Linguistik
Refine
Year of publication
Document Type
- Article (1207)
- Part of a Book (784)
- Conference Proceeding (583)
- Working Paper (254)
- Review (178)
- Preprint (122)
- Book (108)
- Part of Periodical (60)
- Report (58)
- Doctoral Thesis (23)
Language
- English (1815)
- German (1057)
- Croatian (298)
- Portuguese (118)
- Turkish (39)
- French (21)
- Multiple languages (21)
- mis (16)
- Spanish (7)
- Polish (4)
Is part of the Bibliography
- no (3405) (remove)
Keywords
- Deutsch (435)
- Syntax (152)
- Linguistik (125)
- Englisch (123)
- Semantik (112)
- Spracherwerb (97)
- Phonologie (86)
- Rezension (77)
- Kroatisch (68)
- Fremdsprachenlernen (67)
Institute
- Extern (438)
- Institut für Deutsche Sprache (IDS) Mannheim (113)
- Neuere Philologien (43)
- Sprachwissenschaften (43)
- Universitätsbibliothek (5)
- Sprach- und Kulturwissenschaften (3)
- Gesellschaftswissenschaften (2)
- Medizin (2)
- Präsidium (2)
- SFB 268 (2)
Algorithmic approaches to anaphor resolution are known to benefit substantially from syntactic disjoint reference filters. Typically, however, there is a considerable gap between the scope of the formal model of grammar employed for deriving referential evidence and its implementation. While accounting for many subtleties of language, such formal models at most partially address the algorithmic aspects of referential processing. This paper investigates the issue of implementing syntactic disjoint reference for robust anaphor resolution. An algorithmic account of binding condition verification will be developed that, on one hand, captures the theoretical subtleties, and, on the other hand, exhibits computational efficiency and fulfils the robustness requirements. Taking as input the potentially fragmentary parses of a robust state-of-the-art parser, the practical performance of this algorithm will be evaluated with respect to the task of anaphor resolution and shown to be nearly optimal.
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.
This contribution is concerned with integrating the phenomenon of selectional restrictions in HPSG. Firstly, the question of treating selectional restrictions purely in the semantic module is tackled, as there are some contextual (or pragmatic) influences, which can repair the ill-formedness of violated selectional restrictions. Secondly, we present existing approaches to selectional restrictions within the framework and, lastly, make our own proposal which describes the subject as part of the semantics-pragmatics interface. In particular, we show how a semantic ontology can be integrated.
HPSG accounts of filler-gap dependencies hold considerable potential for explaining the cross-linguistic variation in unbounded dependency constructions (UDCs), specifically filler-gap dependencies. This potential comes from the SLASH specifications that are posited in all nodes along the extraction path (the path between filler and gap). However, as Hukari and Levine (1994, 1995, 1996) have observed, the HPSG analysis presented by Pollard and Sag (1994) fails to embody the generalizations required in order to explain key universal properties of UDCs, in particular the ˋregistration' of such dependencies in cases of subject- and adverb-extraction. This demonstration led Bouma et al. (2001) to propose a revised UDC analysis that avoids these difficulties by ˋthreading' the SLASH specfications through all heads within an extraction domain. However, Levine (2002) points out that this analysis encounters a new difficulty concerning the interaction of extraction and coordination. This paper revisits these issues, arguing that a small modification of the BMS analysis provides a solution to the important problem observed by Levine.
This paper investigates the binding of pronouns and reflexives in picture noun phrases, and focuses on data showing that reflexives and pronouns are not in complementary distribution in picture NPs with possessors. In particular, we discuss data showing that whereas reflexives can take either the possessor or the subject of the sentence as antecedent, pronouns are restricted to an antecedent other than the possessor phrase. We suggest that this asymmetry can be straightforwardly explained if we assume that (1) the possessor of a picture NP is not part of the head noun's argument structure and (2) Binding Theory is stated over dependents structure, the representation encompassing both a head's argument structure and other phrases dependent on it in various ways. If the possessor of a picture NP (PNP) is not part of the head's argument structure, it follows that reflexives in PNPs with possessors will be exempt from Binding Theory, which paves the way for an analysis of the reflexive data. Furthermore, we also show that if BT is regarded as defined over dependents structure, it follows that a pronoun in a picture NP with a possessor must be disjoint from that possessor phrase.
Remarks on binding theory
(2005)
We propose some reformulations of binding principle A that build on recent work by Pollard and Xue, and by Runner et al. We then turn to the thorny issue of the status of indices, in connection with the seemingly simpler Principle B. We conclude that the notion of index is fundamentally incoherent, and suggest some possible approaches to eliminating them as theoretical primitives. One possibility is to let logical variables take up the explanatory burden borne by indices, but this turns out to be fraught with difficulties. Another approach, which involves returning to the idea that referentially dependent expressions denote identity functions (as proposed, independently, by Pollard and Sag and by Jacobson) seerms to hold more promise.
It has been commonly assumed since Chomsky (1981) that the distribution of reflexive pronouns is subject to Binding Condition A. Reinhart and Reuland (1993) formulate Condition A in terms of the notion of syntactic predicate. The proposal I will develop in this paper is to factor out semantic and syntactic conditions on the occurrence of reflexive pronouns and to reduce them to independently motivated semantic and syntactic mechanisms. The semantic part is attributed to a theory of semantic composition recently developed by Chung and Ladusaw (2004), while the syntactic residue falls into the proper characterization of syntactic chains, as proposed by Reinhart and Reuland (1993) and Reuland (2001). To the extent that this approach is successful, Binding Condition A is rendered superfluous.
The process of turning a hand-written HPSG theory into a working computational grammar requires complex considerations. Two leading platforms are available for implementing HPSG grammars: The LKB and TRALE. These platforms are based on different approaches, distinct in their underlying logics and implementation details. This paper adopts the perspective of a computational linguist whose goal is to implement an HPSG theory. It focuses on ten different dimensions, relevant to HPSG grammar implementation, and examines, compares, and evaluates the different means which the two approaches provide for implementing them. The paper concludes that the approaches occupy opposite positions on two axes: faithfulness to the hand-written theory and computational accessibility. The choice between them depends largely on the grammar writer's preferences regarding those properties.
Order domains were originally proposed to deal with constituent order, but have recently been concerned with more than just linearization. This paper seeks to contribute to this discussion by considering the possibility of analysing word forms in terms of order domains. We focus on the distribution of the English relative and interrogative pronouns who and whom. It is shown that a small number of constraints can accommodate the seemingly complex body of data. In particular, a linearization-based constraint can provide a straightforward account for the quite puzzling distribution which who and whom show in one of the register types.
In this paper, we claim that the filler-gap linkage in Korean UDCs needs to be handled at the level of syntax and that unbounded dependencies represented by traces, resumptive pronouns, and resumptive reflexives in Korean can be simply captured - without posing any extra mechanisms - in the traditional HPSG analysis of UDCs following Pollard and Sag (1994). It is because in HPSG traces are not all required to have the same feature, unlike in other movement-based approaches including the minimalist program and GB theory. In addition, we argue that the three kinds of Korean UDC elements appearing in gap positions do not form separate categories from their corresponding forms appearing in non-UDCs based on the same semantic and pragmatic properties such as logophoricity and contrastiveness. We also investigate some controversial issues of island constraints and strong crossover with respect to filler-gap linkage in Korean UDCs.
This paper presents an analysis of constructions involving the l-form of the verb in Polish, including primarily the past tense, the conditional mood, and the future tense. Previous approaches have attempted to treat these uniformly as auxiliary verb constructions. We argue against a unified treatment, however, in light of synchronic and diachronic evidence that indicates that only the future tense and the conditional still involve auxiliaries in modern Polish. We show that the past tense is now a simple tense, although the l-forms appear in combination with agreement affixes that can appear in different places in the sentence. We provide an account of the common linearization properties of the past tense markings and the conditional auxiliary. We present a detailed HPSG analysis of the past tense construction that relies on the introduction of two interacting agreement features. We then discuss the consequences of our proposals for the analysis of the conditional and future auxiliary constructions, and finally, we offer a treatment of constructions involving inflected complementizers in Polish.
We propose a formulation of 'relative tense theory' as applied to the complex tenses in Japanese using an HPSG grammar with its semantics represented based on Discourse Representation Theory. Specifically, a mechanism of tense interpretation which consecutively combines Discourse Representation Structures in parallel with an otherwise motivated hierarchical syntactic structure is presented. The framework provides a concise and comprehensive account of the Japanese simplex and complex tenses; it is made possible by the use of a hypothetical temporal point temporal axis, which relates the location time of the relevant clause to an outer temporal point, and a modal semantic operator woll.
This paper discusses how the English Resource Grammar (ERG) captures the optionality of certain complements of verbs based on a single lexical entry coupled with an ontology of markings distinguishing optional from obligatory as well as unrealized from realized elements. Subject-head and head-complement structures are modified accordingly, but due to the lack of a possibility to express and use relational goals in grammars implemented in the LKB system, the ERG encoding falls short of the goal of treating optional complements in a general way. Instead, it requires two new types of ˋauxiliary' phrases which are otherwise unmotivated. We show that the problem can be overcome by using a recursive relation selecting a member from a list. The use of a lean implementation platform not supporting such relational goals, such as the LKB, thus results in a loss of generality of the grammars that can be expressed, which undermines the closeness of the implemented grammar to current linguistic analyses as one of the hallmarks of HPSG-based grammar implementation. The case study presented in this paper thus supports the position argued in Götz and Meurers (1997) that a system for the implementation of HPSG-based grammars should include both universal implicational principles as well as definite clauses over feature terms.
In this paper, I first make an observation that there is a certain parallelism in the scope interpretation possibilities of adverbs and quantifiers with respect to different types complex predicates in Japanese, drawing on a comparison of the light verb construction and the causative construction. I will then argue that previous approaches to complex predicates in Japanese in the lexicalist tradition (Matsumoto 1996; Manning et al. 1999) fail to capture this generalization successfully. Finally, building on a novel approach to syntax/semantics interface in HPSG by Cipollone (2001), I develop an analysis of the semantic structure of complex predicates that accounts for the empirical observation straightforwardly.
We argue here for a lexicalist analysis of the Korean copula (following Kim, Sells and Wescoat (2004)), on the basis of different properties of sequences of noun-plus-copula, which shows word-like behavior, in contrast to noun and negative copula, which are independent syntactic units. The interactions of these items with various copy constructions brings out their clear differences. The analysis is formalized in HPSG using Lexical Sharing, from Wescoat (2002).
According to the Projection Principle (Chomsky 1981), expletives have no semantic content and thus cannot occur in theta-marked positions. However, there are many examples where expletive it appears as a direct object, in violation of the Projection Principle. The various attempts that have been made to account for such cases (e.g. the case-based analysis of Authier (1991), the predication analysis of Rothstein (1995), and the Specifier analysis of Stroik (1991, 1996)) all posit movement of the expletive from a non-theta marked position to direct object position. However, these analyses have so far been unsuccessful in capturing several important contrasts, e.g. variable optionality of the expletive it. This paper argues that such contrasts (and the complex behavior of expletive it more generally) follow straightforwardly from a lexicalist, constraint-based analysis in which lexical information and independently motivated constraints interact in subtle ways.
Focusing on the examples of multiple degree modification, this paper argues that the class of degree expressions in English is syntactically and semantically diverse, subdivided both according to the semantic effects of its members and according to the extent to which they permit, and participate in, multiple layers of modification. We argue that these two factors are linked, and result in (at least) a three-way distinction between ˋtrue degree morphemes', which map gradable adjectives to properties of individuals and combine with their arguments in a Head-Specifier structure; ˋintensifiers', which are syntactic and semantic modifiers of properties constructed out of gradable adjectives; and ˋscale modifiers', which are also syntactic and semantic modifiers, but which combine with ˋbare' gradable adjectives (relations between individuals and degrees) rather than properties formed out of gradable adjectives.
The present paper investigates a certain subset of clause linkage phenomena and develops a constraint-based account to the empirical fact that clauses need to be distinguished with respect to their degree of integratedness into a potential matrix clause. Considering as example German, it is shown that the generally assumed twofold distinction between main and subordinate clauses (or root and embedded clauses) does not suffice to deal with the presented data. It is argued that the discussed linkage phenomena originate from syntactic, semantic and pragmatic properties of the clauses involved, and should hence be encoded in grammar.
The paper reviews basic patterns of reflexive binding in Norwegian, and explores a possible implementation of them in an HPSG grammar using the LKB platform. Norwegian has two reflexive elements, with distinct constraints and corresponding 'anti-binding' effects; they can cooccur but also occur independently. As over-all strategy for resolving reflexive binding we use one resembling the 'slash' procedure for wh-dependencies. Binding constraints are imposed partly through lexical specification, partly through phrasal combination rules. Challenges are noted residing in the possibility for sentences to contain an unbounded number of reflexives.
We examine how a large-scale computational grammar can account for the complex nature of Japanese verbal compounds. Previous computational Japanese grammars have tried to avoid the problem by simple solutions such as enumerating as many verbal compounds in the lexicon as possible. In contrast, we develop the analysis that is linguistically adequate and computationally tractable and thus meets the requirement of a syntactically and semantically precise natural language processing of Japanese like Bond et al. (2005). Our analysis distinguishes between two kinds of verbal compounds: syntactic compounds, which are fully productive; and lexical compounds, which are of varying productivity.
We present a novel well-formedness condition for underspecified semantic representations which requires that every correct MRS representation must be a net. We argue that (almost) all correct MRS representations are indeed nets, and apply this condition to identify a set of eleven rules in the English Resource Grammar (ERG) with bugs in their semantics component. Thus we demonstrate that the net test is useful in grammar debugging.
On binding domains
(2005)
In this paper I want to explore reasons for replacing Binding Theory based on the anaphor-pronoun dichotomy by a Binding Theory allowing more domains restricting/defining anaphoric dependencies. This will, thus, have consequences for the partitioning of anaphoric elements, presupposing more types of 'anaphors'/'pronouns' than standard Binding Theory offers us.
The Grammar Matrix is a resource for linguists writing grammars of natural languages; however, up to this point it has not included support for coordination. In this paper, we survey the typological range of coordination phenomena in the world's languages, then detail the support, both syntactic and semantic, for those phenomena in the Grammar Matrix. Furthermore, we describe the concept of a Matrix module and our software that enables grammar writers to easily produce an extensible starter grammar.
Syncretism in German: A unified approach to underspecification, indeterminacy, and likeness of case
(2005)
In this paper I address the phenomenon of syncretism in German and show how Flickinger (2000)'s approach to related issues in English can be adapted to provide a compact, disjunction-free representation of German nominal paradigms by means of combined case/number/gender type hierarchies. In particular, I will discuss the issue of case identity constraints in German coordinate structures, which has so far prevented successful application of Flickinger's proposal to German, and show how likeness constraints targetting individual inflectional dimensions of a combined type hierarchy can be expressed by means of typed lists that abstract out the relevant dimension.
I further show that current type-based approaches to feature neutrality are unable to combine the treatment of this phenomenon with the virtues of underspecification. I will then propose a revised organisation of the inflectional type hierarchies suggested by Daniels (2001), drawing on a systematic distinction between inherent and external (case) requirements.
This paper proposes a distinct approach to local binding effects for reflexives and pronominals in English whereby the nature of local binding domains is a by-product of the incremental interpretation of syntactic derivations (Uriageraka 1999, Chomsky 2000, 2001), emphasizing the role of the Conceptual /Intentional interface and the computational system (i.e. bare output conditions) in shaping general principles of grammars. A significant development of the Minimalist framework is the proposal that derivations operate through phases or multiple spell outs, which allows to reduce the strict cyclicity of derivations, and related locality effects of movement, to interface (bare output) conditions and economy conditions. In this paper I propose that incremental interpretation can further capture local binding domains effects of conditions A and B of Chomsky's (1981, 1986) Binding Theory. Basically, local binding domains are shown to correspond to accessible phase domains. Our proposal hence contrasts with standard analyses (e.g. Reinhart and Reuland 1993, Pollard and Sag 1992) that define co-argumenthood as the core factor from which binding conditions are developed. Our proposal also provides a new perspective on the core contrasts between A-chain and A-bar chain w.r.t. binding and scope reconstruction effects and argues that checking of the uninterpretable feature Case is what defines potential phase domains.
In this paper we argue that at least for some languages, when there are suitable o-commanders of its selectional domain, a reflexive in the bottom of its obliqueness hierarchy escapes exemption via a reshuffling of its local binding domain. The outcome of such reshuffling is that the local domain extends to include o-commanders of the reflexive in the subcategorization domain immediately upstairs, that is in the domain whose head predicator directly subcategorizes the domain headed by the predicator directly subcategorizing the reflexive.
In this paper we present a proposal to integrate pragmatic information, both from the preceding discourse and the extra-linguistic context, in the grammar. We provide an analysis of elliptical fragments according to how they are anchored to the context and the kind of resolution they require. We also present an alternative view about the syntax of fragments.
I examine the semantic contrasts exhibited by argument/oblique alternations (argument realization alternations where one or more participants may be realized either as a direct argument or an oblique). Previous HPSG accounts of these have proposed that alternating verbs are ambiguous, where each variant has a structured semantics that makes different participants more or less structurally prominent in the semantic representation. I argue that such accounts fail to capture the full richness of the contrasts exhibited by such alternations, and propose instead a model that derives alternations from the lexical entailments each verb associates with the alternating participant.
As has been shown in other Polynesian languages, in Tongan, adnominal elements can modify incorporated nouns in the noun incorporation construction. Two analysis are considered in this paper for understanding this construction within HPSG. The first, lexical sharing (Kim and Sells, this volume), views the verbs that include incorporated nouns as being single words corresponding to two syntactic atoms. However, this analysis makes incorrect predictions on the transitivity of incorporation clauses. A second analysis, extending Malouf (1999), views these words as verbs, but with some of the combinatorial properties of nouns. This offers both a better account of the data, and preserves the more restrictive theory of the morphology-syntax interface.
Georgian is a language allowing reflexives to be marked by ergative. The subject use of the Georgian reflexive phrase was first documented with causative verbs by Asatiani (1982). The later works such as (Amiridze and Everaert, 2000), (Amiridze, 2003), (Amiridze, 2004) discuss the use with object-experiencer verbs and transitive verbs on non-agentive reading. The present paper offers the first hand data on subject uses of the Georgian reflexive phrase with transitive verbs on their agentive reading in special contexts (such as a twin context, Madame Tussaud context, etc.) which are problematic for the Binding Theory of Chomsky (1981) as well as for the Reflexivity Theory of Reinhart and Reuland (1993). The data could be accounted for within the approach developed in (Reuland, 2001). However, the subject uses of the Georgian reciprocal ertmanet- leave the issue of subject anaphors open.
The principal aim of this paper is to present a comprehensive theory of coordination of unlikes, i.e., a theory that is capable of dealing with every phenomenon resulting from coordination of unlikes. The proposed theory accounts not just for standard cases of coordination of unlike arguments and coordination of unlike functors but also for cases involving single-conjunct agreement and what will be called each-conjunct agreement. In the course of the argumentation, it is also shown that, even in a language like English, predicate-argument agreement needs to be described in terms of a relational constraint that is not simply an identity requirement.
To model the pied piping in interrogative and exclamative clauses Ginzburg & Sag (2000) proposes a nonlocal head-driven treatment, thus emphasizing the resemblances with extraction. This treatment has a number of drawbacks: it relies on poorly motivated lexical rules and nonbranching phrase structure rules, it makes false predictions about pied piping in PPs, and it presupposes an implausible structure for NPs with predeterminers. To solve these problems I propose an alternative in which pied piping is treated as a local functor-driven dependency. Technically, the WH feature is integrated in the CATEGORY objects, and the propagation of its values is modeled by constraints which are independently needed for the treatment of other phenomena.
This paper focuses on aspects of the licensing of adverbial noun phrases (AdvNPs) in the HPSG grammar framework. In the first part, empirical issues will be discussed. A number of AdvNPs will be examined with respect to various linguistic phenomena in order to find out to what extent AdvNPs share syntactic and semantic properties with non-adverbial NPs.Based on empirical generalizations, a lexical constraint for licensing both AdvNPs and non-adverbial NPs will be provided. Further on, problems of structural licensing of phrases containing AdvNPs that arise within the standard HPSG framework of Pollard and Sag (1994) will be pointed out, and a possible solution will be proposed. The objective is to provide a constraint-based treatment of NPs which describes non-redundantly both their adverbial and non-adverbial usages. The analysis proposed in this paper applies lexical and phrasal implicational constraints and does not require any radical modifications or extensions of the standard HPSG geometry of Pollard and Sag (1994).
Relative clauses (RCs) in Persian are head-modifying constituents, all typically introduced by the invariant complementizer ke. Persian RCs are Unbounded Dependency Constructions (UDCs), containing either a gap or a resumptive pronoun (RP). In some positions only gaps are allowed, and in other positions only RPs. There are also some positions where both gaps and RPs are alternatively allowed. Illustrating the striking similarities between Persian gaps and RPs, I will provide an HPSG unified approach to take care of the dependency between the licensing structure and the gap/RP with a single mechanism, using only the SLASH feature. Similar to Pollard and Sag s (1994) approach to the bottom of the dependency, I will assume a special sign at the bottom. However, my sign may have a nonempty PHON value. I will introduce a feature called GAPTYPE which is a NONLOCAL feature whose value can be either trace or rp. I will introduce two constraints to capture the pattern of distribution of RPs and traces. At the top of the dependency, I will bind the nonempty SLASH at the complementizer point. I will propose a lexical entry for the complementizer ke that will account for the binding of SLASH by the feature BIND, which has a nonempty set as value.
In this contribution we propose a new module for handling idioms and distributional idiosyncrasies. Based on the concept by Richter/Sailer (1999) the new feature COLL (context of lexical licensing) plays the central role in our approach. We provide a way to handle decomposable and nondecomposable idioms and idioms containing bound words. Our module guarantees the co-occurrence of all idiom parts and of bound word and licensing context, respectively. A prerequisite for our analysis is a means to select for particular elements in the lexicon. We introduce another feature, LISTEME, which gives each lexical item its unique identifier andmakes it possible to select for a particular lexical word or phrase. Finally, we compare our proposal with alternative approaches and give some ideas regarding further applications beyond idiomaticity.
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.
Japanese is often taken to be strictly head-final in its syntax. In our work on a broad-coverage, precision implemented HPSG for Japanese, we have found that while this is generally true, there are nonetheless a few minor exceptions to the broad trend. In this paper, we describe the grammar engineering project, present the exceptions we have found, and conclude that this kind of phenomenon motivates on the one hand the HPSG type hierarchical approach which allows for the statement of both broad generalizations and exceptions to those generalizations and on the other hand the usefulness of grammar engineering as a means of testing linguistic hypotheses.
This paper discusses a special kind of syntax-semantics mismatch: a noun with a relative clause is interpreted as if it were a complement clause. An analysis in terms of Lexical Resource Semantics is developed which provides a uniform account for ''normal'' relative clauses and for the discussed type of relative clause.
This paper summarizes the architecture of Lexical Resource Semantics (LRS). It demonstrates how to encode the language of two-sorted theory (Ty2; Gallin, 1975) in typed feature logic (TFL), and then presents a formal constraint language that can be used to extend conventional description logics for TFL to make direct reference to Ty2 terms. A reduction of this extension to Constraint Handling Rules (CHR; Fruehwirth & Abdennadher, 1997) for the purposes of implementation is also presented.
In Müller, 2005 I provide evidence that suggests that linearization approaches that analyze German clause structure with discontinuous constituents cannot account for the German clause structure in an insightful way. In order to eliminate the very powerful concept of linearization domains and discontinuous constituents from the grammar, analyses of other phenomena which also rely on discontinuous constituents should therefore be revised.
In this paper, I develop an analysis of German depictive secondary predicates that differs from the one suggested in Müller, 2002 by assuming binary branching structures, verb movment, and continuous constituents instead of a linearization approach. Some shortcomings of previous analyses are pointed out and it is shown how linearization constraints regarding depictive predicate and antecedent can be modeled.
During the past fifty years sign languages have been recognised as genuine languages with their own syntax and distinctive phonology. In the case of sign languages, phonetic description characterises the manual and non-manual aspects of signing. The latter relate to facial expression and upper torso position. In the case of manual components these characterise hand shape, orientation and position, and hand/arm movement in three dimensional space around the signer's body. These phonetic charcaterisations can be notated as HamNoSys descriptions of signs which has an executable interpretation to drive an avatar.
The HPSG sign language generation component of a text to sign language system prototype is described. The assimilation of SL morphological features to generate signs which respect positional agreement in signing space are emphasised.
In this study we show that constituency is of limited importance for a proper treatment of the interaction between the linear position of a wa-marked nominal in a Japanese sentence and possible domains of contrastive focus, and that constraints concerning contrastive focus should be represented in terms of linear order and not constituency. Linearisation HPSG, where linear order is independent from constituency, provides a good basis for an analysis. Some constraints are provided in terms of order domains, and it is shown that these constraints can deal with the phenomena in question, and that the cases problematic for the constituency-based analyses can also be accounted for by our analysis.
In this paper I present two classes of double object constructions in Modern Greek, i.e., the genitive, as well as the double accusative, ditransitive constructions. I show that these two classes differ from one another in that not both of them permit derivational processes such as the formation of adjectival passives. I also look at the case properties associated with the verbs which head Modern Greek genitive and double accusative ditransitive constructions. Finally, the analysis I propose for these constructions in Modern Greek are formalized using the Minimal Recursion Semantics (MRS) framework of Copestake et al. (2001) and Copestake et al. (2003).
The Russian data presented in Perlmutter and Moore (2002) seem to call into question the standard analysis of raising within Head-driven Phrase Structure Grammar (HPSG): In Russian, the case marking of the raising target and raising pivot does not seem to be shared. In this paper, we show that the phenomena described by Perlmutter and Moore can receive another analysis, fully compatible with HPSG's theory of raising. We argue in addition that our account leads to a slightly simpler model of the Russian data than Perlmutter and Moore's. Crucially, our analysis is only available if we avail ourselves of a rich network of language-specific constructional schemata, a stance recently advocated within HPSG, following the lead of Construction Grammar.
Linking in constructions
(2004)
In this paper I will make an attempt to show how the linking normally done in the lexicon also can be done in constructions. The motivation behind this is the flexibility it gives the grammar writer in underspecifying lexical entries. Being too rigid about linking in the lexicon may lead to unsatisfying results such as multiple lexical entries for what one intuitively feels is just one lexical entry, or alternatively, lexical rules which are not morphologically motivated. The aim is to show that this can be avoided by letting constructions introduce the linking information instead.
We present a constraint-based syntax-semantics interface for the construction of RMRS (Robust Minimal Recursion Semantics) representations from shallow grammars. The architecture is designed to allow modular interfaces to existing shallow grammars of various depth - ranging from chunk grammars to context-free stochastic grammars. We define modular semantics construction principles in a typed feature structure formalism that allow flexible adaptation to alternative grammars and different languages.
Recent analyses of mismatches at the syntax-semantics interface investigate e.g. modification of agentive nouns (Larson, 1998), modification of quantifying pronouns (Abney, 1987), or recursive modification (Kasper, to appear). Each of these analyses is tailored to a specific set of data, and it is not immediately obvious how they could be generalised to cover a larger set of data. I propose a unified analysis for these mismatches that attempts to bring out their common ground. This analysis shares some of its basic intuitions with the one of Kasper, but is more general because the mismatches are handled locally in the CONT feature. Its pivot is an elaborate syntax-semantics interface that is based on a surface-oriented syntactic analysis. This analysis generalises easily to the mismatches at the morphology-semantics interface for German separable-prefix verbs that were discussed in (Müller, 2003).
Negation and negative indefinites raise problems for the principle of compositionality of meaning, because we find both double and single negation readings in natural languages. De Swart and Sag (2002) solve the compositionality problem in a polyadic quantifier framework. The syntax-semantics interface exploits an extension of the Cooper storage mechanism that HPSG uses to account for scope ambiguities. In de Swart and Sag (2002), all negative quantifiers are collected into an N-store, and are interpreted by means of iteration (double negation) or resumption (negative concord) upon retrieval. This puts the ambiguity between single and double negation readings in the grammar, rather than in the lexical items. This paper extends the earlier analysis with a typology of negation and negative indefinites using bi-directional optimality theory (OT). The constraints defined are universal, but their ranking varies from one language to the next. In negative concord languages, the functional motivation for the marking of 'negative variables' wins out, so we use n-words. Double negation languages value first-order iteration, so we use plain indefinites or negative polarity items within the scope of negation. The bi-directional set-up is essential, for syntactic and semantic variation go hand in hand.
Linearization-based HPSG theories are widely used for analyzing languages with relatively free constituent order. This paper introduces the Generalized ID/LP (GIDLP) grammar format, which supports a direct encoding of such theories, and discusses key aspects of a parser that makes use of the dominance, precedence, and linearization domain information explicitly encoded in this grammar format. We show that GIDLP grammars avoid the explosion in the number of rules required under a traditional phrase structure analysis of free constituent order. As a result, GIDLP grammars supportmore modular and compact grammar encodings and require fewer edges in parsing.
In this paper, I shall discuss the semantic attachment of intersective modifiers in German coherent constructions. I shall show that a purely syntactic solution to the observable attachment ambiguity is undesirable for reasons of processing e ciency and/or massive spurious ambiguity. Instead, I shall follow Egg and Lebeth (1995) and propose an extension to Minimal Recursion Semantics, permitting the expression of underspecified semantic attachment. This rather trivial move, as we shall see, will not only be preferable for processing reasons, but it will also be more in line with the spirit of underspecified semantics, e ectively providing a compact representation of purely semantic distinctions, instead of unfolding these distinctions into a rain forest of tree representations and derivations. I will present an implementation of the underspecification approach integrated into the German HPSG developed at DFKI and compare its e ciency to an alternative implementation where semantic attachment is unfolded by means of retrieval rules.
Recent syntactic theory has highlighted the importance of peripheral constructions such as the comparative correlative construction. This construction involves a pair of filler-gap constructions with unusual properties. The first is a subordinate clause and the second a main clause. The construction has a number of related constructions. A version of HPSG, which assumes hierarchies of phrase types, can provide satisfactory analyses both for the comparative correlative constructions and for the related constructions. The two clauses in the comparative-correlative construction can be analysed as non-standard head-filler phrases differing from standard headfiller phrases in certain respects. The construction as a whole can be analyzed as a non-standard head-adjunct phrase, in which the head and the phrase have different categories.
This article proposes a semantics of directional expressions in Norwegian and German, regarded as VP modifiers. The analysis uses Minimal Recursion Semantics, as an integrated part of Matrix-based HPSG grammars. Directional expressions are analyzed as modifying an individual, the 'mover'. Context dependent directionals like here receive a decomposed analysis. Telicity values reflecting various types of directional and locative expressions are computed.
Within the tradition of Categorial Grammar, so-called 'non-constituent' coordination ('argument cluster' coordination and 'right node raising') has been analyzed in terms of the coordination of nonstandard constituents produced by the operations of type raising and composition. This highly successful research has expanded the domain of data that modern analyses of coordination must take into account. Recent HPSG work by Yatabe (2002) and Crysmann (2003) provides an interesting alternative approach to this problem in terms of the coordination of familiar, but 'elliptical' constituents. We argue that this approach is empirically superior to the Categorial Grammar analysis, both in terms of empirical coverage and cross-linguistic predictions. We reassess the relevant English data in small but important ways, and develop our own ellipsis analysis, building on Yatabe's and Crysmann's insights.
This paper presents an account of English non-restrictive ('appositive') relative clauses (NRCs) in the framework of 'construction based' HPSG. Specifically, it shows how the account of restrictive relative clause constructions presented in Sag (1997) can be extended to provide an account of the syntax and semantics of NRCs and of the main differences between NRCs and restrictive relatives. The analysis reconciles the semantic intuition that NRCs behave like independent clauses with their subordinate syntax. A significant point is that, in contrast with many other approaches, it employs only existing, independently motivated theoretical apparatus, and requires absolutely no new structures, features, or types.
Of all French functional elements, the form de has without question the widest variety of uses, and presents the greatest challenge for linguistic description and analysis. Historically a preposition, it still has a number of prepositional uses in modern French, but in many contexts it calls for an altogether different treatment. We begin by outlining a general distinction between oblique and non-oblique uses of de. We then develop a detailed account of constructions where de combines with an N'. We provide a unitary analysis of de in three constructions (quantifier extraction, "quantification at a distance", and negative contexts) which have been not been considered to be related in previous accounts.
Specificational pseudoclefts (SPCs) have been a great challenge for a syntactic theory, because, despite the surface division between the pre- and post-copular elements, the post-copular 'pivot' behaves as if it occupied the gap position in the precopular wh-clause. This paper argues that movement-based or deletion-based syntactic approaches and purely semantic approaches have problems in dealing with syntactic properties and connectivity problems of SPCs in English. Observing the parallelism between SPC pivots and short answers to questions, it proposes an HPSG account based on a non-deletion-based QDT (Question-in-disguise theory) approach and on the equative analysis of the specificational copular sentences. The paper shows that SPCs must be handled by an integrated account of the syntactic, semantic, and pragmatic properties of the construction, and argues that the connectivity problems should be approached from such an integrated view.
On the notion 'determiner'
(2003)
Following a common practice in generative grammar, HPSG treats the determiners as members of a separate functional part of speech (Det). The status of the functional parts of speech is a matter of debate and controversy. The auxiliaries, for instance, are commonly treated as members of a separate functional category (Aux or Infl) in many variants of generative grammar, including GB, MP and LFG, but in GPSG and HPSG, it is a matter of equally common practice to treat them as members of V and to reject the postulation of a separate functional category, see Pullum & Wilson (1977) and Gazdar, Pullum & Sag (1982). This text makes a similar case for the determiners; more specifically, it argues that the determiners are categorially heterogeneous, in the sense that some are members of A, whereas others are members of N. The argumentation is mainly based on inflectional morphology and morpho-syntactic agreement data. The consequences of the categorial heterogeneity are hard to reconcile with the specifier treatment of the determiners in Pollard & Sag (1994) and with the Det-as-head treatment in Netter (1994), but it can smoothly be integrated in the functor treatment of the prenominals in Allegranza (1998) and Van Eynde (2002).
We present an approach to the interpretation of non-sentential utterances like B's utterance in the following mini-dialogue:
A: "Who came to the party?"
B: "Peter."
Such utterances pose several puzzles: they convey 'sentence-type' messages (propositions, questions or request) while being of non-sentential form; and they are constrained both semantically and syntactically by the context. We address these puzzles in our approach which is compositional, since we provide a formal semantics for such fragments independent of their context, and constraint-based because resolution is based on collecting contextual constraints.
Wasow (1977) argues that linguistic theory should recognize two qualitatively distinct types of rules: syntactic rules, which can affect more "superficial" grammatical function properties; and lexical rules, which affect deeper lexical semantic properties of lexical items. However, lexicalist theories of grammar have replaced syntactic rules with lexical rules leaving Wasow's dichotomy potentially unexplained. Our goal in this paper is to recapture Wasow's insight within a lexicalist framework such as HPSG. Building on Sag & Wasow's (1999) distinction between lexeme and word, we claim that there is a contrast between lexical rules that relate lexemes to lexemes (L-to-L rules) and lexical rules that relate words to words (W-to-W rules) and that these differences follow from the architecture of the grammar. In particular, we argue that syntactic function features (ARGST, VALENCE, etc.) are not defined for lexemes, while lexical semantic features (CONTENT) are. From this it follows that L-to-L rules can affect lexical semantic features, and not syntactic function features. In addition, since words are defined for syntactic function features, W-to-W rules can change them. In this paper, we support this hypothesis by examining certain differences between two types of Noun Incorporation construction, and their relation to other rules in the grammar. We argue that Compounding Noun Incorporation is an L-to-L type and that Classifier Noun Incorporation is a W-to-W type; we base our argument on the interaction of Noun Incorporation and Applicative Formation in the Paleo-Siberian language Chukchi and the isolate language Ainu.
This paper examines reprise questions: questions which request clarification of the meaning intended by a speaker when uttering a word or phrase. As such they can act as semantic probes, providing information about what meaning can be associated with word and phrase types. We present corpus evidence regarding the meaning of nouns and noun phrases, and argue that this evidence runs contrary to the usual treatments of semantics in HPSG, and to the traditional generalised quantifier view of NPs as sets of sets. Instead we outline an analysis of NPs as (possibly functional) sets of individuals.
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 shows that the Gerund Phrase (GP) in the Spanish Gerund Construction (e.g., El jefe entró a su oficina corriendo, lit. The boss entered his office running ) is sometimes a complement (in SGCC) and sometimes an adjunct (in SGCA). Although in both cases, the GP expresses a non-argument of the main lexical verb's denotation, it is a syntactic adjunct in SGCA and a syntactic dependent of the main clause s head in SGCC. We argue that there is a semantic correlate of this syntactic difference and propose a general principle that constrains the semantic relations that can hold between the denotata of heads and added members of their ARG-ST lists: The two denotata must be part of a larger macro-event in the sense of Talmy (2000). We further show that the relation between the events denoted by the gerund and main verbs involves four semantic conditions and that which subset of those four conditions are satisfied in a particular SGCC sentence determines what subkind of SGCC is involved.
It is a much-debated issue whether one should assume separate lexical entries for participles used in passive and perfect constructions or whether there is just one lexical entry that is used in different ways depending on whether a passive or perfect auxiliary is present in the clause.
In previous work I criticized approaches trying to analyze the passive with one lexical entry for making empirically wrong predictions and suggested a lexical-rule based approach were two different lexical items for the participle are licensed.
In this paper I show how Heinz and Matiasek's (1994) formalizations of Haider's (1986) ideas can be extended and modified in a way that both modal infinitives and control constructions can be captured correctly. The suggested analysis needs only one lexical item for participles, base form infinitives, and zu infinitives irrespective of their usage in active or passive like structures.
In Japanese, as in other classifier languages like Chinese and Malay, numerals do not directly quantize nouns, but first combine with a classifier to form a measure phrase (MP; cf. Aikhenvald 2000). From the perspective of constraint-based approaches to syntax/semantics, the mutual selective restriction between classifiers and nouns can be stated in terms of information-sharing and featural identity, to some extent parallel to the treatment of gender/number agreement (between determiner and noun, for instance) (cf. Pollard and Sag 1994; Kathol 1999). There are, however, data that challenge this line of approach to noun-classifier matching. We demonstrate in this paper that it is possible that a single noun is associated with different types of classifier, and show why they are problematic for unification-based approaches, similar to the situation with case syncretism in European languages (Ingria 1990 and others). Later in the paper, we argue that information-sharing between noun, predicate and classifier is not completely transitive, and present a formal analysis which models multiple selectional requirements with sets.
This paper compares transformation-based and constraint-based treatments of unbounded filler-gap dependencies, the latter specifically as articulated in terms of HPSG, and argues, contrary to the commonly made allegations of notational variance , that there is purely empirical evidence that is consistent with only the constraint-based account. Recent proposals to deal with parasitic gaps in terms of null pronominals and empty operators are unable to account for the phenomenon of symbiotic gaps, the apparent case mismatches found in parasitic gap constructions, or (in general) for the well-known across-the-board effects within coordinate structures.
Most researchers now agree that subcategorization correlates significantly with semantics. But this semantic component of linking has proved elusive. Most, if not all, theories of linking have, in pratice, resorted to constructs that are syntactic diacritics. We show in this paper that the implicit syntactic diacritics that plague the basic linking constraints posited in at least some of these theories can be eliminated, provided that (i) the metalanguage in which linguistic constraints are written allows for true implicational statements; (ii) one is willing to slightly increase the number of linking constraints. We focus in particular on the linking theory presented in Davis and Koenig 2000, Davis 2001, and Koenig and Davis 2000, but we maintain that our arguments apply, mutatis mutandis, to many other linking theories. We note some of the consequences of this view of linking, including: linking constraints are stated in terms of semantically natural classes of situations, a single entailment of a verb's argument is sufficient to determine its linking, and interaction among linking constraints restricts the range of possible lexical items.
In this paper we investigate the factors conditioning a morphological alternation on verbal heads in Lai. We show that this alternation eludes a simple characterization and instead exhibits a many-to-many form-function mapping. We will further show that the facts can be given a straightforward analysis in terms of default conditions based on valence and polarity, together with various constructional overrides. Our analysis thus follows recent proposals in HPSG, in particular Malouf (forthcoming), in using a constructional type hierarchy with defaults (cooperating constructions) as an alternative to an Optimality Theoretic system of ranked violable constraints.
In Jaeger (to appear) I have described clitic doubling in Bulgarian wh-interrogatives which constitutes a type of Superiority violation that cannot be accounted for by any existing analyses. By showing that clitic doubling of object wh-phrases marks topicality, I raised the hypothesis that many (or maybe all) so called Superiority effects in Bulgarian are due to topic-fronting of wh-phrases. Here, I provide further support for this hypothesis and show that there is also evidence for topic-fronting of non-object wh-phrases. Differences between colloquial and formal Bulgarian are restricted to how topical objects have to be realized at the source of the extraction (i.e. the VP), which also makes the account readily extendable to other multiple fronting languages. The complex ordering constraints on the left periphery are captured in a Linear Syntax approach (similar to but different from Kathol 2000).
In this article, the so-called wh-relative clause construction is investigated. The German wh-relative clauses are syntactically relevant as they show both, root clause and subordinate clause properties. They matter semantically because they are introduced by a wh-anaphor that has to be resolved by an appropriate abstract entity of the matrix clause. Additionally, the wh-relative clause construction is discourse-functionally peculiar since it evokes coherence. Besides these interesting empirical characteristics, whrelatives raise important theoretical questions. It is argued that the standard HPSG theory has to be extended to account for non-restrictive relative clauses in general, and to cope with the particular properties of the wh-relative construction.
This paper provides a constraint-based account of information-prosody correspondence within the HPSG framework. The starting point of the paper is Klein's (2000) account of prosodic constituency in HPSG. However, it departs from the standard syntactocentric architecture of grammar, and adopts a grammar design in which syntax, phonology, and information structure are generated in parallel, with all three applying to a common list of domain objects. It is shown that this theoretical architecture elegantly captures many of the various constraints that have been shown to hold in classical views of grammar.
This paper is concerned with such concepts as topic, focus and cognitive status of discourse referents, which have been included under the label information structure (alternatively information status), as they are related in some sense to the distribution of given and new information. It addresses the question of which information structural properties are best accounted for by grammatical constraints and which can be attributed to non-linguistic constraints on the way information is processed and communicated. Two logically independent senses of given-new information are distinguished, one referential and the other relational. I discuss some examples of linguistic phenomena that pertain to each of these different senses and show that both are linguistically relevant and must be represented in the grammar. I also argue that phenomena related to both senses have pragmatic effects that do not have to be represented in the grammar since they result from interaction of the language system with general pragmatic principles that constrain inferential processes involved in language production and understanding.
In this paper we present an analysis of English measure noun phrases. Measure noun phrases exhibit both distributional idiosyncrasy, in that they appear in positions normally filled by degree adverbs: "a ten inch long string"; and agreement discord: "ten inches is enough", "it is ten inch/*inches long". The analysis introduces one idiosyncratic construction, the Measure Phrase Rule, which links together syntax and inflectional morphology. Combined with existing rules, in particular the Noun-noun Compound Rule, the new rule accounts for the both the distributional and agreement idiosyncrasies. The rule has been implemented and tested in the ERG, a broad-coverage grammar of English. Our analysis supports the position that broad-coverage grammars will necessarily contain both highly schematic and highly idiosyncratic rules.
The paper investigates a complex word order phenomenon in German and the interaction of syntax and information structure it exemplifies: the occurrence of subjects as part of a fronted non-finite constituent and particularly the so-called definiteness effect excluding (many) definite subjects from this position. We explore the connection between focus projection and the partial fronting cases and show that it is the subject of those verbs which allow their subject to be the focus exponent that can be included as part of a fronted verbal constituent. In combination with the observation by Webelhuth (1990) that fronted verbal constituents need to be focused, this provides a natural explanation of the definiteness effect in terms of the information structure requirements in these sentences. Interestingly, the generally ignored exceptions to the definiteness effect are predicted by our analysis; we show that they involve definite noun phrases which can bear focus, which allows them to be part of a fronted verbal constituent. Finally, building on the integrated grammatical architecture provided in De Kuthy (2002), we formulate an HPSG theory which captures the interaction of constraints from syntax, information structure and intonation.
Much literature in syntax has assumed that all noun phrases are categorically headed by the determiner or the noun, with well-formedness categorial in nature. In this paper I develop a theory of noun phrase structure in which both categories project noun phrases, arguing that this better fits the indeterminacy of the criteria often cited for determining headedness (Zwicky, 1985, inter alia). The only categorial differences between determiners and nouns are their semantics and selectional restrictions, and the conditions that determine well-formedness are semantic in nature. Specifically, a well-formed noun phrase must have some restrictive semantics associated with nouns coupled with some operational semantics associated with determiners (e.g. as a generalized quantifier), and from this I show how we can derive structural well-formedness. Thus the need for categorial well-formedness is nullified, providing an analysis with greater cross-linguistic import, being compatible with languages without determiners.
While French degree words in French have been assigned several syntactic categories, we show that they are rather highy polymorphic adverbs (they occur in all syntactic domains), which select the expression they modify on a purely semantic basis. Like French adverbs in general, they occur both to the left and to the right of the head they modify. Following previous work (a.o. van Noord and Bouma 1994, Abeillé and Godard 1997, Bouma et al. 2000), we assign them two different grammatical functions, adjuncts and complements. Semantically, they differ from quantifiers. We follow Kennedy (2000) who analyzes them as scalar predicate modifiers. Finally, the specific syntactic constraints that characterize a subset of them can be shown to follow from, or be related to, their weight properties (Abeillé and Godard 2000). We conclude that their apparently idiosyncratic properties fit into a more general theory of grammar.
Leaving aside elliptical coordinations, it is striking that no agreement has been reached on the structure of basic coordinate constructions. We propose that:
- coordinate constructions are structurally asymmetric: the conjunction makes a subconstituent with one of the conjuncts.
- such constituents can have several functions: coordinate daughter, adjunct daughter or main clause.
In order to show that some conjuncts should be analysed as adjuncts, we focus on asymmetric cases of coordination, in which the order of the conjuncts cannot be reversed, taking examples from French, Welsh and Korean. We present an HPSG analysis which treats the coordinating conjunctions as weak heads, with lexical subtypes, and coordinate phrases as multi-headed constructions, with different subtypes.
It is generally accepted among psycholinguists that real-time human sentence processing proceeds incrementally from left to right. Recently proposals have been made in the domain of syntax to reduce phenomena which have hitherto been accounted for in terms of linguistic performance to linear structures given at the level of competence. Keeping in line with this tendency in research, this paper tries to reestablish the much discussed relationship between the two aspects of language, competence and performance: the issue of processing difficulty dependent on sorts of multiple clause embedding is addressed by incorporating into HPSG a mechanism reflecting left-to-right processing and memory costs calculated at each processing step.
This paper deals with case marking in auxiliary verb constructions (AVCs) in Korean, and investigates how the case marking pattern in AVCs can be explained in terms of structural case resolution in the spirit of Pollard 1994, Heinz & Matiasek 1994, and Przepiórkowski 1999. In this paper, a new set of data involving various combinations of auxiliary verbs is presented to point out problems for both transformational analyses based on head movement and previous HPSG analyses in which the final auxiliary verb solely determines the case of the complements of the whole complex predicate. This paper shows that while most auxiliary verbs "inherit" the case marking property of the preceding verb, the auxiliary verbs siph- 'want' and ha- 'act like' have an additional property of assigning nominative and accusative case, respectively, to their complements. The actual case assignment by these auxiliary verbs is made possible, however, depending on what other kind of auxiliary verbs they are combined with. Based on the complex predicate analysis of AVCs, this paper proposes that complicated case patterns in AVCs can be accounted for by classification of verbs/auxiliary verbs via distinct feature values and by the mechanism of structural case resolution.
The aim of this paper is to present a theory that explicitly characterizes patterns of summative agreement. The proposed theory builds on my own theory of PNR, presented in Yatabe 2001, and is based on the view that agreement results from a non-lexical constraint that regulates under what circumstances a domain object can be merged with other domain objects by the compaction operation.
As the name of the framework suggests, one of the driving forces behind traditional HPSG analyses is the notion of head. With the exception of a few non-headed constructions (i.e., mostly coordination), constructions are typically seen as being headed by a particular word or phrase with the nonhead constituting a complement, specifier or adjunct. The head determines the internal composition of a phrase and is responsible for its external distribution. Moreover, syntactic headedness, as determined by morphosyntactic criteria, is typically assumed to coincide with semantic headness. In the case of NPs, for instance, this means that the semantic contribution (including the index) of the entire phrase is provided by the element that is the head by morphosyntactic criteria (typically the noun).
In this paper, we intend to challenge this view of heads on the basis of two constructions from English. In both instances, we will argue that the constituents that are responsible for the internal combinatorial make-up of the phrase do not constitute heads because they fail both to determine the external distribution of the phrase and to contribute the semantic index of the projected phrase. At the same time, however, we will show that it is possible to view these cases not as random departures from more well-behaved headed constructions, but instead as particular instantiations of more general construction types which do not impose strict conditions on external headedness.
In this paper we investigate the phenomenon of verb-particle constructions, discussing their characteristics and the challenges that they present for a computational grammar. We concentrate our discussion on the treatment adopted in a wide-coverage HPSG grammar: the LinGO ERG. Given the constantly growing number of verb-particle combinations, possible ways of extending this treatment are investigated, taking into account the regular patterns found in some productive combinations of verbs and particles. We analyse possible ways of identifying regular patterns using different resources. One possible way to try to capture these is by means of lexical rules, and we discuss the difficulties encountered when adopting such an approach. We also investigate how to restrict the productivity of lexical rules to deal with subregularities and exceptions to the patterns found.
This paper presents a descriptive overview of liaison, giving an idea of the scope of the phenomenon and possible approaches to its analysis. As for the contextual conditions on liaison, in many cases, the traditional notions of obligatory and prohibited liaison do not reflect speakers' actual behavior. It turns out that general syntactic constraints cannot determine the systematic presence or absence of liaison at a given word boundary. At best, specific constraints can be formulated to target particular classes of constructions. To express such constraints, I propose a system of representation in the framework of HPSG. The use of EDGE features (introduced by Miller (1992) for a GPSG treatment of French) provides the necessary link between phrasal descriptions and the properties of phrase-peripheral elements.
The structure Accusativus cum Infinitivo (AcI) has been observed in a number of languages, amongst them Latin. Morphologically it consists of an NPacc and a VPinf . In Latin however, a finer distinction has to be drawn, as was already noticed by Bolkestein (1976) who differentiates "between actual accusative cum infinitive clauses and constructions existing of an object-noun in the accusative caseform and a complementary infinitive"(1976:263).
In this paper I develop a unified analysis of the Japanese passive, which provides a uniform syntactic/semantic representation of the alleged varieties of passives (direct, indirect, possessive) as a complex predicate that encodes the triadic relation of "lack of control" among an agent, undergoer and event. Various differences among the direct, possessive, and indirect passives (the adversative effect implicature, the possibility of reflexive binding, the animacy constraint on the subject, etc.) are explained as cooperative effects of the core syntactic/sematic properties of the passive morpheme -(r)are and functional/pragmatic factors like conversational implicature and empathy constraints.
Bulgarian Vocative in HPSG
(2003)
Crosslinguistically vocatives are an underexplored linguistic phenomenon and in different languages they can be highly idiosyncratic and complex (Levinson, 1987, p.71). Therefore, the problem, which is discussed in this paper, is not a language-specific one, in spite of the fact that most of the languages have their own repositories for marking the role of the addressee in the communicative utterances.
In our opinion this linguistic phenomenon needs its adequate treatment in HPSG because of three main reasons:
1) The vocative is supposed to be present on two levels: syntax and pragmatics. Therefore it needs more elaborate interpretation on the interface side, which, in HPSG, is more developed for morphology/syntax and syntax/semantics than syntax/pragmatics. Note that a challenge for the theory is the semantic weight of the vocatives with respect to the head sentence.
2) It will be useful for HPSG-oriented implementations, especially treebanks and dialogue systems.
3) On prosodic grounds the vocatives are often viewed as being 'side or extended parts' of the sentence and therefore - very close to the parenthetical constructions. From our point of view, both phenomena are pragmatic and hence, the treatment of vocative, presented here, could be generalized to cover other phenomena of pragmatic nature.
In our work the vocatives are viewed through the possibility of the integration/separation of their pragmatic, syntactic and semantic properties.
I examine Spanish and French agreement in sentences with "affective" N/A de N constructions, in terms of an agreement theory growing out of Pollard and Sag (1994, §2) and Kathol (1999), with a distinction between two kinds of agreement relations: index agreement and morphosyntactic concord. The application of this theory to hybrid nouns (Wechsler and Zlati'c, 2000) extends straightforwardly to affective constructions. Furthermore, Kathol's characterization of the difference between hybrid nouns in Spanish and French, which I pair with an interpretation in terms of the default unification mechanism of Lascarides and Copestake (1999), turns out to make correct predictions about subtle differences in predicate agreement with affective constructions in the two languages.
Free relatives in German basically behave as NPs. As is first noticed by Groos and Riemsdijk (1981), an interesting property of free relatives that they do not share with ordinary relative clauses is that the relative pronouns are sensitive to matrix case requirements as well as to subordinate ones.
The aim of this paper is to provide a semantic account of valence alternations in Modern Greek of the following general form:
(1) NPk V NPi [P NPj] —> NPk V NPj [P NPi]
In other words, the valence alternations in Modern Greek we focus on in this paper are the ones involving direct internal arguments (i.e., objects) and indirect prepositional complements. Such alternation patterns in Modern Greek characterize mainly the behaviour of verbal predicates which participate in the so-called Locative Alternation phenomena.
Ever since Chomsky's "On Wh-Movement" (Chomsky 1977) it has been assumed that topicalization and wh-question formation can be analyzed as instances of the same operation. Leaving certain features aside, this proposal carries over to the analysis of unbounded dependency constructions in HPSG since structurally, topicalization does not differ from wh-question formation in the analysis suggested in Pollard & Sag (1994: 157-163). In the present paper, we challenge this assumption and suggest an alternative analysis of unbounded dependency constructions. Here, topicalization and wh-question formation are considered as structurally different at least in certain languages. They may, however, be structurally identical in other languages. This difference is empirically reflected in patterns of relative clause extraposition. As has been pointed out by Culicover & Rochemont (1990: 28), an extraposed relative clause must not take an antecedent contained in a VP if the VP is topicalized but the relative clause is not.
Few ideas have proven as influential within the HPSG-based literature on German verb clusters as Hinrichs and Nakazawa's (1989) idea of argument composition. Its basic idea is that in verb clusters, the arguments of a main verb are realized as the dependents of the auxiliary which governs that main verb, and not directly as dependents of the main verb. Thus, for instance in (1a), the tense auxiliary haben governs the transitive main verb gewinnen. As the head of the cluster gewonnen hat, the auxiliary haben effectively takes over the arguments from the main verb.
Prenominals in Dutch
(2003)
For modeling the internal structure of noun phrases (Pollard and Sag, 1994, 385) treats the noun as the head and classi£es its dependents in terms of a three-fold distinction, £rst proposed in Chomsky (1970), between complements, adjuncts and speci£ers. For a phrase like the expensive picture of Sandy the structure looks as follows.
Clitic Climbing Revisited
(2003)
Presently, there is overall consent among researchers on Romance in HPSG (Miller and Sag, 1997, Abeillé et al., 1998, Monachesi, 1996, 1999) that bounded clitic climbing (CC) is best understood in terms of argument composition. Despite the fact that all current analyses of CC are based on the same core idea, individual analyses of this phenomenon differ.
In this paper, I shall propose a unified approach that will be applicable to CC in both French and Italian. The approach will be cast entirely in terms of valence lists, argument structure and slash, such that construction- or language-specific book-keeping devices can be eliminated. As a side-effect, this approach provides a more strengthened view of lexical integrity, in that morphological information, i.e. an argument's mode of realisation, will not be directly accessible for subcategorisation.
In terms of truth conditional meanings, there is no clear difference between (Korean) IHRCs (internally head relative) and EHRCs (externally headed relative). In the analysis of IHRCs, of central interest are thus (a) how we can analyze the constructions in syntax and (b) how we can associate the internal head of the IHRC clause with the matrix predicate so that the head can function as its semantic argument, and (c) what makes the differences between the two constructions. This paper is an attempt to provide answers to such recurring questions within the framework of HPSG.
Downward Unbounded Discontinuities in Korean: An IPSG Analysis of Concord Adverbial Constructions
(2003)
Although there is a lot of literature dealing with the classification and distribution of Korean adverbials, there does not seem to be any satisfactory work. This is partly due to the properties of the adverbials themselves, whose classification and distribution vary depending not only on the lexical properties of each adverbial but also on its distributional environment. However, the distinction between “regular adverbials” and “concord adverbials (CAs)” is very clear and plays a significant role in elucidating the properties of the adverbials as a whole. Theformer have only the function of modifying other phrases, while the latter show a correspondence to some specified elements in the sentence. One of the major differences between them is that a CA and what it corresponds to can be separated from each other unboundedly acrossclausal boundaries.