Linguistik
Refine
Year of publication
Document Type
- Part of a Book (591)
- Conference Proceeding (395)
- Article (377)
- Working Paper (117)
- Preprint (97)
- Report (32)
- Book (26)
- Doctoral Thesis (16)
- Part of Periodical (16)
- Review (16)
Language
- English (1688) (remove)
Is part of the Bibliography
- no (1688)
Keywords
- Syntax (114)
- Englisch (109)
- Deutsch (86)
- Spracherwerb (80)
- Semantik (71)
- Phonologie (64)
- Informationsstruktur (51)
- Phonetik (48)
- Japanisch (43)
- Thema-Rhema-Gliederung (42)
Institute
- Extern (141)
- Institut für Deutsche Sprache (IDS) Mannheim (97)
- Neuere Philologien (26)
- Sprachwissenschaften (8)
- Medizin (2)
- Sprach- und Kulturwissenschaften (2)
- Gesellschaftswissenschaften (1)
- Informatik (1)
- SFB 268 (1)
In this paper, we argue that difficulties in the definition of coreference itself contribute to lower inter-annotator agreement in certain cases. Data from a large referentially annotated corpus serves to corroborate this point, using a quantitative investigation to assess which effects or problems are likely to be the most prominent. Several examples where such problems occur are discussed in more detail, and we then propose a generalisation of Poesio, Reyle and Stevenson’s Justified Sloppiness Hypothesis to provide a unified model for these cases of disagreement and argue that a deeper understanding of the phenomena involved allows to tackle problematic cases in a more principled fashion than would be possible using only pre-theoretic intuitions.
Traditionally, parsers are evaluated against gold standard test data. This can cause problems if there is a mismatch between the data structures and representations used by the parser and the gold standard. A particular case in point is German, for which two treebanks (TiGer and TüBa-D/Z) are available with highly different annotation schemes for the acquisition of (e.g.) PCFG parsers. The differences between the TiGer and TüBa-D/Z annotation schemes make fair and unbiased parser evaluation difficult [7, 9, 12]. The resource (TEPACOC) presented in this paper takes a different approach to parser evaluation: instead of providing evaluation data in a single annotation scheme, TEPACOC uses comparable sentences and their annotations for 5 selected key grammatical phenomena (with 20 sentences each per phenomena) from both TiGer and TüBa-D/Z resources. This provides a 2 times 100 sentence comparable testsuite which allows us to evaluate TiGer-trained parsers against the TiGer part of TEPACOC, and TüBa-D/Z-trained parsers against the TüBa-D/Z part of TEPACOC for key phenomena, instead of comparing them against a single (and potentially biased) gold standard. To overcome the problem of inconsistency in human evaluation and to bridge the gap between the two different annotation schemes, we provide an extensive error classification, which enables us to compare parser output across the two different treebanks. In the remaining part of the paper we present the testsuite and describe the grammatical phenomena covered in the data. We discuss the different annotation strategies used in the two treebanks to encode these phenomena and present our error classification of potential parser errors.
In the recent literature the phenomenon of long distance agreement has become the focus of several studies as it seems to violate certain locality conditions which require that agreeing elements in general stand in clause-mate relationships. In particular, it involves a verb agreeing with a constituent which is located in the verb's clausal complement and hence poses a challenge for theories that assume a strictly local relationship for agreement. In this paper we present empirical evidence from Greek and Romanian for the reality of long distance agreement. Specifically, we focus on raising constructions in these two languages and we show that they do not involve movement but rather instantiate long distance agreement. We further argue that subjunctives allowing long distance agreement lack both a CP layer and semantic Tense. However, since the embedded verb also bears phi-features, these constructions pose a further problem for assumptions that view the presence of phi-features as evidence for the presence of a C layer. Finally, we raise the question of the common properties that these languages have that lead to the presence of long distance agreement.
Distributional approximations to lexical semantics are very useful not only in helping the creation of lexical semantic resources (Kilgariff et al., 2004; Snow et al., 2006), but also when directly applied in tasks that can benefit from large-coverage semantic knowledge such as coreference resolution (Poesio et al., 1998; Gasperin and Vieira, 2004; Versley, 2007), word sense disambiguation (Mc- Carthy et al., 2004) or semantical role labeling (Gordon and Swanson, 2007). We present a model that is built from Webbased corpora using both shallow patterns for grammatical and semantic relations and a window-based approach, using singular value decomposition to decorrelate the feature space which is otherwise too heavily influenced by the skewed topic distribution of Web corpora.
Parsing coordinations
(2009)
The present paper is concerned with statistical parsing of constituent structures in German. The paper presents four experiments that aim at improving parsing performance of coordinate structure: 1) reranking the n-best parses of a PCFG parser, 2) enriching the input to a PCFG parser by gold scopes for any conjunct, 3) reranking the parser output for all possible scopes for conjuncts that are permissible with regard to clause structure. Experiment 4 reranks a combination of parses from experiments 1 and 3. The experiments presented show that n- best parsing combined with reranking improves results by a large margin. Providing the parser with different scope possibilities and reranking the resulting parses results in an increase in F-score from 69.76 for the baseline to 74.69. While the F-score is similar to the one of the first experiment (n-best parsing and reranking), the first experiment results in higher recall (75.48% vs. 73.69%) and the third one in higher precision (75.43% vs. 73.26%). Combining the two methods results in the best result with an F-score of 76.69.
The aim of this paper is to address two main counterarguments raised in Landau (2007) against the movement analysis of Control, and especially against the phenomenon of Backward Control. The paper shows that unlike the situation described in Tsez (Polinsky & Potsdam 2002), Landau's objections do not hold for Greek and Romanian, where all obligatory control verbs exhibit Backward Control. Our results thus provide stronger empirical support for a theoretical approach to Control in terms of Movement, as defended in Hornstein (1999 and subsequent work).
The article analyzes the system of focus-sensitive particles and, to a lesser extent, clefts in Vietnamese. EVEN/ALSO/ONLY foci are discussed across syntactic categories, and Vietnamese is found to organize its system of focus-sensitive particles along three dimensions of classification: (i) EVEN vs. ALSO vs. ONLY; (ii) particles c-commanding foci vs. particles c-commanding backgrounds; (iii) adverbial focus-sensitive particles vs. particles c-commanding argument foci only. Towards the end of the paper, free-choice constructions and additional sentence-final particles conveying ONLY and ALSO semantics are briefly discussed. The peculiar Vietnamese system reflects core properties of the analogous empirical domain in Chinese, a known source of borrowings into Vietnamese over the millennia.
Introduction
(2007)
Semantics
(2007)
Morphology
(2007)
Introduction
(2007)
The annotation guidelines introduced in this chapter present an attempt to create a unique infrastructure for the encoding of data from very different languages. The ultimate target of these annotations is to allow for data retrieval for the study of information structure, and since information structure interacts with all levels of grammar, the present guidelines cover all levels of grammar too. After introducing the guidelines, the current chapter also presents an evaluation by means of measurements of the inter-annotator agreement.
Information structure
(2007)
Traces
(2007)
The present tense is vacuous
(2002)
The meaning of chains
(1998)
This thesis investigates the mechanisms applying in the interpretation of syntactic chains. The theoretical background includes a translation of syntactic forms into semantic forms and a model theoretic explication of the meaning of semantic forms. Simplicity considerations apply to all three stages of the interpretation process: syntactic derivation, translation into semantic forms, interpretation of semantic forms. Three main results are achieved. The first is that trace positions can have semantic content beyond what is needed for the semantic dependency of trace and binder. This extra content is some or all of the lexical material of the head of the chain, as expected on the copy theory of movement. Two independent arguments support this conclusion. One, discussed in chapter 2, is based on the distribution of Condition C effects, where novel interactions between variable binding, antecedent contained deletion and Condition C are observed. The second, developed in chapter 3, is based on conditions on the identity of traces observed in antecedent contained deletion constructions. Both arguments lead to the same generalizations about what lexical material of the head is interpreted in the trace position. The second main result is that lambda calculus is superior to both standard predicate logic and combinatorial logic as the mathematical model for the semantic mechanism mediating the dependency of trace (or bound pronoun) and binder. Chapter 4 argues this on the basis of the distribution of focus and destressing in constructions with bound pronouns. The third main result is that quantification must be allowed to range over pointwise different choice functions. Chapter 5 shows that quantification over individuals is insufficient, and that pointwise different choice functions are required. The result entails that the syntactic difference of A-chains and A-bar chains predicts a semantic difference in the type of the variable involved, which is argued to explain weak crossover phenomena. Chapters 6 argues that the interpretation procedures developed in the preceeding chapters account for all cases. It is shown that only traces of the type of individuals arise, and that scope reconstruction is a phonological phenomenon. The latter result also supports the T-model of syntax.
The lemmings theory of case
(1995)
The epistemic step
(2005)
DP is not a scope island
(2005)
Presupposition
(2007)
On embedded implicatures
(2004)
The Gricean approach explains implicatures by assumptions about the pragmatics of entire utterances. The phenomenon of embedded implicatures remains a challenge for this approach since in such cases apparently implicatures contribute to the truth-conditional content of constituents smaller than utterances. In this paper, I investigate three areas where embedded implicatures seem to differ from implicatures at the utterance level: optionality, epistemic status, and implicated presuppositions. I conclude that the differences between the two kinds of implicatures justify an approach that maintains Gricean assumptions at the utterance level, and assumes a special operator for embedded implicatures.
Intermediate cumulation
(2001)
Implicated presuppositions
(2008)
It is widely believed that existential quantifiers can bring about the semantic effects of a scope which is wider than their actual syntactic scope (See Fodor & Sag (1982), Cresti (1995), Kratzer (1995), Reinhart (1995) and Winter (1995), among many others.) On the other hand, it is assumed that the syntactic scope of universal quantifiers can be determined unequivocally by the semantics. This paper shows that this second assumption is wrong; universal quantifiers can also bring about scope illusions, though in a very specific environment. In particular, we argue that in the environment of generic tense, universal quantifiers can show the semantic effects of a scope which is wider than the one that is actually realized at LF. Our argument has four steps. First, we show that in generic contexts, universal quantifiers escape standard “scope-islands” (Section 1). Second, we show how the effects of wide scope in generic contexts can be achieved without syntactic wide scope (Section 2.1). Third, we show that this result is actually forced on us, once we take seriously certain independent issues concerning the interpretation of generic tense (Sections 2.2 - 2.4). Finally, the semantics of generic tense and, in particular, its interaction with focus, will yield some intricate new predictions, which, as we show, are borne out (Sections 3 - 5).
The paper presents an additional argument for a specific account of semantic binding: the flat-binding analysis. The argument is based on observations concerning sloppy interpretations in verb phrase ellipsis when the binder is not the subject of the elided VP. In one such case, it is important that one of the binders belong to the domain of the other. This case can be derived from the flat-binding analysis as is shown in the paper, while it is unclear how to account for it within other analyses of semantic binding.
Guess how?
(1996)
Quantificational determiners in Japanese can be marked with genitive case. Current analyses (for example by Watanabe, Natural Language and Linguistic Theory, to appear) treat the genetive case marker in these cases as semantically vacuous, but we show that it has semantic effects. We propose a new analysis as reverse partitives. Following Jackendoff (MIT-Press, 1977), we assume that partitives always contain two NPs one of which is phonologically deleted. We claim that, while in normal partitives the higher noun is deleted, in reverse partitives the lower noun is deleted.
We consider evidentials embedded in complement clauses with new data from Bulgarian. For Tibetan, Garrett has shown that embedded evidentials are always shifted to the perspective of the reported speech. In Bulgarian, we show that such a shift is almost never possible. This shows that Bulgarian evidentials should not be analyzed as modals, but rather as presuppositional.
This paper advances a purely presuppositional analysis of intonation. I first show that a inspiring recent article by Geurts and van der Sandt (Theoretical Linguistics, 2004) that pursues the same goal cannot account for multiple foci. Then, I show that if it is assumed that destressed rather than focussed material is semantically marked, multiple foci are accounted for correctly.
Decomposing questions acts
(2006)
Compositionality
(2007)