Acesso Lexical: uma dupla rota para o português brasileiro

Abstract

Identifying which units constitute the expressions of language remains an important issue in contemporary linguistic research. Is the way by which we have access to words influenced by the lexicon structure? Non lexicalist processing models suggest that, at the time of lexical access, words are decomposed into smaller units, i.e. morphemes. The central assumption of these models is that the words are stored in memory in the form of minimal meaningful units (e.g: friend/ly/). Within this research domain, compound words are especially important due to their dual nature: although they have an atomic meaning, they can also have a segmentable structure. Our objective, in lexical decision tests, was to determine whether there was any difference in the way we recognize Brazilian Portuguese (BP) single or compound words. This research may shed some light on how the mind represents these different word categories and if superficial differences on the structure of the languages influence linguistic processing. We used lexical decision tests to ask whether (i) there are differences in mean reaction times and response accuracy between simple and compound words, and (ii) whether these differences correlate with the frequency of use of these words. The results obtained with the lexical decision test I provide evidence that there is a role for decomposition in word recognition in BP, since low and medium frequency compounds yielded significantly lower reaction times than single words. However, for the words with high frequencies in the lexical decision tests I and II, it was the single words that presented shorter response latencies. This kind of evidence is consistent with predictions of full listing models. In order to explain these results we suggest a dual route mechanism of lexical access, in which each word type will be more quickly accessed depending on its frequency and morphological properties.