The number of senses effect in polysemous noun recognition: expanding the database

Abstract

Words with multiple related senses (polysemous words) are recognised faster compared to the words with multiple unrelated meanings (homonymous words) and unambiguous words (Rodd et al., 2002). The measure of ambiguity in polysemous words was the number of senses (NoS), derived from the meanings/senses provided by native speakers, as well as the information theory measures, entropy (sense uncertainty) and redundancy (the balance of meaning probabilities). These measures were significant predictors of reaction time in visual lexical decision task (VLDT) experiments (Filipović Đurđević & Kostić, 2021). In spite of differences, multiple models agree in predicting the observed facilitation. Research in Serbian revealed these effects in noun, adjective, and verb processing (Anđelić, Ilić, Mišić, & Filipović Đurđević, 2021; Filipović Đurđević & Kostić, 2008; 2021; Mišić & Filipović Đurđević, 2021). The aim of this research was to conceptually replicate and further generalise the NoS effect on processing of nouns. Also, the goal was to collect native speakers' intuitions of the senses for the novel set of Serbian nouns and thus expand the existing database (Filipović Đurđević & Kostić, 2016). A novel set of 100 polysemous nouns was selected from the dictionary and then included in the normative study, in which 36 participants were instructed to write all of the senses that they could recall. The senses obtained from the participants were categorised according to the dictionary and the NoS along with the entropy and redundancy of senses was calculated. The same nouns were presented in a visual lexical decision task to a novel group of 87 native speakers. The results indicated that polysemous nouns with higher number of senses were processed faster (β =-.02 , CI = -.03 – -.00, t =-2.78, p = .005), which is in accordance with our hypothesis. The results regarding the information theory measures revealed that the effects of entropy (H) and redundancy (T) indicated a non-significant trend in the predicted direction (H: β =-.00 , CI [-.02 – .01], t =-.597 p = .557, T: β =.01 , CI [-.00 – .03], t =1.66, p = .097). These findings concur with the previous findings from the noun, adjective and verb experiments and the SSD model (Armstrong & Plaut, 2016) and together they converge to the conclusion that the effect of number of senses in the processing of polysemous words facilitates recognition in the visual lexical decision.