Vector based semantic analysis reveals absence of competition among related senses

Abstract

Previous research demonstrated that processing time was facilitated by number of related word senses (polysemy) and inhibited by number of unrelated word meanings (homonymy). The starting point of this research were the findings described by Moscoso del Prado Mart?n and colleagues, who offered a unique account of processing of two forms of lexical ambiguity. By applying the techniques they proposed, for the set of strictly polysemous Serbian nouns we calculated ambiguity measures they introduced. Based on the covariance matrix of the context vectors, we derived entropy of equivalent Gaussian distribution, and based on the context vectors probability density function, we derived differential entropy. Negentropy was calculated as the difference between the two. Based on interpretation that entropy of equivalent Gaussian mirrors sense cooperation, or polysemy, while negentropy mirrors meaning competition, or homonymy, we predicted that in the set of strictly polysemous nouns, negentropy effect would disappear. In accordance with our predictions, entropy of equivalent Gaussian distribution accounted for significant proportion of processing latencies variance. Negentropy did not affect reaction time. This finding is in accordance with the hypothesis that entropy of equivalent Gaussian distribution, as a measure of general width of activation in semantic space, reflects polysemy, that is, the existence of related senses. Therefore, polysemy advantage could be the result of the wide-spread activation in semantic space and reduced competition among overlapping Gaussians.Ranija istrazivanja pokazala su da je porast u broju povezanih znacenja (polisemija) pracen kracim vremenom obrade reci, dok je porast u broju nepovezanih znacenja (homonimija) pracen duzim vremenom obrade reci. U ovom istrazivanju evaluirali smo metod za kvantitativno opisivanje viseznacnih reci na osnovu visedimenzionalne distribucije kontekstualnih vektora kojeg su predlozili Moscoso del Prado Mart?n i saradnici. Na osnovu distribucije kontekstualnih vektora pojedinacnih pojavljivanja polisemicnih reci srpskog jezika izracunata je entropija ekvivalentne Gausove distribucije i negentropija funkcije gustine verovatnoce. Entropija ekvivalentne Gausove distribucije, koja je izvedena iz matrice kovarijanse, predstavlja meru varijabilnosti u visedimenzionalnom prostoru i odslikava postojanje povezanih znacenja reci (polisemija). Negentropija predstavlja razliku izmedju entropije ekvivalentne Gausove distribucije i diferencijalne entropije funkcije gustine verovatnoce kontekstualnih vektora i odslikava postojanje nepovezanih znacenja (homonimija). U skladu sa predvidjanjima, na skupu polisemicnih imenica srpskog jezika, zabelezen je samo efekat entropije ekvivalentne Gausove distribucije. Negentropija nije imala uticaj na vreme reakcije. Ovaj nalaz je u skladu sa pretpostavkom da entropija ekvivalentne Gausove distribucije, kao mera sirine aktivacije u semantickom prostoru, odslikava polisemiju, odnosno prisustvo povezanih znacenja reci. Stoga se prednost u obradi polisemicnih reci moze objasniti sirokom aktivacijom u semantickom prostoru i smanjenom kompeticijom medju Gausovim distribucijama koje se preklapaju u velikoj meri.