Radek Čech

Full Valency. Verb Valency without Distinguishing Complements and Adjuncts

Abstract The aim of the article is to introduce a new approach to verb valency analysis. This approach – full valency – observes properties of verbs which occur solely in actual language usage. The term “full valency” means that all arguments, without distinguishing complements (obligatory arguments governed by the verb) and adjuncts (optional arguments directly dependent on the predicate verb), are taken into account. Because of an expectation that full valency reflects some mechanism which governs verb behaviour in a language, hypotheses concerning (1) the distribution of full valency frames, (2) the relationship between the number of valency frames and the frequency of the verb, and (3) the relationship between the number of valency frames and verb length were tested empirically. To test the hypotheses, a Czech syntactically annotated corpus – the Prague Dependency Treebank – was used.

Testing the Thematic Concentration of Text

Abstract The aim of the article is to evaluate and address the limits of an existing approach to the analysis of the thematic concentration of text. To overcome these limits, the article proposes and applies both a modification of the measurement of thematic concentration – known as secondary thematic concentration and proportional thematic concentration – and methods for their statistical testing. The results show that the modification, as well as the application of the proposed tests, enhances the possibilities for analysing the thematic characteristics of text. The article uses 20 Slovak texts of the same genre written by one author.

Some Geometric Properties of Slovak Poetry

Abstract The article is focused on the analysis of the frequency structure of texts. Specifically, a geometric characterization of the rank-frequency sequence, which is determined by relationships among the highest word frequency, a number of particular word forms in a text, and the so-called h-point, is analysed. We observe that the geometric characterization of the rank-frequency sequence correlates strongly with the other frequency-based text property called the golden section of text which seems to be a result of a universal mechanism controlling ones language usage. These findings can be viewed as a result of self-organization of text.

On the interpretation of complex network analysis of language: comment on "approaching human language with complex networks" by Jin Cong, Haitao Liu.

After a rapid and successful development of the theory of complex networks at the turn of the millennium [1,2], attempts to apply this theory to a language analysis emerged immediately [3,4]. The first results seemed to bring new insights to the functioning of language. Moreover, some authors assumed that this approach can even solve some fundamental problems concerning language evolution [5,6]. However, after a decade of the application of complex network theory to language analysis, the initial expectations have not been fulfilled, in my opinion, and the need for a deeper, linguistically based explanation of observed properties has been still more obvious. Cong and Liu’s review [7] can be seen as a successful attempt to clarify the main aspects of this kind of research from the linguistics point of view. However, I see two problematic aspects in their study relating to the nature of the character of explanation. First, the authors try to interpret Saussure’s conception of the language system as an approach which is consistent with the modern definition of system and, consequently, with the complex network approach. However, Saussure represents a radically different view of the system: according to him, there is an ideal, abstract system of language units and relationships among them (he calls it “langue”), on the one hand, and a manifestation of this ideal system in the reality (he calls it “parole”), on the other [8]. Therefore, Saussure is considered to be a successor of the Platonic tradition, in a more general view. To model language by the complex network, however, does not mean to model language system in Saussurean sense; the complex network is a model of empirically observable characteristics of language, not a model of any ideal, empirically non-observable system. This confusion can lead to a misinterpretation of the complex network approach, especially among linguists; the majority of them have been reluctant to leave the dichotomic, i.e. langue–parole, view of language, cf. a discussion on this topic, e.g. [9–13]. In other words, to present complex network approach to language analysis in the Saussurean sense is not only inappropriate from the theoretical point of view, but also with regard to the practical consequences, if the current state of linguistics is considered. Second, the authors interpret complex networks as a model of human cognition (cf. “The four linguistic networks constructed in the present study (. . .) can be treated as models of how the knowledge of different levels (as sub-systems) of modern Chinese is represented in the human mind.” [7, p. 13]). A more skeptical view would be appropriate, in my opinion, because of the lack of empirical evidence; without this evidence, this interpretation can rightfully be considered a speculation. Further, attempts to connect complex network properties to cognitive characteristics are

Descriptivity in Slovak Lyrics

Abstract The aim of the article is to introduce the measurement of “activity” and “descriptivity” of a text based on the proportions of adjectives and verbs. Functions based on interaction of forces, tests for significance of a property and for comparison of two texts are introduced. The methods are applied to the poetry of the Slovak poetess Eva Bachletová.

Structural conservatism and innovation in texts

In some previous works (Popescu, Ma utek, Altmann 2010; Popescu, ech, Altmann 2011) it has been stated that the lambda-indicator expressing an aspect of lexical structuring of texts – including vocabulary richness – independently of text size has further different aspects whose investigation must be performed stepwise. Needless to say, not all aspects are known – a quite normal situation in any scientific discipline – but one of them will be scrutinized separately in the present article. It has been stated that if one analyzes different works of one and the same author, the number of lambda-similarities among texts abide by the beta function. One can consider it a beta distribution, too, but the difference between the two approaches consists only in the normalization. In order to illustrate the procedure we consider two texts and state the rank-frequency distributions of word forms. For each separately we compute the lambda indicator defined as

Pitfalls of the Transitivity Hypothesis: Transitivity in Conversation and Written Language in Czech

Abstract The aim of the article is to test empirically predictions formulated in the Transitivity Hypothesis framework. Methodological problems of the original approach are discussed and some solutions are offered. For the testing of the hypotheses two corpora of Czech were used (Prague Spoken Corpus and Prague Dependency Treebank). The results question both the predicted impact of the language form on transitivity and, more importantly, the concept of the Transitivity Hypothesis in general.

Quantitative Analysis of Syntactic Dependency in Czech

The article presents a quantitative analysis of some syntactic dependency properties in Czech. A dependency frame is introduced as a linguistic unit and its characteristics are investigated. In particular, a ranked frequencies of dependency frames are observed and modelled and a relationship between particular syntactic functions and the number of dependency frames is examined. For the analysis, the Czech Universal Dependency Treebank is used.

The Development of Context Specificity of Lemma. A Word Embeddings Approach

ABSTRACT The study deals with the application of the neural networks in the linguistic research of word semantics. A new method of measuring context specificity of lemma based on Word Embeddings Word2vec technique is proposed in the first part of the article. Then the method is illustrated in the analysis of the Czech political discourse in the second part. The research is based on the corpus of the Czech journalism consisting of more than 3 billion tokens. The results show that the proposed method is applicable for detecting the semantic development of a lemma and it could have a great potential for linguistic studies if one can apply it with comprehensive explanations.

Syntactic Complex Networks and Their Applications

We present a review of the development and the state of the art of syntactic complex network analysis. Some characteristics of such networks and problems connected with their construction are mentioned. Relations between global network indicators and specific language properties are discussed. Applications of syntactic networks (language acquisition, language typology) are described.