The present work focuses on metalanguage comments in Slovak and Czech texts. The comments occur together with contact-induced phenomena employed by both languages and can be identified in text fragments, such as Áno, taký som aj ostanem, lebo vraj starého psa novým kouskúm nenaučíš - ako hovoria bratia Česi / Nebyl jsem schopen jídla, ale lemtal jsem vodu "ani ťava", jak říkají bratia Slováci. The given issue is presented in the context of research focusing on the non-theoretic (folk, naive) metalinguistic reflection. The usage of metalanguage comments, accompanying deliberately contact-induced phenomena, is a testament to the existence of the Czech-Slovak and Slovak-Czech language contact and it clearly shows how Slovak and Czech language users intentionally employ words, multi-word units, idioms and other linguistic devices from the other language to fulfill their communicative intention. By using the contact-induced phenomena from the closely related language the speaker can achieve various communicativ e goals and effects including actualization and refreshing of his or her utterance, being more expressive and filling in subjectively perceived intro-linguistic lacuna. The starting point of the analysis of the metalanguage comments were the data from the Slovak and Czech corpus. A range of...

Diese Arbeit liefert einen Beitrag zur Erforschung der Negation von <i>bar</i>-Adjektiven. <i>Bar</i>-Adjektive k&ouml;nnen auf zwei verschiedene Weisen negiert werden: durch Pr&auml;figierung mit <i>un</i> wie in <i>undenkbar</i> und durch vorangestelltes <i>nicht</i> wie in <i>nicht essbar</i>. Die Negation mit <i>nicht</i> ist immer m&ouml;glich. Die Negation mit <i>un</i> unterliegt hingegen bestimmten Restriktionen. Diese Restriktionen zu erforschen ist Gegenstand dieser Arbeit. Es wird daf&uuml;r ein integrierter Ansatz aus Korpusanalyse und Data Mining benutzt.

Die Arbeit setzt sich mit verschiedenen Aspekten der Nutzung korpusbasierter Statistiken in quantitativen typologischen Untersuchungen auseinander. Die einzelnen Abschnitte der Arbeit können als Teile einer sprachunabhängigen Prozesskette angesehen werden, die somit umfassende Untersuchungen zu den verschiedenen Sprachen der Welt erlaubt. Es werden dabei die Schritte von der automatisierten Erstellung der grundlegenden Ressourcen über die mathematisch fundierten Methoden bis hin zum fertigen Resultat der verschiedenen typologischen Analysen betrachtet. Hauptaugenmerk der Untersuchungen liegt zunächst auf den Textkorpora, die der Analyse zugrundeliegen, insbesondere auf ihrer Beschaffung und Verarbeitung unter technischen Gesichtspunkten. Es schließen sich Abhandlungen zur Nutzung der Korpora im Gebiet des lexikalischen Sprachvergleich an, wobei eine Quantifizierung sprachlicher Beziehungen mit empirischen Mitteln erreicht wird. Darüber hinaus werden die Korpora als Basis für automatisierte Messungen sprachlicher Parameter verwendet. Zum einen werden derartige messbare Eigenschaften vorgestellt, zum anderen werden sie hinsichtlich ihrer Nutzbarkeit für sprachtypologische Untersuchungen systematisch betrachtet. Abschließend werden Beziehungen dieser Messungen untereinander und zu sprachtypologischen Parametern untersucht. Dabei werden quantitative Verfahren eingesetzt.

Die Arbeit beschäftigt sich mit syntaktischen Strukturen des gesprochenen Gegenwartsdeutschen. Ausgehend von einem Korpus von spontaner Sprache, u.a. aus Talkshows und Reality-TV-Sendungen, wird Videomaterial aus Lehrwerken für Deutsch als Fremdsprache auf gesprochensprachliche syntaktische Strukturen hin analysiert. Es soll der Frage nachgegangen werden, inwieweit Unterrichtsmaterialien die Sprachwirklichkeit angemessen widerspiegeln.

The graduation thesis is dedicated to the binomials in the Czech and Russian languages. The author paid her attention to the substantive binomials, to the determination of their semantic categories and syntactic functions and to the comparison of the results of both languages. The source for the analysis is binomials, which were found in the Czech national corpus and the Russian national corpus.

1 Abstract This paper aims at an empirical corpus driven analysis of a closed set of German comparative idioms with adjectives. Regularities in their variations are examined from several points of view. The contrastive German-Czech perspective contributes to a more objective understanding of the phenomenon itself. Corpus research results are integrated into the context of practical lexicography. The strategy of equivalence in a bilingual dictionary is discussed and reevaluated both theoretically and practically. Keywords: phraseology, comparisons, variability, lexicography, dictionary, corpus

in English Choice of Mood in Spanish Grammatical Constructions Expressing Potentiality (with Reference to Czech and Italian) Key words: Mood, Subjunctive, Potentiality, Corpus, Spanish, Italian, Czech The objective of this paper is to study the systemic possibilities of alternation between the subjunctive and the indicative in Spanish grammatical constructions expressing potentiality, the semantical differences this alternation brings and also the frequency these two modes appear in respective constructions. In order to achieve this objective I studied not only the specialized literature in the topic but also the language corpora. Later I compared the situation in Spanish with Italian and Czech as well. The results of my study brought me to the conclusion that it is necessary to study each expression of potentiality separately because the verbal modes do not appear equally often after these expressions. On the other hand, the use of the conjunctive in Italian is far more formalized. Czech does not posses any systemic tool that could work for the Spanish subjunctive in every context. However, its translation as the conditional or the future indicative is possible.

v anglickém jazyce: The topic of this dissertation is the past participle and its use in the contemporary Spanish language system. The character of the thesis is both theoretical and empirical. The theoretical part delineates the issue of the Spanish past participle in the context of non-finite verbs and explores the general matters of the past participle connected with its morphological and syntactic- semantical features. The empirical part consists of two case studies of selected past participle constructions, dar por + past participle a una vez + past participle, based on language research within the parallel corpus InterCorp and single language corpus CREA. The aim of the research is to identify the particular past participles forming the core of the above-mentioned constructions and, by way of contrastive analysis, describe the possibilities of expressing these constructions in the Czech language. The individual findings are quantified and documented by examples.

This thesis deals with functional equivalents of Italian adjectives with the suffix -bile in present-day Italian based on the Czech-Italian parallel linguistic corpus InterCorp. In the opening theoretical part the main objective of the thesis is introduced - a comparison of the Czech suffix -telný with the Italian suffix -bile. In the next chapters the thesis focuses on corpus material characteristics - on frequency list - from qualitative and quantitative aspects and on Czech equivalents' typology. The last part concentrates on statistical evaluation of Czech translation solutions, on the individual percentage representation of Czech equivalents' variants and on general trends while translating these adjectives with the suffix -bile into Czech.

The Groningen Meaning Bank (GMB) project develops a corpus with rich syntactic and semantic annotations. Annotations in GMB are generated semi-automatically and stem from two sources: (i) Initial annotations from a set of standard NLP tools, (ii) Corrections/refinements by human annotators. For example, on the part-of-speech level of annotation there are currently 18,000 of those corrections, so called Bits of Wisdom (BOWs). For applying this information to boost the NLP processing we experiment how to use the BOWs in retraining the part-of-speech tagger and found that it can be improved to correct up to 70% of identified errors within held-out data. Moreover an improved tagger helps to raise the performance of the parser. Preferring sentences with a high rate of verified tags in retraining has proven to be the most reliable way. With a simulated active learning experiment using Query-by-Uncertainty (QBU) and Query-by- Committee (QBC) we proved that selectively sampling sentences for retraining yields better results with less data needed than random selection. In an additional pilot study we found that a standard maximum-entropy part-of-speech tagger can be augmented so that it uses already known tags to enhance its tagging decisions on an entire sequence without retraining a new model first. Powered by...