The obtained results demonstrate high sensitivity of cognitive system to probabilistic properties of language. Negentropy had no effect on reaction time, indicating the absence of competition between related senses of polysemous words. As predicted, for the set of polysemous nouns only the entropy of equivalent Gaussian distribution accounted for significant proportion of variance of processing latencies. Negentropy represents the difference between the entropy of equivalent Gaussian distribution and the differential entropy of probability density function of context vectors, reflecting unrelated meanings of a word (homonymy). Entropy of equivalent Gaussian distribution is derived from covariance matrix, and represents a measure of general variability in multidimensional space, reflecting related senses of a word (polysemy). Based on their method, entropy of equivalent Gaussian distribution and negentropy of probability density function were calculated for the distribution of context vectors of Serbian polysemous nouns. In the second experiment, two-way-analysis of variance revealed significant main effects of number of senses and redundancy.ДЇinally, a method for quantitative description of ambiguous words based on multidimensional distribution of context vectors proposed by Moscoso del Prado MartГn, KostiД‡ and FilipoviД‡ ДђurД‘eviД‡ was validated (submitted). In both experiments, both multilevel, and by-item regression revealed significant effect of either redundancy, or entropy after the contributions of word length, word frequency, familiarity, and number of senses listed by subjects were partialed out. The four groups of words were matched for length in letters, lemma frequency, familiarity and concreteness. In a factorial design, the number of senses was matched across each level of redundancy, and redundancy was matched across each level of number of senses. In the second experiment, groups of polysemous nouns were selected to have either few, or many senses (listed by subjects), and either balanced (low redundancy) or unbalanced (high redundancy) distributions of sense probabilities. In the first experiment, number of senses (listed by subjects in a previously conducted study), and entropy/redundancy of sense probability distribution (based on the sense frequencies collected in a norming study) were continuous predictors in multiple regression analyses. The collected data were used for further investigation of polysemy processing.We conducted two visual lexical decision experiments. Finally, participants rated word familiarity and word concreteness.

In addition, we collected familiarity and concreteness ratings of each dictionary meaning, and each meaning provided by participants. Based on provided meanings, we calculated number of meanings, proportion of each meaning, entropy and redundancy of meaning probability distribution.

In a meaning collection task, participants listed all meanings of a given word they could think of. Ambiguity measures are derived separately for the dictionary meanings, and the meanings provided by native speakers. With this in mind, we made an attempt to demonstrate that two components of entropy have distinct effects on processing time.ДЇirstly, we collected several measures of ambiguity for 150 Serbian polysemous nouns. the redundancy of a probability distribution) and by the number of senses. Value of entropy is affected by two sources of uncertainty: the balance of sense probabilities (i.e. the balance of sense probabilities also affects processing of polysemous words. The aim of the present study is to demonstrate that in addition to number of senses, entropy of sense probability distribution, i.e.

Najbolji Program Za Pravljenje Matrice