Modelling language contact with diachronic crosslinguistic data - - PowerPoint PPT Presentation

Modelling language contact with diachronic crosslinguistic data

Modelling language contact with diachronic crosslinguistic data

Achim Stein Carola Trips

achim.stein@ling.uni-stuttgart.de ctrips@rumms.uni-mannheim.de

Linguistic Evidence, Tübingen, 10.2.2012

1

Modelling language contact with diachronic crosslinguistic data 1 Introduction

Contents

1 Introduction 1.1 The phenomenon 1.2 Verb second in some Germanic and Romance languages 1.3 Information structure 2 Language contact 2.1 The historical background 2.2 Some facts about Medieval French 2.3 Some facts about Medieval English

• 3. Empirical study

3.1 Corpora 3.2 Medieval French: periods and global data 3.3 Medieval English: periods and global data 3.4 Comparison 3.5 Evidence for language contact (issue #1) 3.6 Cleft sentences (issue #2) 4 Conclusion

2

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.1 The phenomenon

Object topicalisation and left dislocation

Modern examples of topicalisation (1) and left-dislocation (2):

(1) ModF: *Les haricots, je mange. PDE: Beans I eat. (2) ModF: Les haricots, je les mange. PDE: Beans, I eat them.

Medieval examples of topicalisation. OF (3) and OE (4):

(3) Treis three escheles divisions ad has l’emperere the-king Carles. Charles

(Roland,219.3025)

(4) Þæt that hus house hæfdon had Romane Romans to to ðæm the anum

• ne

tacne sign

• geworht. . .

made ‘The Romans had made that house to their sole sign.’ (Or_3:5.59.3.1042)

3

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.1 The phenomenon

Left dislocation with topic continuity. OF: (5), ME: (6):

(5) Ceste This bataille, battle veirement really la it ferum [we]-will-fight

(Roland,70.846)

(6) and slonked her in so hungerly that he lefte neyther flessh ne bone / nomore but a fewe fethers the the smal small fethers feathers he he slange swallowed them them in in wyth with the the flessh flesh

(REYNAR,53.337-8 )

Left dislocation with contrastive focus. OF: (7), ME: (8):

(7) Mais But vos your barons barons en in sa his ballie realm S’ if il he les them trovout found nes them vilonast (Béroul,1106ss) would mistreat (8) and and

• f
• f
• n
• ne

þei they put put

• ute
• ut

his his eyne, eye, þe the

• þir
• ther

þei they broke broke his his bak back

(CAPCHR,249.4166-7) 4

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.1 The phenomenon

The development of these structures

Kroch (1989, 214) “. . . the loss of verb-second word order in French took place via the replacement of topicalization by left-dislocation.”

◮ Phrasal accent forces preposed constituents to occur left-dislocated

⇒ less subject-verb inversion ⇒ more resumptive pronouns

5

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.1 The phenomenon

Questions

Questions:

• 1. What kind of assumptions are necessary to interpret quantitative

developments as language change?

• 2. What kind of information do we have to include to make state-

ments about linguistic levels for which we have no direct evi- dence, like prosody?

• 3. What kind of indicators allow to distinguish between internal

change and contact-induced change?

6

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.2 Verb second in some Germanic and Romance languages

Verb second

OF OE OHG Language type SVO SOV SOV Verb final no yes yes V2 main cl. yes yes yes V2 subord. cl. yes (v. dicendi) yes (v. dicendi) yes (v. dicendi) XP-Spron-Vfin no yes yes (less restr. than OE) Null subjects yes yes (some persons) yes (some persons)

• subj. clitics

no yes yes

• bj. clitics

yes yes yes Loss of V2 yes (residual V2) yes (residual V2) no (full/true V2) Table 1: Comparison of some Germanic and Romance languages

7

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.3 Information structure

Information structure (Krifka, 2007)

Focus “Focus indicates the presence of alternatives that are relevant for the interpretation of linguistic expression”. (Focus often correlates with “new”, “important”, etc., but can’t be defined by these properties.) Topic “constituent which identifies the entity [. . . ] under which the infor- mation expressed in the comment constituent should be stored in the C[ommon] G[round] content”.

(9) A: When did [Aristotle Onassis]Topic marry Jacqueline Kennedy? B: [He]Topic[married her [in 1968]Focus]Comment

8

Modelling language contact with diachronic crosslinguistic data 1 Introduction 1.3 Information structure

Methodological problems

◮ For medieval language, it is difficult or impossible to verify nec-

essary conditions for the felicitous use of non-canonical sentence structures, contrary to modern language (e.g. Birner and Ward 1998; Birner 2004).

◮ Even in modern language, these necessary conditions are not

sufficient: the optionality can depend on text type, code (writ- ten vs spoken), individual preference, etc.

◮ Nevertheless, some discourse-related criteria (cf. Prince 1981,

1992) can be and have been applied to medieval language, as e.g. in Prévost (2003).

9

Modelling language contact with diachronic crosslinguistic data 2 Language contact

Contents

1 Introduction 1.1 The phenomenon 1.2 Verb second in some Germanic and Romance languages 1.3 Information structure 2 Language contact 2.1 The historical background 2.2 Some facts about Medieval French 2.3 Some facts about Medieval English

• 3. Empirical study

3.1 Corpora 3.2 Medieval French: periods and global data 3.3 Medieval English: periods and global data 3.4 Comparison 3.5 Evidence for language contact (issue #1) 3.6 Cleft sentences (issue #2) 4 Conclusion

10

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.1 The historical background

Norman Conquest: Multilingualism since 1066

◮ Evidence: glosses, borrowings, didactic texts, translations, etc. ◮ Middle English (1150-1500):

◮ Anglo-French had gained the role of an official language over

Anglo-Latin

◮ Middle English was predominantly a spoken language ◮ 15th c.: Middle-English took over the role of the national lan-

guage, both spoken and written

11

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.1 The historical background

Methodological consequences

Include sources for multilingualism:

◮ glosses in dictionaries ◮ didactic texts (discussed in Hunt 1991),

grammars of French (e.g. Palsgrave 1530)

◮ business writing, e.g. accounts, inventories:

see the work by Wright (1995; 2003), and also Trotter (2000); Ingham (2010))

◮ direct translations:

e.g. Somme le roi (Laurent, 2008) → Ayenbite of Inwyt (Morris, 1866)

12

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.1 The historical background

Palsgrave’s description for English learners

L’esclaircissement de la langue françoyse:

“The hole reason of theyr accent is grounded chefely upon thre poyntes: fyrst, there is no worde of one syllable whiche with them hath any accent, or that they use to pause upon, [. . . ] they pronounce them nat distinctly a sonder as the latines do, but sounde them all under one voyce and tenour, and never rest nor pause upon any of them, except the commyng next unto a poynt be the cause therof; seconde, every worde of many syllables hath his ac- cent upon the last syllable, but yet that nat withstandynge they use upon no suche worde to pause, except the commyng next unto a poynt be the causer therof: and this is one great thyng whiche inclineth the frenchemen so moche to pronounce that latin tong amysse, which contrary never gyve theyr accent

• n the last syllable. [. . . ]”

(quoted from the edition F. Génin, 1852, XXI)

13

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.2 Some facts about Medieval French

Old French (9th until early 14th c.)

◮ topic position for sentence-initial elements ◮ verb second sentences, mostly in main clauses ◮ null subjects ◮ two cases: cas sujet and cas régime ◮ two accent systems: phrasal and lexical ◮ clitic object pronouns

(subject pronouns became clitics in later OF) Some controversial points:

• 1. How strict is V2 in OF?
• 2. What is the structural position of the topicalized element?
• 3. Does OF have sentence-initial focus elements?

14

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.2 Some facts about Medieval French

Old French: dislocation and topicalisation

In OF, a dislocated element could be the focus:

(10) Li the nies nephew Marsilie,

• f-Marsilius,

il he est is venuz come avant forward sur

• n

un a

• mulet. . .

mule Voici que s’avance sur un mulet le neuveu de Marsilie. (Chanson de Roland, late 11c., quoted from Prévost 2003)

◮ The dislocated element is normally topic, exceptionally focus. ◮ Topicalised objects are unmarked in early OF.

They are increasingly marked as focus towards the 13th c.

15

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.3 Some facts about Medieval English

Medieval English: verb second

(11) Object-Vfin-subject: Þæt hus that hæfdon house Romane had to Romans ðæm to anum the tacne

• ne

geworht sign ... made ‘The Romans had made that house to their sole sign.’ (Or_3:5.59.3.1042) (12) Wh-Vfin-subject: Hwæt what sculon shall we we þæs afterwards nu now ma more secgan? say ‘What shall we afterwards say now more?’ (Bede_2:9.132.1.1253) (13) Neg-Vfin-subject: ne neg cræwþ crows se the hana cock todæg today ær before þu you me me ætsæcst. deny ‘The rooster will not crow today until you deny me.’ (wsgosp,Lk:22.34.5465)

16

Modelling language contact with diachronic crosslinguistic data 2 Language contact 2.3 Some facts about Medieval English

Old English left dislocation

◮ Traugott (2007, 243): left-dislocated constructions involve discourse-

• ld and discourse-new referents.

◮ This is confirmed by our data, but topic dislocations (14) are

much more frequent than focus dislocations (15):

(14) ða these fuglas birds þa then we we hie them ne not

• nweg

away flegdon. drove.

(Alex:21.11.258)

(15) & and Perseuse Perseus Mæacedonia Macedonia’s cyninge king him him wæron were

• n

in fultume help ealle all Thraci Tracians & and Ilirice Ilirians

(Or_4:11.110.12.2305) 17

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

Contents

1 Introduction 1.1 The phenomenon 1.2 Verb second in some Germanic and Romance languages 1.3 Information structure 2 Language contact 2.1 The historical background 2.2 Some facts about Medieval French 2.3 Some facts about Medieval English

• 3. Empirical study

3.1 Corpora 3.2 Medieval French: periods and global data 3.3 Medieval English: periods and global data 3.4 Comparison 3.5 Evidence for language contact (issue #1) 3.6 Cleft sentences (issue #2) 4 Conclusion

18

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.1 Corpora

Empirical study

◮ Syntactically annotated corpora:

◮ YCOE (Taylor et al., 2003) for OE

The York-Toronto-Helsinki Parsed Corpus of Old English Prose

◮ PPCME2 (Kroch and Taylor, 2000) for ME

The Penn-Helsinki Parsed Corpus of Middle English

◮ MCVF (Martineau, 2009) for Medieval French.

Modéliser le changement: les voies du français (Ottawa)

◮ Additional sources: Penn corpora for EME, BME, Penn Tree-

bank.

◮ Extraction and evaluation using CorpusSearch (UPenn), TIGERSearch

(IMS, Stuttgart) and self-made scripts.

19

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.2 Medieval French: periods and global data

Medieval French: variation across periods

period approx. date (first text)

• approx. date (last text)

sentences 12c 1080 (Roland) 1194 (Chièvres) 16841 13c 1200 (Sermon sainte Agnes) 1306 (Joinville) 12294 14c 1371 (Prise Alexandrie) 1405 (Froissart) 27684 15c 1425 (Morchesne) 1491 (Commynes) 15036 Table 2: Definition of four periods in the GTRC-MCVF corpus Null subject Object-V2 Object-LD period fabs frel fabs frel fabs frel 12c 3067 0.1821151 300 0.0178137 53 0.0031471 13c 861 0.0700342 76 0.0061819 30 0.0024402 14c 1995 0.0720633 90 0.0032510 47 0.0016977 15c 1489 0.0990290 17 0.0011306 23 0.0015297 total 7412 0.103152 428 0.006722 153 0.002129 Table 3: OF null subject, V2, and left dislocation

20

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.2 Medieval French: periods and global data

Medieval French: variation across periods

Figure: OF null subject, object V2, and object left dislocation

21

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.3 Medieval English: periods and global data

Medieval English: variation across periods

period compilation date (first text) compilation date (last text) sentences OE before 850 1150 110136 ME-m1 1150 1250 17745 ME-m2 1250 1350 13870 ME-m3 1350 1420 32969 ME-m4 1420 1500 20080 Table 4: Definition of periods in the OE/ME corpus Null subject Object-V2 Object-LD period fabs frel fabs frel fabs frel OE 2073 0.018822 522 0.004740 372 0.003378 ME-m1 257 0.014483 119 0.006706 109 0.006143 ME-m2 45 0.003244 45 0.003244 38 0.002740 ME-m3 56 0.001699 98 0.002972 73 0.002214 ME-m4 27 0.001345 60 0.002988 38 0.001892 total 2458 0.012618 844 0.004333 630 0.003234 Table 5: OE/ME null subject, V2, and left dislocation

22

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.3 Medieval English: periods and global data

Medieval English: variation across periods

Figure: OE/ME null subject, object V2, and object left dislocation

23

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.3 Medieval English: periods and global data

Written vs spoken speech

◮ In PDE, non-canonical constructions like LD are quite rare.

◮ Creswell (2004, 12) compares non-canonical word order in the

Wall Street Journal and the (oral) Switchboard corpus.

◮ Dislocations are almost absent from written speech. ◮ They are about 500 times more frequent in spoken speech.

In medieval corpora:

◮ Dislocation frequencies are similar to the oral Switchboard corpus. ◮ Indicator of spoken speech (in the sense of Biber 1988)

⇒ Dislocation reflects the degree of orality in written texts.

24

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.4 Comparison

Figure: English vs. French variation of object V2 and object left dislocation

compared to Priestley (1955)

25

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.4 Comparison

◮ Interpretation:

◮ French object-V2 develops as predicted ◮ but our data for French LD match Priestley’s only in OF. ◮ Priestley: in Middle French LD occurrences more than double. ◮ In our corpus, LD frequencies decrease steadily after 1300.

◮ Evaluation:

◮ Contra Kroch/Priestley, LD frequencies do not increase, which

implies that the development of V2 and LD are not correlated.

26

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.4 Comparison

Methodology revisited

Two issues:

• 1. LD structures are syntactically heterogeneous. They include:

◮ “heavy” NPs (with modifying clauses):

resumption can be motivated by distance.

◮ unmodified NPs (without modifying clauses):

the resumptive pronoun is an indicator for topichood.

• 2. These diachronic studies (as well as our empirical study) tend to

associate constructions with information values, e.g.

◮ topicalised element ∼ focus ◮ left-dislocated element ∼ topic 27

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.5 Evidence for language contact (issue #1)

Types of left dislocation (OF/OE)

(16) Ki Who mult

• ften

te you sert, serves malvais bad luer reward l’ him en for that dunes. (you) give

(Roland,187.2585)

(free relative) (17) Swa hwilc mon swa whosoever me me timbreð builds gebedhus, chapel sele give þu you mede reward him him

• n

in heofonum. heaven

(Mart-5-[Kotzor]:Jy15,A.14.1166)

(free relative) (18) Tuz all lur their amis friends qu’ whom il they i there unt have morz dead truvet, found Ad to un a carner grave sempres always les them unt (they) have portet. brought

(Roland,212.2943)

(NP+relative) (19) ðæt, that þæt that ic I eow you secge, say þæt that ic I secge say eallum all mannum... men

(ÆLet-6:112.42)

(NP+relative) (20) Mes my escheles, divisions, tutes all les them guiereiz; (you) will lead

(Roland,236.3266)

(unmodified) (21) ða these fuglas birds þa then we we hie them ne not

• nweg

away flegdon drove

(Alex:21.11.258)

(unmodified)

28

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.5 Evidence for language contact (issue #1)

Types of left dislocation (OF/OE)

wh-clause heavy NP NP (CP-FRL) (CP-REL) (other) OETraugott 5.8% OE 36.6% 53.5% 9.9% ME 28.3% 53.5% 18.2% OF 22.9% 27.5% 49.6% Table 7: Types of dislocated elements in English and French

◮ Traugott (2007): only 5.8% of OE dislocations are unmodified. ◮ Our data:

◮ >50% of OE/ME dislocated NPs are modified by a relative. ◮ In French, unmodified NPs are much more frequent.

◮ An indicator of language contact?

◮ In early ME: significant increase of unmodified NPs 29

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.5 Evidence for language contact (issue #1)

Types of left dislocation (OF/OE)

Figure: Types of dislocated elements: free wh-clause (CP-FRL), NP- dependent relative clause (CP-REL), unmodified NPs (other)

30

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.6 Cleft sentences (issue #2)

Diachrony and information structure

How stable is topic/focus marking? Does a given construction (left dislocation or topicalisation) mark a constant information value (topic only or focus only)?

◮ OF topicalisation marks topic and focus

◮ It gradually develops towards a focus marker ◮ V2 loses ground to the (S)VO structures ◮ topicalisation becomes less frequent, and more marked.

◮ OF left-dislocation was a topic marker, but not exclusively. ◮ Decrease of V2: how was focus marking compensated?

31

Modelling language contact with diachronic crosslinguistic data

• 3. Empirical study

3.6 Cleft sentences (issue #2)

Development of French cleft sentences

◮ Cleft sentences? a good candidate:

◮ (real) clefts appear in the 13c., favoured by the loss of V2 ◮ cleft frequency increases in the 14c.

(Marchello-Nizia, 1999; Combettes, 1999)

Clefts (CP-CLF) Object-V2 Object-LD period fabs frel fabs frel fabs frel 12c 6 0.0003563 300 0.0178137 53 0.0031471 13c 7 0.0005694 76 0.0061819 30 0.0024402 14c 17 0.0006141 90 0.0032510 47 0.0016977 15c 27 0.0017957 17 0.0011306 23 0.0015297 Table 6: Cleft sentences (CP-CLF) in French

32

Modelling language contact with diachronic crosslinguistic data 4 Conclusion

Some answers to our questions

• 1. How to interpret quantitative developments as language change

◮ Topic and focus marking is constant over time. ◮ We assume that this is the case at least in spoken speech. ◮ We may assume that our corpora reflect spoken speech (high

frequencies of left-dislocation in our texts)

• 2. How to make statements about linguistic levels for which we

have no direct evidence, like prosody?

◮ meta-linguistic information (like Palsgrave’s grammar) ◮ focus on phenomena at the syntax-prosody-interface

• 3. How to distinguish between internal change and contact-induced

change?

◮ significant changes which are not compatible with internal changes

(unmodified left-dislocated NPs in ME)

◮ focus on texts which clearly show multilingualism 33

Modelling language contact with diachronic crosslinguistic data References Biber, D. (1988). Variation across speech and writing. Cambridge University Press, Cambridge. Birner, B. (2004). Discourse functions at the periphery: Noncanonical word order in english. In Shaer, B., Frey, W., and Maienborn, C., editors, Proceedings of the Dislocated Elements Workshop, ZAS Berlin, November 2003, volume 35 of ZAS Papers in Linguistics, pages 41–62, Berlin. ZAS. Birner, B. and Ward, G. (1998). Information status and noncanonical word order in English, volume 40

• f Studies in language Companion series. Benjamins, Amsterdam.

Bouchard, J., Dupuis, F., and Dufresne, M. (2007). Un processus de focalisation en ancien français: le développement des clivées. In Actes du congrès annuel de l’Association canadienne de linguistique

• 2007. Proceedings of the 2007 annual conference of the Canadian Linguistic Association.

Combettes, B. (1999). Thématisation et topicalisation: leur rôle respectif dans l’évolution du français. In Guimier, C., editor, La thématisation dans les langues. Actes du colloque de Caen, 9-11 octobre 1997, pages 231–245. Peter Lang, Paris. Creswell, C. (2004). Syntactic form and discourse function in natural language generation. Outstanding dissertations in linguistics. Routledge, New York. Hunt, T. (1991). Teaching and Learning Latin in Thirteenth-Century England. Volume I: Texts. Brewer, Cambridge. Ingham, R., editor (2010). The Anglo-Norman Language and its Contexts. Boydell and Brewer. Krifka, M. (2007). Basic notions of information structure. In Féry, C., Fanselow, G., and Krifka, M., editors, The Notions of Information Structure, pages 13–55. Universitätsverlag Potsdam, Potsdam. Kroch, A. (1989). Reflexes of grammar in patterns of language change. Language Variation and Change, (1):199–244. Kroch, A. and Taylor, A., editors (2000). The Penn-Helsinki Parsed Corpus of Middle English, Second Edition (PPCME2). University of Pennsylvania, Philadelphia. Laurent, F. (2008). La Somme le roi. Société des anciens textes français, Paillart, Paris, Abbeville. Marchello-Nizia, C. (1999). Le français en diachronie: douze siècles d’évolution. Collection L’essentiel français. Ophrys, Paris. 34

Modelling language contact with diachronic crosslinguistic data References Martineau, F., editor (2009). Le corpus MCVF. Modéliser le changement: les voies du français. Université d’Ottawa, Ottawa. Morris, R., editor (1866). Dan Michel’s Ayenbite of Inwyt or Remorse of Conscience, volume I of The Early English Text Society 23. N. Trübner und Co., London. Palsgrave, J. (1852). L’éclaircissement de la langue française. Suivi de la grammaire de Giles de Guez. Publiés pour la première fois en France, par F. Génin. Imprimerie nationale, Paris. Priestley, L. (1955). Reprise constructions in french. Archivum Linguisticum, 7:1–28. Prince, E. (1981). Toward a taxonomy of given/new information. In Cole, P., editor, Radical pragmatics, pages 223–254. Academic Press, New York. Prince, E. (1992). The zpg letter: Subjects, definiteness, and information-status. In Mann, W. and Thompson, S., editors, Discourse description: diverse linguistic analyses of a fund-raising text, pages 295–325. Benjamins, Amsterdam. Prévost, S. (2003). Détachement et topicalisation: des niveaux d’analyse différents. Cahiers de praxématique, 40:97–126. Taylor, A. et al. (2003). The York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE). University of York, Heslington, York. Traugott, E. C. (2007). Old english left-dislocations: Their structure and information status. Folia Linguistica, (41/3-4):405–441. Trotter, D., editor (2000). Multilingualism in Later Medieval Britain. D.S. Brewer, Cambridge. Wright, L. (1995). A hypothesis on the structure of macaronic business writing. In Fisiak, J., editor, Medieval Dialectology, number 79 in Trends in Linguistics Studies and Monographs, pages 309–321. Mouton de Gruyter, Berlin. Wright, L. (2003). Models of language mixing: Code-switching versus semicommunication in medieval latin and middle english accounts. In Kastovsky, D. and Mettinger, A., editors, Language contact in the history of English, volume 1 of Studies in English Medieval Language and Literature, pages 363–377. Peter Lang, Frankfurt/Main. 35