Abstracts

 

Katya Alahverdzhieva: Towards an English Tree-Adjoining MetaGrammar

Within this talk I will try to present my masters thesis which is aimed at defining a core tree-adjoining grammar based on the XTAG developed by the University of Pennsylvania. The grammar is set within the context of XMG environment for specifying unification based grammars. The project aims at defining generalisations which are either of type shared structures or diathesis alternations. Within this talk I will try to familiarise the audience with the abstraction levels for efficient grammatical description: tree families (i.e. a main verb and its subcategorisation frame); diathesis alternations (active/passive alternations); syntactic functions (alternative realisations of a syntactic function); tree fragments (elementary building blocks shared among tree families). At the end, I am planning to show its coverage on a specially built test suite of sentences covering various language phenomena.

 

Eric Auer: Max Planck Institute Nijmegen & software

 

Nadine BorchardtNews from field work on Ikaann (Benue-Congo, Nigeria)

After the experience of two months in the field for linguistic research on Ikaann, I would like to present a first glance at the most interesting data I collected as well as valuable hints for future field workers.
Ikaann is an endangered Benue-Congo language in Nigeria, West Africa. Within a HRELP (Hans Rausing Endangered Languages Project) funded project, I conducted a field trip to Ikakumo village, Ondo State, at the beginning of this year in order to collect data for my Master thesis on the numeral system of Ikaann.
First, I would like to outline parts of my work in progress and therewith some of my findings as far as Ikaann numeral bases as well as grammatical number constructions are concerned. Second, I will treat some fundamental and very practical issues of field research: What do I have to expect when going to the field (in West Africa)? How do I find the appropriate environment for my work? How do I choose my informants? What kind of problems am I likely to face? What kind of equipment should I bring? And what should I better not forget?
I will try to answer questions like these and both give very practical hints for all those who aim at going on a field trip and exchange experience with field veterans.

 

Juliane Böttger and Nadine Borchardt: K(l)icks and Quirks in Taa (Southern Khoisan, Namibia)

Even among non-linguists Khoisan languages are famous for their click sounds. But, of course, they have much more to offer than a quirky phoneme inventory. On the basis of an ongoing DoBeS research project on Taa (Southern Khoisan), a dialect cluster spoken between Botswana and Namibia, we will present some globally rare linguistic features found in Taa. After a brief socio-linguistic sketch we will outline the world's largest phoneme inventory, the linear agreement system and the noun class and gender system.

 

Oscar Garcia-Marchena: Spanish Common Noun Architecture in Formal Grammars

In this paper we present a classification of common noun types according to their distributional, morphological and semantic properties, together with a HPSG architecture. We propose an adapted implementation of these data for a MTT grammar currently used for human-machine dialogue with the aim of increasing its performance.

 

Armin Hönen: Similarities between Transeurasiatic and Indoeuropean and didactically inspired implications for the current classification systems

 

Birgit Jänen: Ditransitives in Puma (a Kiranti Language)

Puma is an endangered language belonging to the Sino-Tibetan language family ( > Tibeto-Burman > Himalayish > Kiranti > Eastern ) which is spoken in Eastern Nepal.
Within a cooperation project between the University of Leipzig and the Tribhuvan University Kathmandu, Nepal (Chintang and Puma Documentation Project) I gathered data on ditransitive constructions during a two months stay in Nepal.
Based on elicitations with native speakers I will present preliminary analyses and give an outlook on further steps of my work towards my "Magisterarbeit" on this topic.

 

Kristin Kopf: Gäste, Wörter und Wägen: Der Umlaut als Pluralmarker im Deutschen

In dieser AG soll die Geschichte des Umlauts vom Alt- bis zum Neuhochdeutschen nachgezeichnet werden. Die wichtigen Stufen dabei sind die Entwicklung vom Allophon zum Phonem und schließlich analogischer Ausgleich und analogische Ausdehnung in Nominalparadigmen, die zur Verwendung des Umlauts als Pluralmarker und zu seiner Ausbreitung vom Frühneuhochdeutschen bis heute (vgl. Wägen, Häufen, Köffer) führen.Die AG richtet sich an TeilnehmerInnen mit grundlegenden linguistischen Vorkenntnissen, die sich noch nicht näher mit den Sprachstufen des Deutschen beschäftigt haben.

 

Pierre Lison: Improving the robustness of spoken dialogue systems using grammar relaxation

We present a work in progress exploring the application of grammar relaxation techniques to improve robustness in spoken dialogue comprehension. The capacity to comprehend spoken language is often crucial in achieving natural communication with humans, yet developing dialogue systems able to robustly and accurately interpret spoken utterances remains a significant challenge. This is notably due to the presence of numerous speech disfluencies (pauses, repetitions, repairs, etc.) and to the globally poor performance of speech recognition, especially when operating in noisy, real-world environments.
To overcome this problem, various methods have been proposed in the literature. The most popular ones include “concept spotting” techniques - filling slots in hand-crafted semantic frames -, or statistical models trained on semantically annotated corpora. In this paper we argue for a generic and linguistically sound approach based on a weighted Combinatory Categorial Grammar [CCG]. In order to account for speech disfluencies and recognition errors, the grammatical constraints are relaxed by means of non-standard CCG combinators. A statistical model is applied to select the most likely parse among those licensed by the grammar. The model parameters are estimated using an automatically generated corpus. The proposed approach therefore combines both linguistic and statistical sources of information to yield the most likely utterance interpretation.
The system is currently being implemented as part of a cognitive architecture for mobile robots interacting with humans using spoken natural language to perform a variety of service-oriented tasks. Preliminary evaluation results on a test suite indicate a statistically significant improvement, both in terms of parsing coverage and accuracy, of the relaxed grammar over the non-relaxed one.

 

Daniel Puccini: The modular makeup of the language faculty as reflected through the acquisition of nominal and pronominal reference by children with SLI

The organization of language in the mind has been a widely discussed topic in recent literature in linguistics. On the one hand, researchers such as Fodor , 1983 #50} and Chomsky {, 1986 #51} maintain that language is composed of three distinct sub-modules: the lexicon, the computational system, and the pragmatic system. Within this framework, Specific Language Impairment is considered to affect only the computational system. On the other hand, researchers such as Tomasello (2003) argue that human language emerges only after other more domain-general cognitive processes are in place. Proponents of non-modular theories mind argue that there exists a critical dependency between language acquisition and cognitive development, thus, a disorder that is specific to one’s syntactic knowledge should not even exist in the first place {Johnston, 1994 #6}.
The current project aims to investigate the possibility of a modular composition of the language faculty as reflected by the acquisition of nominal and pronominal reference by English speaking children with Specific Language Impairment. Reference in English provides an interesting testing ground to assess the modular make-up of the mind because speakers must have knowledge of syntactic and pragmatic information in order to refer in an adult-like fashion. By monitoring and comparing longitudinal determiner selections of typically developing and SLI populations, it will be possible to discern whether the deficits of children with SLI can be considered specific to syntactic relations or whether they also lag behind in pragmatic awareness. If Specific Language Impairment (SLI) is a disorder that affects only a child’s grammatical development (as maintained by proponents of modular theories of mind) then one would predict that the children with SLI will perform at age-appropriate levels with respect to the pragmatics of reference. Under the assumption that language emerges as more general cognitive processes mature, one would predict a much stronger relation between pragmatic and syntactic development. Likewise, children with SLI would be expected to experience deficiencies with both the pragmatics syntax of English reference. Results of the project will be discussed with two goals in mind: First, the project should offer insights regarding the underlying source of SLI, and second, results will be discussed with regard to how language is structured in the mind.

 

Uta Reinöhl: Lexicon, grammar and iconicity in German sign language

Sign languages have been recognised as fully-fledged languages with an own grammar since roughly the 1980s. William Stokoe first postulated in the 1960s that sign languages have a grammar and a segmentable morphology distinct from the language spoken in the respective area. Before that, signing was considered a kind of pantomime, that is, a primitively iconic projection of things in the world that could communicate only very simple concepts and needs of everyday life. Today, it is commonly known that sign languages are only partly iconic and that they are equally complex in linguistic structure and equally apt to express abstract ideas as any other human language. However, iconicity is an important and obvious element of sign languages and not only the lexicon but also grammatical functions rely to a certain extent on iconic elements. The relation between grammar, lexicon and iconicity will be explored in the presentation with a focus on the study of verbs in German Sign Language.

 

Patrick Schulz: The Yurok Language

In this presentation, I will talk about Yurok, an Algic Language spoken in northern California by about 10 people. In the rst part of the talk, I will give a brief overview to the Yurok People, their Culture, and their History. In the second part, I will talk about their language in general, and in the last part I will show certain phonological and morpho-syntactical phenomena in the language in more detail, including rhotic vowel harmony, syntactic deponence, and periphrastic agreement.

 

Julia Steiner: Why English has kept what German has dropped. Irregular weak verbs and their influence on the strong-weak-continuum

The Germanic languages historically distinguishes two major classes of verbs: Strong verbs, which form their past tense by means of a vowel change (sing - sang - sung), and weak verbs, which achieve the same through a dental suffix (play - played - played). Both in English and in German, the weak verbs have developed into an extremely productive default inflexion. In German, this regular class sharply contrasts with the strong verbs, which are largely unproductive, yet surprisingly stable. Only a meagre eight verbs do not fit into this classification, as they combine a dental suffix with a vowel change, and sometimes omit the schwa in the suffix, e.g. kennen - kannte - gekannt, senden - sandte - gesandt. These eight verbs are the residue of the once large and even slightly productive class of "Rückumlaut" verbs – long-stemmed, syncopated ja-verbs which amounted to more than 200 types in Early Modern High German, yet were almost completely regularised by the end of that period. In contrast to that, almost half of the irregular verbs in Present-Day English have their origins in the same syncopated subclass of the weak conjugation – examples are keep - kept - kept, meet - met - met, and the best-loved of all, put - put - put. An irregular verb in English can thus sport almost any combination of vowel change, suffix or syncope. As a result, the once neat and clear distinction between "strong" and "weak" has been replaced by a gradient or scale. In my talk I will try to account for this striking divergence, drawing on the findings of traditional grammar as well as on frequency effects, experimental results, and Bybee's relevance theory.
It appears that German eliminates these irregular forms because it has preserved a high degree of stem-internal inflection throughout its morphological system, which leads to markedness conflicts in these verbs. English, however, has completely given up on stem-internal inflection, so that the irregular verbs of both strong and weak origin survive as isolated exceptions. In the absence of any rules, other factors, such as word frequency, phonological schemas, semantic split or rhythmic alternation appear to determine what is preserved and what is not.