New Dissertation Series

Saarbruecken Dissertations in Computational Linguistics and Language Technology is a new series in which outstanding doctoral dissertations are published, which have been submitted at the Computational Linguistics and Phonetics Department of Saarbuecken University or which have been developed in the context of the Language Technology Lab at the German Research Center for Artificial Intelligence (DFKI).

For every thesis you can find an abstract and an order form at http://www.dfki.de/lt/diss/

The following five volumes have appeared so far:

• Günter Neumann: Uniform Computational Model for Natural Language Parsing and Generation

• Hans-Ulrich Krieger: TDL - A Type Description Language for Constraint-Based Grammars. Foundations, Implementation, and Applications

• Klaus Netter: Functional Categories in an HPSG for German

• Gregor Erbach: Bottom-Up Earley Deduction for Preference-Driven Natural Language Processing

• Tania Avgustinova: Word Order and Clitics in Bulgarian

Romance in HPSG

Sergio Balari and Luca Dini: ROMANCE IN HPSG;
ISBN: 1-57586-082-1 (paper), 1-57586-083-X (cloth);
408 pp. CSLI Publications 1998
http://csli-www.stanford.edu/publications/ email:pubs@roslin.stanford.edu.

This volume addresses several aspects of the syntax and semantics of Romance Languages from the constraint-based, lexicalist perspective of Head-Driven Phrase Structure Grammar (HPSG).

The papers in this volume not only broaden the empirical coverage of HPSG, but also discuss the significant implications of Romance languages for the development of HPSG theory. Among the critical topics discussed in this book are: bounded and unbounded dependency constructions, argument structure, the syntax and semantics of quantification, null complements, missing object constructions, cliticization, and the syntax and semantics of nominal expressions.

Sag-Wasow Textbook

is in the final production stages. Advance copies will be on display at the LSA meetings in Los Angeles in early January. The official publication date is in the third week of February, 1999.

PhD project of Emily Bender (Stanford Linguistics)
Syntactic Variation and Linguistic Competence

This dissertation is concerned with representing internal constraints on syntactic variation as part of grammatical competence. A general HPSG analysis is worked out and applied to three case study variables: copula absence in AAVE, negative concord in the speech of white Detroit-area high schoolers, and verb-form honorifics in Japanese.

Speakers deploy linguistic variants for social purposes: the construction of individual identity, of the situation at hand, and of the situation being described. Thus each variant has a certain value or social meaning, and quantitative differences in the distribution of the variants across different linguistic contexts follow from the intensity of the social value of the variants being dependent on the linguistic context. This, in turn, means that grammatical and social information must be simultaneously available.

Two aspects of HPSG are crucial to this enterprise. The first is the concept of the sign, which is easily extended to the pairing of form and social meaning discussed in the last paragraph. The second is the notion of construction. Since it is bundles of formal properties (rather than single formal properties) which are associated with social values, a framework without any concept of constructions would have nothing to hang the social values on.

PhD project of Arne Reimar Kirchner (Universitaet Goettingen)
Eine Kopfgesteuerte Phrasenstrukturgrammatik für spanische Nominalphrasen
[A Head-driven Phrase Structure Grammar for Spanish Nominal Phrases]

In many Spanish grammars determiners are traditionally regarded as adjectives. So they oppose to qualifying adjectives. Consequently the noun will be classified as head and the determiner as adjunct of the nominal phrase. The fact that the definite article is omitted in the list of determiners casts doubt on this assumption, since it is rather the demonstrative pronoun that seems to constitute the counterpart to the qualifying adjectives. A classification, based on bundles of attributes for determiners and adjectives, creates gradual steps between the two categories. We get a mixed category - comparable to the Spanish infinitives - that is founded on syntactic and semantic attributes. Since the use of a determiner is not always necessary and the adjective can occur both before and behind the noun, the following questions need to be answered: What are the grammatical conditions under which a single noun constitutes a NP? Which semantic factors cause pre- or postposition of adjectives? As there is a close semantic relationship between adjectives in postposition and prepositional phrases, this dissertation includes prepositional phrases in the HPSG-discription of Spanish NPs. A first preparatory work has been my exam-paper on Spanish infinitive constructions (Eine kopfgesteuerte Phrasenstrukturgrammatik fuer spanische Infinitivkonstruktionen), which is available at http://www.gwdg.de/~akirchn.

Arne Reimar Kirchner
Seminar f. Romanische Philologie
Humboldtallee 19
D-37073 Goettingen
Tel.(privat): 0551-375774
Email: akirchn@gwdg.de

PhD project of Susanne Riehemann (Stanford Linguistics)
Idiomaticity and Lexical Constructions

This dissertation explores a constructional approach to various aspects of grammar, in particular derivational morphology, idioms and collocations, and phenomena that straddle the lexical/phrasal boundary, such as separable prefix verbs in German. The approach views these as complex patterns with sub-parts, as opposed to separate pieces and ways for assembling them. Neither affixes nor idiomatic words have an existence outside of these complex patterns. As corpus studies show, the patterns are needed both to structure the inventory of lexicalized words and phrases and to account for morphological productivity and the variability of idioms. Syntactic flexibility is handled by specifying the relationship between the parts of the idiom on a semantic level. The approach makes it unnecessary to construct a separate word syntax and dilute the syntactic apparatus with the possibility of `subcategorizing' non-locally for particular lexical items and non-complements.

PhD project of Jesse Tseng (Edinburgh Cog Sci)
The Grammatical Representation of Prepositional Phrases.

My thesis develops an analysis of prepositions and PPs in HPSG. I argue for a syntactic analysis that combines elements of existing head and marker analyses of prepositions. The MARKING attribute takes over the functions of PFORM and CASE, at once merging and eliminating two closely related features and giving a greater role to the MARKING apparatus in HPSG. I am also working on a semantic analysis that allows degrees of contentfulness, to account for the different uses of prepositions, from case-marking to predicative. The grammatical phenomena I am looking at fall into two broad categories: preposition selection and preposition stranding

PhD project of Anna Kupsc (Paris VII)

My main interest is HPSG, both theoretical and implementation aspects. I work on an HPSG grammar of Polish clitics (e.g., reflexives, binding) but I have also some papers (with Adam Przepiorkowski) on negative concord and morphosyntax of Polish negation. Papers co-authored by Adam can be obtained at

http://www.sfs.nphil.uni-tuebingen.de/~adamp/ap-papers.html

Some of my other papers are also available on-line at

http://www.ipipan.waw.pl/mmgroup/papers.html