Lexical Functional Grammar Research Articles

Raising to object in Norwegian and the derived object constraint*

Abstract. Norwegian is traditionally assumed not to have raising to object from infinitival complements with believe type verbs. This article shows that it does, based upon sentences from written texts. It is also shown that this construction is – with some exceptions – subject to the constraint that the raised object cannot be in the regular object position. This constraint was discussed for a group of English verbs in Postal (1974), who called it the Derived Object Constraint. The standard view in the literature is that it is related to case theory. This article proposes a new explanation. With believe type verbs, a raised object is also the subject of an infinitival complement with an individual level predicate. Being the subject of an individual level predicate, the raised object must have a strong reading, which makes it high in topicality. This is the property that can (but does not have to) be grammaticalized as the Derived Object Constraint. The implementation of this constraint in Lexical Functional Grammar (LFG) is discussed, and it is shown how it can be stated as a constraint on functional structure.

A grammar of Jingulu: An Aboriginal language of the Northern Territory (review)

Reviewed by: A grammar of Jingulu: An Aboriginal language of the Northern Territory Jane Simpson A grammar of Jingulu: An Aboriginal language of the Northern Territory. By Robert Pensalfini. (Pacific linguistics 536.) Canberra: Pacific Linguistics, 2003 Pp. xix, 262. ISBN 0858835584. $51.54. This grammar, published by the major publisher of Australian and Pacific language materials, is of an endangered and typologically interesting language. Jingili people live in the Northern Territory of Australia. Their country is on the border of Pama-Nyungan languages (normally suffixing) and non-Pama-Nyungan languages (often prefixing), and their language, Jingulu, shows features of both types. It is part of the proposed Mirndi family, a geographically discontinuous group whose genetic unity is still debated (Green & Nordlinger 2004). Pensalfini builds on an earlier descriptive study done when there were more speakers (Chadwick 1975). Chadwick presents Jingulu as a language with fiendishly complex morphology that interacts with an interesting regressive vowel harmony process (van der Hulst & Smith 1985). Since Chadwick’s study, the language ecology of the area has changed, and Jingili people mostly speak Kriol, code-switching to Jingulu and Mudburra. The bar has been raised as to what a reference grammar of a small language should contain (Himmelmann 1998). The grammarian is torn between the demands of the grammar-reading public (You mean there’s nothing on weak crossover!), the likelihood that this grammar will be the main source of information on the language (Where’s the stuff on gesture/kinship/speech events/information structure?), and the fact that most publishers don’t sell texts and dictionaries of small languages. P’s grammar is characterized by bold analyses. He carried out fieldwork in the context of preparing an MIT Ph.D. dissertation, informed by minimalism, well aware of Rachel Nordlinger’s work on a neighboring and genetically related language, Wambaya, informed by lexical functional grammar (Nordlinger 1998). So, naturally, he asks questions relevant to the theoretical concerns of the time. He carefully tries to gather data on multiple wh-questions; looks for ways to translate sentences with quantifiers and modals, with definiteness, and with indefiniteness; and looks at coordinate and reflexive constructions and complex constructions. These are hard questions and P is open about the difficulties that can arise, especially in a situation of language death. He is careful to note the absence of confirming or disconfirming data. The result is a richer understanding of what is happening in the language—for instance, the presence of a kind of indefinite suffix, quite unusual in Australian languages (203–5). How does P face the challenge of language documentation? His field tapes and transcripts are held in a public archive (Australian Institute of Aboriginal and Torres Strait Islander Studies), and a separate collection of texts and a dictionary has been submitted for publication. He appends eleven texts (Ch. 7, 241–58), including some that show code-switching into Kriol, and occasionally he gives illuminating Kriol glosses from his informants. This grammar comprises analysis of what his teachers said; he does not draw together the earlier records of the language. It is rich in example sentences, mostly from texts, but he follows too many Australian grammars in omitting the source for each example sentence. This is especially unfortunate in language-loss situations, since which speaker said which sentence is important for understanding the variation that P notes (although he is generally good about commenting on rarity). [End Page 425] Ch. 1 (1–19) has a good discussion of the present-day complex language situation and a section on kinship, but little on the ethnography (no reference to the early-twentieth-century investigations (Spencer & Gillen 1904) or the late-twentieth-century work on the Murranji land claim). No more than fifteen people spoke Jingulu fluently when P learned it, and it was not used much in everyday conversation. So the Jingulu language systems of individuals are no longer constrained by the need to understand and be understood. They may have to dredge their memories (what do I call my mother’s brother?). When a speech community erodes and speakers shift to using other languages, the language systems of the last few speakers may diverge more widely than the language...

The diversity and unity of the accusative with infinitive construction: a Norwegian perspective

In an accusative with infinitive sentence like I have seen faith accomplish miracles, the accusative faith is assigned a thematic role by the embedded verb accomplish, and not by the main verb seen. This kind of sentence has been analyzed as involving either “raising to object” or a “small clause”. The present article presents reasons to prefer the former approach, and especially the understanding of raising to object as structure sharing, as in Lexical Functional Grammar. A main point to be made is that accusative with infinitive sentences in Norwegian do not always involve raising. The accusative with infinitive is syntactically and semantically ambiguous: it is assigned either no thematic role or it is assigned a thematic role by the main verb. When the accusative is assigned a thematic role, the accusative with infinitive is an equi (i.e., control) sentence. Several syntactic arguments are presented showing that some accusative with infinitive sentences are equi sentences and some are raising sentences. However, it is also demonstrated that all accusative with infinitive sentences share important syntactic properties. This supports the argument that syntactic theory must allow equi and raising sentences to be syntactically very similar. The theory of control and complementation of Lexical Functional Grammar is shown to give the correct predictions.

Perfectivity, telicity and Estonian verbs

This paper discusses telicity and perfectivity in Estonian. Neither of these categories corresponds exactly to the Estonian object case alternation, which is argued to reflect predicate or clause aspectual properties and not the NP-related properties of objects. The aim of this account is to accommodate the systematic compatibility of verb classes with certain clausal aspectual object case marking patterns. The paper proposes a way of understanding the interaction between verbal and clausal aspect in terms of boundedness and formalizes it in the framework of Lexical Functional Grammar (LFG). The features of verbs and case markers determine the clausal aspect from different constituents at the syntactic level of constituent structure. The information is unified at another syntactic level, functional structure.

The syntax of locative inversion: The preverbal locative NP in lexical functional grammar

Locative inversion construction in Setswana, as in other Bantu languages, defies straightforward analysis in the sense that the preverbal locative NP does not pass the subject hood test, while the post-verbal NP shows features atypical of objects. In the literature in locative inversion constructions, there is a predominant tendency to analyse the preverbal locative NP as the subject It is argued that the preverbal locative NP in Setswana is not the subject but the topic setting the scene for the focused NP. Data is drawn from some of the findings of the study I conducted on the locative inversion constructions in Botswana in 2003. The article explores information structure analysis of the findings and proposes an analysis within Lexical Functional Grammar, a non-transformational theory that considers languages that are discourse-sensitive and reserve particular positions for pragmatically salient elements, such as topic and focus. LWATI: A Journal of Contemporary Research Vol.3 () 2006: pp.202-222

Evaluating machine translation with LFG dependencies

In this paper we show how labelled dependencies produced by a Lexical-Functional Grammar parser can be used in Machine Translation evaluation. In contrast to most popular evaluation metrics based o...

Derivational Morphology of Italian: Principles for Formalization

The present paper investigates the major derivational strategies underlying the formation of suffixed words in Italian, with the purpose of tackling the issue of their formalization. After having specified the theoretical cognitive premises that orient the work, the interacting component modules of the suffixation process, i.e. morphonology, morphotactics and affixal semantics, are explored empirically, by drawing ample naturally occurring data on a Corpus of written Italian. A special attention is paid to the semantic mechanisms that are involved into suffixation. Some semantic nuclei are identified for the major suffixed word types of Italian, which are due to word formation rules active at the synchronic level, and a semantic configuration of productive suffixes is suggested. A general framework is then sketched, which combines classical finite-state methods with a feature unification-based word grammar. More specifically, the semantic information specified for the affixal material is internalised into the structures of the Lexical Functional Grammar (LFG). The formal model allows us to integrate the various modules of suffixation. In particular, it treats, on the one hand, the interface between morphonology/morphotactics and semantics and, on the other hand, the interface between suffixation and inflection. Furthermore, since LFG exploits a hierarchically organised lexicon in order to structure the information regarding the affixal material, affixal co-selectional restrictions are advatageously constrained, avoiding potential multiple spurious analysis/ generations.

日本語ＬＦＧを用いた副助詞の多義性解消

日本語ＬＦＧを用いた副助詞の多義性解消

Control and semantic resource sensitivity

This paper examines tensions between the syntax of control and semantic resource sensitivity. Structure sharing of controller and control target leads to apparent RESOURCE DEFICIT under certain circumstances. An analysis is presented using Glue Semantics for Lexical Functional Grammar. It demonstrates that structure sharing and resource sensitivity can be reconciled without giving up or relaxing either notion. It is shown that the analysis can handle either property or propositional denotations for controlled complements. The analysis is extended to finite controlled complements, which raise the opposing problem of RESOURCE SURPLUS. A solution is proposed and its typological implications discussed. The syntax and semantics of control as structure sharing is compared to a recent anaphoric control analysis by Dalrymple (2001). Based on facts of exhaustive and partial control, the present analysis is argued to be superior.

Large-Scale Induction and Evaluation of Lexical Resources from the Penn-II and Penn-III Treebanks

We present a methodology for extracting subcategorization frames based on an automatic lexical-functional grammar (LFG) f-structure annotation algorithm for the Penn-II and Penn-III Treebanks. We extract syntactic-function-based subcategorization frames (LFG semantic forms) and traditional CFG category-based subcategorization frames as well as mixed function/category-based frames, with or without preposition information for obliques and particle information for particle verbs. Our approach associates probabilities with frames conditional on the lemma, distinguishes between active and passive frames, and fully reflects the effects of long-distance dependencies in the source data structures. In contrast to many other approaches, ours does not predefine the subcategorization frame types extracted, learning them instead from the source data. Including particles and prepositions, we extract 21,005 lemma frame types for 4,362 verb lemmas, with a total of 577 frame types and an average of 4.8 frame types per verb. We present a large-scale evaluation of the complete set of forms extracted against the full COMLEX resource. To our knowledge, this is the largest and most complete evaluation of subcategorization frames acquired automatically for English.

Representation Theory (review)

Reviewed by: Representation theory by Edwin Williams Norbert Hornstein and Andrew Nevins Representation theory. By Edwin Williams. Cambridge, MA: MIT Press, 2003. Pp. 281. ISBN 0262731509. $28. This provocative and original book covers much interesting territory. Our allotted space requires us (sadly) to restrict our attention to two themes that representation theory (RT) explores: shape conservation and embedding. [End Page 757] Shape conservation. RT divides syntax into multiple levels of representation which map to each other according to global principles, some of which may conflict. Many theories share the view of syntax as a set of interrelated levels: witness lexical functional grammar, which maps f-structure and c-structure (with some optimality-theoretic variants); government and binding theory, which maps d-structure and s-structure; Ray Jackendoff’s recent theories, which map syntax, semantics, and phonology onto one another with global matching principles; and minimalism, which maps syntax to both PF (phonetic form) and LF (logical form) through a locally deterministic set of operations on lexical items. These views assume that each level of representation has different principles and constitutes a distinct ‘sublanguage’ that codes specific kinds of information. RT differs from these in a few important respects: There are seven levels (theta structure, case structure, predicate structure, surface structure, quantifier structure, focus structure, and accent structure), and in principle ‘the limiting case is an RT with exactly the same number of levels as there are functional elements in the structure of the clause in the corresponding Checking Theory’ (58). The levels are not related to one another by transformations of movement but rather by mapping principles, whose central principle is shape conservation. Global shape conservation requires that two levels within a representation relation maximally correspond, modulo the introduction of functional material. Thus, the grammar privileges mappings that conserve the linear and hierarchical relations between 1 and 2, and hence, exceptional case-marking constructions 3 and 4 constitute a bracketing paradox. (1) Theta structure: [Agent [Predicate Theme]] (2) Case structure: [Nominative [Case-Assigner Accusative]] (3) Theta structure: [believe [Mary to be alive]] (4) Case structure: [believe Mary [to be alive]] Shape conservation is global in nature. One instance can be seen in Anders Holmberg’s (1985, 1999) generalization, where the order of phrases within the verb phrase (including verb, one or more objects, and particles) must be preserved if scrambled or head-moved within the middle-field. For RT, Holmberg’s generalization results from a single principle of shape conservation rather than as the conspiracy of different independent movements as occurs with localist theories such as minimalism. In the latter, when more than one dependency occurs for a single lexical item, the transformation is modeled as movement subject to principles of locality or relativized minimality. In RT, only wh-movement is so analyzed (though at certain points, e.g. p. 31, English topicalization is suggested to be as well). All other apparent cases of displacement (e.g. heavy NP shift, topicalization, scrambling, passivization, raising, object-shift) result from a mismapping between two independently generated levels of representation, not transformationally. RT has no movement for passives: Case and theta structures simply happen to be nonisomorphic. One reason for this is that surface structure must map faithfully onto both topic structure and case structure, creating a conflict resolved by ranking. Shape conservation is thus a violable principle. Crosslinguistic variation results from different ranking of relative representational faithfulness: for example, English surface order allows ambiguous quantifier scope, but disallows scrambling because for s-structure to map case structure is more important than for s-structure to map quantifier structure. German has the reverse ranking. Though the global formulation of shape conservation captures generalizations about ‘conspiracies’ of independent movements that preserve linear order, we believe (pace Edwin Williams) that a principle like shape conservation need not imply a new architecture, one that employs RT-like sublanguages (1). For example, Fox and Pesetsky (2003) propose a derivational mechanism of phrase structure building, with intermediate levels of representation (‘phases’) that require consistent linear orderings across the derivation. Here, intermediate levels of representation are governed by a consistency principle and do not assume autonomous sublanguages at each stage. Consider one particularly interesting effect of RT’s...

Relational Nouns, Pronouns, and Resumption

This paper presents a variable-free analysis of relational nouns in Glue Semantics, within a Lexical Functional Grammar (LFG) architecture. Relational nouns and resumptive pronouns are bound using the usual binding mechanisms of LFG. Special attention is paid to the bound readings of relational nouns, how these interact with genitives and obliques, and their behaviour with respect to scope, crossover and reconstruction. I consider a puzzle that arises regarding relational nouns and resumptive pronouns, given that relational nouns can have bound readings and resumptive pronouns are just a specific instance of bound pronouns. The puzzle is why is it impossible for bound implicit arguments of relational nouns to be resumptive? The puzzle is highlighted by a well-known variety of variable-free semantics, where pronouns and relational noun phrases are identical both in category and (base) type. I show that the puzzle also arises for an established variable-based theory. I present an analysis of resumptive pronouns that crucially treats resumptives in terms of the resource logic linear logic that underlies Glue Semantics: a resumptive pronoun is a perfectly ordinary pronoun that constitutes a surplus resource; this surplus resource requires the presence of a resumptive-licensing resource consumer, a manager resource. Manager resources properly distinguish between resumptive pronouns and bound relational nouns based on differences between them at the level of semantic structure. The resumptive puzzle is thus solved. The paper closes by considering the solution in light of the hypothesis of direct compositionality. It is argued that a directly compositional version of the theory is possible, although perhaps not desirable. The implications for direct compositionality are considered.

Treebank-Based Acquisition of Multilingual Unification Grammar Resources

Deep unification- (constraint-)based grammars are usually hand-crafted. Scaling such grammars from fragments to unrestricted text is time-consuming and expensive. This problem can be exacerbated in multilingual broad-coverage grammar development scenarios. Cahill et al. (2002, 2004) and O’Donovan et al. (2004) present an automatic f-structure annotation-based methodology to acquire broad-coverage, deep, Lexical-Functional Grammar (LFG) resources for English from the Penn-II Treebank. In this paper we show how this model can be adapted to a multilingual grammar development scenario to induce robust, wide-coverage, PCFG-based LFG approximations for German from the TIGER Treebank. We show how the architecture of LFG, in particular the distinction between c-structure and f-structure representations, facilitates multilingual, treebank-based unification grammar induction, allowing us to cross-linguistically reuse the lexical extraction and parsing modules from O’Donovan et al. (2004) and Cahill et al. (2004), respectively. We evaluate our grammars against the PARC 700 Dependency Bank (King et al., 2003), against dependency structures for 2000 held-out sentences from the TIGER Corpus as well as against a hand-crafted dependency gold standard for 100 TIGER trees. Currently, our resources achieve 81.79% f-score against the PARC 700, a 2.19% improvement over the best result reported for a hand-crafted grammar in Kaplan et al. (2004), 74.6% against the 2000 held-out TIGER dependency structures and 71.08% against the 100-sentence TIGER gold standard, with substantially improved coverage compared to hand-crafted resources. We have since applied our methodology to induce wide-coverage LFG resources for Chinese (Burke et al., 2004b) from the Penn Chinese Treebank (Xue et al., 2002) and for Spanish from the CAST3LB Treebank (Civit, 2003).

The Feature Space in Parallel Grammar Writing

This paper discusses the methodology and tools applied in the Parallel Grammar project (ParGram) to support consistency and parallelism of linguistic representations across multilingual Lexical Functional Grammar (lfg) grammars. A particular issue is that the grammars in the ParGram project are developed at different international sites. The approach that was established over several years relies on (i) a technical tool for checking adherence to the best-practice feature declaration for linguistic representations, (ii) the coordinated, systematic use of templates for expressing generalizations across lexicon entries and grammar rules, and (iii) a grammar code reviewing committee in which extensions to the existing representations are critically discussed.

Information structuring in Akan question-word fronting and focus constructions

Wh-question fronting and focus constructions in Akan have three structural characteristics in common: constituent fronting, introduction of a clitic morpheme after the fronted constituent, and pronoun resumption in a canonical clause position. In comparing these constructions to each other and to related canonical constructions, one is confronted with the question whether the same discourse-contextual information is consistently expressed in both constructions. Using the framework of Lexical-Functional Grammar, we show that both whquestion fronting and focus constructions share representations in the constituent and functional structure. Considering the individual discourse-contextual information expressed in wh-question fronting and focus constructions, as compared to the discourse-contextual information expressed in the respective in situ and canonical clause counterparts, however, we show that a variance is drawn between them in the information structure. In a further constraint-based analysis, Optimality-Theoretic LFG is used to clarify the proposals made.

On the nature of Welsh VSO clauses

Welsh is a VSO language, in which the verb precedes the subject in finite clauses. It is widely assumed that such clauses have a basic SVO structure. This idea is not limited to Principles and Parameters/Minimalist work, but is also found in Lexical-Functional Grammar and it could also be incorporated into Head-driven Phrase Structure Grammar. Other things being equal, an analysis which involves an SVO structure is more complex than one that does not. Thus, the burden of proof is on advocates of an SVO structure. Arguments for such an analysis have been advanced or might be advanced on the basis VP-like constituents, anaphora, coordination and ellipsis. A close inspection of the arguments suggests that none is very compelling. Moreover, some SVO analyses face problems with agreement, negation or mutation. It seems, then, that an analysis which does not involve an SVO structure is preferable.

Evaluating Automatic LFG F-Structure Annotation for the Penn-II Treebank

Lexical-Functional Grammar (LFG: Kaplan and Bresnan, 1982; Bresnan, 2001; Dalrymple, 2001) f-structures represent abstract syntactic information approximating to basic predicate-argument-modifier (dependency) structure or simple logical form (van Genabith and Crouch, 1996; Cahill et al., 2003a) . A number of methods have been developed (van Genabith et al., 1999a,b, 2001; Frank, 2000; Sadler et al., 2000; Frank et al., 2003) for automatically annotating treebank resources with LFG f-structure information. Until recently, however, most of this work on automatic f-structure annotation has been applied only to limited data sets, so while it may have shown ‘proof of concept’, it has not yet demonstrated that the techniques developed scale up to much larger data sets. More recent work (Cahill et al., 2002a,b) has presented efforts in evolving and scaling techniques established in these previous papers to the full Penn-II Treebank (Marcus et al., 1994). In this paper, we present a number of quantitative and qualitative evaluation experiments which provide insights into the effectiveness of the techniques developed to automatically derive a set of f-structures for the more than 1,000,000 words and 49,000 sentences of Penn-II. Currently we obtain 94.85% Precision, 95.4% Recall and 95.09% F-Score for preds-only f-structures against a manually encoded gold standard.

The Lexical Basis of Sentence Processing: Formal, Computational, and Experimental Issues

1. Preface 2. Words, numbers and all that: The lexicon in sentence understanding (by Stevenson, Suzanne) 3. The lexicon in Optimality Theory (by Bresnan, Joan) 4. Optimality-theoretic Lexical Functional Grammar (by Johnson, Mark) 5. The lexicon and the laundromat (by Fodor, Jerry) 6. Semantics in the spin cycle: Competence and performance criteria for the creation of lexical entries (by Weinberg, Amy) 7. Connectionist and symbolist sentence processing (by Steedman, Mark) 8. A computational model of the grammatical aspects of word recognition as supertagging (by Kim, Albert E.) 9. Incrementality and lexicalism: A treebank study (by Lombardo, Vincenzo) 10. Modular architectures and statistical mechanisms: The case from lexical category disambiguation (by Crocker, Matthew) 11. Encoding and storage in working memory during sentence comprehension (by Stowe, Laurie A.) 12. The time course of information integration in sentence processing (by Spivey, Michael J.) 13. The lexical source of unexpressed participants and their role in sentence and discourse understanding (by Mauner, Gail) 14. Reduced relatives judged hard require constraint-based analyses (by Filip, Hana) 15. Predicting thematic role assignments in context (by Altmann, Gerry T.M.) 16. Lexical semantics as a basis for argument structure frequency biases (by Argamann, Vera) 17. Verb sense and verb subcategorization probabilities (by Roland, Doug) 18. Author index 19. Item index

Clausal complementation in Norwegian

The purpose of this article is to give a description of the syntax of clausal complementation in Norwegian within the framework of Lexical-Functional Grammar (LFG). The data are mostly authentic sentences from text corpora and web pages. The description given is rather different from traditional descriptions. An important point to be made is that Norwegian (finite and non-finite) clausal complements are not homogeneous with respect to their syntactic properties. Even if most clausal complements behave like objects, Norwegian also has clausal complements that do not behave like objects. This phenomenon is well known from other languages, but has never been observed in the literature on Norwegian. The structure and function of non-object clausal complements are discussed, and the traditional idea that they are PP obliques is rejected.

Against ConjP

It has been widely assumed within Principles and Parameters theory that coordinate structures are Conjunction Phrases, with the first conjunct the specifier, the conjunction the head, and the second conjunct the complement. However, this analysis has not been assumed in other frameworks such as Head-driven Phrase Structure Grammar, Lexical Functional Grammar and Categorial Grammar. There is considerable evidence that proponents of other frameworks have been right to be sceptical. Among other things, the distribution of coordinate structures, coordinate structures containing more than two conjuncts, and coordinate structures in which the conjuncts are words pose serious problems for the analysis. It has sometimes been suggested that the analysis receives some support from cases of unbalanced coordination where just one conjunct triggers agreement on some external head or bears the case assigned by an external head. However, a close look suggests that there is no support here for the analysis.