Abstract
The dissertation presents a functional linguistic model of grammaticality and investigates methods for applying this notion in empirical work. The use of the notion of grammaticality in generative grammar has been criticized by functionalists (Harder, 1996; Lakoff & Johnson, 1999), but attempts to constructively define what it then means that a sentence is grammatical from a functional linguistics perspective have been limited. The notion of grammaticality that will be put forward here is based on Langacker’s (2000) dynamical usage-based model. Langacker’s model is extended to accommodate the general finding from research on sentence processing that sentences are processed incrementally. Empirical methods for establishing grammaticality status are discussed and applied in relation to non-WH extraction phenomena in Danish.
In Chapter 2, I discuss the use of the notions of grammaticality and acceptability in the generative tradition and the functionalist criticism of this use. Langacker’s model is presented, and it is concluded that it is possible and desirable to distinguish between grammaticality and acceptability within this model.
In Chapter 3, I discuss various methods for establishing grammaticality. It is concluded that the intuitions of linguists should in principle be considered hypotheses of grammaticality, and that such hypotheses need to be tested with independent data. Such data can for example take the form of corpus data or acceptability judgment experiments. It is furthermore argued that reading-time methods can be used to obtain relevant data since difficulties with comprehending ungrammatical sentences presumably translate into prolonged reading-times.
In Chapter 4, the phenomenon of non-WH extraction in Danish is presented. Such extraction is restricted, and linguistic theories concerning the restrictions in Danish and extraction in general are discussed (Deane, 1991; Erteschik-Shir, 1982; Van Valin & LaPolla, 1997). The data on this phenomenon in Danish are, however, not clear, specifically with regard to extraction out of adverbial clauses, which Jensen (2001) has suggested is possible under certain conditions.
In Chapter 5, I present an overview of recent sentence processing research on production and comprehension, with a special section on the processing of extraction. I argue that the usage-based model is largely compatible with constraint-satisfaction models of sentence processing, and I illustrate how the model can implement incremental processing. I then suggest which processing predictions can be derived for the linguistic theories of extraction constraints presented in Chapter 4.
Chapter 6 presents two sets of acceptability-judgment and reading-time experiments on constraints on extraction out of adverbial clauses. The first set of experiments shows that difficulties with extraction out of adverbial clauses appear around the clause boundary. The second set of experiments investigated whether difficulties with extraction out of adverbial clauses are ameliorated when there is a high degree of semantic cohesion between the clauses (Jensen, 2001). The results provided evidence against such selective effects of cohesion on extraction.
The collective results suggest that extraction constraints are the result of the processor committing to non-extraction interpretations of the sentence in the matrix clause, rather than because the right dependencies cannot be established in the clause from which a constituent has been extracted (Erteschik-Shir, 1982). Results from the acceptability-judgment and reading-time experiments converged in both sets of experiments, thus supporting the idea that reading-time data can be used to investigate grammaticality.
In Chapter 2, I discuss the use of the notions of grammaticality and acceptability in the generative tradition and the functionalist criticism of this use. Langacker’s model is presented, and it is concluded that it is possible and desirable to distinguish between grammaticality and acceptability within this model.
In Chapter 3, I discuss various methods for establishing grammaticality. It is concluded that the intuitions of linguists should in principle be considered hypotheses of grammaticality, and that such hypotheses need to be tested with independent data. Such data can for example take the form of corpus data or acceptability judgment experiments. It is furthermore argued that reading-time methods can be used to obtain relevant data since difficulties with comprehending ungrammatical sentences presumably translate into prolonged reading-times.
In Chapter 4, the phenomenon of non-WH extraction in Danish is presented. Such extraction is restricted, and linguistic theories concerning the restrictions in Danish and extraction in general are discussed (Deane, 1991; Erteschik-Shir, 1982; Van Valin & LaPolla, 1997). The data on this phenomenon in Danish are, however, not clear, specifically with regard to extraction out of adverbial clauses, which Jensen (2001) has suggested is possible under certain conditions.
In Chapter 5, I present an overview of recent sentence processing research on production and comprehension, with a special section on the processing of extraction. I argue that the usage-based model is largely compatible with constraint-satisfaction models of sentence processing, and I illustrate how the model can implement incremental processing. I then suggest which processing predictions can be derived for the linguistic theories of extraction constraints presented in Chapter 4.
Chapter 6 presents two sets of acceptability-judgment and reading-time experiments on constraints on extraction out of adverbial clauses. The first set of experiments shows that difficulties with extraction out of adverbial clauses appear around the clause boundary. The second set of experiments investigated whether difficulties with extraction out of adverbial clauses are ameliorated when there is a high degree of semantic cohesion between the clauses (Jensen, 2001). The results provided evidence against such selective effects of cohesion on extraction.
The collective results suggest that extraction constraints are the result of the processor committing to non-extraction interpretations of the sentence in the matrix clause, rather than because the right dependencies cannot be established in the clause from which a constituent has been extracted (Erteschik-Shir, 1982). Results from the acceptability-judgment and reading-time experiments converged in both sets of experiments, thus supporting the idea that reading-time data can be used to investigate grammaticality.
| Original language | English |
|---|
| Place of Publication | Copenhagen |
|---|---|
| Publisher | Department of Nordic Studies and Linguistics, University of Copenhagen, 2005. |
| Publication status | Published - 2005 |