Programme area 5 "Incremental construction of syntactic and (discourse) semantic representations", which expired at the end of 2011, dealt with 2 sub-areas: A) L(eft)R(ight)-incremental grammar formalisms and B) linear effects in the marking of sentence type, sentence mode and illocution in the sentence periphery.
Central work in area A was (i) a basic study on the implementation of "cluster analysis" of multiple constituent questions (Grewendorf 2001; Sabel 2001) in the context of Stabler's "minimalist grammars" (Gärtner and Michaelis 2010b; to appear). (ii) A deepening of forward compositional analysis of negation scope expansion in English (Blaszczak and Gärtner 2005) in the context of Steedman's "Combinatorial Categorial Grammar" (CCG) (Gärtner to appear-b). As well as (iii) a formal semantic dialogue analysis in the context of Noor van Leusen's "Logical Discourse Grammar" (2007) (van Leusen 2009).
In addition to work on understanding the assertion concept in the analysis of verb second declarative clauses under disjunction (Gärtner and Michaelis 2010a) and the analysis of quotative-evidential uses of the german auxiliary verb wollen (Gärtner to appear-a), two important studies were carried out in Area B:
(i) A comprehensive typological work on the complementary distribution of languages with embeddable non-finite interrogative sentences and languages with "robust" ambiguity of indefinite and interrgative pronouns (Gärtner 2009a). Here, the hypothesis was made for German that the said ambiguity together with the OV property blocks a sufficiently clear sentence type characterization of the left periphery of "W-infinitives". This was proposed as a (partial) explanation for the absence of a question interpretation of the subordinate clause in forms such as Er überlegte sich, wen an*(zu)rufen.
(ii) An investigation of the cleft sentence-like no-construction in Malagasy (Gärtner 2009b). Here, Matthew Pearson's proposal to analyze this construction paratactically was formalized and criticized within the framework of event semantics. It could be shown that, although a paratactic analysis has advantages in the analysis of quanta in "cleavage position" and predicts the variability of adverb placement, such an analysis has substantial weaknesses in the treatment of motion locality and binding behavior. The theory was therefore rejected. Instead, based on information-structural facts, a plea was made for a monoclausal pre-placement analysis approximating the German Vorfeld occupation.
An experimental study on the difference diagnosed by Bach & Zaefferer (2010) between forward typing - sentence type marking in the left periphery - and backward typing - sentence type marking in the left periphery - languages was prepared in 2011 by Kazuko Yatsushiro (with the collaboration of Eva-Maria Saur). Here, the weighting of sentence-introducing markings and sentence end particles in the understanding of illocution will be tested for Japanese. Methodologically, this is a reaction time experiment that tests the congruence of verbal and pictorial information by varying interrogative and declarative stimuli. The experiment is scheduled to be conducted in early 2012.
Our production and understanding of linguistic utterances takes place in real time. Moreover, psycholinguistic studies and computational models assume that these processes unfold incrementally: Some parts of an utterance are already being pronounced while others are still under planning. Equally, interpretation on the part of the addressee begins before complete utterances have been received.
The linguistic competence underlying these processes is usually referred to as grammar, and the most successful theories of grammar so far abstract from this temporal-incremental dimension. This has led to a situation in which assumptions about linguistic structures and interpretation in computational- and psycholinguistics regularly diverge noticeably from assumptions made by theorists of grammar.
This project will focus on more recent attempts at bridging this divide by incorporating incrementality into the theorist's grammars directly. Such frameworks have primarily been inspired by successes in dynamic semantics (Kamp) in treating anaphoric dependencies. As is well-known, temporal order is a core ingredient in this domain, as the different possibilities for resolving the pronoun he in (a) and (b) show.
a. A man walked through the park. He whistled.
b. He walked through the park. A man whistled.
Structural and word order asymmetries (Kayne), such as the ones appearing in question formation, constitute a second source of inspiration for incremental grammars. Thus, if question words have to be displaced to the sentential periphery, this affects the beginning of the sentence in most known languages, as shown in (c) and (d) for English.
c. I met a colleague.
d. Who did you meet?
One goal of this project consists in a theoretical and formal analysis and unification of the currently rather heterogeneous and programmatic frameworks promoting incremental grammars. Another goal is to test these frameworks empirically in the domain of sentence types (declarative, interrogative, imperative), where we will look at how insightfully form and function of (phenomena affecting) sentential peripheries can be modeled incrementally. Here we will focus on the contrast between Germanic languages and Asiatic languages like Japanese and Korean and their respective tendencies to place markings of sentence type at sentence beginnings vs. sentence endings.