Refine
Year of publication
Document Type
- Conference Proceeding (14) (remove)
Language
- English (14)
Has Fulltext
- yes (14) (remove)
Is part of the Bibliography
- no (14)
Keywords
- Grammatiktheorie (14) (remove)
Comparison of the ellipsis-based theory of non-constituent coordination with its alternatives
(2012)
In this paper, I compare the ellipsis-based theory of non-constituent coordination proposed in Yatabe (2001) with three of its alternatives, namely the theory that has been widely accepted within the context of Categorial Grammar, Mouret's HPSG-based theory, and the theory proposed by Bachrach and Katzir in the framework of the Minimalist Program. It is found (i) that the CG-based theory of non-constituent coordination cannot deal with medial RNR, i.e. a subset of right-node raising constructions in which either all or a part of the right-node-raised material is realized at a location other than the right edge of the final conjunct, (ii) that Mouret's theory encounters similar difficulties when applied to RNR, and (iii) that Bachrach and Katzir's theory cannot be applied to left-node raising in English, has difficulty capturing the semantic inertness of medial RNR, and overgenerates in several ways. The ellipsis-based theory, on the other hand, appears to be consistent with all the observations.
A compound matrix
(2004)
This paper presents a suplement to the Grammar Matrix, namely what I call a Compound Matrix ; in reality, it is not a matrix, since the type file includes a fully specified cross-linguistic inventory of compound types. The idea is that the grammar writer can comment out the ungrammatical types for his or her own language. The theory behind the typology is presented here in a bottom-up fashion, from the basic assumptions to the actual linguistic types.
We present a method for automatic RMRS semantics construction from dependency structures, following the semantic algebra of Copestake et al. (2001). We have applied this method to a subset of the TIGER Dependency Bank for German (Forst et al., 2004) to obtain a semantic treebank for (HPSG) parser evaluation. We describe the semantics construction mechanism and give evaluation figures from manual validation of the treebank. These indicate high precision of the automatic RMRS construction process.
Progress toward distinguishing clearly between generative and model-theoretic syntactic frameworks has not been smooth or swift, and the obfuscatory term 'constraint-based' has not helped. This paper reviews some elementary subregular formal language theory relevant to comparing description languages for model-theoretic grammars, generalizes the results to trees, and points out that HPSG linguists have maintained an unacknowledged and perhaps unintended allegiance to the idea of strictly local description: unbounded dependencies, in particular, are still being conceptualized in terms of plugging together local tree parts annotated with the SLASH feature. Adopting a description language with quantifiers holds out the prospect of eliminating the need for the SLASH feature. We need to ask whether that would be a good idea. Binding domain phenomena might tell us. More work of both descriptive and mathematical sorts is needed before the answer is clear.
This paper seeks to improve HPSG engineering through the design of more terse, readable and intuitive type signatures. It argues against the exclusive use of IS-A networks and, with reference to the English Resource Grammar, demonstrates that a collection of higher-order datatypes are already acutely in demand in contemporary HPSG design. Some default specification conventions to assist in maximizing the utility of higher-order type constructors are also discussed.
This paper points out certain flaws in the semantics for lexical rule specifications developed in Meurers (2001). Under certain circumstances, certain words may not be licit inputs to a rule according to this semantics while one would expect them to be from inspecting the specification of the rule. The reasons for this are shown to be that whether properties of paths should be transferred from the input of a rule to its output is decided considering only the respective paths and their properties in isolation, ignoring the ‘non-local’ effects that transferring their properties can have. Furthermore, the semantics is insensitive to the possible shapes of inputs to the rule, which also makes it possible that inputs of certain shapes are unexpectedly not accepted. An alternative semantics is developed that does not suffer from these deficits.
In this paper, we report on a transformation scheme that turns a Categorial Grammar, more specifically, a Combinatory Categorial Grammar (CCG; see Baldridge, 2002) into a derivation- and meaning-preserving typed feature structure (TFS) grammar.
We describe the main idea which can be traced back at least to work by Karttunen (1986), Uszkoreit (1986), Bouma (1988), and Calder et al. (1988). We then show how a typed representation of complex categories can be extended by other constraints, such as modes, and indicate how the Lambda semantics of combinators is mapped into a TFS representation, using unification to perform perform alpha-conversion and beta-reduction (Barendregt, 1984). We also present first findings concerning runtime measurements, showing that the PET system, originally developed for the HPSG grammar framework, outperforms the OpenCCG parser by a factor of 8–10 in the time domain and a factor of 4–5 in the space domain.
1. There are two classes of theories of Universal Grammar: (1) Formalist theories, such as the widespread varieties of generative grammar. These theories start from the assumption that certain strings of linguistic forms are grammatical while other strings are ungrammatical. A grammar of this type produces grammatical strings and does not produce ungrammatical ones. All theories of this class fail in the same respect: they do not account for the meaning of the strings. (2) Semiotactic theories, which describe the meaning of a string in terms of the meanings of its constituent forms and their interrelations. The only elaborate formalized theory of this class presently available is the one advanced by C.L. Ebeling (Syntax and Semantics, Leiden: Brill, 1978). I shall discuss some of its mathematical properties here.
This paper desribes four areas in which grammar engineers and theoretical linguists can interact. These include: using grammar engineering to confirm linguistic hypotheses; linguistic issues highlighted by grammar engineering; implementation capabilities guiding theoretical analyses; and insights into architecture issues. It is my hope that we will see more work in these areas in the future and more collaboration among grammar engineers and theoretical linguists. This is an area in which HPSG and LFG as a distinct advantage, given the strong communities and resources available.
This paper presents a left-branching constructionalist grammar design where the phrase structure tree does not correspond to the conventional constituent structure. The constituent structure is rather reflected by embeddings on a feature STACK. The design is compatible with incremental processing, as words are combined from left to right, one by one, and it gives a simple account of long distance dependencies, where the extracted element is assumed to be dominated by the extraction site. It is motivated by psycholinguistic findings.