Abstract
Danish, a V2-language, allows liberal extraction of non-WH elements from a variety of clause types to sentence-initial position, e.g. from relative clauses. Extractions from complement clauses, see (1), are more frequent (Jensen 2001) than extractions from adverbial clauses as in (2).
(1) De sokker tror jeg at han køber [gap] på udsalget i morgen.
Those socks think I that he buy [gap] at sale-DET in tomorrow.
“I think that he will buy those socks at the sale tomorrow”.
(2) De sokker besvimer jeg hvis han køber [gap] på udsalget i morgen.
Those socks faint I if he buy [gap] at sale-DET in tomorrow.
“I will faint if he buys those socks at the sale tomorrow”.
Theories of this phenomenon hold that constraints on extraction in Danish are dependent on coherence (Jensen, 2001) or pragmatic dominance (Erteschik-Shir, 1982) between the main and the dependent clause, both presupposing processing of the dependent clause for judgment of the acceptability of the extraction. Thus assuming that detection of unacceptability causes processing difficulties, these theories predict reading time increases for unacceptable extractions at the earliest late within the dependent clause when evaluation of the pragmatic/semantic status is possible.
I outline an alternative account based on valency expectations associated with the main-clause verb. Since extraction from complement clauses is frequent, when processing sentences with verbs taking clauses as objects (comp-verbs), the parser doesn’t expect to have linked all mentioned arguments with the main verb at the clause boundary. For clauses with intransitive verbs, on the other hand, the parser expects to be able to find a role for all constituents within the clause, i.e. the parser should experience difficulties if unintegrated constituents are present at a clause boundary. Thus this theory predicts longer reading times at ‘if’ in (2) compared to ‘that’ in (1). Thus with regard to time course, this account is similar in spirit to work on island constraints in English indicating that extraction constraints work immediately at island boundaries (e.g. McKinnon & Osterhout, 1996; Traxler & Pickering, 1996).
Self-paced reading experiments were conducted to test these alternative accounts. Type of verb (comp vs. intransitive) was fully crossed with an extraction/no-extraction control factor (3-4). An anova revealed a significant interaction between verb type and extraction at the position of ‘if’/’that’, with longer RTs in the intransitive condition (2) compared to the complement (1) and control conditions (3-4).
(3) Jeg tror at han køber de sokker på udsalget i morgen.
I think he buy those socks at sale-DET in tomorrow.
“I think that he will buy those socks at the sale tomorrow”.
(4) Jeg besvimer hvis han køber de sokker på udsalget i morgen.
I faint if he buy those socks at sale-DET in tomorrow.
“I will faint if he buys those socks at the sale tomorrow”.
The relevance of the experimental manipulations to theories about acceptability of extraction in Danish was confirmed in a separate acceptability judgment experiment with the same stimuli. The intransitive extraction condition was rated significantly lower than the other conditions.
Together, the results of the two experiments are interpreted as supporting the verb expectation theory over theories based on semantic/pragmatic evaluation of the content of the dependent clause, because difficulties arise before evaluation of this content is possible.
References
Jensen, A. (2001). Sentence intertwining in Danish. In E. Engberg-Pedersen & P. Harder (Eds.), Ikonicitet og struktur (pp. 23-39): Preprint from Netværk for Funktionel Lingvistik, Department of English, University of Copenhagen.
Erteschik-Shir, N. (1982). Extractability in Danish and the pragmatic principle of dominance. In E. Engdahl & E. Ejerhed (Eds.), Readings on unbounded dependencies in Scandinavian languages (pp. 175-191). Stockholm: Almqvist & Wiksell International.
Traxler, M., & Pickering, M. (1996). Plausability and the processing of unbounded dependencies: an eye-tracking study. Journal of Memory and Language, 35, 454-475.
McKinnon, R., & Osterhout, L. (1996). Constraints on movement phenomena in sentence processing: evidence from event-related brain potentials. Language and Cognitive Processes, 11(5), 495-523.
(1) De sokker tror jeg at han køber [gap] på udsalget i morgen.
Those socks think I that he buy [gap] at sale-DET in tomorrow.
“I think that he will buy those socks at the sale tomorrow”.
(2) De sokker besvimer jeg hvis han køber [gap] på udsalget i morgen.
Those socks faint I if he buy [gap] at sale-DET in tomorrow.
“I will faint if he buys those socks at the sale tomorrow”.
Theories of this phenomenon hold that constraints on extraction in Danish are dependent on coherence (Jensen, 2001) or pragmatic dominance (Erteschik-Shir, 1982) between the main and the dependent clause, both presupposing processing of the dependent clause for judgment of the acceptability of the extraction. Thus assuming that detection of unacceptability causes processing difficulties, these theories predict reading time increases for unacceptable extractions at the earliest late within the dependent clause when evaluation of the pragmatic/semantic status is possible.
I outline an alternative account based on valency expectations associated with the main-clause verb. Since extraction from complement clauses is frequent, when processing sentences with verbs taking clauses as objects (comp-verbs), the parser doesn’t expect to have linked all mentioned arguments with the main verb at the clause boundary. For clauses with intransitive verbs, on the other hand, the parser expects to be able to find a role for all constituents within the clause, i.e. the parser should experience difficulties if unintegrated constituents are present at a clause boundary. Thus this theory predicts longer reading times at ‘if’ in (2) compared to ‘that’ in (1). Thus with regard to time course, this account is similar in spirit to work on island constraints in English indicating that extraction constraints work immediately at island boundaries (e.g. McKinnon & Osterhout, 1996; Traxler & Pickering, 1996).
Self-paced reading experiments were conducted to test these alternative accounts. Type of verb (comp vs. intransitive) was fully crossed with an extraction/no-extraction control factor (3-4). An anova revealed a significant interaction between verb type and extraction at the position of ‘if’/’that’, with longer RTs in the intransitive condition (2) compared to the complement (1) and control conditions (3-4).
(3) Jeg tror at han køber de sokker på udsalget i morgen.
I think he buy those socks at sale-DET in tomorrow.
“I think that he will buy those socks at the sale tomorrow”.
(4) Jeg besvimer hvis han køber de sokker på udsalget i morgen.
I faint if he buy those socks at sale-DET in tomorrow.
“I will faint if he buys those socks at the sale tomorrow”.
The relevance of the experimental manipulations to theories about acceptability of extraction in Danish was confirmed in a separate acceptability judgment experiment with the same stimuli. The intransitive extraction condition was rated significantly lower than the other conditions.
Together, the results of the two experiments are interpreted as supporting the verb expectation theory over theories based on semantic/pragmatic evaluation of the content of the dependent clause, because difficulties arise before evaluation of this content is possible.
References
Jensen, A. (2001). Sentence intertwining in Danish. In E. Engberg-Pedersen & P. Harder (Eds.), Ikonicitet og struktur (pp. 23-39): Preprint from Netværk for Funktionel Lingvistik, Department of English, University of Copenhagen.
Erteschik-Shir, N. (1982). Extractability in Danish and the pragmatic principle of dominance. In E. Engdahl & E. Ejerhed (Eds.), Readings on unbounded dependencies in Scandinavian languages (pp. 175-191). Stockholm: Almqvist & Wiksell International.
Traxler, M., & Pickering, M. (1996). Plausability and the processing of unbounded dependencies: an eye-tracking study. Journal of Memory and Language, 35, 454-475.
McKinnon, R., & Osterhout, L. (1996). Constraints on movement phenomena in sentence processing: evidence from event-related brain potentials. Language and Cognitive Processes, 11(5), 495-523.
| Originalsprog | Engelsk |
|---|---|
| Publikationsdato | 2004 |
| Status | Udgivet - 2004 |
| Begivenhed | Architectures and Mechanisms for Language Processing - Aix-en-provence, Frankrig Varighed: 16 sep. 2004 → 18 sep. 2004 |
Konference
| Konference | Architectures and Mechanisms for Language Processing |
|---|---|
| Land/Område | Frankrig |
| By | Aix-en-provence |
| Periode | 16/09/2004 → 18/09/2004 |