The words and senses were printed in three random orders in separate booklets, making nine random orders in total. Having in mind the method of collecting polysemy ratings by the participants, three groups of techniques can be distinguished: On the other hand, polysemy represents a systematic process of expanding the semantic field in the language. These characteristics activate nodes on the letters level, which activate inputs on the word level. They started with the model suggested by Masson Masson, ; A big difference in presence of polysemy, to its advantage, could have led to a much larger number of polysemous words in the set of polysemous stimuli, and therefore to obtaining facilitation effect of ambiguity. 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 view of the fact that random variable y has larger number of outcomes, entropy of random variable y is bigger than entropy of random variable x.
Special attention will be given to balance of probabilities of certain senses. Quantifying language phenomena which are manifested at the level of orthography is not a large problem. Although most of the ambiguity measures were significantly correlated with reaction times, correlation coefficients were somewhat larger for the measures based on the raw senses listed by participants. For the given value of redundancy, entropy increases with the increase of number of outcomes, while for the given number of outcomes, entropy decreases with the increase of redundancy. One possibility lowest-threshold hypothesis is that all the logogens of the subject word cumulate information simultaneously, but the logogen corresponding to the most frequent meaning is the first to reach the activation threshold. Reaction time, in milliseconds, and error frequency were dependent variables. The task of the participants was to read the word and rate the extent of the possibility to experience the object denoted by a word using senses, i.
For example, this way, a word “gluma” act had four senses.
The remaining downside of counting the raw senses listed by participants is the possibility that a sense listed by only one participant is not distinct enough to be treated as a separate sense. These models are based on the assumptions that every word is represented as a unique activation pattern in the set of orthographic, phonological and semantic units cf. The role of words-dimensions, or gilm context is most often given to the certain pkodovi of the most frequent words in the language.
Therefore, for each of the measures of number of senses we estimated partial correlation with reaction time, controlling for word length, log lemma frequency and word familiarity.
At the same time, the effect of number of sense was not significant after partialling out the variance explained by the model which included entropy, along with plodovo relevant variables. We will investigate whether, apart from number of senses, balance of sense frequencies influences processing of polysemous words.
morski plodovi ceo film pored
According to this opinion, word meaning activation could come down to the activation of contextual distribution of the given word based on the available co. It is possible to distinguish three classes of models of distributed processing, based on the way they explain faster processing of ambiguous words.
For example, it was shown that out of dictionary entries in plodovk “Wordsmyth dictionary”, with frequency higher than 10 in a million in CELEX lexical base, only 7.
Besides, network was capable of producing approximately correct probability distribution from the corpus on which it was trained, therefore avoiding “blend states”. However, number of dictionary senses that are rated as familiar by participants, is larger than number of dictionary senses listed by participants. Morsi and average number of senses was determined in two ways.
PREPORUČUJEMO ZA DANAS
More than two thirds of the tested words were rated as concrete, while less than a third was rated as abstract. As an alternative source of data about the number of word meanings, they suggest the speakers of the language in question Rubenstein, Garfield and Millikan, ; Gilhooly and Logie, ; Kellas, Ferraro and Simpson, ; Millis moreki Button, ; Ferraro and Kellas, ; Gawlick- Grendell and Woltz, ; Twilley et al.
The main assumption is that processing time will correlate with entropy of word sense probability distribution, i. Single words to which a role of context is assigned — context words are usually the most frequent words in a certain language.
Special attention will be given to balance of probabilities of certain senses. Distributions of collected measures are listed in Table 5. Intuitively observed, the existence of multiple mapping should bring about competition cfo activation patterns, that is mofski and not facilitation ambiguity effect Borowsky and Masson, ; Rodd, Gaskell and Marslen-Wilson, Slightly higher correlation coefficient was recorded between word concreteness ratings, and dominant sense concreteness ratings: Units at the word level are connected to the units at the meaning level by two-way connections.
This finding is in accordance with the assumption that participants are not able to list all of the familiar senses in a short period of time cf. Word concreteness ratings, and average sense concreteness ratings were positively correlated: Since the redundancy of random variable y is lower smaller differences in the probabilities of outcomeentropy of the random variable y is higher than the entropy of random variable z.
For the given value of redundancy, entropy increases with the increase of number of outcomes, while for the given number of outcomes, entropy decreases with the increase of redundancy.
If the sense is partially familiar, i.
The task was the same as in the rating the familiarity of the senses listed in the dictionary. In the second experiment, groups of polysemous nouns were selected plodvi have either few, or many senses listed by subjectsand either balanced low redundancy or unbalanced high redundancy distributions of sense probabilities.
Average number of senses that are familiar to participants is 5. In addition to that, under assumption that everything else is balanced, logogen in big cluster will reach threshold level sooner than the logogen in small cluster.
The Polysemy Effect in Processing of Serbian Nouns | Dusica Filipovic Durdevic –
This principle led to a “higher resolution”, and consequently a larger number of senses. Partial effects of predictors in the model based on the senses with mod familiarity judgment greater than 1.
Lexicographers adopted standard according to which certain meanings of homonyms are treated as different entries, while all the meanings of polysemous words plovovi given under one entry cf. With this in mind, we made an attempt to demonstrate that two components of entropy have distinct effects on processing time.
Recent researches indicated that the distinction between the two types of ambiguity has a status of cognitive relevancy Rodd, Gaskell and Marslen-Wilson, ; Klepousniotou, Word a is characterized by unbalanced sense probabilities, i. Finally, participants rated word familiarity and word concreteness.
Apart from the fact that words with higher number of meanings were processed faster, within the group of polysemous words, processing time was shorter for morsli words with related meanings.