Word Grammar 理論の研究

Due to the programmatic nature of Minimalism, there are many directions that variations of this formalism can take. This paper demonstrates that one trend in Minimalism is steering toward Dependency Grammar (DG): bare phrase structure, label-less trees, and specifier-less syntax are all aspects/developments of/within Minimalism that are steering toward a dependency-based(as opposed to a constituency based) understanding of syntax and grammar. With this trend in mind, a version of Merge is introduced that completes the development. This version generates dependency-based structures in a strictly left-to-right manner, i.e., in the manner that all natural language is spoken and processed. The particular innovation that makes this concept of Merge possible is the catena .T he catena is an ovel unit of syntactic analysis that is unique to dependency-based structures. A number of empirical considerations support catena-based left-to-right Merge, e.g., garden path sentences, restrictions on center embeddings, aspects of coordinate structures, subject-auxiliary inversion, the distribution o fp ronouns, etc.

Bare phrase structure, label-less trees, and specifier-less syntax. Is Minimalism becoming a dependency grammar?

The transfer or share of knowledge between languages is a potential solution to resource scarcity in NLP. However, the effectiveness of cross-lingual transfer can be challenged by variation in syntactic structures. Frameworks such as Universal Dependencies (UD) are designed to be cross-lingually consistent, but even in carefully designed resources trees representing equivalent sentences may not always overlap. In this paper, we measure cross-lingual syntactic variation, or anisomorphism, in the UD treebank collection, considering both morphological and structural properties. We show that reducing the level of anisomorphism yields consistent gains in cross-lingual transfer tasks. We introduce a source language selection procedure that facilitates effective cross-lingual parser transfer, and propose a typologically driven method for syntactic tree processing which reduces anisomorphism. Our results show the effectiveness of this method for both machine translation and cross-lingual sentence similarity, demonstrating the importance of syntactic structure compatibility for boosting cross-lingual transfer in NLP.

/pdf/isomorphic-transfer-of-syntactic-structures-in-cross-lingual-192yxa0jug.pdf

Isomorphic Transfer of Syntactic Structures in Cross-Lingual NLP

The article examines the Universal Dependencies (UD) annotation scheme. The UD project is an international initiative to produce treebanks of the world’s languages, whereby the treebanks have been annotated in a cross-linguistically consistent manner. A central aspect of the UD annotation scheme is its analysis of function words. The scheme advocates subordinating function words to content words. This article discusses linguistic and practical motivations behind the UD decision to subordinate function words to content words. It demonstrates that UD choices in this area are not supported linguistically. At the same time, the near convertibility of the UD treebanks to a more linguistically motivated annotation format means that the UD initiative remains of great value to linguistics in general.

https://hal.inria.fr/hal-02450313/document

The status of function words in dependency grammar: A critique of Universal Dependencies (UD)

/pdf/exploring-language-structure-a-student-s-guide-46q5qsu8ll.pdf

Exploring language structure—A student’s guide

This paper introduces a novel unit of syntactic analysis, the catena (Latin for chain; plural catenae). The catena is defined in a dependency-based grammar as a word or a combination of words that is continuous with respect to dominance. According to this definition, any dependency tree or any subtree (complete or partial) of a dependency tree qualifies as a catena. The paper demonstrates that idioms are stored as catenae and that the elided material of ellipsis mechanisms (e.g., answer fragments, gapping, stripping, VP ellipsis, pseudogapping, sluicing, and comparative deletion) is a catena. Constituents are always catenae, but many catenae are not constituents. Based on the flexibility and utility of the catena concept, the claim is put forth and defended that the catena is the fundamental unit of syntax, not the constituent.

Catenae: Introducing a Novel Unit of Syntactic Analysis

The 'do-so'-diagnostic has been employed for decades to motivate two central aspects of constituency-based theories of syntax: the binary subject–predicate division of the clause (S→NP VP) and the layered analysis of VPs in general. Scrutiny of the diagnostic reveals that it actually supports neither of these two central aspects of most constituency grammars. This manuscript demonstrates that the distribution of the pro-verb do so can be accurately characterized in terms of the catena. The catena is a unit of syntactic analysis associated with dependency grammar (DG). It is a more inclusive and versatile unit than the constituency grammar constituent and it is therefore better suited than the constituent to serve as the fundamental unit of syntactic analysis. In the end, the catena-based account of do so delivers support for the flat analysis of (non-finite) VPs.

The 'Do-So'-Diagnostic: Against finite VPs and for flat non-finite VPs

The Universal Stanford Dependencies (USD) subordinates function words to content words. Auxiliaries, adpositions and subordinators are positioned as dependents of full verbs and nouns, respectively. Such an approach to the syntax of natural languages is contrary to most work in theoretical syntax in the past 35 years, regardless of whether this work is constituencyor dependency-based. A substantial amount of evidence delivers a strong argument for the more conventional approach, which subordinates full verbs to auxiliaries and nouns to adpositions. This contribution demonstrates that the traditional approach to the dependency status of auxiliary verbs is motivated by many empirical considerations, and hence USD cannot be viewed as modeling the syntax of natural languages in a plausible way. 1 The dependency status of function words The Universal Stanford Dependencies (USD), as presented in de Marneffe et al. (2014), advocates a scheme for parsing natural languages that categorically subordinates function words to content words. Auxiliary verbs, adpositions (prepositions and postpositions), subordinators (subordinate conjunctions), etc. are subordinated to the content words with which they co-occur. A more traditional dependency-based analysis assumes the opposite, i.e. most function words dominate the content words with which they co-occur. 1 The following diagrams illustrate both approaches: 1 Determiners are one area of disagreement among linguists. (1) waiting V(Aux)

The Dependency Status of Function Words: Auxiliaries

. The paper presents a new solution to an old problem, the problem of antecedent containment and the infinite regress that it gives rise to. The solution is couched in a dependency grammar (DG) framework and is possible by acknowledging DG catenae. The catena is a novel unit of syntactic analysis. Pronouns that appear to be contained within their antecedents actually take non-constituent catenae as their antecedents. These catenae quite naturally exclude the pronouns, so that antecedent containment never occurs. This account acknowledges surface configurations only. No appeals are made to derivational mechanisms or abstract representations.

Timothy Osborne

Papers

Catenae: Introducing a Novel Unit of Syntactic Analysis

The 'Do-So'-Diagnostic: Against finite VPs and for flat non-finite VPs

The Dependency Status of Function Words: Auxiliaries

Antecedent Containment: A Dependency Grammar Solution in Terms of Catenae