Vol. I: Einleitung und Text. Vol. II: Glossar und Abriss der Grammatik. Freiburg: J.C.B. Mohr. Otfrid von Weissenburg 1973 Otfrids Evangelienbuch. Edited by Oskar Erdmann. 6th edition by Ludwig Wolff. (Altdeutsche Textbibliothek Nr. 49, Ed. 6). Tübingen: Niemeyer.
Linguistic work Büring, Daniel 1997 The Meaning of Topic and Focus. The 59th Street Bridge Accent. London and New York: Routledge. Donhauser, Karin and Roland Hinterhölzl 2003 Die Rolle der Informationsstruktur bei der Herausbildung von Wortstellungsregularitäten im Germanischen. In Informationsstruktur. Die sprachlichen Mittel der Gliederung von Äußerung, Satz und Text. Finanzierungsantrag 2003-2007, ed. Caroline Féry, Potsdam und Berlin: Universität Potsdam. Frascarelli, Mara, and Roland Hinterhölzl 2007 Types of Topics in German and Italian. In Information Structure and the Architecture of Grammar: A Typological Perspective, eds. Susanne Winkler and Kerstin Schwabe, 87–116. Amsterdam/Philadelphia: John Benjamins. Hinterhölzl, Roland, Svetlana Petrova and Michael Solf 2005 Diskurspragmatische Faktoren für Topikalität und Verbstellung in der althochdeutschen Tatianübersetzung (9. Jh.). In Approaches and Findings in Oral, Written and Gestural Language, eds. Shinichiro Ishihara, Michaela Schmitz and Anne Schwarz, 143–182. (Interdisciplinary Studies on Information Structure 3.) Potsdam: Universitätsverlag Potsdam. Jacobs, Joachim 2001 The dimensions of topic-comment. Linguistics 39: 641–681. Lenerz, Jürgen 1984 Syntaktischer Wandel und Grammatiktheorie. Tübingen: Niemeyer.
Verb placement and information structure in the Gospel Harmony by Otfrid
Lötscher, Andreas 2005 Linksperiphere Adverbialsätze in der Geschichte des Deutschen. Beiträge zur Geschichte der deutschen Sprache und Literatur 127: 347–376. 2006 Die Formen der Sprache und die Prozesses des Verstehens. In Text – Verstehen. Grammatik und darüber hinaus, eds. Hardarik Blühdorn, Eva Breindl and Ulrich H. Waßner, 19–45, Berlin: de Gruyter. Näf, Anton 1979 Die Wortstellung in Notkers Consolatio. Berlin/New York: Mouton de Gruyter. Nemitz, Werner 1962 Zur Erklärung der sprachlichen Vertöße Otfrids von Weißenburg. Beiträge zur Geschichte der deutschen Sprache und Literatur (T) 84: 358–432. Petrova, Sevtlana and Michael Solf this vol. On the methods of the information-structural analysis of historical texts. A case study on Old High German. In New Approaches to Word Order Variation and Change in the Germanic Languages, eds. Roland Hinterhölzl and Svetlana Petrova. Berlin: Mouton de Gruyer. Ramers, Karl Heinz 2005 Verbstellung im Althochdeutschen. Zeitschrift für Germanistische Linguistik 33: 78–91. Robinson, Orrin W 1997 Clause subordination and verb placement in the old high German “Isidor” translation. Heidelberg: Universitätsverlag C. Winter, cop. Schrodt, Richard 2004 Althochdeutsche Grammatik. Tübingen: Max Niemeyer. Sperber, Dan and Deirdre Wilson 1986 Relevance: communication and cognition. Oxford: Blackwell. Tomaselli, Alessandra 1995 Cases of Verb Third in Old High German. In Clause Structure and Language Change, eds. Adrian Battye and Ian Roberts, 345–369. Oxford: Oxford University Press. Wunder, Dieter 1965 Der Nebensatz bei Otfrid. Heidelberg: Carl Winter. Zifonun, Gisela (et al.) 1997 Grammatik der deutschen Sprache. Vol. 3. Berlin/New York: de Gruyter.
Translating information structure: A study of Notker’s translation of Boethius’s Latin De Consolatione Philosophiae into Old High German Rosemarie Lühr
Abstract In this paper we are studying information structure on the basis of Notker’s Old High German translation of Boethius’s Latin De Consolatione Philosophiae. Our questions are: How and to which extent is the information structure of Latin converted into Old High German? What insights into the information structure of Old High German do we gain from this? To answer these questions we have to describe specific semantic characteristics of the Latin and of the Old High German language and to compare Classical Latin with Late Classical Latin. Furthermore we must discuss whether there are any specific characteristics of the information structure in Notker’s translation of rhetorically marked word structures which are particularly prominent in his poetic source. It will be shown that Notker did not convert the hyperbaton into Old High German, but, where appropriate, he used other figures of speech or rhetorical word order, especially in exclamative sentences and causal sentences introduced with wanda, which are mostly in explanatory parts of the text. Concerning focus and topic, all the linguistic means which mark these information structural entities are investigated: focus particles, emphatic pronouns, word order, contrastive structures, which resemble I-topicalisation in modern German vs. initial position, anaphoric and demonstrative pronouns, the dissolution of the relative connection of sentences, a construction which does not exist in Old High German. In sum, it will be shown that Notker’s representation of the information structure of the Latin original is first of all contingent on his didactic purposes. In addition Notker’s handling of the information structure demonstrates a fundamental difference between the two languages compared here with regard to the positioning of the kinds of foci: Old High German: structural focus – verb – emphatic focus vs. Latin: verb – emphatic focus – structural focus.
324 Rosemarie Lühr 1. Introduction Translations are a possible means of coping with the methodological problems arising from the study of the syntax of historical linguistic corpora. For the lack of (native) speaker competence can be compensated in parts by comparing the source language and the target language, cf. Petrova and Solf (in this volume). In the historical stages of German the relevant contrast is the one between Old High German and Latin as it is reflected in the language of the translators. One Old High German author who explicitly comments on this is Notker Labeo (Notker III of St Gall; Notker the German). In her groundbreaking study of Notker’s translation of Martianus Capella Glauch therefore rightly quotes Notker’s letter to the Bishop of Sitten: “Da die Klosterschüler ohne das Vorstudium gewisser Disziplinen die kirchlichen Bücher nicht vollständig verstehen könnten und er wünsche, dass sie Zugang zu diesen Büchern hätten, wage er es, lateinische Texte in unsere Sprache zu übertragen und das syllogistisch, figürlich und rhetorisch Ausgedrückte mittels Aristoteles, Cicero oder eines anderen artes-Schriftstellers zu erhellen.“ [As the pupils of the convent school could not fully understand the ecclesiastical books without having previously studied certain subjects and as he wished them to have access to these books, he was taking the liberty to transfer Latin texts into our language and to use Aristotle, Cicero or other writers of the artes poeticae to clarify what has been expressed by syllogisms, figures of speech and by rhetoric.] (2000: 29). And indeed, “Notker is unique in his time with his pedagogically motivated work as a translator and with his attempt to use German prose for educational and academic purposes”. Strictly speaking, Notker’s individual achievement lies in the “Verbindung von Texterklärung und Text überhaupt mit Übersetzung” [in combining explanations of texts or texts in general with translation] (Schröbler 1953: 153). It is generally assumed that in his translations from the Latin, Notker “always takes the original as his starting point and (...) never really departs from its contents or structure” (Glauch 2000: 60). The question is, however, whether this holds true for all levels of language. Of particular interest in this context is the information structure as one “organizational level”, because we can assume a priori that a good translator always tries to imitate the informational structure of the original. To answer this question, we need to contrast the focusbackground and the topic-comment structure of the Latin and of the Old High German text. For although we can assume that the categories of information structure are universal concepts, this, however, does not apply to
Translating information structure 325
the ways different languages realize them. Therefore the specific lexical, morphosyntactic and topological strategies for marking categories of information structure of specific languages are particularly interesting for us. This problem becomes particularly pertinent if there is no fixed syntactic position, for example for marking focus: Because if we have no focus in situ (Hetland and Molnár 2001: 621; cf. also Jacobs 1986: 123), focussing one of the constituents means that we have to adapt syntax structures accordingly. Therefore the first object of our research is: How and to which extent is the information structure of Latin converted into Old High German? From this the second question arises: What insights into the information structure of Old High German do we gain from this? In order to learn more about this, we will begin with describing specific semantic characteristics of the Latin and of the Old High German language. In this context we will also compare Classical Latin with Late Classical Latin. We will also have to discuss whether there are any specific characteristics of the information structure in Notker’s translation of rhetorically marked word structures which are particularly prominent in his poetic source1. In Latin, the hyperbaton is a particularly important linguistic device expressing information structure. As the hyperbaton has a potential for distinct effects of the information structure, which may have inspired imitation even in nonclassical languages, Notker may also have used these kinds of structures for representing the information structure of Latin in Old High German. Moreover, we have to examine both languages as to whether there are syntactic differences of position with regard to realizing various kinds of focus. Linguistic research in general distinguishes between a structural focus in the innermost embedded phrase of a sentence on the one hand, and a different form of narrow or emphatic focus on the other hand (Abraham 1986 and 1992; Doherty 2002: 30–43). We have evidence that in Latin both the structural focus and the emphatic focus are usually serialized on the right hand side of the verb, whereas in Old High German one focus is placed on the left hand side and the other one on the right hand side. In special cases, in addition, focus in Latin seems to be positioned in the left periphery. We also have to keep in mind the structuring of given and inferable information2 – aspects such as continuous, shifting3, familiar4, aboutness5, discourse6 and contrastive topics7 – as Notker may have used these topics in various forms and degrees. In general, however, we can say that due to Notker’s linguistic creativity each act of translation as such is a unique act. For example, for pedagogical reasons Notker frequently changed the syn-
326 Rosemarie Lühr tactic structure of his source when translating it into Old High German8. To make sentences more comprehensible, he often added short subordinate clauses – especially causal clauses – which can also be interpreted as part of the information structure. But we need to look at the translations in their entirety in order to get a more general picture about Notker’s syntax. We choose Boethius’s De consolatione Philosophiae as our basic text, for as a platonic dialogue and a disputation, this text uses the devices of rhetoric and dialectics in almost every part of the text9. The text was written in Late Classical Roman Prose, but it also contains poems, so it is a prosimetrum. Boethius wrote the original manuscript shortly before his execution in AD 524 or 525. In the Early Middle Ages, the text was hidden for three centuries; from the second half of the 9th century onwards it was used as a textbook in schools. Today there are still 500 manuscript copies (Gruber 1978: 13 ff.). It has been shown that Notker edited the manuscripts available in the Abbey of St. Gall in connection with the commentaries prevalent at the time (Tax 1986: XIXff.) and that he used the Alemannic variety of Old High German10. As a rule, the basis for Notker’s translations was the specific, slightly modified version of the Latin text which he prefixed to his translations, whereas commentary manuscripts were only used occasionally (Glauch 2000: 145). Therefore we can compare the word order of the individual Latin text with that of the immediately following Old High German translation.
2. Particularities of the Latin and the Old High German syntax pertinent to the information structure 2.1. Classical Latin vs. Late Classical Latin 2.1.1. Classical Latin We will compare those parts of the syntax of Classical Latin which are relevant for the information structural arrangement to contemporary German in order to highlight the differences. These will be grammatical vs. rhetorical word order and the hyperbaton11. First, let us look at word order in Classical Latin. As is generally known, the usual or grammatical or traditional word order is as follows: Das Subjekt eröffnet den Satz, das Prädikat beendet ihn, die Objektsund Adverbialbestimmungen werden zwischen diese beiden Satzteile
Translating information structure 327
eingeschlossen, und zwar so, dass sie um so näher am Subjekt bzw. Prädikat stehen, je enger sie zu einem von beidem gehören (sog. SOP-Stellung). (...) Diese gewöhnliche Wortstellung ist nur selten streng eingehalten. Häufig richtet sich die Stellung der einzelnen Satzteile nach der Betonung, der Menge an neuer Information, der Deutlichkeit, dem Wohlklang, der Abwechslung oder nach anderen stilistischen Gesichtspunkten. So entsteht die sog. rhetorische (okkasionelle oder invertierte) Wortstellung, die von den Absichten des Sprechers bestimmt ist. Die für den Gedanken wichtigsten Satzteile nehmen häufig die am stärksten betonten Stellen im Satz ein, meistens Satzanfang und Satzende. [The sentence begins with the subject and ends with the predicate; object and adverbial clauses are inserted between these two parts in a way which moves them closer to the subject or to the predicate respectively, depending on which of these two they are more closely referring to (so-called SOP-position) (...) This common rule for word order is rarely followed. Often the position of the individual parts of the sentence is determined by emphasis, the amount of new information, clarity, melodiousness, variation or by other stylistic aspects. Thus we arrive at the so-called rhetorical (inverted or okkasionell [occasional]) word order which is determined by the intentions of the reader. The parts of the sentence which are most relevant to the ideas frequently take the most emphatic positions in the sentence, which is mostly at the beginning or the end.] (Menge 2000: 575) If you look at these word order patterns from the point of view of information structure, you have to examine whether Classical Latin can be described as “discourse configurational”12. In this case, the word order is not determined by the grammatical relation between the respective constituents of the sentence but serves special discourse needs. The topic position, for example, can be occupied by subjects, objects, indirect objects, etc., if these constituents function as the topic on the information structural level. The syntactic structure may be the following: (1)
X pragmatically unmarked
With respect to topics, modern German, for example, is said to be “discourse configurational”, for it is assumed that modern German has a special topic position in the middle field (Frey 2000). Second, a hyperbaton is a figure of speech in which words that syntactically belong together, such as noun and attribute, verb and adverb, etc., are separated from each other for emphasis. This kind of unnatural or
328 Rosemarie Lühr rhetorical separation is possible to a much greater degree in highly inflected languages, where sentence meaning does not depend closely on word order. In Latin and Ancient Greek, the effect of hyperbaton is usually to emphasize the first word. It has been called “perhaps the most distinctively alien feature of Latin word order”13. In Classical Latin, the hyperbaton is considered as grammatical, if that kind of separation is possible in an elevated prose style: “Man kann sogar von einer Tendenz in der gehobenen Prosa (und erst recht in der Poesie) sprechen, syntaktisch zusammengehörende Wörter zu trennen, falls dies zwanglos möglich ist, ohne den Eindruck der Künstlichkeit hervorzurufen” [One can even speak of a propensity of elevated prose (and even more so of poetry) to separate words if this is informally possible without creating an impression of artificiality] (Menge 2000: 581). The insertion of enclitics such as pronouns is considered to be a point in case, but also the use of conjunctions such as autem, enim, igitur, quoque, -ne in the second position in the sentence, all of which are actually instances of Wackernagel particles. But in principle, any kind of word can be placed between words which syntactically belong together: (2)
Cicero, fam. 3,9,3 Tuis incredibiliter studiis your-Abl incredibly-Adv studies-Abl ‘I’m made incredibly happy by your studies’
delector I’m made happy
If that construction is translated directly into English we would expect sentences such as (3a,b) which violate Ross’s Left Branch Condition. In focus constructions in English, the right branch of a noun phrase cannot remain in situ. Note that (3c), which is grammatical, is not a split construction in which raw is a modifying adjective of the noun oysters but a case of secondary predication: (3)
a. *Which has he invited friend to dinner? b. *The RED he bought car last week, the BLUE he has had car for years c. Raw he used to eat oysters. (Devine and Stephens 2000: 4f.)
Apart from the SVO-order, (3a) and (3b) are also ungrammatical in contemporary German. It follows that the hyperbaton in form of the Latin example (2) is not a possible syntactic structure of New High German14.
Translating information structure 329
2.1.2. Late Classical Latin So far, there is no comprehensive description of Late Classical Latin which constitutes the beginning of Middle Latin (from 500). “Antike Syntax ist [in der Regel] eine moderne Ableitung aus antiken Autoren.” [So we can still say as a rule, ancient syntax is a modern derivation from ancient authors.] (Kindermann 1998: 42)15 The question now is, whether “discourse configurationality” actually applies in Late Classical Latin. If you go through what the handbooks tell us, you will find: In order to achieve a special effect, the end of the sentence is frequently – especially in the writings of poets – reserved for the main term which dissolves the tension or suspense which was built up before. Zwischenstellung [sandwiched position] of the verb, positioning the verb somewhere between other constituents, especially between subject and object, is said to be natural if there is a syntactic connection between the final word and the following sentence, as for example in Petronius. Apuleius is also said to frequently put the verb in the third position from the end of the sentence if it is followed by a noun with an adjective or a preposition (Hofmann and Szantyr 1965: 404). Therefore there seems to be no fixed position for topic and focus in Late Classical Latin. Regarding the hyperbaton the handbooks will tell you: Rhetorical schooling, which took parts of its rules from the poetry of the classical period, and the Klauseltechnik, the periodic style with various forms of ending periods, resulted in a frequently rather unnatural way of handling this kind of word order in post-classical prose. Petronius, for example, frequently uses this word order in the vulgar passages of his novel Satyrica. This is interpreted as a sign that the use of this order had become a general tendency16.
2.2. Old High German Syntax The position of the structural focus depends on the position of the verb. Näf (1979: 114) describes the following rule for the “personal form” of the verb, i.e. the finite verb, in declarative sentences in the translation of Boethius’s Consolatio philosophiae: (4)
Rule 1: In declaratives the personal form of the verb is in second position
This rule also applies in sentences opened by a conjunction or a connective such as unde, uuanda, noch, aber, nube, ioh, alde, sunder (Näf 1979: 125
330 Rosemarie Lühr ff). This is illustrated in (4)–(5) which display no possible pattern in modern German. (5)
I 6,18f. Et rigant and they moisten .i. i.e. Únde and
non not fúllent fill
ora elegi. i. miseri. faces-Acc mourner-Gen i.e. poor-Gen fictis false-Abl sie they
mîniv óugen. mit my-Acc eyes-Acc with
érnestlichên drânen honest-Dat tears-Dat
Lat. ‘And they moisten the face of the mourner, i.e. the poor, with true, i.e. not false tears’ OHG ‘And they fill my eyes with honest tears.’ (6)
I 9,9ff. Et abstulerant and they had dragged away quisque each of them Únde and stúc- /chen. pieces-Dat
particu-/ las pieces-Acc
poterat. could sie they
dîe îogelicher Rel-Acc each of them
besuérben dragg off
Lat. ‘And they had dragged away as many pieces as each of them could.’ OHG ‘And they went away with the (their) pieces which each of them could drag.’
Nonetheless, we can rightly claim for Notker’s language that, apart from a few relics, we have verb-second position in declaratives17. Näf (1979: 187f.) also puts up a rule for the verbal Satzklammer [sentence bracket]18: a structure where the finite and the infinite parts of a compound verb are separated from each other and placed at some distance within a sentence. Thus they form a kind of bracket for the other parts of the phrase: (7)
Rule 2: “Im Hauptsatz steht die Infinitform auf der dritten oder einer späteren Stelle, und zwar (...) vor oder nach substantivischem Subjekt und substantivischen kasuellen Ergänzungen (...) vor oder nach (pronominalen oder substantivischen) Präpositionalgruppen (...) vor oder nach Satzadjektiven und Adjektivadverbien.” [In the main clause the infinite form is in the third
Translating information structure 331 or in a later position, either before or after subject noun and case-determined noun complements (...) before or after a prepositional group (pronominal or nominal) (...) before or after sentence adjectives and adjective-adverbs.]
For the complete Satzklammer, however, we have to add examples of sentences with AcI-construction: For instance, in contrast to the unmarked word order of Classical Latin, in the following sentence from the Latin original the agent of an AcI-construction is in final position. The result is a focus on the right margin of the sentence. In (7) Notker does not adopt this word order but converts it into a different Old High German construction with unmarked word order, in this case with a verbal bracket: (8)
I 6,28f. Et and
léid hábet sorrow has
inesse suam approach his-Acc míh/ me
álten getân. old-Acc made
Lat. ‘And the pain made his (old) age approach.’ OHG ‘And the sorrow has made me old.’
We also have to note that sometimes exbraciation, i.e. the positioning of words outside the sentence bracket, may also be due to rhythmical reasons: (9)
II 100,14ff. Et relicta prestantia and left aside-Prt.Abl preference-Abl uirtutisque. virtue-Gen Únde and
postulatis you demand
conscienitĊ/ sermunculis. conscience-Gen idle talk-Abl
premia de alienis rewards-Acc of strange-Abl
nehéina/ uuára tûondo déro stíuri no-Acc attention paying-Prt-Adv Art-Dat leading-Dat
geuuízzedo. álde conscience-Gen or
uuéllent ir will/want you
déro túgede./ Art-Dat virtue-Dat
Lat. ‘And, after the preference of conscience and of virtue has been left disregarded, you demand rewards for strange idle talk.’ OHG ‘And although you did not pay attention to the precedence of conscience and of virtue, you want to be thanked for the words which came from others.’
332 Rosemarie Lühr Presumably for rhythmical reasons, the voluminous genitive object déro stíuri déro geuuízzedo. álde déro túgede as part of the focus here is positioned after the adverb of the present participle, whereas in nehéina uuára tûondo [not paying attention] the other part of the focus nehéina uuára again is placed on the left side, before the present participle. In contrast to the position of the one-word verb, Näf’s “either ... or”- rules do not point to a distinct change of position of complex predicates in comparison to early Old High German. But the following sentences clearly demonstrate the obvious differences to New High German: (10) a. Er hat seinen Großvater besucht He has his grand-father visited b. *Er hat besucht seinen Großvater He has visited his grand-father c. Er hat ihn besucht He has him visited d. *Er hat besucht ihn He has visited him Whereas in present-day German only a. and c. are acceptable, for Notker a., b. and c. are possible options. Näf does not give any further rules for the word order which could be relevant for the information structural arrangement of Latin and of Old High German. Therefore further particularities can only be demonstrated by working with a text, in our case a continuous passage from the consolatio, mainly Book I, Chapter 1.
3. Information structure in the Latin original and in the Old High German translation In analyzing the topic-comment and the focus-background structure we implicitly adopt the scalar representation of features attested to the informational status of discourse referents, cf. Petrova and Solf (in this volume). The following features are constitutive for topicality: givenness/accessibility, referentiality, definiteness21, early position in the sentence. We also agree on the differentiation of a new information focus and a contrastive focus. One kind of focus is also the I-Topicalisation (see below). Let us say in advance that we can take it for granted that Notker did not only command the rules for word order in Old High German, but of course also those for Latin. His knowledge of Latin was brilliant, as we can see from his Latin insertions, see (10). In his amendment, Notker retains the
Translating information structure 333
aboutness topic22 at the beginning of the sentence. This correlates to the linguistic usage in Latin (Menge 2000: 576). (11)
I 10,15f. Hunc uero innutritum eleaticis studiis. atque this-Acc scilicet brought up-Acc eleatic-Abl studies-Abl and achademicis academic-Abl
s. non pa-/tior mihi not I tolerate me
subtrahi. taken away-Inf.Pass
Lat. ‘to wit/scilicet, I do not tolerate that he who was brought up with eleatic and academic studies will be taken away from me.’
3.1. Hyperbaton In the hyperbaton of the Classical languages, the modifier can either be topicalized as focus and other words can be positioned between modifier and head or the topicalized head as topic is separated from the focal modifier which follows later. A further option is splitting a wide focus into a primary focus containing the head and, succeeding other words, a second focus – the modifier23. One example of a focal modifier is: (12)
trôlicho séhendíu. threatening-Adv looking
Lat. ‘and with sinister looks inflamed’
In Latin the past participle (passive) inflammata refers to philosophy which appears in the shape of a woman. This participle separates a preceding adjective from the word it refers to. Notker uses a syntactic structure with a familiar vernacular word order for this, an adverb preceding a present participle (active) “and threateningly looking”. In this case the adverb takes the position reserved for the structural focus.
334 Rosemarie Lühr The difference between these two structures shows the map of information structure onto syntactic structure: (12)
c. AP A
d. AP AdvP
Consider also the following case: (13)
I 10,4ff. HĊ sunt these are
enim quĊ namely who
necant / infructuosis kill infertile-Abl
affectuum . affections-Gen
ézisg seed-Acc mít with
den uuûocher Art-Acc harvest-Acc rationis reason-Gen dórnen thorn-Dat
fructibus rationis. fruit-Abl reason-Gen únde and
ertémfent. / stifle
The semantically corresponding translation of the Latin sentence is: Because it is them who stifle the fertile seed of reason with the infertile briars
Translating information structure 335
and thorns of the affections24.The literal translation, however, is: ‘To wit, they are those who kill with infertile thorns of affections the rich seed by fruits of reason.’ So “the rich seed by fruits of the reason” takes the place of “the seed of reason made fertile / rich by fruits”: The modifying adjective uberem (fertile, rich) is separated from the word it refers to – segetem (seed) – by the adjective complement “by fruits of reason” and receives special emphasis by appearing in front. The regular order would have been putting the whole adjective phrase at the end (Rubenbauer, Hofmann and Heine 1977: 327): segetem fructibus rationis uberem (Menge 1961: 536). At this point of his rendition Notker doesn’t go into such fine details and translates: “These are who stifle the harvest and the seed of reason with the thorns of desire.” Now, in Latin the phrase uberem segetem fructibus rationis, which contains the structural focus, stands on the right-hand side of the finite verb after the ablative phrase infructuosis spinis affectuum: In a broader sense, this is an adverb of manner and constitutes a set of alternatives25 which is why it has special emphasis. So we can say that in Latin we have emphatic focus here. In contrast, the corresponding phrase in Old High German mít tîen dórnen uuíllônnes appears immediately right of the finite verb – also an emphatic focus – while the phrase with structural focus uuûocher únde dén ézisg tero rationis is placed on the left. With regard to the position of foci in the languages compared here, this means that the order in Latin may be: verb – emphatic focus – structural focus whereas that in Old High German would be: structural focus – verb – emphatic focus. This distribution of the foci also applies to the subordinate clause: (14)
I 7,20ff. ... dum … Et when and
lacrimabilem signarem I would chronicle tearful-Acc
…Únz … Únde íh …when and I temo screíb mít wrote with Art-Dat
sús âmerlicha so wailful-Acc grífele. pen-Dat
querimoniam . complaint-Acc
Lat. ‘...when I ....and the tearful complaint chronicled with the help of a pen’ OHG ‘...when I ... so/such wailful complaint wrote with the pen’
A further example of a hyperbaton, this time with a possessive pronoun as modifier, is the following:
336 Rosemarie Lühr (15)
I 9,19ff. Et di-/ctantes and dictating Únde and
meis uerba my-Dat words-Acc
trâne récchende . tears-Acc producing
mít / with
fletibus. weeping-Dat íro uuórten. her-Gen words-Dat
Lat. ‘and dictating the words to my flood of tears’
First of all, we have to note again that in Latin the structural focus uerba again appears after the predicate while in OHG the accusative trâne is placed before the predicate. But as far as the hyperbaton is concerned, in Latin meis – the possessive pronoun in dative – which refers to the dative word fletibus (flood of tears) is positioned between the participle dictantes and its object uerba. Notker’s translation is different: “únde mír trâne récchende mít íro uuórten” [and me tears producing/bringing forth with her words“] with “mír” [me] as a pertinent or possessive dative before the object trâne [tears]. Admittedly, the separation of possessive pronoun and reference word has again been converted into a familiar vernacular structure of Old High German. However, Notker imitated the emphasis put on the Latin possessive pronoun in the hyperbaton by deciding on a pertinent dative rather than a possessive pronoun for its representation. But the “mít íro uuórten” outside the Satzklammer here is an emphatic focus which parallels the structural focus of the Latin uerba. So what we actually have here is a re-structuring of the different kinds of focus: The Latin structural focus uerba is turned into the emphatic focus mít íro uuórten which is put outside the Satzklammer, whereas the phrase meis ... fletibus, which had been emphatically focussed by the hyperbaton, now appears as structural focus trâne with pertinent dative. To resume: The construction of the hyperbaton cannot be converted directly into Old High German26. In Latin, it does not indicate a fixed focus position. But it shows that this language is not “discourse configurational”, as the hyperbaton is not only optional but also appears in various types. With regard to Old High German, the examples discussed in (11) to (14) clearly show the extent to which and the techniques by which Notker tries to convert information structural distinctions of Latin into Old High German. The examples show that it is highly probable that there are distinct differences between Latin and Old High German with regard to the order of the various kinds of focus: in Old High German: structural focus – verb – emphatic focus; in Latin: verb – emphatic focus – structural focus.
Translating information structure 337
3.2. Rhetorical word order When converting complex sentences, participle constructions, imperative sentences and interrogative sentences in poetic texts, Notker shows no particularities of information structure in comparison to that of prose texts. (By the way, here Middle Latin doesn’t show any differences to Classical Latin either). Therefore, we can disregard the difference between poetry and prose hereafter. But like causal clauses, exclamative sentences are part of the rhetorical word order patterns, which is why they are of particular interest with regard to the distribution of the information structural entities.
3.2.1. Exclamative sentences For a start, we find exclamative sentences in which Notker uses verb-final position in Old High German instead of the Latin verb-second position. Verb-final position also occurs in that kind of exclamative sentences in New High German which are consistent with the order in subordinate clauses. (16)
Poetry I 7,6f. Eheu . quam oh how Áh oh
surda deaf-Abl sêre .
aure auertitur ear-Abl he scorns uuîo how
ér / he
gehôret. listens to
Lat. ‘Oh, how he with deaf ear the poor scorns’ OHG ‘Oh alas, how bad(ly) he to the poor listens.’
While in Latin the structural focus, accusative miseros, is again placed on the right side of the verb, in Old High German die uuênegen is placed on the left27, like in a subordinate clause. On the other hand, we can assume that like in New High German the main accent of the focus is on the adjective adverb. But sometimes Notker also follows the word order of his source; in the following example this is the postposition of the subject as emphatic focus after the verb, i.e. VS-position : (17)
Poetry I 11,10f. Heu quam hebet mens . oh how is faint soul
precipiti into the depth
338 Rosemarie Lühr Áh oh
síh / itself
mûot . soul
mísse-hábet lets (be)disconcert(ed)
Lat. ‘Oh how faint is the soul fallen in the bottomless depth’ OHG ‘Oh how much lets itself be disconcerted man’s soul suddenly fallen into the abyss’
Compared to (16), Old High German in (17) uses exbraciation, positioning the subject after the verb group. There are cases with evidence for postposition of the subject (2.3.1.). But sometimes Notker complete re-structures exclamations: In (18), for example in Latin the focus of the sentence consists of the exclamation “felix mors hominum” [blessed death of man], while Notker uses a categorical sentence (Steube and Späth 2002: 238). He explicitly introduces a discourse referent which functions as topic of the statement in the following sentence: (18)
Poetry I 7,3ff. Felix mors blessed death dul-/cibus sweet-Dat Táz that
hominum . men-Gen
annis . years-Dat
quĊ Rel saepe often
nec not uocata called-Prt
se itself uenit comes
inserit inserts mestis sad-Dat
sâlig tôd . blessed death
lústsa- / mên pleasant-Dat
nechúmet . not comes
léit-sámên / painful-Dat
netuélet. not hesitates
Lat. ‘Blessed death of men which does not push itself into the sweet years and, if he is often called, comes to the sad ones.’ OHG ‘That is a blessed death which does not come in pleasant times and on painful times if you wish for it does not hesitate.’
Considering its form, the Old High German sentence could be an answer to the question “How do you define a blessed death?” In this, the copulative construction in Táz íst sâlig tôd displayst he information structural distribution: The finite verb marks the beginning of the focus domain in the sentence.
Translating information structure 339
Exclamatory sentences are more emphatic and easier to memorize than declarative sentences and so for pedagogical reasons, Notker sometimes also converts Latin declarative sentences into Old High German exclamatory sentences: sentence (19), for example, is an exclamation with postposition of the verb which represents a Latin declarative statement with VO-position: (19)
Poetry I 7,7f. Et sæua . and cruel
negat flentes refuses crying-Prt.Acc
Únde / uuîo and how
Lat. ‘And cruel [death] refuses to close the crying eyes.’ OHG ‘And how unwillingly he cruel(ly) closes her/their crying eyes.’ 28
Notker’s rephrasings also underline his didactically motivated endeavour to make the Latin text readily accessible and as comprehensible as a schoolbook. Due to their expressive character exclamatory sentences are particularly suited for that purpose.
3.2.2. Causal clauses Notker often added causal clauses as explanations. So, for example, in (20) and (21) he changes a simple Latin sentence into a complex one. Here the causal clause – wanda-sentences in the function of subordinate clauses29 – follows the matrix sentence. They substantiate statements made in the main clause, that is to say weil connects two propositions (Lühr 2007). From the point of view of information structure, that kind of causal clauses add focal elements to the rest of the background of the matrix sentence. They add an overall commentary to this sentence, but they have a topic-comment structure. In (20) and (21) this is a continuous-topic. In the following examples there is no correlate to deswegen in the matrix sentence: (20)
I 8,4ff. Nam nunc for now
sese ad herself to
340 Rosemarie Lühr Uuánda éina uuîla . for Art-Acc time-Acc
méze . measure-Dat
uuánda / si as she
Lat. ‘For now certainly she contracted herself to the common measure of men/mankind’. OHG ‘For a short time she contracted herself hither/down to our measure as she sometimes regards human things’. (21)
I 8,6ff. Nunc uero now but
uidebatur she seems
cacumine / top-Abl
uerticis. crown (of the head)-Gen
uuîla tûohta time-Acc seemed
den hímel rûoren. Art-Acc sky-Acc touch-Inf uuánda si as she
cĊlum . sky
astronomiam uuéiz. astronomy-Acc knows
Lat. ‘But now she seems the sky with the top of the crown of her head to touch’ OHG ‘The other time seemed she (to) me the sky with the crown of her head to touch as she astronomy knows’.
We find evidence of a different structure in the following sentence: (22)
I 10,25ff. At ego and I
cuius acies Rel-Gen eye
lacrimis . tears-Abl
quĊnam esset who was/would be auctoritatis . dignity-Gen
caliga- / rat . was dark
dinoscere recognize hĊc this
obstipui. I became silent
possim . I could-Subj.Pres
muli-/er tam imperiosĊ woman so great-Gen
Translating information structure 341 Áber ut
míh tô myself there
uuîb uâre woman as/would be
bechénnen . recognize-Inf
uuánda mír as me
nemáhta not could daz Art
dés . / uuér of-Gen who
geuuáltîgo powerful-Adv sia / she/her-Acc óuga eye
tímbereta . became dim
Lat. ‘And I, whose eyes dark of tears were dark with tears, so that I could not recognize who this woman of such great dignity was, became silent’. OHG ‘But I feared myself for the following reason: Who this woman was who acted so powerful(ly) couldn’t I recognize because my eye became dim full of tears.’
In Old High German the indirect interrogative clause uuér dáz uuîb uuâre sô geuuáltîgo uárentíu is placed before the matrix verb, whereas in Latin it appears after the matrix verb. And in place of the Latin relative clause cuius acies caligarat ‘whose eye was dark, immersed in tears’ after the ego at the head of the sentence, Notker uses a causal clause uuánda mír daz óuga tímbereta . fóllez trâno ‘because my eye became dim, full of tears’ at the end of the compound sentence. But this uuánda-sentence states the reason why Boethius didn’t recognize the woman. That is to say, the causal clause again refers to the propositional level of the matrix sentence and consists mainly of focal elements (only mír reverts to íh from the matrix sentence). The best New High German translation for the conjunction is “deswegen weil” because Notker uses the word dés in the cataphoric construction Áber íh erchám míh tô dés ‘But I became silent because of that’ which precedes the compound sentence. So Notker broke down this passage into its components and represented the train of thoughts step by step. Thus he arrives at a much more precise presentation of the logical structure, but at the same time at a different distribution of information structural units concerning the contents of the wanda-sentence: In Latin, the relative clause cuius acies caligarat ‘whose eye was dark’ serves to set the frame while the causal clause uuánda mír daz óuga tímbereta . fóllez trâno ‘because my eye became dim, full of tears’ in Old High German has focal function. From this pragmatic analysis it follows
342 Rosemarie Lühr that what we have here is a wide focus. This confirms data from language typology which suggest that in many languages that kind of focus is positioned on the right periphery of the sentence. Yet another case is the following: (23)
I 35, 20ff. Qui tum who then
credidit has trusted
sulcis . furrows-Dat
sydus sign of the zodiac
elusus fide mock(s) faith-Abl
Tér sô . he/who so
dô / when
pergat shall proceed
in cancro méistûn in cancer most-Acc
téta fílo caused much
ácher . field-Acc
únzît uuás . tér untime(ly) was he
chórnlôsêr without corn
déro of them-Gen
inĊstuat radiis phĊ- /bi rages rays-Abl Phoebus-Gen
semina / ... seeds-Acc
ze to the
nére may nourish
únuuílligen unwilling-Acc gánge may go
éi-/chelôn . acorns collect-Inf
síh . himself
Lat. ‘Who, then trusted many seed to the reluctant furrows when heavy from the rays of the sun the sign of the cancer rages (with heat), he will go disappointed in his faith to Ceres to the oak trees.’ OHG ‘He who, when the sun in the sign of the cancer caused most heat, sowed much in the unwilling field. As it was untimely he may go therefore without corn to the wood to collect acorns and with these may nourish himself.’
The wanda-sentence here is also to be found in a commenting passage, but in this example it provides background information: uuánda iz únzît uuás ‘as it was untime(ly)’ names the heat described before as the cause for the failure of the seed to grow. And this sentence is placed before the matrix sentence.
Translating information structure 343
So, as far as information structure is concerned, Notker, on the one hand, uses newly added or rephrased causal clauses as focus material: The wanda-sentences connect propositions, they appear without or with a correlate and they constitute a wide focus on the right periphery. On the other hand, such wanda-sentences contain background information; in that case they precede the rest of the sentence.
3.3. Marked foci In the topic-comment and in the focus-background structure, focus and topic are the central terms. Therefore we have to look for linguistic devices which distinguish these two information structural entities in the two languages we are comparing here. The pertinent elements for the marking of foci are focus particles or emphasizing pronouns, the word order and instances of contrast.
3.3.1. Focus particles, emphatic pronouns Like the Latin quoque, the word auch [also/too] functions as a focus particle in Old High German. But while in Latin quoque follows the word it refers to, in Old High German auch precedes its word of reference30: (24)
I 40,25ff. Tu / quoque you too
claro clear-Abl Úbe óuh if also
lumine. light-Abl tû uuél-/lêst you want
diu uuârhéit the-Acc truth-Acc
cernere to see
Lat./OHG ‘If you also want to see the truth with clear eyes’
In some instances, however, there is no equivalent of Old High German óuh in the Latin original. Admittedly, in the following example it is not quite clear whether we have a modal particle or a focus particle meaning “...self”31:
344 Rosemarie Lühr (25)
Poetry I 11,22ff. Et ui- / ctor and winner quĊcumque whatever
numeris . numbers-Abl
ua- / gos cursus . wandering-Acc course-Acc
flexa per uarios orbes . curved through various circles-Acc Únde and
uuíssa er knew he
zálo . number-Dat tûot . does
óuh too tîe Art-Acc
tîe / uérte be Art-Acc courses after dehéin some-Nom
ánder-/ ro planetarum other-Gen planets-Gen
uérte . course-Acc
Lat. ‘And it [the spirit] had as winner condensed into numbers which star traverses the wandering courses which is curved through various circles’ OHG ‘And he knew also the courses by numbers which some star makes by departing into other planets’ courses’
In contrast, examples of evidence with a preceding sélb ‘self’ in combination with the definite article as a means of emphasis are undisputed: here, sélb comes close to the meaning of the focus particle NHG sogar ‘even’. In contrast to NHG, however, it is still flectional. So, for example, in the commentary passage below, the focussing sélb is the equivalent of Latin ipse ‘self’: (26)
II 131,17ff. Nam quid because what regum . kings-Gen
ego I cum as
demonstrem . may show
disseram shall speak ipsa selves-Acc
tantĊ imbecillitatis? so much-Gen weakness-Gen ságen / fóne say of
dero chúningo . uuîo uuéih Art-Gen kings-Gen how weak sélben die selves Art-Acc
de familiaribus about household-Abl regna / kingdoms-Acc
dîen gesuâsôn Art-Dat relatives sîn . are geóuget shown
íh I hábeo have
Translating information structure 345 sô so
Lat. ‘Because what shall I speak about the household of the King when I show that the kingdoms [the reign of kings] (them)selves are full of weakness?’ OHG ‘What shall I say of the relatives of the kings, how weak they are, as I the kings themselves have shown so weak?/ ... as I have shown even the kings so weak?’
But sélb also occurs independent of an immediate Latin source. In a free translation of the Latin original, for example, we find: (27)
III 185,17ff. Uel re- / currat or may run back
pingitur. is decorated
gestíge-/ nez . climbed
astri . quocumque star-Gen wherever
Lat. ‘or it [the spirit] wanders through the circle of stars, wherever the twinkling night is decorated’ OHG ‘and also when it has climbed higher, even reached the sky ‘
It is mainly its use in the exclamatory sentence, which shows that Notker employed the focussing function of sélb quite deliberately – he is about to explain the usage of the Latin word vallum which has just been used: (28)
I 37,26f. Tér Art-Nom
bóuma hîezen trees-Nom were called
hîez was called
uallum . wall/rampart
ualli ! entrenchment piles-Nom
‘The fence was called wall. Even the trees were called “ualli”, i.e. entrenchment piles!’
346 Rosemarie Lühr 3.3.2. Word order A special feature of OHG syntax is that subjects are placed late, e.g. at the very end of the sentence, if they introduce a new discourse referent which is taken up as topic in the following sentence. When these new discourse referents are introduced, they are in the focus domain of the sentence in which they are established. Ex. (29)–(30) illustrate that it is not only the position of the pre-verbal accusative aboutness topic which is maintained, but also that of the subject at the end of the sentence in the Old High German translation: (29)
I 9,7ff. Eandem this-Acc
tamen32 ue- / stem . yet garment
Tîa sélbûn the same-Acc
hábe- / ton had
sciderant had torn to pieces
Lat. ‘This garment, however, had [been] torn to pieces [by] some violent hands.’ OHG ‘The self-same garment [by] some thugs had [by] torn.’ (30)
I 6,26f. inopina Uenit enim came namely unexpectedly
senectus old age
malis. / misfortune-Abl
uuánda mír for me
árbéiten / trouble
zûo geslúngen . happened
properata hastened úngeuuândo . unexpected
álti. (old) age
Lat. ‘There came namely unexpectedly old age, hastened by misfortune.’ OHG ‘For to me unexpectedly because of pain came fast old age .’ [Old age, which comes fast, came to me unexpectedly because of trouble.]
The same word order pattern involving postverbal subjects can be also found in interrogative sentences:
Translating information structure 347 (31)
II 50,13ff. Nos alli- / gabit ad us binds to alienam . foreign-Acc Sól shall
getûon . make-Inf
constantiam . constancy-Acc
cu- / piditas greed
uuíder mînemo against my-Dat tero Art-Gen
nostris moribus our-Dat customs-Dat hominum? mankind-Gen
síte customs-Dat ménniskôn mankind-Gen
stâta / constant
gîrhéit? (Näf 1979: 197) greed Lat. ‘Us shall to a constancy, which is foreign to our customs/character, bind the insatiate greed of mankind?’ [Shall the insatiate greed of mankind bind us to constancy which is foreign to our character?] OHG ‘Shall me against my customs constant make mankind’s unfulfilled greed?’ [Shall mankind’s unfulfilled greed make me constant/resistant against my usual customs?]
3.3.3. Cases of contrast
Contrast is an important means of structuring information for Notker and he uses it in various forms. In Latin, it is not only focussed accusative objects which are positioned right of the verb, but also focussed dative objects. Notker adopts this information structure but not the syntax: In place of the dative object he uses a directional compound construction: (32)
I 10,8ff. Hominumque men-Gen-and
mentes assuefaci- / unt senses they accustom
Únde ménniskôn and men-Gen sîe they
morbo . non liberant. illness-Dat not they liberate
stôzent sie ín_ dia / súht. drive them in Art-Acc illness-Acc siæ them-Acc
Lat. ‘And men’s senses accustom them to the illness, not liberate (them).’ OHG ‘And men’s senses drives them into the illness, they release them not.’
348 Rosemarie Lühr What we have here is the rhetorical figure disiunctio, a form of the isocolon: Coordinated sentences show a difference of meaning which is based on the negation of a positive term33. Thus we arrive at a contrast between the wide foci stôzent sie ín_dia súht and nelôsent siæ nîeht. Similarly, conjunctions can be used for describing contrasts: (33)
II 50,10ff. Ius est right is
mari . ocean-Dat
Ċquore . surface-Abl
blandiri stra- /to flatter-Inf smooth-Abl
sîn . be
strûben síh soar-Inf itself
uuínde . únde wind-Dat and
fluctibus. floods-Abl Ter Art
mé-/ re mûoz ocean must
ébene ! surface-Dat
uuéllôn . waves-Dat
Lat. ‘The right of the ocean is it, now to flatter with (a) smooth surface, now to shudder with storms and floods.’ OHG ‘The ocean also has to be calm with smooth surface, sometime soar with the wind and the waves.’
In Old High German óuh – óuh are set parallel to each other to connect structures with antithetical meaning. Compare, for example: (34)
I 10,2ff. QuĊ non who not
eius. uerum his-Acc but
sîn sêr nîeht his-Acc pain-Acc not
nullis remediis no-Abl remedy-Abl insuper moreover
fouerent . would cure ale- / rent would nourish
nehéillent . not-cure
Translating information structure 349 núbe / but
mêront . mít increase with
Lat. ‘Who not only with no remedy his pain can cure, but moreover want to would nourish him with sweet poison.’ OHG ‘Who his pain not only not cure, but even increase it with the sweet poison of their words.’
With the double conjunction non modo (...) verum insuper, OHG nîeht éin (...) núbe ióh ‘not only (...) but even/also’ after a negative statement, another – a contrasting – statement is emphasized. Through the negation of the first phrase, the contrastive parallel structures Lat. remediis foverent dolores (...) alerent dulcibus venenis, OHG sêr (...) héillent (...) mêront mít sûozemo éitere ío uuórto gain particular weight. Notker retained this structure in OHG because the combination of negation, contrast and parallelism makes it easier for his pupils to memorize the gist of his statements. In other cases, expressions of contrast in parallel structures in Old High German are placed in initial positions; positions, which in New High German are typical of a contrastive accent, i.e. I-topicalisation34. It is possible that these expressions already had this particular prosodic quality in Old High German. So, for example, in (35) the introductory phrase án dero zéseuuûn – án dero uuínsterûn describes frames which function as topics of contrast; therefore their pragmatic effect could have been a contrastive accent: (35)
I 9,13ff. Et and uero but
gestabat carried sceptrum. sceptre-Acc
dero zéseuuûn Art-Dat right-Dat
dero uuínsterûn Art-Dat left-Dat
quidem dextra libellos . now the right books-Acc
trûog si carried she
sinistra the left
bûoh..../ án book-Acc in
sceptrum ! sceptre-Acc
Lat. ‘And it carried now the right [hand] books, the left, however, a sceptre.’ OHG ‘In the right [hand] she carried books ... in the left a sceptre.’
350 Rosemarie Lühr Compare also to ze_níderost and ze_óberôst as frames in (36): (36)
a. I 8,27ff. ... Harum these-Gen legebatur was read Ze_ At
extremo / extreme-Abl ʌ ʌ
intextvm weaved into-Prt
níderost the bottom
margine . margin-Abl grecum. Greek
dero uuâte . / stûont Art-Dat garment-Dat stood
p. ... / p
Lat. ‘At its extreme margin a woven-in Greek ʌ was [to be] read.’ OHG ‘At the bottom of the garment was written the Greek ʌ .’ b. In in Ze_ at
legebatur was read
óberôst stûont / theta. the top stood theta
Lat. ‘on the upper one, however, was [to be] read ș.’ OHG ‘at the top stood theta.’
In a similar way, the subjects taz chrîecheska p und theta at the end of the sentences express contrasts, that is to say alternatives from a set of comparable entities. Both, in Latin and in Old High German we are dealing a focus – as can also be seen from Latin uero ‘but’. However, it is not only in cases as these, but also in other contexts that the end of the sentence in Old High German is a position for a focus which forms a contrast with another expression of focus. What is added in (37), for example, are the contrastive frames êr and nû which Notker uses as parallels (unlike the use of quondam in the Latin passage): (37)
I 6,13ff. Qui peregi who I completed
quondam carmina formerly songs-Acc
studio . zeal-Abl ...flebilis ...lamentable
cogor I am forced
mestos modos. sad-Acc tunes-Acc
Translating information structure 351 Íh-tir I-who
êr téta formerly did
máchôn nû make now
frôlichív / joyous-Acc
sáng . songs-Acc
Lat. ‘[me] who I completed formerly songs in flowering lamentabl(y) I am forced to begin sad tunes.’ OHG ‘I, who formerly made joyous songs I make now perforce dirges.’
A further example of the reinforcement of the meaning of contrastive expressions of Latin by parallel syntactic structures in Old High German is the following one: (38)
I 6,23ff. Gloria fe- / licis glory happy-Gen solantur they comfort
iuuentĊ . youth-Gen
nunc35 mea fata . mesti now my-Acc fate-Acc sad-Gen
Êr uuâ- / ren formerly were
trôstent sie comfort they
gûollichi mînero ornament my-Gen
senis. old man-Gen iúgende youth-Gen
álten . mî- / nero old-Acc my-Gen
mísseskíhte. misfortune-Gen Lat. ‘The glory of formerly happy and blooming youth now comforts the fate of the sad old man.’ OHG ‘Formerly they were the ornament of my youth. Now they comfort me old (man) in my misfortune.’
In this example, Notker uses contrasts by building two main clauses, in which now the frames êr and nu (Lat. olim and nunc) function as contrastive topics at the head of the respective sentences. A further contrast consists in the contrast foci iúgende and álten. So Notker employs contrastive structures, which underline and emphasize the existing contrasts of the Latin original, in order to clarify the respective passage for didactic reasons.
352 Rosemarie Lühr 3.3.4. Left peripheral foci in Latin
In the Latin emphasized pronoun vos and in the left peripheral focus recte facere in (39) we could have a structure comparable to I-topicalisation: (39)
II 100,12ff. Uos autem you but
nisi recte facere nescitis . rightly act-Inf you don’t understand except
ad in front of
populares auras . / et inanes the people’s-Acc ears-Acc and inane-Acc
rumores. gossip-Acc Ír you
neuuéllent don’t want
úmbe / líuto for people’s-Gen úppigen empty-Acc
réhto fáren . rightly act únde and
Lat. ‘But you don’t understand (how) to act rightly, except in front of the ears of the people and inane gossip.’ OHG ‘But you don’t want to act rightly, except for the praise of the people and the empty fame.’
In contrast to Latin, in Old High German the finite verb appears before the infinitive (Bolli 1975: 167). But similar to Latin, the prepositional phrase âne úmbe líuto lób . únde úmbe úppigen líument has been extraposed.
3.4. Constructions with marked topic positions
In this context we have to look at word order and pronouns.
3.4.1. Word order The Old High German sentence in the following example deviates from the Latin original: (40)
I 10,22f. His ille this-Abl that
increpitvs . / scolded
Translating information structure 353 humi to the ground Tô there
mestior rather sad
snífta níder lowered down
uultum. look-Acc sús so
erstóuta ge- / zuâhte. scolded flock
Lat. ‘That by this scolded multitude lowered rather sad the look to the ground [The multitude scolded by that, looked sadly to the ground].’ OHG ‘There lowered down (the look) the thus scolded flock.’
The Old High German sentence contains a given discourse referent after the adverbial dô and the finite verb. The adverbial dô in initial position can also be found in other Old High German texts, for example in the OHG Tatian translation. This pattern typically appears in contexts in which previously mentioned material does not function as the aboutness topic of the sentence36. The topic function of Latin ille chorus incretitvs or OHG dáz sús erstóuta ge is indicated by the deixis pronoun ille which Notker translates with a definite article.
3.4.2. Pronouns Even apart from examples as (40) above, we generally have to consider the various forms of pronouns in Latin when looking at their translations into Old High German. As a rule, subject pronouns are dropped in Latin when they represent continuous topics. The accusative object also frequently remains unexpressed due to object drop. The oblique cases are represented by the forms of the pronouns is, ea, id. While these anaphorically allow the implication of nouns in sentences further away, they can also have a demonstrative function and refer back to immediately preceding contents words or facts, just as the demonstrative pronoun hic, haec, hoc does. This means that the Latin pronouns is, ea, id are ambiguous. Notker, however, can disambiguate the reference by choosing a proper equivalent in OHG. After the disastrous influence of the sirens has been described, the discourse continues with the sentences quoted in (41): (41)
I 10,13ff. Nihil quippe not certainly Án at/of
in eo lede- / rentur shall be belittled in the
démo neinfûore the-Dat not may be lost
operĊ nostrĊ . troubles ours
nîeht mînero nothing (of) my-Gen
354 Rosemarie Lühr ár- / béito. troubles-Gen Lat. ‘Not, certainly, shall by this our troubles be belittled.’ OHG ‘Of this may not be lost any of my troubles.’
In Latin, the anaphoric in eo is placed on the right of the verb, Notker uses án démo (instead of the less emphatic dârána) and places at the beginning of the sentence. While Notker retains the position of the focal subject at the end of the sentence, he decided on a topic marking which deviates from the Latin one. In keeping with the position of the demonstrative pronoun dieser, the position is at the head of sentence. (42)
Poetry I 11,19ff. Hic quondam liber . assuetus aperto this one once free used open-Abl ire go-Inf
Tíser uuás this one was
meatus . ... paths-Acc
ke- / uuón used
hímel-férte … celestial paths-Acc Lat. ‘This one was once free(ly) used to go through the open space in ethereal paths....’ OHG ‘This one was used to think of the celestial paths ...’37
Of particular interest, however, is how Notker translates Latin relative pronouns used in a continuative function (relativischer Anschluss), for this structure is unfamiliar in German. In case a subordinate and a matrix clause share a common subject, it is expressed by a relative pronoun put before the conjunction in Latin. Notker does not retain this structure but rather substitutes the relative pronoun by a personal pronoun as the referent is taken up again due to topic promotion, see (43): (43)
I 8,8ff. QuĊ cum she when ipsum cĊlum. (it)self sky-Acc Sô when
altivs / higher
extulisset caput . had raised head-Acc
penetrabat. penetrated dáz hóubet Art-Acc head-Acc
hô / high
ûf erbúreta. raised up high
Translating information structure 355 sô so
úber slûog surmounted
HÍmel . sky-Acc
Lat. ‘When she higher had raised the head, she even penetrated the sky.’ OHG ‘When she the head raised up high, it surmounted the sky.’
Sometimes he chooses a possessive pronoun for a substitute: (44)
I 8,23ff. Quarum speciem their-Gen appearance-Acc uetustatis age-Gen
obduxerat . covered
quĊdam neglectĊ / certain neglected-Gen
Íro bílde uuâ- / ren its-Gen appearance were
álti uersáleuuet ... age-Dat darkened
Lat. ‘Their outer appearance had covered a certain, caused by neglected age, mist [a certain mist caused by neglected age]’ OHG ‘Its outer appearance by age was darkened ...’
More frequently we find a demonstrative pronoun as a substitute: (45)
I 9,1ff. … gradus in steps after …
léi- / ter-sprózen ... rungs of a ladder
Quibus esset on these would be superius higher-Acc Áfter on demo Art-Dat
modum / Art-Acc
scalarum. ladder-Gen álde or
ascensus . / ascent
stégon stûofa. steps of a ladder ab from
inferiori ad lower-Abl to
dîen these-Dat níderen lower-Dat
man stîgen one climb-Inf pûohstabe zu demo letter-Dat to Art-Dat
máhti . fó- / ne could from óberen. higher-Dat
Lat. ‘... steps in the manner of ladders. On these would an ascent from the lower to the higher part (be possible).’ OHG ‘... rungs or steps of a ladder. On these could one from the lower letter to the higher climb.’
356 Rosemarie Lühr But there are even more alternatives: In the following example, Notker seems to take the relative sentence connection with Lat. quas for a relative pronoun and places the noun phrase at the head of the sentence: (46)
I 8,20ff. Quas which-Acc Tîa Art-Acc
ipsa self uuât garment
texuerat she had woven sî she
manibus hands-Abl sélbiu self
suis . her-Abl uuórhta ! made
Lat. ‘Which (the garment) she herself with her own hand had woven.’ OHG ‘The garment she herself made!’
But as we can see from the exclamation mark, Notker changes this sentence into an exclamative sentence with verb-final position. As Lat. quas serves both as a relative and an interrogative pronoun, the sentence could be transformed into a type of sentence which contains an interrogative pronoun. As we expected, the translation of pronouns with topic function, especially the relative connection of sentences again demonstrates the sophistication of Notker’s translations.
4. Summary The first objective of our research was to answer the question of how Notker converted the information structure of Latin into Old High German. The result is a complex picture: Notker did not convert the hyperbaton, a typical linguistic phenomenon of Latin, into Old High German ((12) to (15)), but, where appropriate, he used other techniques to imitate the emphasis implied in the separation of elements which belong together (15). However, we do find evidence of rhetorical word order in Notker: We discussed exclamative sentences and causal sentences introduced with wanda. There is an increased use of exclamative sentences in the poetic parts of the Latin original. In these sentences, Notker sometimes changed the position of the verb (16), sometimes retained the word order of the original (17) and sometimes completely restructured the sentence, as in the instance of the pre-position of an all-focus sentence (18). Notker also transforms Latin declarative sentences into exclamative sentences ((19), (46)).
Translating information structure 357
As far as causal sentences are concerned, he frequently added them as an explanation of the text ((20), (21), (23)) or to make a passage more stringent (22). If they appear after the matrix sentence, they contain focus material and form a wide focus ((20), (21), (22)). If they are put in front, they provide background information (23). As topic and focus are the marked elements of the opposition in the topic-comment and the focus-background structure, all the linguistic means which mark these information structural entities are of particular interest for us. Concerning the focus, these are focus particles and emphasizing pronouns: OHG óuh, Lat. quoque ((24), (25)) or sélb, which Notker uses not only as an equivalent to Lat. ipse (26), but also independent of any Latin source ((27)–(28)). Word order also serves as a marker of focus as long as we have a subject as a new discourse referent. In this function it appears as an emphatic focus at the end of the sentence ((17), (29), (30), (31)). A special characteristic of Notker’s style as a translator is his incorporation of foci into contrastive structures38. We find the imitation of the rhetorical figure of disiunctio (14), the use of parallel conjunctions ((33)–(34)), but also examples which resemble the New High German Itopicalisation with two contrast topics at the beginning of the sentence and two contrast foci at the end of the sentence ((35)–(38)). In contrast to this, in Latin you also find foci on the left periphery (39). For marked topic constructions (ausgezeichnete Topik-Konstruktionen), word order is also relevant, as is suggested by sentences with covert initial position with adverbial dô (40).We also have to take into account the use of pronouns in the languages we are comparing here: the representation of Lat. ille by the definite article (40), the clarification of the ambiguous Lat. is by a demonstrative der at the head of the sentence (41), a position which is occupied by demonstrative pronouns in general (42), and the dissolution of the relative connection of sentences. For this construction, which does not exist in Old High German, Notker chooses various forms of translation, the personal pronoun (43), the possessive pronoun (44), and, most frequently, the demonstrative pronoun (45). Finally, since in Latin the relative pronoun and the interrogative pronoun can be identical, there is the option to transform a Latin declarative sentence with relative connection into an exclamative sentence (46). All in all, what we can definitely say about Notker’s representation of the information structure of the Latin original is, that he certainly had a firm grasp of its structures. First and foremost, this becomes obvious in cases where the allocation of the informational entity focus coincides in the two languages, but where Notker structures the foci independent of the
358 Rosemarie Lühr Latin source. If he doesn’t imitate the information structure, he has good reasons for doing so: it is either because of the basic differences between the two languages or because of his didactic purposes. For Notker’s primary aim always was to make the text comprehensibly to his pupils. The second objective of the research points beyond Notker; for from Notker’s handling of the information structure we can draw a conclusion for Old High German; i.e. that it is more than likely that there is a fundamental difference between the two languages compared here with regard to the positioning of the kinds of foci: Old High German: structural focus – verb – emphatic focus vs. Latin: verb – emphatic focus – structural focus ((13), (15)). In addition, our discussion shows that neither Latin nor Old High German is a ‘discourse configurational’ language.
So far there is no general agreement as to whether Notker actually translated the metrical passages in a metrical form. In some passages, however, one can certainly detect a particular rhythm (Glauch 2000: 170). 2. Prince (1981); Frascarelli & Hinterhölzl (2007). 3. In the case of a “continuous topic” the topic of the preceding sentence is retained, whereas in the case of a “shifting topic” there is a change of topic (Speyer 2007). 4. Gundel (1988). The dialogue partners are already familiar with these topics. 5. Cf. Frey (2000: 138) following Reinhart (1981): “Topiks sind die Ausdrücke, über deren Referenten durch die Sätze Aussagen gemacht werden – Topik ist eine Kategorie des pragmatischen ‘Worüber’” [Topics are expressions about whose referents the sentences make statements – topic is a category of pragmatic ‘aboutness’]. 6. The discourse topic is about a new topic (Frey 2000; Späth 2005). 7. Contrastive topics can be found, for example, in the so-called I-topicalization. Cf. Lang and Umbach (2002). 8. There is verifiable evidence that Notker based this on a tract on grammatical and syntactical problems (in four parts) from St. Gall (Tax 1986: XXII). 9. Glauch (2000: 172). For Christian influences cf. Mohrmann ( 1984: 302ff.). 10. The Old High German text was written around 1025; even though most of it was written by a scribe he was probably working under the aegis of Notker (Tax 1986: XXVI; XLIII). 11. A further particularity occurs in the compound or periphrastic tenses, the socalled conjugatio periphrastica of the type scripturus fui, eram, etc., and pas-
Translating information structure 359
12. 13. 14. 15. 16. 17.
18. 19. 20. 21. 22. 23. 24. 25.
sive verbal paraphrases of the type amatus fui [I have been loved], amatus fueram [I had been loved] (as opposed to the synthetically constructed active amavi, amaveram). These forms increase in the course of the development of Middle Latin and of Romance languages: From Vitruvius onwards, the est factus-type more and more prevails until in Vulgar Latin and in the Romance languages it completely replaces its rival factus est (Hofmann and Szantyr 1965: 405). Moreover, the victurus sum - and tradendus sum-types begin to rival the simple future; in addition, paraphrases with auxiliaries, modal verbs and aspect verbs with infinitive begin to spread; e.g. habeo, possum, volo, debeo, incipio (Hofmann and Szantyr 1965: 313ff.; Stotz 2004: 323ff.; 445f.). Furthermore, in Middle Latin the infinitive is also used after censere, putare, credere, iudicare, consentire, petere, rogare, permittere, timere, as well as in a final function in abiit manducare, or after facere, curare in the sense of “have something done; make someone do something” (Kindermann 1998: 42. For the use in Boethius cf. Dienelt 1942: 114ff. and 132). This construction is also known in German. For the following discussion of the information structure, however, the compound tenses are irrelevant. Dik 1995; for Ancient Greek see Matiü (2003: 578ff.) following Kiss (1995: 2001). Devine and Stephens (2006: 524). For potentially comparable structures in German cf. Krisch (1998: 373ff.). For particular aspects of syntax cf. now Stotz (2004). Hofmann and Szantyr (1965: 690). But as Krisch (1998: 353ff.) shows, the hyperbaton is a vernacular phenomenon even in Old Latin. While we find verb second, verb first and verb third position of the finite verb in the Old High German Isidor and in the Old High German Tatian, “scheinen ca. 200 Jahre nach der Isidor-Übersetzung und ca. 170 Jahre nach der TatianÜbersetzung alle Stellungsmöglichkeiten außer der Zweitstellung verloren gegangen zu sein” [except for verb second position, all the other potential positions seem to have been lost about 200 years after the Isidore translation and about 170 years after the Tatian translation] (Näf 1979: 146). Cf. also Borter (1982: 51ff.). Ms. Ċtatem. Näf (1979: 137 and 196). Cf. Umbach (2001). Cf. also Büring (1997); Chierchia (1995); Asher and Lascarides (1998) on the concept of aboutness. Cf. Devine and Stephens (2000) for the Greek language. “Sind sie es doch, die mit dem unfruchtbaren Dorngestrüpp der Leidenschaften die fruchtreiche Saat der Vernunft ersticken” (Gothein 1932: 9). Speyer (2003: 15).
360 Rosemarie Lühr 26. The reproduction of the position of Wackernagel-particles would result in other “non-German” structures. So Notker places the “light” pronoun tíz after the conjunction and the subject pronoun íh (Näf 1979:338ff.): I 7,20f. ipse. Haec dum mecum tacitus reputarem this-Acc when with me silent-Prt I would think self Únz íh tíz suîgendo in mîne- / mo when I this-Acc silently in my-Dat
mûote áhtota. mind-Dat thought over
OHG “Als ich dies schweigend in meiner Vorstellung überdachte” [When I this silently in my mind thought over] A word order such as: “Dies als ich schweigend in meiner Vorstellung überdachte” [This when I silently in my mind thought over] would certainly have been ungrammatical in Old High German. 27. But in Latin you can have pronouns between the verb and the structural focus: I 6,15f. scribenda. Ecce lacerê camenê dictant / mihi see sad muses dictate me (what) to write Síh see
no . now
léidege musê . lêrent sad muses teach
Lat. ‘See, sad muses dictate me what to write’ OHG ‘See now, sad muses teach me to write’ Notker in contrast, chooses an AcI in unmarked word position. 28. Cf. also: Poetry I 11,11ff. Et / relicta propria luce ... tendit ire and left behind-Prt.Abl own-Abl light-Abl strives go-Inf externas / external-Acc Únde and demo Art-Dat
uuîo gnôto / how urgent-Adv lîehte in light-Dat in
iz tánne it then dia Art-Acc
îlet . ûzer hurries out uínstri. darkness-Acc
Lat. ‚and (after) the own light been left behind, strives [the mind] to go out into the darkness’ OHG ‘And how urgently it then from the light into the darkness hurries.’ 29. Because of the verb final or verb late position, we can be certain here that we have a subordinate clause; cf. Lötscher (in this volume). 30. Cf. also I 14,15f.
Translating information structure 361 An part
mecum with me
rea . accusation-Acc
agiteris may be moved
falsis criminationibus? false-Abl accusations-Abl Ínno . part
óuh too kemûot tortured
gescúldigotív . fóne accusations-Acc from
/ uuérdêst . may be
Lat., OHG ‘Or that you, too, from false accusations may be tortured’. 31. For the difference between óuh/auch ‘also’ as a focus particle and as a modal particle in declarative sentences cf. Thurmair (1989: 155). 32. Here a pronoun has been separated from its word of reference by the interposition of tamen [however] in the Wackernagel position (Hofmann and Szantyr 1965: 398). Strictly speaking, we therefore do not really have a hyperbaton here. 33. Lausberg (1990: 114). 34. For the intonation pattern cf. ¥Otto ist ins \Kino gegangen (und ¥MarIA in die \Oper [¥Otto went to the \cinema (and ¥MarIA to the \opera] (Frey 2000: 149 with further reading). 35. In Latin, an adverb can be placed between the verb and the structural focus on the right hand side. Cf. the position of the adverb nunc [now] before mea fata. 36. Hinterhölzl and Petrova (2005); Donhauser, Solf and Zeige (2006: 9f.). 37. Cf. also: I 6,30f. uertice intempestiui cani. Funduntur pour crown of the head-Abl untimely hairs Fóne from
dîen / these-Dat
grâuuên turn grey
ze_ únzite. at “untime“-Dat Lat. ‘There pour from the crown of my head the untimely hairs.’ OHG ‘From these things I turn grey before time.’ For word order in Latin cf. Menge (2000: 577). 38. For contrast in general cf. Molnár (2002).
362 Rosemarie Lühr References Abraham, Werner 1986 Word order in the middle field of the German sentence. In Topics, focus, and configuarionality, eds. Werner Abraham and Sjaak de Meij, 15–38. Amsterdam: John Benjamins. 1992 Clausal focus vs. discourse rhema in German: A programmatic view. Language and Kognition 2: 1–19. Asher, Nicholas and Alex Lascarides 1998 Bridging. Journal of Semantics 15: 83–113. Bolli, Ernst 1975 Die verbale Klammer bei Notker. Untersuchungen zur Wortstellung in der Boethius-Übersetzung. Berlin/New York: Mouton de Gruyter. Borter, Alfred 1982 Syntaktische Klammerbildung in Notkers Psalter. Berlin/New York: Mouton de Gruyter. Büring, Daniel 1997 The 59th Street Bridge accent. On the Meaning of Topic and Focus. Phil. Diss. Tübingen. Chierchia, Gennaro 1995 Dynamics of Meaning. Anaphora, Presupposition and the Theory of Grammar. Chicago/London: The University of Chicago Press. Devine, Andrew M. and Laurence D. Stephens 2000 Discontinuous Syntax. Hyperbaton in Greek. New York/Oxford: Oxford University Press. 2006 Latin Word Order: Structured Meaning and Information. Oxford: Oxford University Press. Dienelt, Karl 1942 Sprachliche Untersuchungen zu Boethius’ Consolatio philosophiae. Glotta 29: 98–128; 129–138. Dik, Helma 1995 Ancient Greek word order. Amsterdam: Gieben. Doherty, Monika 2002 Language Processing in Discours. A key to felicitous translation. London: Routledge. Donhauser, Karin, Michael Solf and Lars Erik Zeige 2006 Informationsstruktur und Diskursrelationen im Vergleich: Althochdeutsch – Altisländisch. In Grenzgänger. Festschrift zum 65. Geburtstag von Jurij Kusmenko, eds. Antje Hornscheidt, Kristina Kotcheva, Michael Rießler and Tomas Milosch, 73–90. Berlin: Nordeuropa-Institut.
Translating information structure 363 Frascarelli, Mara and Roland Hinterhölzl 2007 Types of Topics in German and Italian. In On Information Structure. Meaning and Form, eds. Susanne Winkler and Kerstin Schwabe, 87– 116. Amsterdam/Philadelphia: John Benjamins. Frey, Werner 2000 Über die syntaktische Position der Satztopiks im Deutschen. ZAS Papers in Linguistics 20: 137–172. Glauch, Sonja 2000 Die Martianus-Capella-Bearbeitung Notkers des Deutschen. Bd. I: Untersuchungen. Tübingen: Max Niemeyer. Gothein, Eberhard 1932 (transl.): Anicius Manlius Severinus Boethius: Trost der Philosophie. Berlin: Die Runde. Gruber, Joachim 1978 Kommentar zu Boethius ‚De consolatione philosophiae‘. (Texte und Kommentare 9). Berlin/New York: Mouton de Gruyter. Gundel, Jeanette K. 1988 Universals of topic-comment structure. In Studies in Syntactic Typology, eds. Michael Hammond, Edith A. Moravcsik and Jessica R. Wirth, 209–239. Amsterdam: John Benjamins. Hetland, Jorunn and Valéria Molnár 2001 Informationsstruktur und Reliefgebung. In Language typology and language universals: An international handbook, eds. Martin Haspelmath, Ekkehard König and Wulf Oesterreicher, 617–633. Berlin/New York: Mouton de Gruyter. Hinterhölzl, Roland and Svetlana Petrova 2005 Rhetorical Relations and Verb Placement in Early Germanic Languages. Evidence from the Old High German Tatian Translation (9th century). In Salience in Discourse. Multidisciplinary Approaches to Discourse, eds. Manfred Stede, Christian Chiarcos, Michael Grabski and Lunk Lagerwerf, 71–78. Amsterdam: Stichting, Münster: Nodus. Hofmann, Johann Baptist and Anton Szantyr 1965 Lateinische Syntax und Stilistik. München: Beck’sche Verlagsbuchhandlung (Handbuch der Altertumswissenschaft II, 2,2). Jacobs, Joachim 1986 The Syntax of Focus and Adverbials in German. In Topic, Focus, and Configurationality, eds. Werner Abraham and Sjaak de Meij, 103–128. Amsterdam: John Benjamins. Kindermann, Udo 1998 Einführung in die lateinische Literatur des mittelalterlichen Europa. Turnhout: Brepols-Verlag.
364 Rosemarie Lühr Kiss, Katalin É. 1995 Discourse configurational languages. Oxford: Oxford University Press. 2001 Discourse configurationality. In Language typology and language universals: An international handbook, Vol. 2, eds. Martin Haspelmath, Ekkehard König and Wulf Oesterreicher, 1442–1455. Berlin: Mouton de Gruyter. Krisch, Thomas 1998 Zum Hyperbaton in altindogermanischen Sprachen. In Sprache und Kultur der Indogermanen. Akten der X. Fachtagung der Indogermanischen Gesellschaft, Innsbruck, 22.–28.September 1996, ed. Wolfgang Meid, 351–384. Innsbruck: Institut für Sprachwissenschaft. Lang, Ewald and Carla Umbach 2002 Kontrast in der Grammatik: Spezifische Realisierungen und übergreifender Konnex. In Sprachtheoretische Grundlagen der Kognitionswissenschaft: Sprachliches und nichtsprachliches Wissen, ed. Anita Steube, special issue of Linguistische Arbeitsberichte 79: 145–186. Lausberg, Heinrich 1990 Elemente der literarischen Rhetorik. Eine Einführung für Studierende der klassischen, romanischen, englischen und deutschen Philologie. 10München: Max Hueber Verlag Lötscher, Andreas this vol. Verb placement and information structure in the OHG Gospel Harmony by Otfrid von Weissenburg. In New Approaches to Word Order Variation and Word Order Change, eds. Roland Hinterhölzl and Svetlana Petrova. Berlin: Mouton de Gruyter. Lühr, Rosemarie 2007 Information Structure in Ancient Greek. In The discourse potential of unspecified structures, ed. Anita Steube, 487–512. Berlin: Mouton de Gruyter. Matiü, Dejan 2003 Topic, focus, and discourse structure. Ancient Greek Word Order. In Studies in Language 27 (3): 573–633. Menge, Hermann 1961 Repetitorium der lateinischen Syntax und Stilistik. München: Max Hueber. 2000 Lehrbuch der lateinischen Syntax und Semantik. Völlig neu bearb. von Thorsten Burkard und Markus Schauer. Darmstadt: Wissenschaftliche Buchgemeinschaft.
Translating information structure 365 Mohrmann, Christiane 1984 Some remarks on the language of Boethius ‘consolatio philosophiae’. In Boethius, eds. Manfred Fuhrmann and Joachim Gruber, 302–310. Darmstadt: Wissenschaftliche Buchgesellschaft (Wege der Forschung CDLXXXIII). Molnár, Valéria 2002 Contrast from a contrastive perspective. In Information Structure in a cross-linguistic perspective, ed. Hilde Hasselgård, 147–162. Amsterdam/New York: Rodopi. Näf, Anton 1979 Die Wortstellung in Notkers Consolatio. Untersuchungen zur Syntax und Übersetzungstechnik. Berlin/New York: Walter de Gruyter. Notker der Deutsche 1986 Boethius: De consolatione Philosophiae. Buch I/II. Hg. von Petrus W. Tax. Neue Ausgabe. Tübingen: Niemeyer. Petrova, Svetlana and Michael Solf this vol. On the Methods of Information-Structural Analysis of Texts from Historical Corpora. A Case Study on the OHG Tatian. In New Approaches to Word Order Variation and Word Order Change, eds. Roland Hinterhölzl and Svetlana Petrova. Berlin: Mouton de Gruyter. Prince, Ellen F. 1981 Toward a Taxonomy of Given-New Information. In Radical Pragmatics, ed. Peter Cole, 223–255. New York: Academic Press. Reinhart, Tanya 1981 Pragmatics and Linguistics: An Analysis of Sentence Topics. In Philosophica 27: 53–94. Rubenbauer, Hans, Johann B. Hofmann and Rolf Heine 1977 Lateinische Grammatik. Bamberg: Buchners Verlag. Schröbler, Ingeborg 1953 Notker III. von St. Gallen als Übersetzer und Kommentator von Boethius’ De Consolatione Philosophiae. Tübingen (Heraema N.F. 2). Späth, Andreas 2005 Die linke Satzperipherie und ihr semantischer Beitrag zur Diskurseinbettung. Ms Leipzig. Speyer, Augustin 2003 A Prosodic Factor for the Decline in Topicalisation in English. Ms. Philadelphia. 2007 Die Bedeutung der Centering Theory für Fragen der Vorfeldbesetzung im Deutschen. Zeitschrift für Sprachwissenschaft 26: 83–115.
366 Rosemarie Lühr Steube, Anita and Andreas Späth 2002 Semantik, Informationsstruktur und grammatische Modularität. In Sprachtheoretische Grundlagen der Kognitionswissenschaft: Sprachliches und nichtsprachliches Wissen, ed. Anita Steube, special issue of Linguistische Arbeitsberichte 79: 235–254. Stotz, Peter 2004 Handbuch zur lateinischen Sprache des Mittelalters. Band 4: Formenlehre, Syntax und Stilistik: München: Beck’sche Verlagsbuchhandlung. Tax, Petrus 1986 cf. Notker der Deutsche. Thurmair, Maria 1989 Modalpartikeln und ihre Kombinationen. Tübingen: Max Niemeyer (Linguistische Arbeiten 223). Umbach, Carla 2001 (De)accenting definite descriptions. Theoretical Linguistics 27: 251– 280. Wackernagel, Jacob 1892 Über ein Gesetz der indogermanischen Wortstellung. Indogermanische Forschungen 1: 333–436.
Aspects of word order and information structure in Old Saxon Sonja Linde
1. Introduction Current research on the development of word order in the Germanic languages suffers from the lack of a consistent description of the situation in Old Saxon (henceforth OS) which offers one of the earliest writing traditions of the Germanic group at all and includes various types of word order patterns. Standard descriptions of OS contain very little or no information about word order. While e.g. the grammar book by Gallée (ed. Tiefenbach 1993) only refers to issues of phonology and inflectional morphology, Behaghel’s (1897) comprehensive monograph mainly reflects the syntactic behaviour of different word classes with respect to their property to organize word groups and phrases but lacks any observations on the most intriguing question with respect to our investigation, the description of the linear order of constituents in the clause. In the preface to his monograph, Behaghel himself addresses the incompleteness of his book with respect to word order, cf. Behaghel (1897: vi). A notable exception in this respect is the survey of OS provided by Rauch (1992) who pays attention to the sentence structure in OS records. Rauch (1992: 24–31) accounts for a high degree of variation in word order and establishes a distinction between marked and unmarked patterns on the basis of their relative frequency in the records. In the case of independent declarative sentences, she argues that the (X)VSO order with all arguments following the verb and an optional single non-argument preceding it displays the most common pattern which therefore is defined as unmarked, while other patterns like (X)SVO or (X)SOV are said to be less frequent and therefore marked (1992: 24). In accounting for the functional differences among marked and unmarked patterns, Rauch addresses the issue of their textual role and pragmatic value in the discourse. E.g., she explains the pattern in which the verb precedes all arguments as a device of “concatenation or continuation in a narrative sequence” (Rauch 1992: 30). As shall become clear later, this is a major consideration in the approach pur-
sued in the present study as well. However, the analysis provided by Rauch raises a couple of questions. First, she discusses early verb positioning in the pattern (X)VSO accounted by her as unmarked on the basis of the example gaf it is iungarum forð ‘[he] gave it [=the bread] further to his disciples’ (Rauch 1992: 28–30). Admittedly, this is an elliptical coordinate conjunct which has no overt subject itself but shares the subject of the previous conjunct expressed in the phrase manno drohtin ‘the Lord of men’. Hence, this example is no suitable representative of the (X)VSO pattern, which really abounds in OS. Second, Rauch explains the initial placement of the verb gaf ‘[he] gave’ in relation to Behaghel’s Rule 2 according to which old information precedes new one in a sentence. In Rauch’s analysis of the example quoted, the verb denotes one of a series of expected actions and therefore carries of old information. At the same time, she seems to overlook that the old information is actually conveyed in the subject manno drohtin ‘the Lord of men’ dropped in this sentence and that the verb merely introduces the new information assigned to it. All in all, Rauch’s discussion on the matter is very brief. It neither results in an attempt to provide a general model of the underlying syntactic structure of the OS sentence nor does it systematically explore all factors leading to variation in the linear order of sentence constituents. Moreover, the classification of marked vs. unmarked order in OS given by Rauch (1992) contradicts the observations made approximately one hundred years earlier by Ries (1880: 5–11). In his view, the basic or regular word order pattern in OS involves the preverbal position of the subject in (S…V…) while the pattern in which the subject (including all arguments) follows the verb (…VS...) is regarded to be secondary, and derived from the basic order for special rhetorical and textual purposes. On closer look, it turns out that the pattern viewed as ‘unmarked’ in Rauch’s terms corresponds to what Ries counts as a secondary, or occasional pattern, and vice versa. Ries builds his proposal on language-internal as well as genealogical considerations. First, he accounts for the fact that the reverse pattern is typical for interrogatives, requests, exclamations etc., i.e. for sentences that bear a special, occasional usage with respect to simple declaratives. Therefore, it is not likely that the syntactic patterns occurring in those functionally ‘marked’ cases should present the basic order in the language. He further points at the fact that the preverbal position of the subject is the basic one in the Indo-European proto-language and a common property of all ancestors except of Celtic which was the only one to generalize verb-first as its basic order. Above all, Ries (1880: 11) explicitly refers to the high frequency of sentences with the so-called reverse order
Aspects of word order and information structure in Old Saxon
but argues that quantity of occurrences alone is not decisive for the identification of the basic word order in the system of a language. Apart from these contradictory views, the elaborations in Ries (1880) and Rauch (1992) share one important property: they both promote the idea that word order in OS – although exposing a variety of competing word order patterns – is by no means random but obeys specific rules and principles influencing the positioning of sentence constituents, and especially the placement of the finite verb. They further relate these principles to the broad field of pragmatics and discourse organization, though a more precise characterization of the functions and categories taking primary influence on the syntactic realization of sentence constituents is still missing. In more recent times, a small investigation on the issues of word order in OS appeared aiming to evince that the principles of word order in OS are explainable on purely syntactic grounds, cf. Erickson (1997). This account is nested within the ‘Government and Binding’ model of generative theory and shows that methods and concepts provided for the explanation of the syntactic regularities in other early Germanic languages like Lenerz (1984) for Old High German or van Kemenade (1987) for Old English are adaptable to the situation in OS. Following this approach, OS exhibits a base-generated SOV order retained in embedded clauses introduced by an overt complementizer whereas independent clauses display a variety of features related to a generalized V2-pattern like in modern German or Dutch. The aim of the present study is to account for both views on OS syntax in more detail and to trace back the influences of both syntactically as well as pragmatically based factors for the explanation of the word order patterns attested in OS records. The study is based on the observation of a large amount of empirical data that exceed the examples discussed in the previous literature. The results of the investigation give proof that both types of governing principles – discourse-configurational as well as purely syntactic ones – can be identified as motivating word order in OS. These observations allow for the assumption that OS is a language in which two competing systems govern the linear ordering of constituents.
2. The Source OS is attested in two major texts, Heliand and Genesis, which are both alliterative rhyme poems dated back to the 9th century. Besides, there are some prose texts, e.g. some liturgical texts or tax lists (Heberegister). Un-
fortunately, the prose texts are of a rather small size, and research on OS is based on the two major texts of the Heliand and the Genesis. Of these, the Heliand is the more extensive one by far including about 6000 long verses. Therefore, it was taken as the basis of this investigation. The OS Heliand was handed down in two major manuscripts, C (Ms. Cotton. Calig. A. VII, London, British Library) and M (Cgm 25, München, Bayerische Staatsbibliothek) as well as in four fragments. The text itself is subdivided into 72 chapters called ‘fits’ which in the original manuscripts are indicated by initial capital letters. In manuscript C, fits are additionally enumerated by Roman numbers.1 Quite often, the beginnings of smaller episodes or periods show initial capitalization as well. It is clear that the Heliand manuscripts expose a sophisticated system of capitalization according to which initial capitals function as indicators of the text organization provided by the writer. Unfortunately, none of the current editions of Heliand reflect these potentially valuable text structuring devices but represent the graphical layout of the manuscripts normalized according to modern German usage. Since no complete facsimile edition and no diplomatic edition exist, evidence on the formal means of text organization provided in the manuscripts is not available to us in full extent. The OS data are taken from the old but rather well-accepted Heliand edition by Sievers (1878). It reflects both major manuscripts in parallel columns and presents the original writing in glosses in case the edition differs from the manuscript. In terms of punctuation and capitalization, Sievers applies the rules of modern German usage. All examples cited are taken from manuscript M according to Sievers (1878) and from manuscript C only if there is no equivalent in M.
3. Properties of V2 in OS? Well-known generative work (e.g. cf. Lenerz (1984) for Old High German, van Kemenade (1987) for Old English) assumes that the early Germanic languages expose a base-generated verb final order. In independent clauses, the inflected verb is raised regulary to a vacant C0-position, and additionally, the SpecC-position is usually filled thus yielding a generalized V2-pattern like in modern German and Dutch. A similar view has been proposed by Erickson (1997) for OS as well. In fact, the data presented in his investigation strongly support this view. In this section, we shall investigate in more detail the assumed V2-properties of OS and discuss some further points concerning this question.
Aspects of word order and information structure in Old Saxon
A great part of the observable data indeed prompts the view that OS displays some typical features of a SVO- language. In dependent clauses like complement clauses (1) or relative clauses (2) the finite verb appears at the end of the sentence following all its arguments: (1)
[so uuas thero liudio thau = ‘so it was the custom with the people’] that that erlo gihuulic obean scolde COMP that-ACC noblemen-GEN everyone celebrate should (Hel 2732) ‘(…) that everyone of the noblemen should celebrate that (i.e. the birthday of the monarch)’
[buide imu be theru brudi = ‘He was living with the woman’] thiu er sines broder uuas the-FEM before his-GEN brother-GEN was (Hel 2706) ‘(…) who had been his brother’s bride before.’
As an exception, PPs and CP-complements are often extraposed, cf. the adjunct phrase an Galilæo land in (3): (3)
tho he im mid is gesidon giuuet eft an COMP he he-DAT with his companions went back to Galilæo land [for im (…)] (Hel 2290) Galilee land went he-DAT ‘When he came back to Galilee with his company, he went …’
In root clauses, on the other hand, the finite verb occurs in the second position like in modern V2-languages, where it is typical that the specifier position of the C-domain is occupied by any constituent regardless of its grammatical relation to the governing verb or to its information-structural value in the context. This property of V2-languages may be traced back in OS data as well. In (4), the constituent in front of the verb is the pronominal subject siu ‘she’. In case that an object or adjunct is preposed, the subject remains post-verbally. This is shown in (5) and (6) accordingly where the pronominal subject follows the verb:
siu uuelde tho ira geba egan she wanted then her gift have ‘She wanted to have her reward’
that scoldun sea fiori thuo fingran that-ACC should they four then fingers-DAT
scriban write (Hel 32)
‘These four were to write it down with their fingers’ (6)
so helde he thea haltun man so healed he the lame men ‘so healed the lame people’
Furthermore, the preverbal position may host a syntactic operator as the wh-phrase in a direct question (7), the aboutness-topic of an utterance (8) as well as newly added information, i.e. focal material (9): (7)
huui uueldes thu thinera modar manno liobosto gisidon how wanted you your mother man dearest do sulica sorga (Hel 821) such worries ‘How could you cause your mother such worries, dearest sun?’
thea liudi stodun umbi that helaga hus (Hel 101) the-PL people-PL stood-PL around the holy house ‘The people stood around the temple’
Gabriel bium ic hetan Gabriel am I called ‘I am called Gabriel’
However, the straightforward analysis of OS as a V2-language faces a number of methodological and empirical problems which shall be discussed briefly below. One of them concerns the proper differentiation of root vs. embedded clauses in the data from OS. Bearing in mind that the system of complementizers is not fully grammaticalized at that period in the sense that subordinating conjunctions are not distinguishable from adverbials and pronouns in each case, a high number of sentences can be interpreted both as root and dependent clauses. This mainly concerns some adverbial clauses (10) as well as relative clauses (11):
Aspects of word order and information structure in Old Saxon
uuarun im an Nazarethburg thar the neriondio were they-DAT in Nazareth there the saving Krist uuohs undar them uuerode (Hel 782) Christ grew among the people ‘They were in Nazareth, where the saving Christ grew up among the people.’ ‘They were in Nazareth. The saving Christ grew up there among the people.’
Sie ni uueldun it thoh farlaten ac hetun thar they NEG wanted-PL it yet leave-off but ordered-PL there ledien ford en uuif for themu uuerode guide away a woman in front of the people thiu habde uuam gefrumid (Hel 3840) the-FEM sin sin comitted ‘They did not want to give up but instead they ordered to bring a woman in front of the crowd, who had comitted sin.’ ‘They did not want to give up but instead they ordered to bring a woman in front of the crowd. She had comitted sin.’
Synchronically, the position of the finite verb cannot be used to distinguish root from subordinated clauses. There are clauses which are unambigliously subordinated ones but exhibit post-verbal arguments like subjects (12) or objects (13): (12)
[thit sculun gi uuitan alle iungaron mine = ‘You shall know it all, my disciples’] huand iu fargeben habad uualdand thesaro uueroldes because you forgiven had ruler this-GENworld-GEN (Hel 2434) ‘(…) because the ruler of the world had forgiven you’
[habda them heriscipie herta gisterkid = ‘He strengthened the heart of their army’] that sie habdon bithuungana thiedo gihuilica (Hel 55) COMP they had defeated folk every ‘(…) so that they had conquered every nation.’
Moreover, matrix clauses show much more variation with respect to verb placement than is expected in a pure V2-language. In OS, the verb is often found in initial position (14) as well as in later positions (15)2–(16): (14)
habda im the engil godes al giuuisid had them the angel god-GEN everything shown torhtun tecnun (Hel 427) bright-DAT.PL signs-DAT.PL ‘The angel of God had shown them everything with bright signs’
thuo hie sia an is era antfeng thuru hluttran then he her in his protection took through clear hugi (Hel 5619) mind ‘Then he took her under his protection due to his clear mind.’
nu ik theses thinges getruon now I this-GENthing-GEN trust ‘Now I trust in this matter’
Patterns like (14)–(16) are frequent in the Heliand. Therefore, it seems unjustified to exclude them from the analysis of OS syntax as isolated exceptions picturing archaisms3. The attested data induce most notably two questions: First, how can we explain this confusing coexistence of different word order patterns syntactically, and second, how are the constituents in patterns organized which do not display V2 in the surface. According to Kiparskys (1995) well-known analysis, Old Germanic syntax just started to establish the category of CP. As a consequence, the shift from adjunction to embedding by means of subordinating comple0 mentizers in C took place and so did V-to-C-movement in main clauses as well. Since this syntactic change is supposed to proceed in steps, early attested data may still show variations between CP and S main clauses as reflexes of this development. In fact, the differences in Old Saxon word order shown above can be best explained by assuming this variation, this means that the C-projection is not fully established yet; and therefore V-toC-movement did not occur in all cases. Bearing in mind an analysis like this, still the question arises how to explain the order of elements in non-V2 main clauses since it seems to be dissatisfying just to notice that the finite verb could not move to the C0 slot due to the fact that these clauses have no CP at all. Furthermore, still an
Aspects of word order and information structure in Old Saxon
asymmetry between V1 and V-late main clauses occurs. Although in particular V1 clauses can be explained as cases of V-to-C-movement with an empty focus-operator in SpecC4, I want to investigate alternatively whether V1 and V-late in independent declaratives may be attributed to information-structural conditions instead of confining on syntactical factors only. Taking as a starting point some influential ideas in early philological work on the function of different word patterns (Ries 1880, Delbrück 1911) and some more recent approaches on this matter (cf. Donhauser and Hinterhölzl 2003), I will pursue an approach to discerning pragmatic or information-structural factors which account for variation in OS verb placement.
4. A discourse-pragmatic analysis 4.1.
Some basic assumptions
Recent work on the role of information structure and discourse organization for the explanation of different word order patterns in Old High German (Hinterhölzl and Petrova 2005) reveals some regular correlations between pragmatic factors and verb placement in early Germanic languages. In this paper I will show that OS word order also depends to a large extent on pragmatic rules which can be consistently described in terms of information structure. Information structure itself is a complex phenomenon which comprises at least two aspects concerning the organization of an utterance. On the one hand, the utterance can be divided into a topic and a comment about this topic (topic-comment structure), on the other hand, the most relevant information in the utterance makes up the focus as opposed to the background (focus-background structure) (Lambrecht 1994: 117ff and 206ff).5 The topic-comment structure affects the predicational seperations of an utterance and is deeply connected with the concept of aboutness in the sense of Hockett (1958) and Reinhart (1981) or with the notion of familiarity in the sense of Gundel (1988). Usually, the topic expression refers to an entity which is given or presupposed in the particular context. By contrast the level of focus-background is defined in terms of the speaker’s attitude towards the informational relevance of sentence parts. Novelty often correlates with the referent of the focus expression, but also presupposed elements can be emphasized under certain conditions. The pragmatic categories of topic and focus should not be understood as complementary ones. However, there are conditions which prevent the
establishment of a topic-comment structure in an utterance as is the case in so-called thetic sentences (Drubig 1992, Sasse 1987). On the level of focus-background structure these sentences are analysed as all-focus structures, i.e. the focus domain comprises the entire utterance (cf. Lambrecht 1994: 137ff and 233ff). For the identification of the pragmatic categories of topic and focus in a historical text, features that are grammatical correlates of these categories are most important. So e.g. topics are assumed to be expressed by anaphoric means while focus correlates with prosodic prominence which in the data analysed here may be detected by virtue of rhyme, rhythm and stress. Furthermore, factors of discourse organization play an important role in word order regularities. I refer to the basic assumptions in two of the most accepted models of discourse analysis, the Rhetorical Structure Theory (RST) by Mann and Thompson (1988) and the Segmented Discourse Representation Theory (SDRT) by Asher and Lascarides (2003)6. As a basic principle both share the assumption that utterances in discourse hold certain logical relations among each other yielding a hierarchical structure in the discourse. Although the catalogue of discourse relations is subject to extended discussion, two main types of relation can be broadly distinguished: (i) (ii)
An utterance can hold a dependency relation to a previous one in supplying more information on it; Two utterances can display no dependency relation among each other, i.e. they belong to the same level of discourse organization.
According to Asher and Lascarides the first type of relation is called subordination prototypically instanciated by ‘elaboration’, while the second one is called coordination with ‘narration’ as its prototypical representative. Each structural unit within a discourse is associated either to coordination or to subordination. The types of discourse relations are assumed to be linked to formal correlates such as prosody, tense usage, anaphoricity, linear word order etc. which are governed by parametrical variation. Comparing the discourse function and the information structure of an utterance, one can realize that these correlate in some important points. Based on the definition of the subordinating type of linking, it should be evident that elaboration involves a topic-comment structure. By contrast to that, sentences of the coordinative type help to develop the main line of the narration. Following these considerations, one can assume that sentences of the coordinative type open a new sequence or signal a change in the
Aspects of word order and information structure in Old Saxon
narrative setting but do not necessarily display a topic-comment structure. In other words, they share properties of thetic judgements. In a recent application of this model to data from OHG Hinterhölzl and Petrova (t.a.) and Petrova (2006) point at a correlation between the word order and the discourse function of a sentence. Their suggestion especially concerns the placement of the finite verb. To summarize, verb first patterns are said to appear in sentences opening a new text sequence or denoting a change in the narrative setting, thus correlating with coordinating discourse relations. On the other hand, the verb second pattern occurs in contexts elaborating on an already established discourse referent and is therefore prototypical for subordinating discourse relations. In the following section, this model will be applied to OS in order to look whether properties of discourse organization correlate with verb placement in OS as well.
Subordinating Discourse Relations in the OS Heliand
Here, we investigate sentences which provide more information on a discourse-given or presupposed referent in supplying additional information on it. Due to their pragmatic nature, we expect these sentences to expose a topic comment division, i.e. to qualify as categorical judgements. In sentences establishing such kind of relations, the V2 pattern dominates. The preverbal constituent in this case is a referential expression conveyed in different forms such as full NPs (17), pronouns (18) and elliptical material (19). These correlate with the topic referent: (17)
[The three Magi enter the place of the birth of Christ] uurekion] fellun te them kinde an kneobeda TOP[thea the-PL strangers fell-PL to the child on knees (Hel 671) ‘The strangers fell on their knees in front of the child.’
[An old woman named Anna came into the temple] [siu] habde ira drohtine uuel githionod te thanca TOP she had her Lord well served to gratitude (Hel 505) ‘She had served her Lord well in gratitude.’
[‘There was an old man’] hugi [that uuas fruod gomoi] TOP[proi] habda ferehtan this was wise man had experienced mind [proi] uuas fan them liudeon Leuias cunnes (Hel 73) TOP was from the people Levi-GEN dynasty-GEN ‘He was a wise man and had much experience and was descended from the dynasty of Levy.’
In most of these cases, the topic is the grammatical subject, however, an object which functions as the aboutness topic is also found in the preverbal position (20): (20)
[‘His name was Simon’] habda giuuisid uualdandas craft langa TOP[Im] led rulers power long-ACC him-DAT had huila (…) (Hel 469) time-ACC ‘The power of the ruler had led him for a long time (…)’
The pattern in (17)–(20) is a prototypical topic comment structure with the topic in the left periphery and the comment following it. The structure Top+V2 is therefore firmly associated with the type of subordinating linking. Syntactically expressed these OS data show instances of V-to-C-movement and SpecC filled by a topic expression for discourse internal reasons. However, besides V2-clauses with a topic expression in the preverbal position, we find a number of main clauses establishing a subordinating discourse relation to the previous context which exhibit a word order which is different from the one described above. More precisely, these sentences involve more than one constituent to the left of the finite verb, see (21)7 and (22).8 (21)
endi siu an iro breostun forstod and she in her heart understood ‘In her heart she realized’
thar ina thiu modar fand sittean under them there him-ACC the mother found sit-INF among the gisidea (Hel 818) crowd ‘The mother found him there sitting in their company.’
Aspects of word order and information structure in Old Saxon
In (21) and (22) both arguments and adjuncts are placed in front of the finite verb. Looking at the informational status of the preverbal elements, we notice that they are clearly arranged according to their pragmatic status9. In both sentences, one of the preverbal constituents is the aboutness topic of the utterance, cf. siu in (21) and ina in (22). The remaining constituents left to the verb are also contextually given elements (e.g. thar, thiu modar in (22)) or elements related to already established entities (an iro breoston in (21)). In other words, all preverbal elements share the property of being presupposed. Therefore they constitute the background of the utterance. This is in clear contrast to the information represented by the verb, which is new. This leads to the assumption that in all verb-later cases above, the finite verb seperates background and focus material in the utterance.10 Further evidence for a postverbal position of focus is provided by the syntactic placement of new referents (23) or given entities which are highlighted for reasons of contrast or emphasis (24): (23)
fan adrom Thar imu tegegnes quam new[en idis there him-DAT towards came a woman from other thiodun] (Hel 2984) people ‘There a woman from another region came towards him.’
endi thar an them uuiha afstod emphasis[mahtig barn and there at the temple stayed powerful child godes] (Hel 797) god- GEN ‘And the mighty son of God stayed in the temple.’
It is important to note that in most of the verb-late main clauses discussed above an adverbial is placed in the initial position. These adverbials bear referential properties as indications for time, place and manner. Therefore they function as frame elements belonging to the background of the utterance. To sum up, syntactic structures occurring in sentences of the subordinating type and showing a verb-late pattern can be represented in the following scheme given in (25). (25)
Background – Vfin – Focus
The background domain is structured in the way that the frame elements precede the aboutness topic and this is followed by the remaining presupposed elements if any are given. From the perspective of pragmatic principles motivating the syntactic placement of sentence constituents in OS, we can also provide an explanation for structures being ambiguous between root and embedded clauses, cf. (10)–(11) above. Both sentences serve to provide more information on a previously established referent, i.e. they are prototypical cases of subordination in discourse. Quite consequently, they expose the pattern in (25). In this sense, it is not appropiate to speak about parataxis and hypotaxis in our modern understanding of these terms; instead, utterances in OS often obey the rule of organizing the material according to pragmatic principles. Since patterns with the finite verb in a later position than the second one are too frequent to be exceptions and some rules concerning their contextual appearance as well as their linear word order can be identified, one can assume that they reflect an older stage in which CP was not developed yet so that V-to-C-movement cannot take place; the finite verb is expected to remain in its base position. Above we showed that the elements in these verb-late patterns are ordered with regard to their information structural status rather than exclusive of their syntactical status, although word order is surely based on it. Focus elements are assumed to be extraposed. Even though we stated that these structures are frequent, one cannot ignore the fact that the V2-pattern is the more common one in sentences of the subordinated type of discourse relations. Here the preverbal constituent represents the aboutness topic regularly. This is consistent to the fact that subordinate discourse relations add more information about an already established discourse referent, which functions then as the aboutness topic of the utterance. Therefore these patterns depict prototypical topiccomment strutures. All in all, these observations lead to following shemes representing patterns occurring in sentences of the subordinating type: (26)
a. TOP=BG[NP] COM=FOC[Vfin …] b. BG[(Frame)[Top](XP ...)] FOC[Vfin(XP)]
However, if two utterly different patterns, namely a prototypically V2structure involving V-to-C-movement (26a) and a rather pragmatic organized structure analysed as some kind of relict without CP (26b), constitute subordinating discourse relations, the question arises how their discourse internal counterpart – the coordinated discourse relations – is structured.
Aspects of word order and information structure in Old Saxon
Coordinating Discourse Relations in the OS Heliand
Also in instances belonging to the coordinating type of linking two particular patterns occur. Here, structures in which an adverbial particle precedes the verb in second position (27) and V1-structures (28) prevail. Note that all sentences have an overt subject in postverbal position: (27)
Tho uuard that heuencuninges bodon then became the-ACC heaven.king-GEN messenger-ACC harm an is mode (Hel 159) sorrow-NOM in his heart ‘Then the messenger of the king of heaven felt sorrow in his heart.’
Bidun allan dag that uuerod for them wait-PRT.PL all day the crowd in front of the uuiha (Hel 174) temple ‘The people waited all day long in front of the temple.’
Both sentences establish a new situation which is settled on the main line of narration. For all that I can tell they seem to be functional equivalents. The sentence in (27) opens a new fit. In an analogous way, (28) opens a new sequence involving a change of place and participant: whereas the preceding context informed us on the events concerning Zacharias inside the temple, the narration switches to a characterization of the people waiting outside. As to the quantitative distribution of these patterns, cases of tho+V2 occur more than twice as often as V1 in this respect. Next to the adverbial particle tho, we find V2 clauses associated to coordinating discourse relations with other adverbial elements in the initial position such as so (29): (29)
So uurogdun ina mid uuordun uuerod Iudeono so accused him-ACC with words crowd Jews-GENc thurh hotean hugi (Hel 5245–46) through full of hatred mind ‘So the crowd of Jews accused him with words full of hatred thoughts.’
Furthermore, there is a group of verbs which inherently have an eventreporting status and focus the entire proposition. As instances of such
predicates we interpret verba dicendi11, motion verbs and verbs of sensual perception, the latter especially in inchoative meaning. Hinterhölzl and Petrova (t.a.) observe that in Old High German these verbs regularly occur in V1 or tho+V2-patterns respectively. It is argued that these lexemes signal a change in the overall deictic orientation of the situation with respect to place, time and participant. Therefore, instances with these verbs are analysed as typical cases of all-focus sentences inducing a change in the narrative situation. In OS, the situation is quite similar in sentences with verba dicendi (30), motion verbs (31) or perception verbs (32): (30)
tho het he sie an thana sid faran then told he them to that way go ‘He then instructed them to go on their journey.’
quamun managa Iudeon an came-PL many Jews into ‘Many Jews came to the great hall.’
uuissun that thoh managa liudi aftar them landa knew-PL that yet many people in the land (Hel 855) ‘But many people all over this country knew that.’
gastseli (Hel 2736) hall
The placement of the verb in the initial position to highlight the entire content of the sentence is assumed to be common to all Germanic languages (Fourquet 1974: 316; cf. also Ries 1880). In other words that the verb opens a maximal focus domain constitutes a very old pattern. Surface-V1-declaratives allow for two types of explanations. On the one hand, one may assume that they derive from general V-to-C-movement with a SpecC slot remaining empty or hosting a silent focus operator, and synchronically nothing contradicts this point of view. On the other hand, some functional contexts were analysed where verb movement to the left periphery seems to have taken place much earlier than overall V-to-Cmovement. In general, imperatives, negated sentences and wh-interrogatives are closely associated to verb fronting in Old Germanic (see e.g. Eyþórsson (1996: 111) and Axel (2007: 52f)). In this paper we do not want to discuss if CP itself or the movement of the finite verb to C0 is optional in Old Germanic. However, it is crucial for our aim to perceive verb fronting under certain circumstances as mentioned above since all of them are related to the purpose of focusing the proposition of the utterance.
Aspects of word order and information structure in Old Saxon
V1-patterns in OS Heliand always focus the entire utterance. For all we know structures with a fronted verb associated to focus operations are much older than general V-to-C-movement with additional XP-fronting. In accordance with V-late structures discussed above V1 declaratives seem to represent some kind of reflex of a prehistorical stage of language. On the other hand, tho+V2 corresponds to the need for a lexically filled SpecC slot in declarative main clauses like in modern V2- languages. According to this, all-focus structures in OS are most often found with the adverbial tho, which gradually looses its temporal semantics and develops to a text-structuring discourse-particle. It signals the continuation of the narration and guarantees progress in discourse. The structure tho+V2 is the most wide-spread verb second pattern in the Heliand by far. Besides tho as a focus particle we also find tho functioning as a referential temporal adverbial element in the middlefield which can be topicalised optionally. Therefore, the patterns (i) [[thoParticle] FOC[Vfin ...]] and (ii) BG[thoFrame] FOC[Vfin ...] cannot always be distinguished with certainty. Less frequently, other adverbs as so, nu etc. also function as discourse particles in initial position of an all-focus V2-sentence, cf. (29) above. However, it is interesting to note, that besides adverbs, also the subject can occupy the initial position of V2 sentences associated with coordinating discourse relations. In this case, the subject represents the focus exponent of the sentence, as is illustrated in (33). There are several criteria supporting this analysis: 1) the expression is stressed by virtue of its position in the metrical structure of the verse line; 2) the expression bears the alliteration rhyme (alliterative syllables are given in bold type for clarity); 3) the expression conveys new (or relevant) information. All these criteria apply for the initial element Iohannes in (33); the sentence is analysed as representing a coordinating relation because it introduces a new episode within on the main line of narration: (33)
[A chain of five coordinate conjuncts with V1, which promote the progress of narration as an intermediate sequence] geng thes geres gital Iohannes quam went that-GEN year-GEN quickly John came an liudeo lioht lik uuas im sconi to mankind-GEN light body was him beautiful (Hel 198) ‘The year passed by quickly. John came to the light of mankind. He was beautiful.’
However, while V2-patterns with an initial discourse particle, especially tho, are extremely frequent, patterns like the one in (33) occur very rarely. Summarizing our findings concerning sentences attributed to the coordinating type of linking we can conclude that the following patterns appear in the OS Heliand in this function: (34)
a. b. b.’
[Vfin ...] [Particle] FOC[Vfin ...] [ [NP] [Vfin ...]] FOC Fexp FOC
As SpecC is surely sensitive to focus elements, it does not surprise that these can be found in this position. The quantitative distribution of these patterns indicates that the verb second pattern with an initial focus particle (34b) exceeds by far. Bearing in mind the overall development of word order in continental WestGermanic, one can assume that the pattern in (34b) replaces the one in (34a).
5. Conclusion It is assumed that the confusing variation of linear word order patterns in Old Saxon Heliand is due to the mixture of two different grammars; however, our text reflexes “the elimination of the variation of CP and S main clauses” (Kiparsky 1995: 162) whereas a good part of main clauses construed as simple S are synchronically reanalysed as CP with a vacant SpecC slot, i.e. V1-declaratives. This point of view is supported by the well established assumption that language change in general takes place gradually and therefore reflexes of an older stage are still found later in changed environments with the possibility of reanalysing “old” properties with regard to the new conditions. The scenario supposed carries out as follows: In a prehistoric language period, say to simplify matters in Protogermanic, only simple S clauses occur. Here the finite verb appears mostly in its base position following all arguments. Moreover, subordinate clauses are adjoined, and linear word order is not only related to the syntactic properties of constituents but also to information structural factors. Background elements are organized preverbally due to their salience and focus expressions are extraposed (see as
Aspects of word order and information structure in Old Saxon
a reflex of this protogermanic property the postverbal focus position in OS, e.g. (12), most likely (13), (23) and (24)). Furthermore, in order to focus the whole utterance V1 appears connected to an overt or covert focusoperator triggering verb raising and bearing one of the following features [+wh], [+imperative], eventually [+negation] and, for opening new contents in a discourse, [+coordinative]. This leads to the assumption of a leftperipheral operator-position. However, once CP is established in Germanic, this operator position is reanalysed as SpecC which then opens successively for other elements than focus-operators only; first and foremost SpecC allures the aboutness topic which is predestinated for this position due to the fact that it occurred always preverbally representing the prototypical background element. And in fact, the presumption that SpecC becomes sensitive for any XP gradually is confirmed by the observation that other elements occupying the SpecC slot except for the so-called discourse particle tho and the aboutness topic are extremely rarly found in Old Saxon. All in all, the OS Heliand mirrors a period of syntactic change; this is the establishment of CP in general associated to operations like overall Vto-C-movement, embedding of subordinate clauses in alliance with the development of the category of complementizers and – chronologically later – the generalisation of SpecC hosting any constituent.
6. Acknowledgements The present paper relates to the information-structural analysis provided for my 2006 master thesis ‘Verbstellung und Informationsstruktur im Altsächsischen’, Humboldt Universtität Berlin. The thesis was supervised by the research project SFB 632/B4 ‘Die Rolle der Informationsstruktur bei der Herausbildung von Wortstellungsregularitäten im Germanischen’. I am thankful to the principle investigators of this project Karin Donhauser and Roland Hinterhölzl for giving me the chance to complete my research on OS. I also thank Michael Solf for helpful comments concerning this paper. I am grateful to Svetlana Petrova for help and fruitful discussions during writing this paper.
Cf. Bästlein (1991: 213ff). As Bästlein points out, the distribution of initial capitals differs in the single manuscripts. Bästlein challenges the traditional editorial usage to subdivide the Heliand into fits according C only, because only this youngest manuscript structures the text in this way. 2. (15) cannot be analysed as a temporal subclause as it is clearly divided from the previous sentence and is continued as follows: (…) so im is herro gibod ‘as his master told him’. 3. Eyþórsson (1996: 116) describes OHG Vlate declaratives as “isolated (…) examples reflect[ing] a more archaic stage of Germanic” and therefore excludes them from his analysis. OS he states to be regularly V2. 4. E.g. Eyþórsson (1996: 123) considers this solution. 5. For a more detailed discussion see also Petrova and Solf (in this vol.). 6. I follow the model presented in Hinterhölzl and Petrova (t.a.) 7. (21) is definitely not conjoined to a subordinate clause. 8. See also (15) an (16) above. 9. One can be inclined to explain the order of the preverbal elements e.g. in (22) as a realisation of Behaghel’s law of placing longer constituents after shorter ones (Behaghel 1932: 6). However, we think that principles of linear order stated by Behaghel are a correlate of pragmatic distinctions: in the cited OS data the preverbal elements are clearly ordered with respect to their salience. 10. Again, so-called long constituents like PP-complements may be extraposed independently of their information structural status. 11. See also Önnerfors (1997: 120ff).
Primary Text [Hel]
Heliand. Titelauflage vermehrt um das Prager Fragment des Heliand und die Vaticanischen Fragmente von Heliand und Genesis. Ed. by E. Sievers, Paderborn 1878.
References Asher, Nicholas and Alex Lascarides 2003 Logics of conservation. Cambridge: Cambridge University Press. Axel, Katrin 2007 Studies on Old High German Syntax: Left Sentence Periphery, Verb Placement and Verb-Second. Amsterdam/Philadelphia: John Benjamins.
Aspects of word order and information structure in Old Saxon
Bästlein, Ulf Christian 1991 Gliederungsinitialen in frühmittelalterlichen Epenhandschriften: Studie zur Problematik ihres Auftretens, ihrer Entwicklung und Funktion in lateinischen und volkssprachlichen Texten der Karolinger- und Ottonenzeit. Frankfurt a.M.: Peter Lang. Behaghel, Otto 1897 Die Syntax des Heliand. Wien: Tempsky. 1932 Deutsche Syntax. Band IV. Heidelberg: Winter. Delbrück, Berthold 1911 Germanische Syntax II: Zur Stellung des Verbums. Leipzig: Karl Teubner. Donhauser, Karin and Roland Hinterhölzl 2003 Die Rolle der Informationsstruktur bei der Herausstellung von Wortstellungsregularitäten im Germanischen. In Informationsstruktur: Die sprachlichen Mittel der Gliederung von Äußerung, Satz und Text. Finanzierungsantrag 2003–2007, ed. Caroline Féry, 173–193. Potsdam: Universitätsverlag. Drubig, Bernhard 1992 Zur Frage der grammatischen Repräsentationen thetischer und kategorischer Sätze. In Informationsstruktur und Grammatik. Linguistische Berichte, Sonderheft 4, ed. Joachim Jacobs, 142–195. Opladen: Westdeutscher Verlag. Erickson, Jon 1997 Some observations on word order in Old Saxon. In Sprache im Fokus, eds. Christa Dürscheid, Karl Heinz Ramers and Monika Schwarz, 95–105. Tübingen: Max Niemeyer. Eyþórsson, Þórhallur 1996 Functional Categories, Cliticization, and Verb Movement in the Early Germanic Languages. In Studies in Comparative Germanic Syntax. Vol. II, eds. Höskuldur Thráinsson, Samuel D. Epstein and Steve Peter, 109–139. Dordrecht/Boston/London: Kluwer Academic Publishers. Fourquet, Jean 1974 Genetische Betrachtungen über den deutschen Satzbau. In Studien zur deutschen Sprache und Literatur des Mittelalters, eds. Werner Besch, Günther Jungbluth, Gerhard Maissburger and Eberhard Nellmann, 314–323. Berlin: Erich Schmidt. Gallée, Johan Hendrik 1993 Altsächsische Grammatik. Tübingen: Max Niemeyer. Gundel, Jeanette 1988 Universals of topic-comment structure. In Studies in Syntactic Typology, eds. Michael Hammond, Edith A. Moravcsik and Jessica R. Wirth, 209–239. Amsterdam: John Benjamins.
Hinterhölzl, Roland and Svetlana Petrova to app. Rhetorical Relations and Verb Placement in Old High German Tatian. In Salience: Multi-disciplinary perspectives on its function in discourse, eds. Christian Chiarcos, Berry Claus and Michael Grabski. Berlin: de Gruyter. Hockett, Charles F. 1958 A Course in Modern Linguistics. New York: Mac Millan. Karttunen, Lauri 1976 Discourse Referents. In Syntax and Semantics 7: Notes from the Linguistic Underground, ed. James McCawley, 363–385. New York: Academic Press. Kemenade, Ans van 1987 Syntactic Case and Morphological Case in the History of English. Dordrecht: Foris. Kiparsky, Paul 1995 Indo-European Origins of Germanic Syntax. In Clause Structure and Language Change, eds. Adrian Battye and Ian Roberts, 140–169. New York/Oxford: Oxford University Press. Lakoff, George 1976 Pronouns and Reference. In Syntax and Semantics 7: Notes from the Linguistic Underground, ed. James McCawley, 275–335. New York: Academic Press. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus and the Mental Representation of Discourse Referents. Cambridge: Cambridge University Press. Lenerz, Jürgen 1984 Syntaktischer Wandel und Grammatiktheorie. Tübingen: Max Niemeyer. Mann, William and Sandra Thompson 1988 Rhetorical Structure Theory: Toward a functional theory of text organization. Text 8 (3): 243–281. Önnerfors, Olaf 1997 Verb-erst-Deklarativsätze: Grammatik und Pragmatik. Stockholm: Alqvist & Wiksell. Petrova, Svetlana and Michael Solf this vol. On the Methods of Information-Structural Analysis in Historical Texts: A Case Study on the OHG Tatian. Rauch, Irmengard 1992 The Old Saxon Language. New York: Peter Lang. Reinhart, Tanya 1981 Pragmatics and Linguistics: An Analysis of Sentence Topics. Philosophica 27: 53–94.
Aspects of word order and information structure in Old Saxon Ries, John 1880
Die Stellung von Subject und Prädicatsverbum im Hêliand. Strassburg/London: Karl J. Trübner. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25: 511–580.
alliteration, 383 analogy, 136, 146, 194, 199 anaphoric use, 60, 263 Behaghel’s Law (Gesetz der wachsenden Glieder), 51–53, 58 clause type, 7, 194, 253 adverbial clause, 153, 228, 230, 327, 372 complement clause, 26, 228, 230, 252, 371 wanda-clause, 323, 339, 341–343, 356 copular construction, 173, 224, 242, 255, 266–267, 273 demonstrative pronoun, 60, 62, 73, 91, 97, 99–101, 104–107, 111– 113, 126, 145, 216, 228–229, 309, 324, 353–355, 357 directionality, 194 discourse adverb, 8, 91, 94, 96–98, 112–113 discourse marker, 109, 112, 310, 311, 317 fronting of, 7, 20, 22, 28, 306, 311 discourse particle, 105, 112–113, 383–385 discourse relation, 228, 376–378, 380–381, 383 discourse status given (discourse-old), 2, 8–10, 21–22, 48, 51, 59–61, 73, 83, 91, 96–98, 100–103, 108, 133–141, 145–148, 152, 200, 204, 238, 240, 243, 252, 258–260, 263, 265, 268–270, 284–287, 298, 306, 311, 325, 332, 353, 375, 377, 379–380
new (information), 8, 10, 21, 48, 51–52, 59, 62, 67, 72–75, 76, 78– 79, 83, 96, 100–101, 133–136, 139, 141–142, 145–147, 149–153, 238–241, 243, 258, 260, 265–274, 284, 298, 327, 332, 368, 379, 383 double base hypothesis (DBH), 4, 46, 49 ellipsis, 8, 96, 191–197, 200–208, 210–212 expletive es, 19, 30, 33–34, 36 focus ambiguity, 274 focus background structure, 332, 343, 357, 375–376 focus type, 271–274 contrastive focus, 52, 141, 149, 150, 153, 239, 243, 273, 276, 332 emphatic focus, 323, 325, 335– 337, 357–358 narrow focus, 73–74, 86, 263, 271 new information focus (presentational focus), 52, 149– 150, 266, 268–270, 273–274 sentence focus (thetic judgement), 24, 152, 302, 342, 356, 367, 377, 382–383 structural focus, 52, 323, 325, 329, 333, 335–337, 358 VP-focus, 149, 264, 269, 274 gapping backward gapping, 8, 191, 193– 194, 198–200, 202–203 forward gapping, 8, 191, 193– 194, 196, 198–200, 204 grammar competition, 45
head parameter, 46 hyperbaton, 11, 323, 325–329, 333, 335–336, 356 information-structural category (IScategory), 6–9, 11, 45, 49, 51, 58, 64, 132–133, 138, 142, 144–145, 154, 264 background, 3, 9, 51–52, 122, 133, 140, 144, 148, 238, 243, 245–246, 252, 261, 263, 265, 268–269, 273, 283–286, 299, 304–306, 309, 314, 324, 332, 339, 342, 343, 357, 375–376, 379–380, 384–385 contrast, 10, 21–22, 52, 73, 141, 149, 252, 258, 272, 274, 309, 343, 347–351, 357, 375, 379 focus, 3–7, 20, 23, 48, 51, 58, 67, 73–74, 86, 97–98, 132–133, 140– 142, 144, 148–153, 200, 224–225, 235, 238–246, 252, 259, 261–264, 268–274, 284–286, 310, 323–325, 327–329, 331–333, 336–368, 342–347, 350, 352, 357, 361, 375–376, 379–385 topic, 3–7, 22, 30, 63, 73, 111– 112, 137–140, 199–200, 269, 284–287, 302–307, 309–312, 314–315, 317, 323–325, 327, 329, 332–333, 338–339, 343, 346, 349, 352–354, 356–357, 375–380, 383 implicature, 192–193, 197, 287 language change, 1, 63, 67, 74, 78– 81, 87, 122, 315, 384 change in the external language, 86 change in the internal language, 79 language (stage)s, Early Middle English (EME), 4, 45, 62–63, 94, 113
Early New High German (ENHG), 61, 72, 274 Greek, 8, 121, 165, 191, 196–198, 200–201, 207–209, 212, 328 Hittite, 8, 191, 194–195, 197–198, 200, 202, 209–210 Latin, 8–11, 17–19, 48, 121–131, 161, 163–164, 191, 195, 198, 200, 203–205, 211, 227, 229, 253, 325–329, 331–339 Middle English (ME), 4, 8, 62, 67, 83, 91, 94, 96–97, 99, 112–113, 223, 276 Modern German, 10, 285, 295, 300–301, 332, 337, 341, 344, 349, 357 Modern Icelandic, 24, 68–71, 82– 83, 86–87 Old English (OE), 3–5, 7–8, 46– 47, 51, 54, 62–63, 67, 71–73, 91– 92, 94, 96–99, 101, 109–113, 121, 143, 165, 223, 251–252, 267, 316, 369–370 Old(er) Icelandic (OI), 2, 4–5, 7, 46, 51, 62, 67–72, 74–75, 78, 82– 84, 86–87 Old High German (OHG), 2–5, 17–18, 46, 51–52, 58, 122–127, 142–144, 148, 154, 161, 164–178, 180–181, 211, 223–229, 237, 244, 251–268, 270–274, 281–283, 302– 304, 315–316, 323–326, 332, 335–343, 348–353, 356–358, 369–370, 375–377, 382 Old Saxon (OS), 2, 63, 122, 143, 181, 367–372, 374–378, 380–385 Sanskrit, 8, 191, 197–198, 200, 202, 208 Yiddish, 52, 72, 273–274 Latin influence, 5, 123, 227, 229, 230 deviation from Latin, 125, 226– 227, 229, 235 syntactic loan, 124, 131
Subject index left dislocation, 137, 147, 300 left periphery, 5, 6, 7, 11, 82, 106, 111, 113, 325, 357, 378, 382 multiple foci, 270, 273–274 Notker’s Anlautgesetz, 181 null object, 203, 208–212 null subject (pro), 26, 28–29, 32, 295 object deletion (see null object) object shift, 63, 82–83, 83, 86–87 operator movement, 19, 22, 36 palaeographic clue, 8 accent diacritics, 8, 143, 165–176, 179, 181 capitalization, 141, 143, 152, 180, 181, 370 (text) initials, 180–181 punctuation, 161, 180–181, 370 word separation, 8, 161–162, 177–179, 181 phase impenetrability condition (PIC), 57 phase, 53, 57–58, 61, 63 phonological phrase, 50, 56 prosodic constraint (prosodic condition), 52–53, 57–58, 61–62 prosody, 1, 6, 8, 63, 143–144, 168, 170, 180–181, 376 quotation syntax, 223, 227–230, 235, 241, 244 Rahmenbildung (see sentence bracket) reconstruction, 198, 208, 211 remnant movement, 71 rhythmic activation principle, 50 right periphery, 7, 10, 251–252, 268, 270, 273, 342–343
Satzklammer (see sentence bracket) scrambling, 8, 59, 70, 72, 73, 107, 108, 109, 110, 226, 260 sentence bracket, 5, 7, 10, 31, 62, 223, 317, 330–331, 336 exbraciation (see extraposition) extraposition, 5, 10, 32, 45, 48– 49, 224, 235, 245–246, 251–252, 268, 270–271, 273, 297–299, 331, 338 sloppy identity, 197, 201–202 specificity, 59, 72–73, 106–108, 273 stylistic fronting, 7, 19, 22–25, 29– 30, 35–37, 68, 82 subject position, 32, 103–104, 106– 107, 111 texts Heliand, 11, 143, 181, 369–370 374, 377, 381, 383–385 Isidor, 7, 10, 17–19, 22, 123–125, 162, 224–227, 238, 240, 246, 252, 315–316, 318 Notker’s Consolatio, 7, 9–11, 17, 21, 27, 60, 62, 123, 143, 164 – 168, 170–171, 175, 181, 282, 293, 317, 323–326, 330–333, 335–339, 341, 343, 345, 347, 349, 350–351, 353–354, 356–358 Otfrid’s Gospel Harmony (Evangelienbuch), 10, 60, 123, 125, 143, 164, 166, 168–172, 175–176, 180–181, 281–283, 285, 287–288, 290, 293, 296–297, 299, 301–302, 304, 306, 310, 312–313, 315–318 Tatian, 9–10, 17–18, 20–22, 27, 33, 35, 48, 51, 123–127, 131, 143–144, 150, 154, 166, 171–173, 175– 176, 178–181, 226–228, 252–253, 256, 258, 264, 271, 273, 353
topicalization, 57–58, 68, 102, 139, 317 topic comment structure, 133, 138, 147, 284, 324, 339, 375, 376, 377 topic type aboutness topic, 21, 140, 151, 306, 333, 346, 353, 372, 378–380, 385 continuous topic, 325, 339, 353 contrastive topic, 314, 325, 351 discourse topic, 34, 325 empty topic (topic drop), 33–34 familiar topic, 138–140, 146, 325, 375 frame setting topic, 140, 341 hanging topic, 137, 147 shifting topic, 325 typological field, Mittelfeld (middle field), 8, 10, 22–26, 29, 50, 54, 56, 58, 63, 226, 295, 297, 327 Nachfeld (see exbraciation, extraposition) Vorfeld (prefield), 17–36, 138, 147, 299–301, 303–305, 307–308, 310–312, 370, 375, 378, 382–385 universal base hypothesis (UBH), 50, 52 verb movement, 67, 70, 82, 111, 382 verb projection raising (VPR), 236, 237, 251, 255, 273 verb raising (VR), 251, 255, 273, 385 V-to-C-movement, 8, 19, 191, 200, 374–375, 378, 380, 382–383
verb placement residual verb second, 3 verb early, 223, 229, 230, 231, 243, 244 verb first (V1), 19, 26–36, 227, 281, 291, 293, 299, 301–302, 310, 312, 368, 375, 377, 381–385 verb last (Vend), 253, 255–256, 259, 263–264, 273, 293–294, 312, 313 verb late, 296–297, 312–313, 360, 379–380 verb second, (V2), 17–18, 21, 24, 26–30, 32–37, 45, 63, 101, 223– 224, 228–229, 256, 281, 293, 299–302, 310, 312, 317, 330, 337, 369–372, 374, 377–378, 380–384 verb third (V3), 17, 224, 290 VP-intraposition, 54 VP-topicalization, 57 Wackernagel-position, 256, 294–297, 303, 305–306, 312–315, 318 word order, basic, 2, 5, 45, 79, 212, 369 change, 2, 6–7, 45, 50–51, 58, 62– 64, 91 mixed, 8, 46, 49, 52, 71–72 OV order, 2, 8, 45, 68, 71–72, 74, 80–83, 252, 273, 369 unmarked, 7–8, 50–51, 58, 60, 63, 331 variation, 1, 2, 6–10, 45–46, 49, 51, 58, 121–122, 153, 224, 251– 252 VO order, 2, 4–5, 8, 45–46, 49, 68, 71–72, 252, 328
comes after a sonorant instead of , the sonorant is in fact followed by a punctuation sign.