Refine
Year of publication
- 2006 (49) (remove)
Document Type
- Preprint (49) (remove)
Has Fulltext
- yes (49)
Is part of the Bibliography
- no (49)
Keywords
- Chatten <Kommunikation> (3)
- Deutsch (3)
- Dialektologie (3)
- Lexicalized Tree Adjoining Grammar (2)
- RHIC (2)
- Schweizerdeutsch (2)
- Syntax (2)
- syntax (2)
- ADD (1)
- African Diaspora (1)
Institute
- Physik (24)
- Frankfurt Institute for Advanced Studies (FIAS) (16)
- Extern (10)
- Informatik (1)
- Medizin (1)
- Sportwissenschaften (1)
- Universitätsbibliothek (1)
- Zentrum für Weiterbildung (1)
Using a qualitative analysis of disagreements from a referentially annotated newspaper corpus, we show that, in coreference annotation, vague referents are prone to greater disagreement. We show how potentially problematic cases can be dealt with in a way that is practical even for larger-scale annotation, considering a real-world example from newspaper text.
In this paper, we investigate the usefulness of a wide range of features for their usefulness in the resolution of nominal coreference, both as hard constraints (i.e. completely removing elements from the list of possible candidates) as well as soft constraints (where a cumulation of violations of soft constraints will make it less likely that a candidate is chosen as the antecedent). We present a state of the art system based on such constraints and weights estimated with a maximum entropy model, using lexical information to resolve cases of coreferent bridging.
In recent years, research in parsing has extended in several new directions. One of these directions is concerned with parsing languages other than English. Treebanks have become available for many European languages, but also for Arabic, Chinese, or Japanese. However, it was shown that parsing results on these treebanks depend on the types of treebank annotations used. Another direction in parsing research is the development of dependency parsers. Dependency parsing profits from the non-hierarchical nature of dependency relations, thus lexical information can be included in the parsing process in a much more natural way. Especially machine learning based approaches are very successful (cf. e.g.). The results achieved by these dependency parsers are very competitive although comparisons are difficult because of the differences in annotation. For English, the Penn Treebank has been converted to dependencies. For this version, Nivre et al. report an accuracy rate of 86.3%, as compared to an F-score of 92.1 for Charniaks parser. The Penn Chinese Treebank is also available in a constituent and a dependency representations. The best results reported for parsing experiments with this treebank give an F-score of 81.8 for the constituent version and 79.8% accuracy for the dependency version. The general trend in comparisons between constituent and dependency parsers is that the dependency parser performs slightly worse than the constituent parser. The only exception occurs for German, where F-scores for constituent plus grammatical function parses range between 51.4 and 75.3, depending on the treebank, NEGRA or TüBa-D/Z. The dependency parser based on a converted version of Tüba-D/Z, in contrast, reached an accuracy of 83.4%, i.e. 12 percent points better than the best constituent analysis including grammatical functions.
This paper presents an approach to the question whether it is possible to construct a parser based on ideas from case-based reasoning. Such a parser would employ a partial analysis of the input sentence to select a (nearly) complete syntax tree and then adapt this tree to the input sentence. The experiments performed on German data from the Tüba-D/Z treebank and the KaRoPars partial parser show that a wide range of levels of generality can be reached, depending on which types of information are used to determine the similarity between input sentence and training sentences. The results are such that it is possible to construct a case-based parser. The optimal setting out of those presented here need to be determined empirically.
This paper presents a comparative study of probabilistic treebank parsing of German, using the Negra and TüBa-D/Z treebanks. Experiments with the Stanford parser, which uses a factored PCFG and dependency model, show that, contrary to previous claims for other parsers, lexicalization of PCFG models boosts parsing performance for both treebanks. The experiments also show that there is a big difference in parsing performance, when trained on the Negra and on the TüBa-D/Z treebanks. Parser performance for the models trained on TüBa-D/Z are comparable to parsing results for English with the Stanford parser, when trained on the Penn treebank. This comparison at least suggests that German is not harder to parse than its West-Germanic neighbor language English.
Relative quantifier scope in German depends, in contrast to English, very much on word order. The scope possibilities of a quantifier are determined by its surface position, its base position and the type of the quantifier. In this paper we propose a multicomponent analysis for German quantifiers computing the scope of the quantifier, in particular its minimal nuclear scope, depending on the syntactic configuration it occurs in.
Our paper aims at capturing the distribution of negative polarity items (NPIs) within lexicalized Tree Adjoining Grammar (LTAG). The condition under which an NPI can occur in a sentence is for it to be in the scope of a negation with no quantifiers scopally intervening. We model this restriction within a recent framework for LTAG semantics based on semantic unification. The proposed analysis provides features that signal the presence of a negation in the semantics and that specify its scope. We extend our analysis to modelling the interaction of NPI licensing and neg raising constructions.
This paper compares two approaches to computational semantics, namely semantic unification in Lexicalized Tree Adjoining Grammars (LTAG) and Lexical Resource Semantics (LRS) in HPSG. There are striking similarities between the frameworks that make them comparable in many respects. We will exemplify the differences and similarities by looking at several phenomena. We will show, first of all, that many intuitions about the mechanisms of semantic computations can be implemented in similar ways in both frameworks. Secondly, we will identify some aspects in which the frameworks intrinsically differ due to more general differences between the approaches to formal grammar adopted by LTAG and HPSG.
The work presented here addresses the question of how to determine whether a grammar formalism is powerful enough to describe natural languages. The expressive power of a formalism can be characterized in terms of i) the string languages it generates (weak generative capacity (WGC)) or ii) the tree languages it generates (strong generative capacity (SGC)). The notion of WGC is not enough to determine whether a formalism is adequate for natural languages. We argue that even SGC is problematic since the sets of trees a grammar formalism for natural languages should be able to generate is difficult to determine. The concrete syntactic structures assumed for natural languages depend very much on theoretical stipulations and empirical evidence for syntactic structures is rather hard to obtain. Therefore, for lexicalized formalisms, we propose to consider the ability to generate certain strings together with specific predicate argument dependencies as a criterion for adequacy for natural languages.
In the recent literature there is growing interest in the morpho-syntactic encoding of hierarchical effects. The paper investigates one domain where such effects are attested: ergative splits conditioned by person. This type of splits is then compared to hierarchical effects in direct-inverse alternations. On the basis of two case studies (Lummi instantiating an ergative split person language and Passamaquoddy an inverse language) we offer an account that makes no use of hierarchies as a primitive. We propose that the two language types differ as far as the location of person features is concerned. In inverse systems person features are located exclusively in T, while in ergative systems, they are located in T and a particular type of v. A consequence of our analysis is that Case checking in split and inverse systems is guided by the presence/absence of specific phi-features. This in turn provides evidence for a close connection between Case and phi-features, reminiscent of Chomsky’s (2000, 2001) Agree.
The causative/anticausative alternation has been the topic of much typological and theoretical discussion in the linguistic literature. This alternation is characterized by verbs with transitive and intransitive uses, such that the transitive use of a verb V means roughly "cause to Vintransitive" (see Levin 1993). The discussion revolves around two issues: the first one concerns the similarities and differences between the anticausative and the passive, and the second one concerns the derivational relationship, if any, between the transitive and intransitive variant. With respect to the second issue, a number of approaches have been developed. Judging the approach conceptually unsatisfactory, according to which each variant is assigned an independent lexical entry, it was concluded that the two variants have to be derivationally related. The question then is which one of the two is basic and where this derivation takes place in the grammar. Our contribution to this discussion is to argue against derivational approaches to the causative / anticausative alternation. We focus on the distribution of PPs related to external arguments (agent, causer, instrument, causing event) in passives and anticausatives of English, German and Greek and the set of verbs undergoing the causative/anticausative alternation in these languages. We argue that the crosslinguistic differences in these two domains provide evidence against both causativization and detransitivization analyses of the causative / anticausative alternation. We offer an approach to this alternation which builds on a syntactic decomposition of change of state verbs into a Voice and a CAUS component. Crosslinguistic variation in passives and anticausatives depends on properties of Voice and its combinations with CAUS and various types of roots.
The globalized Western culture of innovation, as propagated by major aid institutions, does not necessarily lead to empowerment or improvement of the well-being of the stakeholders. On the contrary, it often blocks viable indigenous innovation cultures. In African societies and African Diasporas in Latin America, cultures of innovation largely accrue from the informal, not the formal sector. Crucial for their proper understanding is a threefold structural differentiation: between the formal and informal sector, within the informal sector, according to class, gender or religion, and between different transnational social spaces. Different innovation cultures may be complementary, mutually reinforcing, or conflicting, leading in extreme cases even to a ‘clash of cultures’ at the local level. The repercussions of competing, even antagonistic agencies of innovative strategic groups are demonstrated, analyzing the case of the African poor in Benin and the African Diasporas of Brazil and Haiti.
The article provides a critical discussion of the literature on “patrimonialism” and "neopatrimonialism” as far as the use in Development Studies in general or African Studies in particular is concerned. To overcome the catch-all use of the concept the authors present their own definition of “neopatrimonialism” based on Max Weber’s concept of patrimonialism and legal-rational bureaucracy. However, in order to make the concept more useful for comparative empirical research, they argue, it needs a thorough operationalisation (qualitatively and quantitatively) and the creation of possible subtypes which, in combination, might contribute to a theory of neopatrimonial action.
The properties of the outer crust of non-accreting cold neutron stars are studied by using modern nuclear data and theoretical mass tables updating in particular the classic work of Baym, Pethick and Sutherland. Experimental data from the atomic mass table from Audi, Wapstra, and Thibault of 2003 is used and a thorough comparison of many modern theoretical nuclear models, relativistic and non-relativistic ones, is performed for the first time. In addition, the influences of pairing and deformation are investigated. State-of-the-art theoretical nuclear mass tables are compared in order to check their differences concerning the neutron dripline, magic neutron numbers, the equation of state, and the sequence of neutron-rich nuclei up to the dripline in the outer crust of non-accreting cold neutron stars.
We study the effect of neutrino trapping on the phase diagram of dense, locally neutral three-flavor quark matter within the framework of a Nambu--Jona-Lasinio model. In the analysis, dynamically generated quark masses are taken into account self-consistently. The phase diagrams in the plane of temperature and quark chemical potential, as well as in the plane of temperature and lepton-number chemical potential are presented. We show that neutrino trapping favors two-flavor color superconductivity and disfavors the color-flavor-locked phase at intermediate densities of matter. At the same time, the location of the critical line separating the two-flavor color-superconducting phase and the normal phase of quark matter is little affected by the presence of neutrinos. The implications of these results for the evolution of protoneutron stars are briefly discussed. PACS numbers: 12.39.-x 12.38.Aw 26.60.+c
We discuss the phase diagram of moderately dense, locally neutral three-flavor quark matter using the framework of an effective model of quantum chromodynamics with a local interaction. The phase diagrams in the plane of temperature and quark chemical potential as well as in the plane of temperature and lepton-number chemical potential are discussed.
Event-by-event fluctuations of the net baryon number and electric charge in nucleus-nucleus collisions are studied in Pb+Pb at SPS energies within the HSD transport model. We reveal an important role of the fluctuations in the number of target nucleon participants. They strongly influence all measured fluctuations even in the samples of events with rather rigid centrality trigger. This fact can be used to check different scenarios of nucleus-nucleus collisions by measuring the multiplicity fluctuations as a function of collision centrality in fixed kinematical regions of the projectile and target hemispheres. The HSD results for the event-by-event fluctuations of electric charge in central Pb+Pb collisions at 20, 30, 40, 80 and 158 A GeV are in a good agreement with the NA49 experimental data and considerably larger than expected in a quark-gluon plasma. This demonstrate that the distortions of the initial fluctuations by the hadronization phase and, in particular, by the final resonance decays dominate the observable fluctuations.
Based on the UrQMD transport model, the transverse momentum and the rapidity dependence of the Hanbury-Brown-Twiss (HBT) radii R_L, R_O, R_S as well as the cross term R_OL at SPS energies are investigated and compared with the experimental NA49 and CERES data. The rapidity dependence of the R_L, R_O, R_S is weak while the R_OL is significantly increased at large rapidities and small transverse momenta. The HBT "life-time" issue (the phenomenon that the calculated sqrt R_O^2-R_S^2 value is larger than the correspondingly extracted experimental data) is also present at SPS energies.