contributing to the extraction/parenthesis debate: judgement studies and historical data

44
Contributing to the extraction/parenthesis debate: judgement studies and historical data Katrin Axel & Tanja Kiziak [email protected] [email protected] International Conference on Linguistic Evide 2-4 February 2006, University of Tübingen Linguistic Data Structures Linguistic Data Structures

Upload: gavin-gamble

Post on 03-Jan-2016

19 views

Category:

Documents


0 download

DESCRIPTION

International Conference on Linguistic Evidence 2-4 February 2006, University of Tübingen. Linguistic Data Structures. Contributing to the extraction/parenthesis debate: judgement studies and historical data. Katrin Axel & Tanja Kiziak [email protected] - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Contributing to the extraction/parenthesis debate: judgement studies and historical data

Katrin Axel & Tanja Kiziak

[email protected]@uni-tuebingen.de

International Conference on Linguistic Evidence2-4 February 2006, University of TübingenLinguistic Data StructuresLinguistic Data Structures

Page 2: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 2

Talk Outline

1. Two competing analyses for critical construction

• long extraction analysis

• parenthetical analysis

2. Judgement studies of present-day German

3. Historical corpus data

4. Conclusions

Page 3: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 3

The critical structure

Simple subordinate clauses:

(1a) Du glaubst, dass Ede Peter angerufen hat. = that-Vlast

you believe that Ede Peter called has

(1b) Du glaubst, Ede hat Peter angerufen. = V2you believe Ede has Peter called

Wh-constructions:

(2) Weni glaubst du, ti dass Ede ti angerufen hat? who believe you that Ede called has?

= extraction construction

(3) Wen glaubst du, hat Ede angerufen? who believe you has Ede called?”

= extraction or parenthetical construction?

Page 4: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 4

Two analyses of critical structure

(3) Wen glaubst du hat Ede angerufen? who believe you has Ede called ‘Who do you believe Ede has called?’ / ‘Who - do you believe - has Ede called?’

Analysis I: Extraction constructionderived from dependent V2-clause with intermediate tracein its ‘Vorfeld‘ (SpecCP)

Weni glaubst du ti hat Ede ti angerufen?

Analysis II: Parenthetical constructionparenthetical insert: glaubst duhost: Wen hat Ede angerufen?

Page 5: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 5

Two analyses of critical structure

Extraction analysis:[taken from Grewendorf/Hamm/Sternefeld 1989:242]

Parenthetical analysis:

Page 6: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

The judgement studies

- Methodology: Magnitude Estimation

- Theoretical assumptions

- The study series

Page 7: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 7

Gathering judgements experimentally

• Magnitude Estimation (Bard et al 1996)

• elicited grammaticality judgements, but ...

– judgements in numerical form

– relative to a reference item and relative to own previous judgements

– scale is open-ended and has no minimum division

• task: “you gave this one a 20, and that one a 30, so what score will you give this one?”

• “How natural do these examples sound?”

• online questionnaire (WebExp-Tool, Keller et al 1998)

Page 8: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 8

Three similar constructions

I.) Critical construction (4) Wen glaubst du hat Ede angerufen? = (3)

‘Who do you believe Ede has called?’

II.) Extractions from dass-clause = (2) (5) Wen glaubst du, dass Ede angerufen hat? ‘Who do you believe that Ede has called?’

III.) Verb-first parentheticals in other insertion slots: (6a) Wen hat Ede angerufen, glaubst du? (6b) Wen hat Ede, glaubst du, angerufen? ‘Who has Ede (do you believe) called (do you believe)?’

Central question of our experiments:Do critical constructions behave like uncontroversial extractionsor like uncontroversial parentheticals?

Page 9: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 9

Test case: predicate restrictions

Reis (2002) claims: Some predicates occur in dass-extractions but not in integrated parentheticals– (strong) factive predicates– negative/negated predicates– preference predicates– adjectival predicates

An Example [Reis 2002, (55b)]:(10a) Wohin wünschte er, dass sie zu Fuß ginge? dass-extraction

(10b) *Wohin ginge sie, wünschte er, zu Fuß? V1-parenthetical

(10c) *Wohin wünschte er ginge sie zu Fuß? critical structure

‘Where does he wish (that) she should go on foot?’

Page 10: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 10

Study Idass-extraction vs. critical construction

2 syntactic conditions

(a) the critical construction Welchen Bewerber PRED du stellt unser Chef ein? Which applicant PRED you hire our boss in

(b) long extraction over dass Welchen Bewerber PRED du dass unser Chef einstellt? Which applicant PRED you that our boss in-hire

‘Which applicant do you PRED (that) our boss will hire?’

18 predicates

(I) predicates of thought & speech

(II) negative, preference & adjectival predicates

Page 11: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 11

Predicates in study I

(I) Predicates of thought & speech:

sagen say

glaubenbelieve

hoffen hope

behaupten claim

fürchen fear

erzählen tell

erklären explain

(II) Negative, preference & adjectival predicates:

bezweifeln doubtwünschen wishwollen wantvorziehen preferbevorzugen favourlieber sein be preferablebesser sein be betterratsam sein be advisabledas Beste sein be the bestklar sein be clearbekannt sein be known

Page 12: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 12

Predictions of study I

The judgements of the critical construction and the dass-extraction

will respond to the continuum of predicates in a similar way

if they are both extraction constructions.

Two schematic pictures showing “similarity”

Page 13: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 13

Results of study Idass-extraction vs. critical construction

Page 14: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 14

Impact of dependent V2-clause?

On extraction analysis:

critical construction = extraction from dependent V2-clause

Note: Dependent V2-clauses more restricted than dass-clauses Only a subset of the predicates which select a dass-clause also select a V2-clause.

Question:

Does acceptability of dependent V2-clauses with certain predicates influence judgements of critical construction?

Page 15: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 15

Study IIImpact of dependent V2-clause?

2 syntactic conditions

(a) the critical construction

Welchen Bewerber PRED du stellt unser Chef ein? Which applicant PRED you hire our boss in

‘Which applicant do you PRED our boss will hire?’

(b) dependent V2-clause

Du PRED, unser Chef stellt diesen Bewerber ein. You PRED our boss hire this applicant in

‘You PRED, our boss will hire this applicant.’

12 predicates, a subset of the earlier tested predicates

Page 16: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 16

Results of study II – dependent V2-clause vs. critical construction

No direct correlation between critical construction anddependent V2-clauses throughout range of tested predicates

Page 17: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 17

Intermediate result

Differences between critical construction and dass-extraction cannot be related back to the factor “V2-subordination”.

Does the critical construction pattern like …

… uncontroversial extractions? no

… uncontroversial parentheticals? ?

Page 18: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 18

Study III parenthetical vs. critical construction

3 syntactic conditions (a) the critical construction

Welchen Bewerber PRED du stellt unser Chef im März ein? “Which applicant do you PRED our boss will hire in March?”

(b) clear parenthetical in post-subject position Welchen Bewerber stellt unser Chef PRED du im März ein? “Which applicant will our boss - do you PRED - hire in March?“

(c) long extraction over dass Welchen Bewerber PRED du dass unser Chef im März einstellt? “Which applicant do you PRED that our boss will hire in March?”

12 predicates, a subset of the earlier tested predicates

Page 19: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 19

Predicates in study III

(I) Predicates of thought & speech:

glaubenbelieve

hoffen hope

fürchen fear

erzählen tell

(II) Negative, preference & adjectival predicates:

bezweifeln doubt

wünschen wish

wollen want

vorziehen prefer

bevorzugen favour

lieber sein be preferable

ratsam sein be advisable

klar sein be clear

Page 20: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 20

Predictions of study III

The judgements of the critical construction and the clear parentheticals will respond to the continuum of predicates in a similar way if they are both parenthetical constructions.

The earlier finding concerning dass-extraction and critical construction should also be replicated.

Page 21: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 21

Results of study IIIdass-extraction vs. critical construction

- findings of first study replicated

- predicate-class dependent pattern

Page 22: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 22

Results of study IIIparenthetical vs. critical construction

- no predicate-class dependent pattern- both Pearson‘s correlation and ANOVAs confirm findings

Page 23: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 23

Tentative Conclusion & Objections

In predicate restrictions, the critical construction behaves:

- differently from dass-extractions

- similarly to clear integrated parentheticals

Possible objections:

- preference predicates as a special verb class (Frank 1998, Meinunger 2004)?

- the role of the complementizer dass in marginal structures?

- can speakers have two analyses of same surface structure?

Page 24: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

The historical study

Question:

Extraction or parenthetical analysis better motivated from historical perspective?

Page 25: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 25

Corpus

Old High German (OHG)

- oldest attested period of German (8th century - ca. 1050)

Corpus

- Isidor translation (ca. 790 A.D.)

- The Monsee Fragments (ca. 790 A.D.)

- Otfrid‘s Gospel Harmony (poem) (ca. 870 A.D.)

- Notker III.: ›Consolatio philosophiae‹, (ca. 1000 A.D.) ›Martianus Capella‹,› De Interpretatione‹, ›Psalter‹, ›Categoriae‹,

Page 26: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 26

Critical construction

With wânen ‘believe’

(11) uuéderêr déro uuânest tu gemág mêr whichnom of-those believe you is-capable more

‘Which one, do you believe, is capable of more?’ (N Cons IV 189,18)

(12) Was wánet werde thánne themo úmbitherben wáldewhatnom believe2.pl becomes then thedat unfit forest ‘What do you believe becomes then of the unfit forest?’ (O IV 26,51)

With quedan ‘say, speak, tell‘

(13) Fone uuîu chist dû nâhent? from whatinstr.. say you approach3.pl

‘From where do you say they are approaching?’ (N Ps. 54 189,14)

Page 27: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 27

Extraction analysis

(14) uuéderêr déro uuânest tu gemág mêr = (11) which of-those believe you is-capable more

‘Which one of those, do you believe, is capable of more?’

Extraction analysisderived from dependent V2-clause:(15) [uuéderêr déro]i uuânest tu [ ti gemág ti mêr ]?

Preconditions for extraction analysis (i) long wh-extraction(ii) dependent V2-clauses (as argument clauses)

QuestionAre both (i) and (ii) present in Old High German?

Page 28: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 28

Long wh-extraction over thaz?

Long wh-extraction

(16) wazi wánist [ ti thaz er ti wérde] whatnom believe2.sg that he becomesconj..

‘What do you believe that he becomes?’ (O I

9,29)

(17) [uuélên uuéhsel]i múgen uuír chéden. [ ti dáz tîe ti lîdên]?

which transitionakk can we say that those undergo ‘Which transition can we say that they undergo?’ (N Cons IV 216,1)

Page 29: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 29

Long wh-extraction over thaz

• Secondary Literature

Behaghel (1928: 547-552)Paul (1919: 319-324)Blatz (1896: 929-932)Lenerz (1985: 112-114)

Isidor Monsee Frag.

Otfrid Notker

• OHG corpus

Page 30: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 30

Extraction hypothesis – preconditions

long wh-extraction

dependent V2-clauses ?

Page 31: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 31

Thaz-less complement clauses

With wânen ‘believe’

(18)Uuânest [tu dehéin mûot keuéstenôtez ába stéte eruuékkêst]? believe2.sgyou any disposition determined from position softenconj..

‘Do you believe that you can soften such a determined disposition?’ (N Cons II 90.25)

(19) wánu [ sie ouh thaz rúzin] believe1.sg they also thatacc. weptconj.

‘I believe that they also wept for that’ (O IV 26,6)

With quedan ‘say’

(20) ih chido, [ iz tempus bezeichenne] I say it ‘tempus’ denoteconj.

‘I tell you that it denotes ‘tempus’’ (N Int. I 10,12)

Page 32: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 32

Thaz-less complement clauses

Thaz-less complement clauses in OHG corpus

• are verb last (!)

• frequently occur with uuânen, quedan as matrix predicates

• are attested with volitional predicates (e.g. wellen ‚want‘) and after negated matrix predicates (versus modern V2-complement clauses)

→ possible analysis: silent complementizer (‚Comp drop‘):

(21) ih chido, [ ø iz tempus bezeichenne ] = (20)I say it ‘tempus’ denoteconj.

Page 33: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 33

Dependent V2-clauses

In OHG corpus

• dependent V2-clauses (as argument clauses) not attested

• preconditions of extraction analysis

long wh-extraction

dependent V2-clauses —

Page 34: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 34

Long extraction from thaz-less complement clauses

Long Wh-extraction

(22) wasi wanist [ ti Ø thémo ti irgange (the …)] what believe the-onedat. faresconj. who ‘What do you believe happens to somebody (who …)?’

(O V 21,10)

↔ Critical Construction of (22)

(22’) wasi wanist [ ti Ø thémo ti irgange ] (23) wasi wanist [ ti irgangek thémo ti tk ] (example made up)

Page 35: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 35

Summary – extraction hypothesis

• one precondition fulfilled in OHG: long wh-movement attested from (verb-last) complement clauses with ±thaz (bridge prediactes: uuânen, quedan etc.)

bridge predicates uuânen, quedan also attested in critical construction

• second precondition:dependent V2-clauses not attested

→ Extraction analysis not very well motivated for OHG

Page 36: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 36

Parenthetical analysis

(24) uuéderêr déro uuânest tu gemág mêr = (11) which of-those believe you is-capable more

‘Which one of those, do you believe, is capable of more?’

Parenthetical analysis:

parenthetical insert: uuânest tu

host: uuéderêr déro gemág mêr?

Page 37: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 37

Parenthetical inserts

• numerous uncontroversial V1-parentheticals in OHG corpus (notably in Otfrid and in Notker’s texts)

• most prominent verbs: wânen ‘believe’, quedan ‘say’

• V1-parentheticals as a typical property of historical German texts (from OHG to NHG period, e.g. Maurer 1924)

• various insertion slots (in the middle field, clause-finally)

Page 38: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 38

Parenthetical inserts – slots

In the middle field (e.g. after the left bracket, after the subject):

(25) … thaz, [ wán ih ] , thu nu fíndes whichrel.pron. believe I you now find

‘… which you will now find here, I believe’ (O IV 18,8)

(26) uuánda dû nû [ uuâno ih ] uuácherôren óugen hábest. because you now believe I more-vigilant eyes haveconj.

‘because you have now, I believe, more vigilant eyes’ (N Cons III 174,16)

In clause-last position:

(27) Ír uuéllent iz sô bríngen [ uuâno íh ] . táz íu nîehtes nebréste you want it so bring believe I that youdat.nothinggen. NEGlack ‘You want to bring it about that you won’t lack anything’ (N Cons II 80,26)

After the ‘Vorfeld’ (→ critical construction):(28) sun [ uuânint ir ] ist er

son believe2.pl. you is he ‘He is the son, you believe’ (N Ps 77 276,3)

Page 39: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 39

Summary – historical study

Constructions attested in OHG corpus

• critical construction (with wânen, quedan)

• extraction of ± wh-phrases out of verb-last complement clauses with ± thaz after a number of bridge predicates (including wânen, quedan)

• V1-parenthetical inserts (with wânen, quedan etc.)

Constructions NOT attested

• dependent V2-clauses (as argument clauses)

→ Extraction analysis not well motivated for OHGParenthetical analysis well motivated for OHG

Page 40: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 40

Objections – historical study

• lack of positive evidence ≠ negative evidence→ dependent V2-clauses (as argument clauses) really ungrammatical in OHG?

• no prosodic evidence

• implication for Modern German?

Page 41: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 41

Conclusion

• both data types – judgement data and historical corpus data – favour parenthetical analysis of critical construction

• alleged V2-extraction analysis of critical construction has implications

- for models of German sentence structure and

- for status of dependent V2-clauses (+/- embedded?)

→ impact on syntactic theories

Page 42: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 42

References

Axel, K. (2005): Studien zur ahd. Syntax. Linke Satzperipherie, Verbstellung und Verb-zweit. PhD thesis, University of Tübingen.

Bard, E./Sorace, A. (1996) „Magnitude Estimation of Linguistic Acceptability“. Language 72(1). 32-68.Behaghel, O. (1928). Deutsche Syntax. Eine geschichtliche Darstellung. Band III. Heidelberg: Winter.Blatz, F. (1896). Neuhochdeutsche Grammatik. Third edition. Vol.II: Satzlehre (Syntax). Karlsruhe: J.

Lang.Frank, N. (1998) „Präferenzprädikate und abhängige Verbzweitsätze“. Arbeitspapiere des SFB 340:128.

Stuttgart-Tübingen.Grewendorf, G./Hamm, F./Sternefeld, W. (1989) Sprachliches Wissen. Eine Einführung in moderne

Theorien der grammatischen Beschreibung. (3rd edition) Frankfurt: Suhrkamp.Haider, H. (1993) „ECP-Etüden: Anmerkungen zur Extraktion aus eingebetteten Verb-Zweit-Sätzen“.

Linguistische Berichte 145.185-203.Lenerz, J. 1985. Diachronic syntax verb position and COMP in German. In Studies in German

Grammar, J. Toman. (ed), 103-132. Dordrecht: Foris.Maurer, F. (1924): Zur Anfangsstellung des Verbs im Deutschen. In Beiträge zur germanischen

Sprachwissenschaft. Festschrift für O. Behaghel, ed. by W. Horn, 141-184. Heidelberg: Winter.Meinunger, A. (2004) „Verb position, verbal mood and the anchoring (potential) of sentences“. In:

Lohenstein, H./Trissler, S. (eds.) The syntax and semantics of the Left Periphery. Berlin, New York: de Gruyter. 313-341.

Paul, H. (1919): Deutsche Grammatik. Vols. III,IV: Syntax. Halle: Niemeyer.Reis, M. (2002) „Wh-Movement and Integrated Parenthetical Constructions“. In: Zwart, J.-W./Abraham,

W. (eds.) Proceedings from the 15th Workshop on Comparative Germanic Syntax. Amsterdam: Benjamins. 3-40.

TITUS 3.0 = Thesaurus Indogermanischer Text- und Sprachmaterialien: http://titus.fkidg1.uni-frankfurt.de/indexd.htm.

Page 44: Contributing to the extraction/parenthesis debate:  judgement studies and  historical data

Conference on Linguistic Evidence, 2-4 February 2006 44

Statistical data from third study

Pearson‘s correlation

dass-extraction and critical construction: 0.738

post-subject parenthetical and critical construction: 0.924

Interaction of factors PREDICATE and STRUCTURE

dass-extraction and critical construction:

highly significant both by subjects and by items

(F1(11,297) = 5.853, p1< 0.001; F2(11,121) = 4.693, p2<0.001)

post-subject parenthetical and critical construction:

no significant interaction either by subjects or by items

(F1(11,286) = 1.79, p1 = 0.056; F2(11,121)= 1.424, p2 = 0.221)