Resource Grammar Research Articles

Morphological marking of constituent questions: A case for nonlocal amalgamation

This paper considers the role of nonlocal amalgamation in a system of analyses for typologically diverse languages. Nonlocal amalgamation (Bouma et al. 2001) was suggested in particular to get rid of extraction rules in Pollard and Sag's (1994) analysis of long-distance dependencies. However, in implemented projects like the English Resource Grammar (Flickinger, 2000, 2011) and the Grammar Matrix (Bender et al., 2002, 2010), the extraction rules have been maintained, while nonlocal amalgamation is used for the analysis of phenomena like the easy-adjectives. Zamaraeva and Emerson (2020) argue that, if extraction rules are kept, then supporting the English easy-adjectives may be an insufficient reason to maintain nonlocal amalgamation in a cross-linguistic system like the Grammar Matrix, as it complicates the analysis of multiple question word fronting with flexible word order (in languages such as Russian [rus]). However, I present here a case of morphological marking of questions (in languages like Makah [myh]) which further motivates nonlocal amalgamation, as the analysis is remarkably more simple with it than it is without it. An analysis of morphological marking of questions needs to be part of a cross-linguistic system such as the Grammar Matrix as well as an analysis of multiple fronting, which adds a new tension at the level of the Matrix "core" and provides concrete material for discussion of issues ranging from empirical implementation of theoretical ideas like nonlocal amalgamation to the big question of how much of typological space a single system of grammar is expected to cover.

HPSG analysis and computational implementation of Indonesian passives

This study aims to analyze and develop a detailed model of syntax and semantics of passive sentences in standard Indonesian in the framework of Head-Driven Phrase Structure Grammar (HPSG) (Pollard & Sag, 1994; Sag et al., 2003) and Minimal Recursion Semantics (MRS) (Copestake et al., 2005), explicit enough to be interpreted by a computer, focusing on implementation rather than theory. There are two main types of passive in Indonesian, following Sneddon et al. (2010, pp. 256-260) and Alwi et al. (2014, pp. 352-356), called ‘passive type 1’ (P1) and ‘passive type 2’ (P2). Both types were analyzed and implemented in the Indonesian Resource Grammar (INDRA), a computational grammar for Indonesian (Moeljadi et al., 2015).

Aligning speech and co-speech gesture in a constraint-based grammar

This paper concerns the form-meaning mapping of communicative actions consisting of speech and improvised co-speech gestures. Based on the findings of previous cognitive and computational approaches, we advance a new theory in which this form-meaning mapping is analysed in a constraint-based grammar. Motivated by observations in naturally occurring examples, we propose several construction rules, which use linguistic form, gesture form and their relative timing to constrain the derivation of a single speech-gesture syntax tree, from which a meaning representation can be composed via standard methods for semantic composition. The paper further reports on implementing these speech-gesture construction rules within the English Resource Grammar (Copestake and Flickinger 2000). Since gestural form often underspecifies its meaning, the logical formulae that are composed via syntax are underspecified so that current models of the semantics/pragmatics interface support the range of possible interpretations of the speech-gesture act in its context of use.

From Abstract Syntax to Universal Dependencies

Abstract syntax is a semantic tree representation that lies between parse trees and logical forms. It abstracts away from word order and lexical items, but contains enough information to generate both surface strings and logical forms. Abstract syntax is commonly used in compilers as an intermediate between source and target languages. Grammatical Framework (GF) is a grammar formalism that generalizes the idea to natural languages, to capture cross-lingual generalizations and perform interlingual translation. As one of the main results, the GF Resource Grammar Library (GF-RGL) has implemented a shared abstract syntax for over 30 languages. Each language has its own set of concrete syntax rules (morphology and syntax), by which it can be generated from the abstract syntax and parsed into it. This paper presents a conversion method from abstract syntax trees to dependency trees. The method is applied for converting GF-RGL trees to Universal Dependencies (UD), which uses a common set of labels for different languages. The correspondence between GF-RGL and UD turns out to be good, and the relatively few discrepancies give rise to interesting questions about universality. The conversion also has potential for practical applications: (1) it makes the GF parser usable as a rule-based dependency parser; (2) it enables bootstrapping UD treebanks from GF treebanks; (3) it defines formal criteria to assess the informal annotation schemes of UD; (4) it gives a method to check the consistency of manually annotated UD trees with respect to the annotation schemes; (5) it makes information from UD treebanks available.

A GF miniature resource grammar for Tswana: modelling the proper verb

The Grammatical Framework (GF) not only offers state of the art grammar-based machine translation support between an increasing number of languages through its so-called Resource Grammar Library, but is also fast becoming a de facto framework for developing multilingual controlled natural languages (CNLs). For a natural language to share maximally in the opportunities that GF-based multilingual CNL support presents, it has to have a GF resource grammar. Tswana, an agglutinating Bantu language, spoken in Southern Africa as one of the eleven official languages of South Africa, does not yet have such a grammar. This article reports on the development of a so-called miniature resource grammar, a first step towards a full resource grammar for Tswana. The focus is on the modelling of the Tswana proper verb as it occurs in simple sentences. The (proper) verb is the morphologically most complex word category in Tswana, and therefore constitutes a notable contribution towards the development of a GF resource grammar for Tswana. The computational model is discussed in some detail, implemented and tested on a systematically constructed treebank.

A multilingual FrameNet-based grammar and lexicon for controlled natural language

Berkeley FrameNet is a lexico-semantic resource for English based on the theory of frame semantics. It has been exploited in a range of natural language processing applications and has inspired the development of framenets for many languages. We present a methodological approach to the extraction and generation of a computational multilingual FrameNet-based grammar and lexicon. The approach leverages FrameNet-annotated corpora to automatically extract a set of cross-lingual semantico-syntactic valence patterns. Based on data from Berkeley FrameNet and Swedish FrameNet, the proposed approach has been implemented in Grammatical Framework (GF), a categorial grammar formalism specialized for multilingual grammars. The implementation of the grammar and lexicon is supported by the design of FrameNet, providing a frame semantic abstraction layer, an interlingual semantic API (application programming interface), over the interlingual syntactic API already provided by GF Resource Grammar Library. The evaluation of the acquired grammar and lexicon shows the feasibility of the approach. Additionally, we illustrate how the FrameNet-based grammar and lexicon are exploited in two distinct multilingual controlled natural language applications. The produced resources are available under an open source license.

Feeling our way to an analysis of English possessed idioms

This paper describes an analysis for possessive idioms in English (e.g. ˋI twiddle my thumbs' ˋˋI am idle''). The analysis relies on matching at the semantic level, to allow for syntactic variation. It has been implemented in the English Resource Grammar, and tested by parsing a subset of the British National Corpus. In addition to the syntactic analysis, we have linked the idioms to entries in the Princeton Wordnet, to allow for further lexical semantic analysis.

High-level methodologies for grammar engineering, introduction to the special issue

Grammar Engineering is the task of designing and implementing linguistically motivated electronic descriptions of natural language (so-called grammars). These grammars are expressed within well-defined theoretical frameworks, and offer a fine-grained description of natural language. While grammars were first used to describe syntax, that is to say, the relations between constituents in a sentence, they often go beyond syntax and include semantic information. Grammar engineering provides precise descriptions which can be used for natural language understanding and generation, making these valuable resources for various natural language applications, including textual entailment, dialogue systems, or machine translation. The first attempts at designing large-scale resource grammars were costly because of the complexity of the task (Erbach et al. 1990) and of the number of persons that were needed (see e.g. Doran et al. 1997). Advances in the field have led to the development of environments for semi-automatic grammar engineering, borrowing ideas from compilation (grammar engineering is compared with software development) and machine learning. This special issue reports on new trends in the field, where grammar engineering benefits from elaborate high-level methodologies and techniques, dealing with various issues (both theoretical and practical).

Ivan A. Sag

Ivan A. Sag

Detecting modification of biomedical events using a deep parsing approach

BackgroundThis work describes a system for identifying event mentions in bio-molecular research abstracts that are either speculative (e.g. analysis of IkappaBalpha phosphorylation, where it is not specified whether phosphorylation did or did not occur) or negated (e.g. inhibition of IkappaBalpha phosphorylation, where phosphorylation did not occur). The data comes from a standard dataset created for the BioNLP 2009 Shared Task. The system uses a machine-learning approach, where the features used for classification are a combination of shallow features derived from the words of the sentences and more complex features based on the semantic outputs produced by a deep parser.MethodTo detect event modification, we use a Maximum Entropy learner with features extracted from the data relative to the trigger words of the events. The shallow features are bag-of-words features based on a small sliding context window of 3-4 tokens on either side of the trigger word. The deep parser features are derived from parses produced by the English Resource Grammar and the RASP parser. The outputs of these parsers are converted into the Minimal Recursion Semantics formalism, and from this, we extract features motivated by linguistics and the data itself. All of these features are combined to create training or test data for the machine learning algorithm.ResultsOver the test data, our methods produce approximately a 4% absolute increase in F-score for detection of event modification compared to a baseline based only on the shallow bag-of-words features.ConclusionsOur results indicate that grammar-based techniques can enhance the accuracy of methods for detecting event modification.

Symmetric and Asymmetric Properties in Korean Verbal Coordination: A Computational Implementation

Of the coordination structures in Korean, the symmetric and asymmetric properties of verbal coordination have challenged both theoretical and computational approaches. This paper shows how a typed feature structure grammar, HPSG, together with the notions of ``type hierarchy`` and ``constructions``, can provide a robust basis for parsing (un)tensed verbal coordination as well as pseudo-coordination found in the language. We show that the analysis sketched here and computationally implemented in the existing resource grammar for Korean, Korean Resource Grammar (KRG), can yield proper syntactic structures as well as enriched semantic representations for real-time applications such as machine translation. (Kyung Hee University and Kangnam University)

Deep Processing of Korean and the Development of the Korean Resource Grammar

Kim, Jong-Bok, Yang, Jaehyung, Song, Sanghoun & Bond, Francis. 2011. Deep Processing of Korean and the Development of the Korean Resource Grammar. Linguistic Research 28(3), 635-672. The Korean Resource Grammar (KRG) is a broad-coverage, linguistically precise, and open-source grammar of Korean that has been developed since 2003. This paper reports how the KRG, aiming at deep processing of Korean, has gone through two main developing phases, in particular from a linguistically-oriented deep grammar to a wider-coverage with grammar customization. The development of the KRG in these two phases, also participating in the international collaboration project DELPH-IN (Deep Language Processing in HPSG), has resulted in promising directions for the improvement of deep parsing as well as generation methods in a more systematic way so that it can be geared toward practical NLP applications such as machine translation.

Cross-Domain Effects on Parse Selection for Precision Grammars

We examine the impact of domain on parse selection accuracy, in the context of precision HPSG parsing using the English Resource Grammar, using two training corpora and four test corpora and evaluating using exact tree matches as well as dependency F-scores. In addition to determining the relative impact of in- vs. cross-domain parse selection training on parser performance, we propose strategies to avoid cross-domain performance penalty when limited in-domain data is available. Our work supports previous research showing that in-domain training data significantly improves parse selection accuracy, and that it provides greater parser accuracy than an out-of-domain training corpus of the same size, but we verify experimentally that this holds for a handcrafted grammar, observing a 10---16% improvement in exact match and 5---6% improvement in dependency F-score by using a domain-matched training corpus. We also find it is possible to considerably improve parse selection accuracy through construction of even small-scale in-domain treebanks, and learning of parse selection models over in-domain and out-of-domain data. Naively adding an 11,000-token in-domain training corpus boosts dependency F-score by 2---3% over using solely out-of-domain data. We investigate more sophisticated strategies for combining data from these sources to train models: weighted linear interpolation between the single-domain models, and training a model from the combined data, optionally duplicating the smaller corpus to give it a higher weighting. The most successful strategy is training a monolithic model after duplicating the smaller corpus, which gives an improvement over a range of weightings, but we also show that the optimal value for these parameters can be estimated on a case-by-case basis using a cross-validation strategy. This domain-tuning strategy provides a further performance improvement of up to 2.3% for exact match and 0.9% for dependency F-score compared to the naive combination strategy using the same data.

Korean comparative constructions: A constraint-based approach and computational implementation

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.

Parsing Korean Comparative Constructions in a Typed-Feature Structure Grammar

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 ner model for contextual information.

The GF Resource Grammar Library

The GF Resource Grammar Library is a set of natural language grammars implemented in GF (Grammatical Framework). These grammars are in a strong sense parallel: they are built upon a common abstract syntax, i.e. a common tree structure. Individual languages are obtained via compositional mappings from abstract syntax trees to feature structures specific to each language. The grammar defines, for each language, a complete set of morphological paradigms and a syntax fragment comparable to CLE (Core Language Engine). It is available as open-source software under the GNU LGPL License. The current coverage is fourteen languages: Bulgarian, Catalan, Danish, English, Finnish, French, German, Italian, Norwegian (bokmål), Polish, Romanian, Russian, Spanish, and Swedish. More languages are under construction.
 The library can be used as a resource for language processing tasks, such as translation, multilingual generation, software localization, natural language interfaces, and spoken dialogue systems. The library may also have some language-typological interest as an experiment showing how much grammatical structure can be shared between languages. The focus of this paper is on the linguistic aspects of the library—in particular, what syntactic structures are covered and how the problems arising in different languages have been solved. We will also discuss the nature of the common abstract syntax with respect to translation equivalence.

GF: A Multilingual Grammar Formalism

AbstractGF (Grammatical Framework) is a grammar formalism designed for dealing with many languages simultaneously. A multilingual GF grammar has a language‐independent abstract syntax that deals with pure tree structures, and a set of language‐dependent concrete syntaxes that specify how the trees are expressed in different languages. Multilingual grammars formalize the idea that the same grammatical categories (such as noun phrases and verb phrases) and syntax rules (such as predication) can appear in different languages. They are also a powerful tool for natural language engineering, used for applications ranging from a translator of mathematical exercises to a dialogue system usable for talking with devices in a car. As a programming language, GF has constructs inherited from functional programming languages, an optimizing grammar compiler and code generation into several other formats. The most ambitious multilingual grammar is the GF Resource Grammar Library, which implements the main grammar rules of 12 languages. This library enables non‐linguist programmers to build linguistically correct applications. It is at the same time a linguistic experiment showing how far one can go in sharing structure between languages.

Modular Grammar Engineering in GF

The Grammatical Framework GF is a grammar formalism designed for multilingual grammars. A multilingual grammar has a shared representation, called abstract syntax, and a set of concrete syntaxes that map the abstract syntax to different languages. A GF grammar consists of modules, which can share code through inheritance, but which can also hide information to achieve division of labour between grammarians working on different modules. The goal is to make it possible for linguistically untrained programmers to write linguistically correct application grammars encoding the semantics of special domains. Such programmers can rely on resource grammars, written by linguists, which play the role of standard libraries. Application grammarians use resource grammars through abstract interfaces, and the type system of GF guarantees that grammaticality is preserved. The ongoing GF resource grammar project provides resource grammars for ten languages. In addition to their use as libraries, resource grammars serve as an experiment showing how much grammar code can be shared between different languages.

Any questions left? Review of Ginzburg & Sag's Interrogative investigations

For better or worse, linguistics is rife with frameworks, each with its own ethos. Two important aspects of Head-driven Phrase Structure Grammar's are: (1) the search for an explicit and exhaustive account of the intricate syntactic and semantic facts that constitute one's grammar and (2) the hypothesis that general/universal and specific aspects of one's grammar are not qualitatively distinct. This book stands as a superb example of this ethos. It illustrates the fecundity of Head-driven Phrase Structure Grammar (henceforth, HPSG) as a framework and advances our knowledge of the syntax and semantics of interrogatives. It is the most explicit description of the semantics and syntax of any area of English syntax of which this reviewer is aware. It thus sets up a healthy empirical benchmark for other theories of interrogatives. The sixty pages of appendices that detail the grammar discussed in the book and its almost complete implementation in the current version of the English Resource Grammar attest to this empirical bent. Any future theory will have to match it in accuracy before any metatheoretical issues (e.g. simplicity or explanatory adequacy) can be meaningfully discussed. Its precision and comprehensiveness will also, hopefully, lead to descriptions of unbounded dependencies and clause-types in other languages that are orders of magnitude more detailed than those currently available. Ginzburg & Sag's (G&S's) book also illustrates how possibly universal principles (such as the requirement that head-daughters and mothers of a local tree share information) and construction-specific requirements (such as the fact that the scope of

In search of epistemic primitives in the English Resource Grammar (or Why HPSG can't live without higher-order datatypes)

This paper seeks to improve HPSG engineering through the design of more terse, readable and intuitive type signatures. It argues against the exclusive use of IS-A networks and, with reference to the English Resource Grammar, demonstrates that a collection of higher-order datatypes are already acutely in demand in contemporary HPSG design. Some default specification conventions to assist in maximizing the utility of higher-order type constructors are also discussed.