Linguistik
Refine
Year of publication
Document Type
- Article (1269)
- Part of a Book (784)
- Conference Proceeding (646)
- Working Paper (254)
- Review (181)
- Preprint (122)
- Book (109)
- Part of Periodical (64)
- Report (58)
- Doctoral Thesis (23)
Language
- English (1931)
- German (1055)
- Croatian (298)
- Portuguese (120)
- Turkish (43)
- Multiple languages (24)
- French (21)
- mis (16)
- Spanish (7)
- Polish (4)
Has Fulltext
- yes (3528) (remove)
Keywords
- Deutsch (436)
- Syntax (152)
- Linguistik (127)
- Englisch (123)
- Semantik (112)
- Spracherwerb (101)
- Phonologie (86)
- Rezension (77)
- Fremdsprachenlernen (69)
- Kroatisch (68)
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)
Recent years have witnessed a renewal of interest in variable morph ordering, the situation where the position of a morph in the word is not constant. These situations present a challenge to extant inferential-realisational approaches to morphology (Stump, 2001), insofar as these adopt implicitly or explicitly an a-morphous approach to morphological composition (Anderson, 1992). In this paper we will first review the typology of known variable morph ordering phenomena in inflection. We then argue that the challenges can be met by making a distinction between paradigmatic opposition classes and syntagmatic position classes, and show that this distinction can readily be implemented in HPSG while keeping the amorphous assumption.
In this paper I explore the logical range of sentential negation types predicted by the theory of HPSG. I find that typological surveys confirm that attested simple negation strategies neatly line up with the types of lexical material given by assuming Lexical Integrity and standard Phrase Structure Grammar dependencies. I then extend the methodology to bipartite negation and derive a space of predicted sentential negation types. I present details of the analysis for each type and relevant examples where possible.
Backshift is a phenomenon affecting verb tense that is visible as a mismatch between some specific embedded contexts and other environments. For instance, the indirect speech equivalent of a sentence like 'Kim likes reading', with a present tense verb, may show the same verb in a past tense form, as in 'Sandy said Kim liked reading'. We present a general analysis of backshift, pooling data from English and Romance languages. Our analysis acknowledges that tense morphology is ambiguous between different temporal meanings, explicitly models the role of the speech time and the event times involved and takes the aspectual constraints of tenses into consideration.
Verbal suffix-repetition construction in Korean: A constraint- and construction-based approach
(2012)
There are various Verbal Suffix-Repetition (VSR) constructions in Korean, where suffixes such as -kena/tun(ci)/tun(ka) are attached to the repeated verbs. Calling the VSR Choice-denying Repeated Verbs construction, Lee (2011) claims that the following verb of the VSR, which can be replaced with mal-, should contain a negative but the preceding verb should be affirmative in the VSR construction which disallows any NPI within it. Unlike Lee (2011), we claim that the verbs in the VSR can freely occur either in the preceding position or in the following one regardless of their Neg value so long as they share the same verbal suffix forms such as -tun(ka). Furthermore, NPIs may occur within the VSR construction if they occur with a negative predicate within the same clause. To implement the findings above into HPSG, we have proposed the two lexical entries for mal-, the VSR Construction Rule, and the NPI Clause-mate Constraint. These tools enable us to account for the idiosyncratic properties of the VSR constructions under this constraint- and construction-based approach.
The Korean double nominative construction exhibits various properties distinguished not only from ordinary subject-object clauses but also from nominative complement constructions. Particularly, the second NP, not the initial NP, triggers the honorific agreement with the verb. I argue that the first NP of the construction is identified as a sentential specifier which exists in addition to the subject (cf. Major subject in Yoon 2004). The sentential specifier can be justified as the characteristic of the topic-prominent language in the sense of Li and Thompson (1976). Specifically I claim that any elements that satisfy the aboutness condition can be the sentential specifier. Finally, I show that HPSG's valence value and an optional lexical rule provides an elegant treatment of the construction; SPR list in a sentence level can be utilized for the sentential specifier (cf. Kim et al. 2007).
This paper presents an analysis of Danish free relative constructions. Fol- lowing Bresnan and Grimshaw (1978) we will adopt a wh-head (in Danish hv-head) analysis where the hv-phrase is the head of an NP. Also following Bresnan and Grimshaw (1978) we will propose an analysis which does not involve a filler-gap dependency between the hv-phrase and the gap in the sis- ter clause. Instead we will propose that the gap in the sister clause is bound off by a constructional constraint. In this way the analysis will be shown to differ from previous HPSG wh-head analyses of free relatives.
In HPSG relative clauses have been analyzed in terms of phonologically empty heads in Pollard and Sag (1994) and in terms of a complex system of phrase types in Sag (1997). Modern Standard Arabic has a distinction between relative clauses with a definite antecedent, which are introduced by a special complementizer, and relative clauses with an indefinite antecedent, which are 'bare' clauses. Analyses eschewing empty heads and assuming a complex system of phrase types face a number of problems. An analysis in which relatives with an indefinite antecedent are headed by a phonologically empty complementizer is more satisfactory. Thus, in the case of Arabic, the approach of Pollard and Sag (1994) seems preferable to the approach of Sag (1997).
This paper describes free relative constructions in Modern Standard Arabic (henceforth, MSA) and aims to provide an HPSG analysis for them. MSA has two types of free relative constructions. One, which is introduced by the complementizer ?allaði, looks just like a relative clause. The other, which is introduced by the elements man and maa, which also appear to be complementizers, does not look like a relative clause. Both types can be analysed in term of unary-branching structures (as NPs consisting just of a CP). In ?allaði free relatives, the NP and the value of SLASH can be coindexed via the value of MOD on the CP. In man and maa free relatives, the NP and the value of SLASH must be coindexed directly.
Korean has two types of answers shorter than full sentential answers: Fragments and null argument constructions. Apparently the two constructions have the same interpretative processes. However, there are some cases where the fragment and null argument construction behave differently: e.g., wh-puzzles, sloppy interpretation. We suggest that the two constructions involve two different types of anaphora and that the sources of sloppy(-like) interpretation are fundamentally distinct. Fragments pattern differently with null arguments in that only the former may display genuine sloppy readings. The latter may yield sloppy-like readings which are pragmatically induced by the explicature that can be cancelled unlike genuine sloppy readings in fragments. Evidence (wh-ellipsis, quantifier ellipsis) all lends substantial support to our claim that fragments are analyzed as an instance of clausal ellipsis while null arguments are analyzed as an instance of null pronoun pro; hence, the former is surface anaphora whereas the latter is deep anaphora in the sense of Hankamer & Sag (1976).
Verb second (V2) word order is determined by considering the absolute position of clausal constituents. Previous accounts of such word order in HPSG have been developed for individual V2 languages (predominantly German) but are often not cross-linguistically applicable. I propose a set of generalized mechanisms in linearization-based SBCG which accounts for cross-linguistic V2 data by use of: (1) a simple two-valued feature rather than many-typed topological domains, (2) domain compaction, and (3) constructionally-determined domain positions. Not only does this analysis account for V2 placement, but it can also model verb third (V3) placement and other positionally-stipulated word orders.
The HPSG binding theory in Pollard and Sag (1994) cannot account for the binding-theoretic interaction between main clause and adjunct-internal elements. Following Hukari and Levine (1995), I claim that structural configurations must be taken into account. In this article, I present a revised version of Hukari and Levine's configurational relation called v(alence-based)-c-command and propose that Principle C must involve this relation in addition to the obliqueness-based relation of o-command. New data are provided that strongly support the proposed revision of the HPSG binding theory. Finally, I argue that Principle C is syntactic rather than pragmatic in nature.
This paper examines the morpho-syntactic puzzle of case suffixes and postpositions that Hungarian displays. Although these two categories show distributional similarities, they are distinguishable from a morphological and a syntactic point of view. Moreover, this language has defective postpositions which are in complementary distribution with case suffixes. I argue that there is no real argument for lumping case suffixes together with postpositions into the same syntactic category, as has been suggested in recent linguistics studies (Trommer, 2008; Asbury, 2007). I rather propose to treat case suffixes and postpositions as two different objects: case suffixes are inflectional material on nominal heads and postpositions as well as defective postpositions are independent words subcategorizing an NP. This distinction straightforwardly accounts for morphological and syntactic differences. Finally, the shared distributional properties between case suffixes, postpositions and defective postpositions are captured by means of the use of the MARKING feature.
This paper presents a Synchronous Tree Adjoining Grammar (STAG) account of Information Structure, whereby Givenness-marking requires a link between nodes on a syntactic tree and LF nodes whose interpretation is supplied by a contextually determined set of Given semantic objects. By hypothesis, the interpretation of linked nodes bypasses a default interpretation principle that requires pragmatic reasoning to disambiguate elements and enrich semantic material. Thus, interpreting Given elements requires less cognitive effort than Focused elements. This, combined with some established insights from Game-theoretic pragmatics, yields empirical advantages over more traditional semantic/pragmatic analyses of equal simplicity.
This paper hypothesizes that transfer-based machine translation systems can be improved by encoding information structure in both the source and target grammars, and preserving information structure in the transfer stage. We explore how information structure can be represented within the HPSG/MRS formalism (Pollard and Sag, 1994; Copestake et al., 2005) and how it can help refine multilingual MT. Building upon that framework, we provide a sample translation between English and Japanese and check the feasibility of the proposals in small-scale translation systems built with the HPSG/MRS-based LOGON MT infrastructure (Oepen et al., 2007). Our experiment shows the information structure-based MT system that we propose in this paper reduces the number of translations 75.71% for Japanese and 80.23% for Korean. The dramatic reductions in the number of translations is expected to make a contribution to our HPSG/MRS-based MT in terms of latency as well as accuracy.
Remarks on sluicing
(2011)
Sluicing is widely regarded as requiring an analysis via deletion operations, a potentially problematic conclusion for non-transformational frameworks like HPSG. We examine critically and reassess the motivation for a deletion analysis of Sluicing, offering cross-linguistic and language-internal evidence in support of a fundamentally semantic constructional alternative like the one proposed by Ginzburg and Sag (2000).
This paper deals with expletives that are inserted into clauses for structural reasons. We will focus on the Germanic languages Danish, German, and Yiddish. In Danish and Yiddish expletives are inserted in preverbal position in certain wh-clauses: In Danish such an insertion is observed when the subject is locally extracted from an SVO configuration in non-assertive clauses. In Yiddish wh-clauses are formed from a wh-phrase and a V2 clause. If no element would be fronted in the embedded V2 clause, an expletive is inserted in non-assertive clauses in order to meet the V3 requirement for embedded clauses. In addition to embedded wh-clauses, declarative V2 clauses also allow the insertion of an expletive. In Danish the expletive fills the subject position and is not necessarily fronted. In German and Yiddish the expletive has to occur in fronted position. In contrast to Danish and Yiddish, German does not insert expletives into embedded wh-clauses. They are inserted only into declarative V2 clauses in order to fulfill the V2 requirement without having to front another constituent. In this paper we try to provide an account that captures the commonalities between the three languages while being able to account for the differences.
My objective here is to assess the relevance of information structural notions for analyzing subject inversion in French. Subject inversion is not a unified phenomenon. In fact, there are three distinct constructions featuring an inverted subject. I show that the sentences do not have the same informational potential (the type of focus-ground articulation they are compatible with) depending on the construction they abide by. I propose a contextual factor – the informational solidarity between the verb and its first argument – to account for those differences. Then, I show that the three constructions share a common feature that pertains to a completely different dimension: the perspective chosen to describe the situation. I adopt Langacker's notion of absolute construal to characterize it. Finally, I present another common feature: the blocking of the referential anchoring of the referent of indefinite and partitive NPs.
Reanalysis of semantically required dependents as complements in the Chinese "ba"-construction
(2011)
The paper aims at a formulation of semantic constraints on the produc- tivity of the Chinese ba-construction and their representation at the syntax-semantics interface. It builds on the observation that requirements on the surface form of the construction may be altered by the choice of the verb. I propose that the semantics of the ba-construction can be treated in terms of a scalar constraint: a ba-sentence must come with a scale and a difference value that holds of the described event. The satisfaction of this constraint largely relies on the lexical semantics of the sentence. Not all verbs are inherently associated with scalar relations; those that are not must combine with an additional dependent which satisfies the scale requirement. Due to the obligatory presence of the additional dependent for some verbs, it is reanalyzed as a complement of ba: being optional on their level of combination with the verb, it becomes obligatory once the verb is used in the ba-construction.
This paper investigates the information-structural characteristics of extraposed subjects in Early New High German (ENHG). Based on new quantitative data from a parsed corpus of ENHG, I will argue that unlike objects, subjects in ENHG have two motivations for extraposing. First, subjects may extrapose in order to receive narrow focus, which is the pattern Bies (1996) has shown for object extraposition in ENHG. Secondly, however, subjects may extrapose in order to receive a default sentence accent, which is most visible in the case of presentational constructions. This motivation does not affect objects, which may achieve the same prosodic goal without having to extrapose. The study has two major consequences: (1) subject extraposition in ENHG demonstrates that there is not necessarily a one-to-one correspondence between syntactic structure and information structural effect (cf. Féry 2007); and (2) the overall phenomenon of DP extraposition in ENHG fits into a broader set of crosslinguistic focus phenomena which demonstrate a subject-object asymmetry (cf. Hartmann and Zimmermann 2007, Skopeteas and Fanselow 2010), raising important questions about the relationship between argument structure and information structural notions.
Much recent work on coordination in the HPSG framework seeks to deal with some of the most intractable issues this phenomenon poses for a constraint-based phrase structure architecture by appealing to the linearization mechanism introduced in Reape 1993. The research in question utilizes the mismatch between linear phonological sequences on the one hand and phrasal configuration on the other to underwrite a particular interpretation of ellipsis in which multiple structural objects with identical or near-identical descriptions are mapped to a single dom-object token. This mapping apparently allows a variety of problematic cases, such as right node raising, dependent cluster coordination, and unlike category coordination to be reinterpreted as instances of ordinary coordination in which structurally present elements receive no prosodic expression, creating the impression that strings which do not correspond to constituents of the same category have nonetheless been conjoined or disjoined. I argue in this paper that such linearization-based ellipsis (LBE) analyses, though plausible when confined to a narrow class of simplest-case data, prove untenable in the face of data sets in which the LBE approach must account for the interaction of nonconstituent coordination and quantification or symmetric predication, symmetrical modification of nominal heads, and a large and varied class of unlike category coordinations that do not admit of any ellipsis-based solutions. I show in addition that various objections offered in the LBE literature to categorial grammar treatments of the problems posed by noncanonical coordinations do not take into account techical resources available to CG which permit straightforward and unproblematic solutions to these problems. One must conclude that despite the general poplularity of LBE accounts of conjunction, there is at the moment no satisfactory HPSG treatment of noncanonical coordinations.
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.
English Binominal NPs (BNP) (e.g., a hell of a problem) are of empirical and theoretical interest due to their complex syntactic and semantic properties. In this paper, we review some basic properties of the BNP construction, focusing on its headedness, semantic relations, and the role of the preposition of. We argue that these properties suggest an account in the spirit of construction grammar. In particular, we show that English BNP is a nominal juxtaposition construction whose special syntactic constraints are linked to semantic relations like a subject-predicate relation.
Focus particles, secondary meanings, and Lexical Resource Semantics: The case of Japanese "shika"
(2011)
Japanese has two exclusive particles 'shika' and 'dake'. Although traditionally, both particles were considered to be exclusive particles like 'only', a recent proposal claims that 'shika' is an exceptive particle like 'everyone except' to account for the necessary co-occurrence of the negative suffix 'na' and 'shika'. We show that this negative suffix lacks two critical semantic properties of ordinary logical negation: It is not downward entailing, nor does it license negative polarity items. We show that both 'shika' and 'dake' are exclusive particles, but that 'shika' encodes an additional secondary meaning. The negative suffix only contributes to the sentence's secondary meaning when it co-occurs with 'shika'. We present an HPSG and LRS analysis that models the co-occurrence of 'shika' and the negative suffix 'na', and their contribution to the sentence's secondary meaning.
This paper presents a descriptive overview and a formal analysis of the syntax of pronominal arguments, pronominal conjuncts and bound pronouns in Arabic. I argue that Arabic allows first conjuncts to be null and that this is an instance of a more general pattern of zero anaphora that may affect pronominal arguments or their first conjuncts. First Conjunct Agreement and constraints on the distribution of zero anaphora are accounted for by a new feature sharing mechanism which allows a uniform treatment without appeal to the internal structure of argument NPs. I then argue that Arabic bound pronouns should be analyzed as affixes and present an analysis of their relation to argument structure and coordination. Finally, it is shown how constraints on case marking in Arabic coordination can be formalized. The analysis is part of an Arabic grammar fragment implemented in the TRALE system.
In this article we show how the HPSG approach to information structure of De Kuthy (2002) and De Kuthy and Meurers (2003) can be extended to capture givenness (Schwarzschild, 1999) and make the right predictions for so-called deaccenting of given information, a widespread phenomenon not previously dealt with in HPSG.
We explore the interaction of sentential negation and word order in Basque using a small experimental implemented grammar based on the Grammar Matrix (Bender et al., 2002, 2010) to test the analyses. We find that the analysis of free word order (Fokkens, 2010) provided by the Grammar Matrix customization system can be adapted to handle the Basque facts, and that the constructional approach taken in that analysis supports the integration of negation.
The present paper proposes an analysis of the asymmetrical distribution of der, 'there', in embedded interrogative and relative clauses, respectively, in standard Danish. The analysis sets itself apart from previous analyses in integrating information structural constraints. We will show that the discourse function of the extracted subject in the clauses in question determines whether der insertion takes place in standard Danish. The analysis will further be shown to support the position that der in interrogative and relative clauses is an expletive subject filler, and that from an information structural point of view, the der in existential, presentational, passives and relative clauses is indeed the same der.
This paper examines the apparently odd location of case-marking formatives found in the Pacific Northwest language, Coast Tsimshian. It first argues that the case-marking formatives are actually affixes on the preceding words, not prosodically-dependent words. Given this morphological analysis, a syntactic analysis is proposed that utilizes the 'informationally-rich' syntactic structure of HPSG. In particular, the analysis proposed uses EDGE features and chained identities between adjacent phrasal sisters to license the clause. This enables a simple analysis of the clausal syntax of Coast Tsimshian while still accounting for the wide array of facts surrounding the connectives.
The use of hand gestures to point at objects and individuals, or to navigate through landmarks on a virtually created map is ubiquitous in face-to-face conversation. We take this observation as a starting point, and we demonstrate that deictic gestures can be analysed on a par with speech by using standard methods from constraint-based grammars such as HPSG. In particular, we use the form of the deictic signal, the form of the speech signal (including its prosodic marking) and their relative temporal performance to derive an integrated multimodal tree that maps to an integrated multimodal meaning. The integration process is constrained via construction rules that rule out ill-formed input. These rules are driven from an empirical corporal study which sheds light on the interaction between speech and deictic gesture.
In this paper, I discuss verb to noun conversion in French. The properties of the input verb and the output noun are presented and a formal representation is proposed using the SBCG framework. The use of such a formalism based on constraints and multiple inheritance highlights the difficulties in defining what exactly is a conversion rule. I propose that the different properties of the input verb and the output noun can be thought of as different dimensions of classification, which characterize the verb to noun conversion rule.
Based on the notion of a lexicon with default inheritance, I address the problem of how to provide a template for lexical representations that allows us to capture the relatedness between inflected word forms and canonically derived lexemes within a broadly realizational-inferential model of morphology. To achieve this we need to be able to represent a whole host of intermediate types of lexical relatedness that are much less frequently discussed in the literature. These include transpositions such as deverbal participles, in which a word's morphosyntactic class changes (e.g. verb ⇒ adjective) but no semantic predicate is added to the semantic representation and the derived word remains, in an important sense, a "form" of the base lexeme (e.g. the 'present participle form of the verb'). I propose a model in which morphological properties are inherited by default from syntactic properties and syntactic properties are inherited from semantic properties, such as ontological category (the Default Cascade). Relatedness is defined in terms of a Generalized Paradigm Function (perhaps in reality a relation), a generalization of the Paradigm Function of Paradigm Function Morphology (Stump 2001). The GPF has four components which deliver respectively specifications of a morphological form, syntactic properties, semantic representation and a lexemic index (LI) unique to each individuated lexeme in the lexicon. In principle, therefore, the same function delivers derived lexemes as inflected forms. In order to ensure that a newly derived lexeme of a distinct word class can be inflected I assume two additional principles. First, I assume an Inflectional Specifiability Principle, which states that the form component of the GPF (which defines inflected word forms of a lexeme) is dependent on the specification of the lexeme's morpholexical signature, a declaration of the properties that the lexeme is obliged to inflect for (defined by default on the basis of morpholexical class). I then propose a Category Erasure Principle, which states that 'lower' attributes are erased when the GPF introduces a non-trivial change to a 'higher' attribute (e.g. a change to the semantic representation entails erasure of syntactic and morphological information). The required information is then provided by the Default Cascade, unless overridden by specific declarations in the GPF. I show how this model can account for a variety of intermediate types of relatedness which cannot easily be treated as either inflection or derivation, and conclude with a detailed illustration of how the system applies to a particularly interesting type of transposition in the Samoyedic language Sel'kup, in which a noun is transposed to a similitudinal adjective whose form is in paradigmatic opposition to case-marked noun forms, and which is therefore a kind of inflection.
This paper primarily presents an analysis of nominal inflection in Hindi within the framework of Distributed Morphology (Halle & Marantz 1993, 1994 and Harley and Noyer 1999). Müller (2002, 2003, 2004) for German, Icelandic and Russian nouns respectively and Weisser (2006) for Croatian nouns have also used Distributed Morphology (henceforth DM) to analyze nominal inflectional morphology. This paper will discuss in detail the inflectional categories and inflectional classes, the morphological processes operating at syntax, the distribution of vocabulary items and the readjustment rules required to describe Hindi nominal inflection. Earlier studies on Hindi inflectional morphology (Guru 1920, Vajpeyi 1958, Upreti 1964, etc.) were greatly influenced by the Paninian tradition (classical Sanskrit model) and work with Paninian constructs such as root and stem. They only provide descriptive studies of Hindi nouns and verbs and their inflections without discussing the role or status of affixes that take part in inflection. The discussion on the mechanisms (morphological operations and rules) used to analyze or generate word forms are missing in these studies. In addition, these studies do not account for syntax-morphology or morphology-phonology mismatches that show up in word formation. One aim of this paper is to present an economical way of forming noun classes in Hindi as compared to other traditional methods, especially gender and stem ending based or paradigm based methods that give rise to a large number of inflectional paradigms. Using inflectional class information to analyse the various forms of Hindi nouns, we can reduce the number of affixes and word-generation and readjustment rules that are required to describe nominal inflection. The analysis also helps us in developing a morphological analyzer for Hindi. The small set of rules and fewer inflectional classes are of great help to lexicographers and system developers. To the best of our knowledge, the analysis of Hindi inflectional morphology based on DM and its implementation in a Hindi morphological analyzer has not been done before. The methods discussed here can be applied to other Indian languages for analysis as well as word generation.
This paper presents a descriptive overview and formal analysis of the use of pronominal clitics for realizing various types of arguments in Persian, with particular emphasis on object clitics in the verbal domain. We argue that pronominal clitics behave more like suffixes than independent syntactic elements; in cases where they take syntactic scope over an NP or a PP, they must be phrasal affixes. We propose an HPSG analysis to account for the morphosyntactic aspects of verbal suffixation of object clitics, possessive clitics, preverbal object clitics, and clitic doubling constructions. Finally, we explore extensions of the analysis to periphrastic verb forms, and we compare our proposals for Persian to previous HPSG work on clitic phenomena in other languages.
In the Cognate Object Construction (COC) a typically intransitive verb combines with a postverbal noun phrase whose head noun is morphologically or semantically cognate to the verb. I will argue that English has a family of COCs which consists of four different types. The COCs share common core properties but differ with respect to some of their syntactic and semantic properties. I will capture the ''cognateness'' between the verb and the noun in all COCs by token identities at the level of their lexical semantic contribution. I will use an inheritance hierarchy on lexical rule sorts to model the family relations among the different COC types.
Mazatec is an Eastern Otomanguean language spoken by about 200,000 people, located in the northeastern part of the state of Oaxaca, Mexico. The present paper aims to shed new light on Mazatec verb inflection within the framework of current research on Otomanguean phonology and morphology. We intend to show that, despite bewildering apparent complexity, mainly due to extensive morphophonological processes, Mazatec inflectional morphology is in fact rather simple and regular. Realizational approaches, in particular Paradigm Function Morphology (PFM) seem especially adequate to capture such regularities.
Korean comparative constructions: A constraint-based approach and computational implementation
(2010)
The complexity of comparative constructions in each language has given challenges to both theoretical and computational analyses. This paper first identifies types of comparative constructions in Korean and discusses their main grammatical properties. It then builds a syntactic parser couched upon the typed feature structure grammar, HPSG and proposes a context-dependent interpretation for the comparison. To check the feasibility of the proposed analysis, we have implemented the grammar into the existing Korean Resource Grammar. The results show us that the grammar we have developed here is feasible enough to parse Korean comparative sentences and yield proper semantic representations though further development is needed for a finer model for contextual information.
Semitic languages exhibit rich nonconcatenative morphological operations, which can generate a myriad of derived lexemes. Especially, the feature rich, root-driven morphology in the Arabic language demonstrates the construction of several verb-derived nominals (verbal nouns) such as gerunds, active participles, passive participles, locative participles, etc. Although HPSG is a successful syntactic theory, it lacks the representation of complex nonconcatenative morphology. In this paper, we propose a novel HPSG representation for Arabic nominals and various verb-derived nouns. We also present the lexical type hierarchy and derivational rules for generating these verb-derived nominals using the HPSG framework.
Usage-based preferences in written sentence production: The role of local and global statistics
(2010)
In this paper, we will discuss the role of different levels of frequency distributions in sentence processing and in written production, looking at French homophones. A comparison of experimental data and corpus statistics will demonstrate that lexical frequencies as well as local and global coherences have to be taken into account to fully explain the empirically established patterns.
Investigating the morphological and syntactic properties of discontinuous negative marking in Hausa, I shall suggest a constructional approach involving edge inflection, accounting simultaneously for the morphologically bound nature of the initial marker and its interaction with the TAM system, haplology of the final marker, and wide scope over coordination. I will argue that the degree of morphological integration of initial markers and haplology of final markers both favour an edge feature approach over phrasal affixation.
There are fascinating problems at the syntax-morphology interface which tend to be missed. I offer a brief explanation of why that may be happening, then give a Canonical Typology perspective, which brings these problems to the fore. I give examples showing that the phenomena could in principle be treated either by syntactic rules (but these would be complex) or within morphology (but this would involve redundancy). Thus 'non-autonomous' case values, those which have no unique form but are realized by patterns of syncretism, could be handled by a rule of syntax (one with access to other features, such as number) or by morphology (with resulting systematic syncretisms). I concentrate on one of the most striking sets of data, the issue of prepositional government in Latvian, and outline a solution within Network Morphology using structured case values.
Coherence generally refers to a kind of predicate formation where a verb forms a complex predicate with the head of its infinitival complement. Adjectives taking infinitival complements have also been shown to allow coherence, but the exact conditions for coherence with adjectives appear not to have been addressed in the literature. Based on a corpus-study (supplemented with grammaticality judgements by native speakers) we show that adjectives fall into three semantically and syntactically defined classes correlating with their ability to construct coherently. Non-factive and non-gradable adjectives allow coherence, factive and gradable adjectives do not allow coherence and non-factive and gradable adjectives are tolerated with coherence. On the basis of previous work on coherence in German we argue that coherence allows the infinitival complement of a verb or an adjective to be "split-up", so that the head and a dependent of this head are associated with different information structural functions. In this respect coherence patterns with extraction structures where the extracted constituent has an information structural function different from the constituent from which it is extracted. Following literature on the information structural basis of extraction islands, we show how the lack of coherence with factive adjectives follows from their complements' being information structurally backgrounded, while the infinitival complements of non-factive adjectives tend to a higher fusion with the matrix clause. We also show that coherence is observed with attributive adjectives as well, arguing that coherence is not a distinct verbal property. Finally we provide an analysis of coherence with adjectives within HPSG.
This work focuses on the syntax and semantics of the expression vice versa, and shows that its syntactic distribution is much more flexible than semantically related expressions. Although vice versa usually appears in clausal coordinate environments, it can in principle occur in any other type of construction. Second, it can occur as an embedded verb phrase or even as a noun phrase, rather than as an adjunct. This suggests that vice versa is a propositional anaphor that corresponds to a converse of a propositional antecedent. Finally, although the predicates singled out to be interchanged are usually nominal, they can in fact be of virtually any part of speech. I argue that a possible account of the interpretation of vice versa lies at the interface between logical form (with rich decompositional lexical semantics along the lines of Pustejovsky (1995)), and pragmatics (drawing from independent work by Hobbs (1990) and Kehler (2002)).
We describe an empirical method to explore and contrast the roles of default and principal part information in the differentiation of inflectional classes. We use an unsupervised machine learning method to classify Russian nouns into inflectional classes, first with full paradigm information, and then with particular types of information removed. When we remove default information, shared across classes, we expect there to be little effect on the classification. In contrast when we remove principal part information we expect there to be a more detrimental effect on classification performance. Our data set consists of paradigm listings of the 80 most frequent Russian nouns, generated from a formal theory which allows us to distinguish default and principal part information. Our results show that removal of forms classified as principal parts has a more detrimental effect on the classification than removal of default information. However, we also find that there are differences within the defaults and principal parts, and we suggest that these may in part be attributable to stress patterns.
Welsh is a language in which unbounded dependency constructions involve both gaps and resumptive pronouns (RPs). Gaps and RPs appear in disjoint sets of environments. Otherwise, however, they are quite similar. This suggests that they involve the same mechanism, and in HPSG that they involve the SLASH feature. It is possible to provide an analysis in which RPs are associated with the SLASH feature but are also the ordinary pronouns which they appear to be.
This papers addresses information-structural restrictions on the occurrence of what is known as "multiple fronting" in German. Multiple fronting involves the realization of (what appears to be) more than one constituent in the first position of main clause declaratives, a clause type that otherwise respects the verb-second constraint of German. Relying on a large body of naturally occurring instances of multiple fronting with the surrounding discourse context, we show that in certain contexts, multiple fronting is fully grammatical in German, in contrast to what has sometimes been claimed previously. Examination of this data reveals two different patterns, which we analyze in terms of two distinct constructions, each instantiating a specific pairing of form, meaning and contextual appropriateness.
A little discussed feature of English are non-restrictive relative clauses in which the antecedent is normally not an NP and the gap follows an auxiliary, as in Kim will sing, which Lee won't. These relative clauses resemble clauses with auxiliary complement ellipsis or fronting. There are a variety of analyses that might be proposed, but there are reasons for thinking that the best analysis is one where which is a nominal filler associated with a gap which is generally non-nominal: a filler-gap mismatch analysis in other words.
This paper addresses the form-meaning relation of multimodal communicative actions by means of a grammar that combines verbal input with hand gestures. Unlike speech, gesture signals are interpretable only through their semantic relation to the synchronous speech content. This relation serves to resolve the incomplete meaning that is revealed by gestural form alone. We demonstrate that by using standard linguistic methods, speech and gesture can be integrated in a constrained way into a single derivation tree which maps to a uniform meaning representation.
Does chain hybridization in Irish support movement-based approaches to long-distance dependencies?
(2010)
Huybregts (2009) makes the claim that hybrid A'-chains in Irish favor derivational theories of syntax over representational ones such as HPSG. In this paper, we subject this assertion to closer scrutiny. Based on a new technical proposal, we will reach the conclusion that, in principle, both derivational and representational accounts can accomodate hybrid dependencies. Thus, no argument against either approach can be made on the basis of the Irish data, disconfirming Huybregts's (2009) claim.
The analysis of the copula as a semantically vacuous word in mainstream HPSG is appropriate for some of its uses, such as the progressive and the passive, but not for its use in clauses with a predicative complement. In such clauses the copula denotes a relation of coreference between the indices of the subject and the predicative complement.
This paper explores the use of HPSG for modeling historical phonological change and grammaticalization, focusing on the evolution of the pronunciation of word-final consonants in Modern French. The diachronic evidence is presented in detail, and interpreted as two main transitions, first from Old French to Middle French, then from Middle French to the modern language. The data show how the loss of final consonants, originally a phonological development in Middle French, gave rise to the grammaticalized external sandhi phenomenon known as consonant liaison in modern French. The stages of development are analyzed formally as a succession of HPSG lexical schemas in which phonological representations are determined by reference to the immediately following phonological context.
The paper aims to present approach to HPSG phonology which would account for underlying forms of phonemes. It shows some of the issues arising in monostratal analyses of phonology, and proposes a solution based on a notion of underlying representations. The approach presented, partly inspired by Optimality Theory, resolves cases of neutralisation and opacity by formulating constraints which either restrict the surface representation or relate it to the underlying form.
In this paper we investigate German idioms which contain phraseologically fixed clauses (PCl). To provide a comprehensive HPSG theory of PCls we extend the idiom theory of Soehn 2006 in such a way that it can distinguish different degrees of regularity in idiomatic expressions. An in-depth analysis of two characteristic PCls shows how our two-dimensional theory of idiomatic expressions can be applied and illustrates the scope of the theory.
This paper discusses ergative case assignment in Hindi and its interaction with aspectual verb complexes or complex predicate constructions. It is shown that ergative case is assigned by the last head in the aspectual verb complex and that ergative case on the subject of intransitive verbs denoting bodily-functions is associated with a counter-to-expectation meaning. It is then shown that aspect complex predicates in Hindi involve two distinct syntactic structures, which have similar semantics. While one syntactic structure involves argument composition, the other involves a head-modifier structure. It is argued that the existence of two structures favor approaches to the interface between syntax and semantics which do not require a uniform isomorphism between the semantics and syntax of aspect.
Preposed negation in Danish
(2009)
In Danish the base position of the negation and negated quantifier phrases is between the subject and the finite verb in embedded clauses. However, in embedded clauses introduced by a non-veridical complementizer such as hvis ('if') or om ('whether') the negation and negated quantifier phrases can also appear between the complementizer and the subject. This phenomenon is referred to as preposed negation. The paper investigates the structure and semantics of this construction. It is argued that preposed negation is no adjunction structure, but a special construction where the negation element is a sister of the complementizer and the filler of a filler-gap-structure. It is further argued that preposed negation is associated with negated verum-focus of a clause lacking an (aboutness-) topic. The negation of a verum predicate explains why preposed negation fails to license strong negative polarity items and to rule out positive ones. The lack of a topic explains why preposed negation is preferred with non-referential subjects and with weak readings of indefinite subjects and why preposed negation is incompatible with topic-binding particles.The final section presents an HPSG-analysis of preposed negation using Minimal Recursion Semantics (MRS).
The present paper gives an account of Serial Verb Constructions (SVCs) in Mandarin Chinese. After a typological presentation of the phenomenon, we give an overview of the Chinese data with examples of the semantic variations of SVCs. The inventory of SVC types is classified according to causal and temporal relations between the components. We also discuss the pragmatic conditions on the use of SVCs as well as alternative, semantically equivalent constructions. A HPSG-analysis is proposed for marked SVCs, which uses the interaction between aspect marking and the set of possible subordinative relations to deduce the extra-lexical meaning of the construction. Particular attention is payed to the syntactically peculiar SVC with shared internal arguments, which is accounted for by a non-cancellation approach to valence requirements.
On predication
(2009)
This paper discusses copula constructions in English, German, and Danish and argues that a uniform analysis of all copula constructions is inappropriate. I provide evidence from German that there should be a raising variant of the copula in addition to an identificational copula. A unary schema is provided that maps referential NPs that can be used as arguments onto predicational NPs. Data from Danish shows that predicational NPs can be subjects in specificational structures. An account for such specificational structures is provided and the different behaviour of predicational and specificational structures with regard to question tags is explained. A similar contrast can be found in German left dislocation structures, which follows from the assumptions made in this paper.
A modified treatment of complex predicate formation allows for a reduction of selectional features (that is abolishing of xcomp or vcomp) and for a uniform treatment of predicational phrases in copula constructions and resultative secondary predicates. This yields an account for constituent order variants that remained unexplained by earlier analyses.
I reconsider the HPSG Raising Principle which is introduced in Pollard & Sag (1994) to constrain the way in which lexical entries describe the SUBCAT lists of the words they license. On the basis of whether a complement is assigned a semantic role in a lexical entry or not, this entry may not or must describe this complement as structure-shared with the unrealised subject of some other (non-subject) complement. The formal status of this principle is still unclear, as it is formulated as a 'meta principle' that does not talk about linguistic objects directly but rather about the lexical entries that license them. I show that, although its meaning cannot be expressed faithfully by the usual kind of constraints employed in HPSG, the Raising Principle can nevertheless be replaced by two such constraints which make largely the same predictions. Most importantly, these constraints interact with the output values of description-level lexical rules in the style of Meurers (2001) in a way that makes predictions available that Pollard & Sag (1994) intended the Raising Principle to make but that it cannot possibly make if description-level lexical rules are employed.
This paper analyzes the interrelation of two understudied phenomena of English: discontinuous modifier phenomenon (so willing to help out that they called early; more ready for what was coming than I was) and the complex pre-determination phenomenon (this delicious a lasagna; How hard a problem (was it)?). Despite their independence, they frequently occur intertwined, as in too heavy {a trunk} (for me) to lift and so lovely a melody that some people cried. This paper presents a declarative analysis of these and related facts that avoids syntactic movement in favor of monotonic constraint satisfaction. It demonstrates how an explicit, sign-based, constructional approach to grammatical structure captures linguistic generalizations, while at the same time accounting for idiosyncratic facts in this seemingly complex grammatical domain.
In this paper we develop an HPSG syntax-semantics of negative concord in Romanian. We show that n-words in Romanian can best be treated as negative quantifiers which may combine by resumption to form polyadic negative quantifiers. Optionality of resumption explains the existence of simple sentential negation readings alongside double negation readings. We solve the well-known problem of defining general semantic composition rules for translations of natural language expressions in a logical language with polyadic quantifiers by integrating our higher-order logic in Lexical Resource Semantics, whose constraint-based composition mechanisms directly support a systematic syntax-semantics for negative concord with polyadic quantification.
The paper discusses the so-called adverbial use of the wh-pronoun was ('what'), which establishes a non-standard interrogative construction type in German. It argues that the adverbial use of was ('what') is based on the lexical properties of a categorically deficient pronoun was ('what'), which bears a causal meaning. In addition, adverbial was ('what') differs from canonical argument was ('what') as it is analyzed as a functor which is generated in clause-initial position.
By means of empirical facts mainly provided by d'Avis (2001) it is shown that was ('what') behaves ambivalently regarding the wh-property: On the one hand, was ('what') can introduce an interrogative clause, but on the other hand it cannot license wh-phrases in situ. While formally analyzing the data against the background of existing accounts on wh-interrogatives couched in the framework of Head-driven Phrase Structure Grammar, an analysis is developed that separates two pieces of information to keep track of the wh-information percolating in an interrogative clause. Whereas the WH-value models wh-fronting and pied-piping phenomena, the QUE value links syntactic and semantic information and thus keeps track of wh-phrases in-situ.
We address three properties of Turkish morphology and VP coordination: the identification of tense and aspect values across conjuncts, the optional omission of affixes on non-final conjuncts coordinated with the word ve and the obligatory sharing of scopal modals across conjuncts in coordination structures with the affix -ip. For the modals in an -ip structure, we propose an analysis that uses syntactic features to trigger the application of a construction at the level of the coordinated VP introducing the scopal predications. Our analysis is implemented in a small HPSG grammar and tested against datasets confirming the functionality and consistency of the analysis.
Although the original framework of HPSG is mostly compatible with independent theoretical claims or analyses in lexical lexeme base morphology (Anderson 1992, Aronoff & Fudeman 2004, Beard 1995, Booij 2005, Carstairs-McCarthy 1992, Fradin 2003, Haspelmath 2002, Matthews 1991, Plag 2003, for example), so far, most research in morphology has been done on inflexional phenomena (Orgun & Inkelas 2002, Bonami & Boyé 2006), and few on derivational morphology (Koenig 1999, Riehemann 1998). Yet, we believe it is worth investigating how the formal and theoretical apparatus of HPSG deals with capturing multilevel constraints that apply in the lexeme formation of French Verb-Noun nominal compounds, such as as GRILLE-PAIN (lit. grill-bread, 'toaster'), PERCE-OREILLE (lit. pierce-ear, 'earwig'), TOURNEVIS (lit. turn-screw, 'screwdriver'), or LÈCHE-VITRINE (lit. lick-window, 'window-shopping'). Contrary to what has often been said, we argue VN lexemes formation comes under morphological constraints but not under syntactic mechanisms. Our analysis integrates VN lexemes into a multiple-dimension typed-hierarchy of lexemes and provides an account for semantic generalizations involved in different types of lexeme formation (compounding, derivation, and conversion).
In this paper, I discuss the case and agreement system of Nias, a language that has been described as a marked-absolutive system by various authors (Donohue and Brown, 1999; Corbett, 2006; Cysouw, 2005; Handschuh, 2008; Wichmann, 2005). I shall argue in particular that the ergativity of this language is highly superficial in nature, showing that hypothesised marked-absolutive arguments fail to display typical subject properties. Extending the linking theory of ergativity by Manning (1994) and Manning and Sag (1999), which assumes an inverse linking pattern for transitive, I shall suggest that Nias transitives are best analysed as a Nominative-Accusative system, attributing the ergative split in Nias to an inverse linking of intransitives instead. Under this perspective, case, agreement, and word order will receive a natural explanation.
Previous HPSG accounts of extraction blur the distinction between valents and adjuncts by allowing verbs to lexically control the modifiers that combine with their phrasal projections. However, assuming that adjuncts are valents runs into various difficulties. This paper argues that the distinction between complements and adjuncts can be maintained, and that certain semantic phenomena that challenge traceless theories of extraction can be seen as an instance of a more general process. Finally, this paper also discusses a uniform mechanism for case assignment to valents and adverbial nominals.
Modern Persian conjugation makes use of five periphrastic constructions. We contrast the properties of these five constructions and argue that they call for different analyses. We propose contrasting analyses relying on the combination of an HPSG approach to feature geometry and syntactic combination, and an approach to paradigm organization and morphological exponence based on Paradigm Function Morphology. This combination of analytic tools allows us to treat the whole array of periphrastic constructions as lexical in origin—no phrasal construction or multi-word lexical entry of any kind is required.
French and Romanian verbless relative adjuncts are incidental adjuncts which have been described as elliptical relative clauses. We show that this analysis is not empirically adequate and propose an alternative non-elliptical analysis. We analyze verbless relative adjuncts as sentential fragments whose head can be a cluster of phrases. They are marked by a functor phrase which displays selection properties with respect to the head phrase and makes an essential contribution to the semantics of the adjunct. The analysis relies on the interaction of grammatical constraints introduced by various linguistic objects, as well as on a constructional analysis of verbless relative adjuncts distinguishing several subtypes.
Transparent free relatives (TFRs) are constituents involving a WH-gap dependency in which the phrase that is predicated of the gap associated with 'what', not the wh-phrase itself, functions as the syntactic and semantic "nucleus." Previous analyses have either treated TFRs as a construction radically different from ordinary FRs, utilizing such mechanisms as parenthetical placement or grafts, or assimilated them to ordinary FRs, relying on abstract/empty head elements and a vague semantic relation holding between the gap and the predicate phrase. In this paper, we investigate how the puzzling properties of English TFRs can be accounted for in HPSG. The paper shows that the transparency effect of TRFs can be handled by feature inheritance from the nucleus predicate phrase, together with a constructional constraint that deals with the exocentric property of TFRs.
Dualist Syntax
(2008)
A dualist syntax has two components: (1) the lexicon, a structured set of formatives ('words'); and (2) rules for combining those formatives into utterances. This paper defends syntactic dualism against three 'monist' challenges. First, evidence for lexical argument structure can be found in deverbal nominalization, which preserves that structure systematically. Second, words represent the smallest units for idiom formation and contextual polysemy effects, which is expected on the dualist view but not if word meanings are composed in the syntax. Third, the count/mass properties of nouns suggest an interleaving of conceptual and grammatical information in semantic composition.
Predicate complements
(2008)
This paper proposes a representation for syllable structure in HPSG, building on previous work by Bird and Klein (1994), Höhle (1999), and Crysmann (2002). Instead of mapping segments into a a separate part of the sign where syllables are represented structurally, information about syllabification is encoded directly in the list of segments, the core of the PHONOLOGY value. Higher level prosodic phenomena can operate on a more abstract representation of the sequence of syllables derived from the syllabified segments list. The approach is illustrated with analyses of some word-boundary phenomena conditioned by syllable structure in French.
Our analysis of pseudopartitives and measure phrases draws on the idea of 'of' as a copula in a pseudopartitive. The copular analysis allows us to avoid the complications caused by treating either the numeral-noun combination before the of-phrase or the of-object as the head of a pseudopartitive on agreement, and hence to account for all the agreement patterns without creating any extra rule. We also outline how we can extend our analysis to handle measure phrases that do not co-occur with of-phrases by treating these measure phrases as anaphoric, an analysis that can adapt to the anaphoric constructions in classifier languages. Such an analysis does not only come closer to the intuition of native speakers but also have an appeal from the perspective of the universality of languages.
We consider two alternatives for memory management in typed-feature-structure-based parsers by identifying structural properties of grammar signatures that may be of some predictive value in determining the consequences of those alternatives. We define these properties, summarize the results of a number of experiments on artificially constructed signatures with respect to the relative rank of their asymptotic cost at parse-time, and experimentally consider how they impact memory management.
The lexical information of verbal lexemes, such as verbs and adjectives, plays an important role in syntactic parsing, because the structure of a sentence mainly hinges on the type of verbal lexemes. The question we address in this research is how to acquire the argument structure (henceforth ARG-ST) of verbal lexemes in Korean. It is well known that manual build-up of type hierarchy usually cost too much time and resources, so an alternative method, namely automatic collection of relevant information is much more preferred. This paper proposes a procedure to automatically collect ARG-ST of Korean verbal lexemes from a Korean Treebank. Specifically, the system we develop in this paper first extracts lexical information of ARG-ST of verbal lexemes from a 0.8 million graphic word Korean Treebank in an unsupervised way, checks the hierarchical relationship among them, and builds up the type hierarchy automatically. The result is written in an HPSG-style annotation, thus making it possible to readily implement the result in an HPSG-based parser for Korean. Finally, the result is evaluated with reference to two Korean dictionaries and also with respect to a manually constructed type hierarchy.
We present an analysis of adjuncts which, while based on the traditional binary adjunction schema, accommodates the phenomena that motivate the alternative Adjunct-as-Complement approach, such as adjunct extraction and case marking. The key idea is to enable the syntactic head (modifiee) to select for its modifier (adjunct) via the new valence feature dedicated for adjuncts, while leaving its values underspecified. Thus the selectional property of the modifiee percolates as well as that of the modifier, dispensing with the need to endow adjuncts a complement-like status.
My objective in this paper is to integrate scalar exclamatives into an HPSG grammar of French. First, a procedure to sort out scalar exclamatives from declaratives and interrogatives is proposed. Then, the main semantic and dialogical properties of exclamatives are presented: veridicity, ego-evidentiality, illocutionary double life and scalarity. Finally, assuming Ginzburg & Sag 2000, the exclamative clause type is defined.
This paper focuses on a specific type of verbless utterance, labeled PVU, which is defined by two properties:
• PVUs are not discourse fragments.
• PVUs can host a phrase in their right periphery which is coreferential with their external argument. This phrase is labeled α-phrase.
PVUs are analyzed as clausal predicative phrases. Although PVUs can have various illocutionary forces, their content type is constrained by their syntactic form. As for α-phrases, they are shown to be right-dislocated phrases. Right-dislocation is analyzed as a local anaphoric phenomenon. This ap proach is consistent with the anaphoric properties of PVUs’ external arguments.
Whether the Coordinate Structure Constraint (CSC) (Ross, 1967) is a syntactic constraint has been discussed much in the literature. This paper reconsiders this issue by drawing on evidence from Japanese and Korean. Our examination of the CSC patterns in relative clauses in the two languages reveals that a pragmatically-based approach along the lines of Kehler (2002) predicts the relevant empirical patterns straightforwardly whereas alternative syntactic approaches run into many problems. We take these results to provide strong support for the view that the CSC is a pragmatic principle rather than a syntactic constraint.
This paper deals with Korean postpositions. They are treated as suffixes in recent lexicalist works. But they differ syntactically from suffixes and we will propose to treat them as clitics, i.e. words combining with a phrase in the syntax and attaching to its last lexical item in the phonology. We treat them as weak syntactic heads, taking into account their head properties and the syntactic similarity between the mother phrase and the host phrase. They take the latter as complement and share most of its syntactic properties. Revising the traditional classification, we divide postpositions into three subtypes: marking, oblique and semantic postpositions, based on their distributional properties, such as optionality, non-nominal marking and stacking, etc. Finally we show how our analysis can be described in the HPSG model.
Preposition-noun combinations (PNCs) are compositional and productive, but not fully regular. In school grammars and many theoretical approaches, PNCs are neglected, but they have recently been addressed in an HPSG analysis by Baldwin et al. (2006). After discussing some basic properties of PNCs, we show that statistical methods can be employed to prove that PNCs are indeed productive and compositional, which again implies that PNCs should receive a syntactic analysis. Such an analysis, however, is impeded by the limited regularity of the construction. We will point out why adding semantic conditions to syntactic schemata might be necessary but not sufficient and turn then to a framework which allows the derivation of syntactic (and semantic) generalizations from linguistic data without taking recourse to introspective judgments.
Coordination in Japanese poses various puzzles which defy the standard notion of syntactic category. On the one hand, one can conjoin structures which one usually would not expect to form any constituent, and on the other hand, there are various conjunction particles that are sensitive to the kind of conjuncts that they combine with. In this paper we argue against abandoning the usual notion of constituency, and redefining the entire grammar of Japanese. We provide a novel construction-based account of the data in which the phenomena result from the interaction of the coordination construction, ellipsis, and allomorphy of the conjunction particle.
Verbs are the centerpiece of the sentence, and understanding of verb meanings is essential for language acquisition. Yet verb learning is said to be more challenging than noun learning for young children for several reasons. First, while nouns tend to denote concrete objects, which are perceptually stable over time, verbs tend to refer to action events, which are temporally ephemeral, and the beginning and the end of the action referred to by the verb are not clearly specified. Second, a verb takes nouns as arguments, and the meaning of a verb is determined as the relation between the arguments. To infer the meaning of a verb, children need to attend to the relation between the objects in the event rather than the objects themselves. In so doing, children make use of a variety of cues such as argument structure, meta-knowledge of the lexicon, and extra-linguistic contextual cues. In this paper, I present two lines of my recent research concerning young children's novel verb learning. Specifically, I first report a cross-linguistic study (Imai et al., 2008) examining how Japanese-, English-, and Chinese-speaking children utilize structural and non-structural, extra-linguistic cues when inferring novel verb meanings. Second, I present another study examining how young children utilize sound-meaning correlates (sound symbolism) in their inference of novel verb meanings. In the end, I evaluate the relative importance of structural cues among different cues children use in verb learning.
This paper presents a constraint-based account of verb form alternations (Short and Long Forms) in Mauritian which, basically, is syntactically driven: Short Forms appear with Canonical complements while Long Forms are expected with no realized complements. However, in specific contexts, Long Forms are unexpectedly authorized in declaratives with canonical complements and expresses Verum Focus.
Transparent heads
(2008)
Head-complement structures in HPSG identify most properties of the phrase with those of the head daughter, except for that valence property (e.g. SUBCAT or COMPS) whose constraints are met by the non-head daughter(s) in the phrase. In this paper I present several phenomena in English syntax where idiosyncratic properties of a non-head daughter in a phrase must remain visible on the phrasal node, in order to preserve the strong version of the principle of locality in subcategorization. I propose a general formal mechanism to effect this occasional transparency of heads with respect to certain properties of their complements.
This paper describes a number of verbal argument marking patterns found in the world's languages and provides HPSG analyses for them. In addition to commonly-occurring variations of morphosyntactic alignment (e.g. nominative-accusative, ergative-absolutive), this paper also presents analyses of more complex phenomena, including ergativity splits, Austronesian-style focus-case systems, and direct-inverse systems and their interaction with case.
Direct quotation raises three major problems for grammatical modelling: (i) the variety of quoted material (which can be a non linguistic behavior, or a sign in a different language), (ii) the embedding of an utterance inside another one, (iii) a special denotation, the content of the quotation being the utterance itself. We propose a unary rule, which turns the quoted material into a linguistic sign whose content is itself a behavior, which entertains a resemblance relation to the behavior demonstrated by the speaker. Syntactically, direct quotation comes in two varieties: it can be the complement of a quotative verb, or constitutes a head sentence, modified by an adjunct containing a quotative verb whose complement is extracted and identified with its local features.
In this paper we address the question of which transitive verbs allow there-insertion in Danish. We propose that two constraints have to be met in order for verbs to appear in Danish there-constructions. Firstly, as have been noted by others, an empty direct object position must be available. This constraint is not sufficient for restricting the set of verbs in there-constructions. We further propose a locative constraint. The transitive verbs allowing there-insertion will be shown to coincide with verbs that allow a locative analysis.
The information-structural status of clitic left dislocated arguments in Spanish has been argued to depend crucially on their thematic role. Earlier HPSG analyses of related phenomena in other languages do not take into account this sort of information. A formalization will be presented which can handle differences in information-structure arising from different thematic roles of clitic left dislocated phrases.
This paper examines the syntactic behaviour of two omnisyndetic coordinations (also called correlative coordinations), i.e. the disjunctive and the conjunctive types in Romanian, by explaining its data in a Romance perspective. Major issue has been whether these structures have symmetric or asymmetric structures. If all these Romance languages share a symmetric analysis for the disjunctive type Conj ... Conj, it is not the case for the conjunctive type. Our aim is to show that the postulation of a conjunctional status for the Romanian structure şi ... şi ('both ... and'), which is the most widespread view in Romanian grammars, is inadequate for the Romanian data.
The word order facts of radically non-configurational languages pose a challenge to HPSG approaches which assume both that the surface order of words is the yield of the (tectogrammatical) tree and standard HPSG-style cancellation of valence lists. These languages allow discontinuous noun phrases, in which modifiers appear separated from their head nouns by arbitrarily many other words from the same clause. In this paper, I explore an analysis which preserves tectogrammatical-phenogrammatical equivalence, and accounts for the word order facts of Wambaya with an analysis based on non-cancellation. This analysis is contrasted with other approaches to discontinuous constituents and analyses of other phenomena based on non-cancellation. Finally, I explore the implications for current models of semantic compositionality.
Non-restrictive relative clauses (NRRCs) can modify constituents which undergo 'pragmatic enrichment' when they appear in answers to questions. For example, in an interchange like: 'A: What did Jo think? B: That you should say nothing, which is surprising.' What B says is surprising is that 'Jo thinks ...' On the face of it, this might seem problematic for approaches to NRRCs which assume 'syntactic integration' and to support an 'orphan' analysis, where NRRCs are combined with purely conceptual representations. In this paper we examine a range of elliptical and anaphoric phenomena, and show that this conclusion is misplaced. In fact, the phenomena argue strongly in favour of a syntactically integrated analysis.
We contrast two types of sentences with a preposed NP in French in a construction based HPSG grammar. They differ with respect to different grammatical aspects (syntax, semantics, pragmatics and phonology), which cluster uniquely into constructions. Both are colloquial, a reason why they have been recognized only recently (see Zribi-Hertz 1986, 1996, Sabio 1995, 2006). Accordingly, we rely for the data on spoken corpora (Corpaix, CFRP) as well as on our intuitions. Both constructions involve a partitioned semantics but this mode of composition is associated with different effects. One construction is characterized semantically: the preposed NP is the theme of a categorical proposition. The other construction is characterized pragmatically: it is associated with an independent declarative clause, a typical use of which is to signal a break in the interaction.
The result of questionnaire studies are presented which shows (i) that conjuncts are scope islands in Japanese and (ii) that left-node raising can nullify such scope islands. This finding confirms the theory advanced in Yatabe (2001), in which semantic composition is almost entirely carried out within order domains, and arguably contradicts the theory proposed in Beavers and Sag (2004), which introduces a mechanism called Optional Quantifier Merger to deal with the fact that right-node raising and left-node raising can have semantic effects.
Based on Krifka (1992) and de Kuthy (2000), this paper develops an architecture for complex topic-comment structures in HPSG and applies it to predicate fronting in English with the goal of capturing the insights of Ward (1988) on this construction. We argue that predicate fronting is a distributed constructional form consisting of an auxiliary occurring in a predicate preposing phrase. The use of predicate preposing is a function of a combination of simultaneous constraints on its theme structure, its background-focus distribution, and its presuppositional structure. It is shown that these constraints can be made explicit within the HPSG architecture developed here.
This paper aims at making a general description of Chinese NPs using Head-Driven Phrase Structure Grammar. The paper introduces the basic and complex structures of Chinese NPs and then shed light on the noun-classifier matching problem when implemented in HPSG. To solve this problem, the paper tries to establish a basic grammar of Chinese NPs in the framework of HPSG, which is implemented in the LKB system. The implementation shows, although the matching problem between noun and classifier can be described in HPSG, especially by the MRS, it is still difficult to efficiently represent the semantic constrains in the LKB system.
The Big Mess Construction
(2007)
There is a construction in English, exemplified by 'how long a bridge', which is so irregular that it has been named the Big Mess Construction (Berman 1974). This paper first sketches its main characteristics and a treatment of the internal structure of the noun phrase which serves as a background for the analysis. It then presents three ways in which the Big Mess Construction can be analysed; two of them are lexicalist and are shown to be implausible; the third is constructivist and is argued to be superior. In a next step, the discussion is extended to two other types of constructions. The first concerns the English adnominal reflexives, as in 'the children themselves', and is shown to require a constructivist analysis which is similar but not identical to the one for the Big Mess Construction. The second concerns the combination of 'such' and 'what' with the indefinite article, as in 'such a pleasure'. In spite of its obvious resemblance with the Big Mess Construction this combination does not require a constructivist analysis; instead, it fits the lexicalist mould of most of the rest of HPSG.
An empirical overview of the properties of English prepositional passives is presented, followed by a discussion of formal approaches to the analysis of the various types of prepositional passives in HPSG. While a lexical treatment is available, the significant number of technical and conceptual difficulties encountered point to an alternative approach relying on constructional constraints. The constructional approach is argued to be the best option for prepositional passives involving adjunct PPs, and this analysis can be extended to create a hierarchy of constructions accommodating all types of prepositional passives in English, and the ordinary NP passive.
This paper aims to provide type hierarchies for Korean passive constructions on the basis of their forms within the HPSG framework. The type hierarchies proposed in this paper are based on the classification of Korean passives; suffixal passives, auxiliary passives, inherent passives, and passive light verb constructions. Verbs are divided into five subtypes in accordance with the possibility of passivization. We also provide type hierarchies for verbal nouns and passive light verbs.
In Sorani Kurdish dialects, the complement of a preposition can generally be realized either as a syntactic item (NP, independent pronoun or PP) or a bound personal morpheme (clitic/affix). However, the affixal realization of the complement gives rise to a range of specific phenomena. First, some prepositions display two different phonological forms depending on the realization of their complement: the variant combining with a syntactic item is referred to as ˋsimple', while the variant combining with an affixal complement is called ˋabsolute'. Furthermore, unlike syntactic complements, which are always realized locally, the affixal complement of an absolute preposition can have a non-local realization, attaching to a host with which it has no morphosyntactic relations. In order to deal with these facts, this paper proposes a classification of Sorani prepositions along two lines: the affixal versus non-affixal realization of the complement on the one hand and its local versus non-local realization on the other hand. All cases of non-local realization receive a lexical account, either in terms of argument composition or in terms of linearization constraints on domain objects.
Negative Polarity Items (NPI) are expressions such as English 'ever' and 'lift a finger' that only occur in sentences that are somehow negative. NPIs have puzzled linguists working in syntax, semantics and pragmatics, but no final conclusion as to which module of the grammar should be responsible for the licensing has been reached. Within HPSG interest in NPI has developed only relatively recently and is mainly inspired by the entailment-based approach of Ladusaw 1980 and Zwarts 1997. Since HPSG's CONTENT value is a semantic representation, the integration of such a denotational theory cannot be done directly. Adopting Discourse Representation Theory (DRT, Kamp and Reyle 1993, von Genabith et al. 2004) I show that it is possible to formulate a theory of NPI licensing that uses purely representational notions. In contrast to most other frameworks in semantics, DRT attributes theoretical significance to the representation of meaning, i.e. to a logical form, and not only to the denotation itself. This makes DRT particularly well-suited to my purpose.
Remarks on locality
(2007)
This paper proposes a modification of HPSG theory—Sign-Based Construction Grammar—that incorporates a strong theory of both selectional and constructional locality. A number of empirical phenomena that give the appearance of requiring nonlocal constraints are given a principled, localist analysis consistent with this general approach, which incorporates certain insights from work in the tradition of Berkeley Construction Grammar, as exemplified by Fillmore et al. (1988), Kay and Fillmore (1999), and related work.