Search

In order to describe changes in the lexical system, M. Dokulil used the metaphor of a deep body of water. While down in the deep, nothing is happening, on the surface the water never stops to move, flow, circulate. In Dokulil’s metaphor, the surface layer symbolises those lexical items closely related to social changes: new words and old words whose meaning has changed, or old words that have become elements of new words. The analysis of neologisms occurring during relevant social changes provides a valuable insight into the users’ attitudes towards contemporary events. This paper deals with neologisms and neosemanticism concerning the migration processes that have taken place in Europe from 2011 to 2018. The neologisms are extracted from Neomat, the database for Czech neologisms. An analysis based on word formation and semantics, supported by papers in Neologizmy v dnešní češtině, is followed by an analysis from a sociolinguistic perspective, relying on the work of V. Dvořáčková, E. Mleziva and the theoretical frame given in the monography of V. Muhvić-Dimanovski. The focus of this article will be on those words which denote concepts related to the great migrations of our time. The productive formation by employing prefixes meaning pro or contra indicates a social urge to express polar attitudes on the subject. Due to the relevance of migration processes in a global context, lexical and semantical borrowing constitute the most frequent word formation types. The meaning of some stable lexical items has changed due to an increased frequency of use and this is carried out in two opposite ways: by blurring lines between closely related items and by filling empty spaces in order to gain higher levels of precision.

Named Entity Recognition (NER) focuses on finding named entities in text and classifying them into one of the entity types. Modern state-of-the-art NER approaches avoid using hand-crafted features and rely on feature-inferring neural network systems based on word embeddings. The paper analyzes the impact of different aspects related to word embeddings on the process and results of the named entity recognition task in Czech, which has not been investigated so far. Various aspects of word vectors preparation were experimentally examined to draw useful conclusions. The suitable settings in different steps were determined, including the used corpus, number of word vectors dimensions, used text preprocessing techniques, context window size, number of training epochs, and word vectors inferring algorithms and their specific parameters. The paper demonstrates that focusing on the process of word vectors preparation can bring a significant improvement for NER in Czech even without using additional language independent and dependent resources. info:eu-repo/semantics/openAccess

The paper discusses the typology of metalinguistic comments in the speech of representatives of the Slavic minority communities residing in the Republic of Srpska (Bosnia and Herzegovina). Using a corpus of examples from the Ukrainian, Polish, Slovene and Czech languages gathered during fieldwork carried out between 2016 and 2019, the authors show that the informants provide both contact elements and intralinguistic units with metalinguistic comments. Such comments reflect the complex ethnic and confessional composition of the area where the field research was conducted, providing information about the linguistic situation in the communities under scrutiny.

The Czech language is in a very particular sociolinguistic situation in comparison with other slavonic languages. In this paper I explain the discussion about the existence of diglossia in Czech language, due to the fact that this language has two different varieties for different contexts, in order to show the different positions in Bohemistics during the last decades. This work analyses some grammatical and lexical phenomena from both varieties

This paper describes the training process of the first Czech monolingual language representation models based on BERT and ALBERT architectures. We pre-train our models on more than 340K of sentences, which is 50 times more than multilingual models that include Czech data. We outperform the multilingual models on 9 out of 11 datasets. In addition, we establish the new state-of-the-art results on nine datasets. At the end, we discuss properties of monolingual and multilingual models based upon our results. We publish all the pre-trained and fine-tuned models freely for the research community., Comment: 13 pages