Ten dwuczęściowy artykuł konsoliduje fakty dotyczące polisemii z zakresu psycholingwistyki oraz leksykografii teoretycznej i obliczeniowej oraz prezentuje wyniki badań ilościowych i jakościowych przeprowadzonych na dużym zbiorze danych polisemicznych w języku polskim, na podstawie których proponujemy uszczegółowioną klasyfikację polisemii oraz opracowujemy poszerzone spektrum podobieństwa sensów słów wieloznacznych. W pierwszej części publikacji omówiono psycholingwistyczne modele reprezentacji polisemii w mentalnym leksykonie oraz wprowadzono podstawowe spektrum podobieństwa sensów, zaproponowane w literaturze przedmiotu. Przedstawiono też metodologię oraz wyniki badań ilościowych przeprowadzonych dla par sensów wyrazów polisemicznych losowo wybranych ze Słowosieci i pozyskanych dzięki wsparciu Centrum Technologii Językowych CLARIN-PL (polskiej sekcji europejskiej infrastruktury badawczej CLARIN ERIC). Wyniki naszego badania pokazują, że najczęściej reprezentowaną polisemią są następujące typy: polisemia gniazdowa, polisemia przez metaforę i polisemia przez metonimię. W drugiej części publikacji poszerzamy spektrum podobieństwa sensów i uszczegóławiamy klasyfikację polisemii na podstawie badania jakościowego. Przedstawiamy także nowe obserwacje na temat różnych typów polisemii wchodzących w skład tego spektrum, ze szczególnym uwzględnieniem polisemii gniazdowej. This two-part paper bridges insights from psycholinguistics and from theoretical and computational lexicography to develop a fine-grained classification of polysemy organized along a wider spectrum of sense remoteness of ambiguous words in Polish based on the investigation of a large collection of linguistic data.1 In the first part, we equip readers with background knowledge on different psycholinguistic views on polysemy and we introduce the basic spectrum of sense remoteness proposed in earlier literature. We also present the methodology of our research and we report the results of our quantitative study based on a large sample of sense pairs randomly extracted from plWordNet This two-part paper bridges insights from psycholinguistics and from theoretical and computational lexicography to develop a fine-grained classification of polysemy organized along a wider spectrum of sense remoteness of ambiguous words in Polish based on the investigation of a large collection of linguistic data.1 In the first part, we equip readers with background knowledge on different psycholinguistic views on polysemy and we introduce the basic spectrum of sense remoteness proposed in earlier literature. We also present the methodology of our research and we report the results of our quantitative study based on a large sample of sense pairs randomly extracted from plWordNet (Słowosieć) thanks to the resources received from the CLARIN-PL Language Technology Center (the Polish section of the European research infrastructure CLARIN ERIC). We show that the most widely represented polysemy types are nested polysemy, polysemy by metaphor and polysemy by metonymy. The second part proposes an extended spectrum of sense remoteness and presents insights on different types of polysemy included in this spectrum with a special attention paid to nested polysemy.

Ten dwuczęściowy artykułkonsoliduje fakty dotyczące polisemii z zakresu psycholingwistyki oraz leksykografii teoretycznej i obliczeniowej oraz prezentuje wyniki badańilościowych i jakościowych przeprowadzonych na dużym zbiorze danych polisemicznych w języku polskim, na podstawie których proponujemy uszczegółowionąklasyfikacjępolisemii oraz opracowujemy poszerzone spektrum podobieństwa sensów słów wieloznacznych. W drugiej części publikacji wykazujemy, że polisemia nie jest zjawiskiem stabilnym, a relacje między sensami mogąbyćoceniane odmiennie przez różnych mówców w zależności od ich postrzegania świata, wiedzy o życiu i skojarzeń. Wskazujemy na kilka parametrów zróżnicowania, które mogąwpłynąćna ocenępodobieństwa sensów i reprezentacjęw mentalnym leksykonie polisemii przez metonimięi metaforę. ABSTRACT The Spectrum of Sense Remoteness in Polysemy: Bridging Computational and Theoretical Lexicography with Psycholinguistics (Part 2) This two-part paper bridges insights from psycholinguistics and from theoretical and computational lexicography to develop a fine-grained classification of polysemy organized along a wider spectrum of sense remoteness of ambiguous words in Polish based on the investigation of a large collection of linguistic data. In the second part, we show that polysemy is not a stable phenomenon and relations between senses may differ across language users. For instance, our fifty-fifty class or borderline cases may be represented differently by different language users depending on their perception of the world, world knowledge, associations. We point to some parameters of variation in the class of polysemy by metonymy and polysemy by metaphor which may affect their sense remoteness and consequently also the way they are represented in the mental lexicon.

As compounding in English is a very productive process, a number of constructions emerged which can have multiple meanings. This paper aims to investigate the extent to which promiscuous compound nouns can be problematic for L2 English learners. The central questions which are addressed here are as follows: i What strategies do Polish speakers use to overcome the production and comprehension difficulties when processing English compounds? ii To what extent can Polish learners acquire the ability to produce and comprehend English compounds effectively? We compare the production and perception of stress patterns in various types of English compounds by L1 and L2 speakers of English on the basis of a reading protocol and a lexical decision experiment. One of the striking observations is that L1s are not always more accurate than L2s. This concerns stress-recognition in Adjective–Noun A–N phrases which are juxtaposed with A–N compounds. Elsewhere, natives lead in terms of stress-placement. On the other hand, the results obtained by L1s show no difference in stress-production and stress-recognition in Noun–Noun N–N attributive compounds, whereas L2s are more accurate in understanding than producing this type of compounds. A similar situation, in the case of L1s, concerns A–N opaque and N–N argument-head compounds but it is less prominent. As for L2s, A–N opaque compounds elicit relatively comparable results, but N–N argument-head compounds are responded to more accurately in the production experiment. A–N phrases, on the other hand, elicit a huge disproportion of accuracy between production and perception among L1s: although almost flawless in production, two-thirds of responses are inaccurate in perception. As for non-natives, they are more or less equally accurate in the case of A–N opaque compounds less than half correct renditions, but they adopt a strong tendency to produce and perceive fore-stress patterns across the board. Our experiments showed certain regularities in both accuracy and inaccuracy of stress placement by L2s. These regularities can be applied in methodology as to the order of teaching different types of compounds and phrases to Polish learners of English.

The aim of this paper is to present the benefits and risks of implementing corpus linguistics for discourse analysis. The authors describe the origins and development of Natural Language Processing (NLP) tools in a historical perspective and provide examples of their application in social sciences, particularly in the methodology of Social Communication and Media Sciences. Fragmentary findings of studies carried out at the Institute of Journalism, Media and Social Communication at the Jagiellonian University in collaboration with the CLARIN-PL consortium illustrate a practical approach to the topic. The article presents the application of NLP tools in the analysis of the corpus of parliamentary discourse from 1989-2019 in terms of determinants for the institutionalization of civic dialogue in Poland and also in the comparative analysis of ulticulturalism in the city council discourse and media discourse in Krakow between 2014–2018 (7 th term of the Krakow City Council). The authors indicate in which phase and at which stage of communication research the use of computational linguistics can support the conclusion.

No abstract