Abstract do Paper apresentado no International Symposium on Bilingualism, realizado em Newcastle, Inglaterra, de 9 a 12 de Abril de 1997

Information-Processing in Bilinguals (Portuguese/French) in a Mixed Language Stroop Task







The organization of words and the various kinds of information about them is presumably more complex in the memory of a bilingual person than in the memory of a monolingual person because the words have to be organized not only according to meaning but also acccording to the structure of the language. As well as this, there are far more words to be stored in the brain of a fluent bilingual, perhaps in the region of 50,000 words in each language(Taylor,1990).

In everyday life, a bilingual may be viewed as translating freely from one language to another (the theory of the common store). He or she may be viewed as using one of the languages independently of the other and with the minimum of interference. This is the position taken by the separate store hypothesis. In the case of the third hypothesis, that of overlapping distributed memory, the semantic information of the two languages is stored in a distributed way.

Various authors explore information processing in relation to speed of response. The slower response to verbal stimuli in multilingual subjects can be explained by (a) the less frequent use of one of the languages and (b) interference between competing or similar linguistic systems. The results of these studies support the view that there is a dependency in the bilingual“s storage of information.

Bilinguals, like monolinguals, store words in the memory according to semantic representation. If we accept that there is a common store, comprehension (decoding) and production (encoding) of words in two languages will cause the activation of the same semantic representation in the memory (Mägiste, 1979).

Bilingual semantic memory, compared with that of monolinguals, has more units and patterns of units, to represent additional information related to the two languages.Some of these patterns, however, may be related to only one of the languages while others may be related to both. This will depend on the use, task, type of word and other factors.

For example, in decoding tasks, Rao (1964) and Dornic (1975) concluded that the speed of response was related to the dominance of one of the languages: slower response time to verbal stimuli and poorer performance in verbal reasoning tasks had a direct correlation with the non-dominant language, especially when the task difficulty was increased.

Nevertheless, the results of investigation to date show that, with regard to comprehension, the decoding process in the two languages expressed in terms of speed of response, takes place more quickly than the encoding process. The acquisition of encoding and decoding competences in L2 develop differentially. Mägiste (1979 ) stated that the the decoding process in the second language takes less time than the encoding process and that in naming tasks the response time is longer in groups of subjects aged 6-11 and also in subjects aged 13-18. She also showed the slower response time of bilinguals to be directlly related to the conflict between alternative responses.

Among studies of tasks involving the comprehension and production of verbal stimuli, many authors, starting from the processing and storage of words in two languages, questioned the degree of dependence and relative independence (e.g. Albert & Obler, 1978; Mack, 1986; Altenberg & Cairns, 1983). However, laboratory results are not comparable to everyday behaviour as the tests used in the experiments may not be generalizable to the everyday interactive communication of bilinguals; for example in normal situations the speed of response is not a relevant parameter (Mägiste,1986).

Lambert(1972) questioned the form in which bilingual is capable of "switching off" an entire linguistic system when he or she is functioning with the second system and then activate it a few moments later at the same time disactivating the other system which was being used.

Macnamara (1967) hypothesizes that a bilingual has two parallel systems of "switching on" and "switching off" to explain the fact that he or she can speak a language while understanding what is being said in the other language. Thus the systems of production and perception are partially independent. Unlike the switching-off system, the switching-on system is under the control of arriving stimuli. In other words, bilinguals cannot ignore the language which is being transmitted to them..

The experimental evidence for this fact was provided by the Stroop test (Hamers & Lambert,1972) .

This test has been used with bilinguals since the 50's to investigate whether rapid changes from one language to another were complete.

An experiment was carried out by Hamers & Lambert on bilingual French/ English speakers . When the subjects were reading, the words could be written in either of the languages. For example, when the word "blue" was written in red ink on a card or,alternatively, the word "jaune" was written in blue ink, the results showed that when the colour of ink and the colour word corresponded there was no difficulty. On the other hand, when the two colours did not correspond, the response was slower and the subject more likely to err.

In a bilingual version the task is more complicated given the fact that two languages are involved. The response might be given in one or other of the languages, depending on the stimulus.

Green(1986), in his bilingual model of the production and perception of the spoken language, distinguishes three states of the language system: selected, switched on and switched off. He considers the selection of a language as the main one to be a way of impeding the switching on of the other and that the interference which arises through the mixture of the two languages is caused by conflict between the two languages systems. For example, aphasics who speak more than one language may combine the languages in various ways, producing hybrid words combining morphemes from both languages or a combination of syllables from both languages. This type of behaviour is also found in normal bilinguals, which shows the difference of opinion with regard to dependence or independence of information storage and the necessity for more studies to be carried out in this area. Such studies should control for the previously mentioned variables as well as for such variables as the possible dominance of one of the languages; the greater or lesser representation of each language at the hemisphere level; relative competences in the 2 languages; the context in which acquisition is taking place; sex and age.

In this context we look at the hypothesis that the performance and the speed of response in decoding tasks is lower in bilinguals than in monolinguals and also among bilinguals especially when the tests involve the two languages simultaneously. We set out to confirm this hypothesis using Stroop tests in Portuguese and French and in both languages.

From the three models of information storage in the semantic memory of bilinguals mentioned above, we consider the common store hypothesis to be the one most consonant with our findings.

¾ ¾


Abstract do Poster apresentado no 5th International Congress of the International Society of Applied Psycholinguistics, Porto, de 25 a 27 de Junho de 1997


Verbal Encoding in Portuguese/French Bilinguals: A Word Fluency Task






An important question in research on lexical access in bilinguals is whether the bilingual possesses one or two internal lexicons. Three types of lexical organisation are accepted since the pioneer work of Weinreich (1953): the compound and subordinative systems with a single underlying conceptual system that is shared by the two languages vocabulary and the coordinate system that defends two conceptual systems, one for each language. The main difference between compound and subordinative systems is that in the first one the access of the two languages to the conceptual system is direct, whereas in the latter one the access to the concepts is done through the identical L1 word.

Paradis (1989) speaks about the probability bilinguals have to use various activation and deactivation procedures to keep their languages separate in the monolingual mode and make them interact in the bilingual mode. Bilinguals, when in a bilingual mode, activate both languages. They usually choose a base-language but they can, within the same language of interaction, decide to switch base languages if the situation, topic, interlocutor, function of the interaction, etc, requires it.(Grosjean, 1995). There are two processes they may use: code-switching and borrowing. Code-switching is shifting completely to the other language for a word, a phrase, a sentence, etc. Borrowing is taking a word or short expression from the other language and (normally phonologically or morphologically) adapting it to the base language (Grosjean, 1995).

Poulisse and Bongaerts in a recent work (apud De Groot, 1993) adapted Levelt“s model of communication (1989) for a bilingual speech-production model and proposed that lexical items belonging to different languages may be related to a shared conceptual representation. This is similar to De Groot“s account of bilingual word representation in which translation equivalents such as English "father" and Dutsch "vader" are related to a common conceptual node (De Groot, 1992).

Soares & Grosjean (1984) investigated the lexical access of base-language words and code-switched words through the Phoneme Triggered Lexical Decision Task (Blank, 1980) and have concluded two main aspects: (1) bilinguals accessed real words in English as quickly as English monolinguals: this finding provided evidence for the activation of the other language when the bilingual is in a monolingual mode; (2) bilinguals took longer to access code-switched words in the bilingual mode. Here Soares & Grosjean (1986) in a later paper suggested that a number of factors can account for the delay, as the access strategy, etc.

Obligatory frequent switch interferes with production of associations , but in freer conditions, the probability of changing language after any given response is greater in one language than in another and so subjects could show better performances in one of the languages (Taylor, 1971). The same author refers that for both languages the probability of changing language was low and subjects evidenced within-language clusters - it was easier for them to continue in either language than to switch languages. However, given the freedom to change languages, use of cross-language option did happen. Albert & Obler (1978) affirm that the spontaneous use of language changing implies some measure of interdependence and cross-language clustering.

In another way, the use of translation equivalents is frequent in association tests. The strategy of giving translation equivalents when naming words seems to be common (Rüke-Dravina, 1971; Riegel & Zivian, 1972). Giving translation equivalents is one way in which language switching may work almost naturally in an association task, but if bilinguals are obliged to change languages at certain times, it will slow down or inhibit their association process (Mac-Namara, 1967; Taylor, 1971). These authors concluded too that when subjects are free to alternate between their two languages in chain association, they will tend to produce clusters in each language.

In the present work we pretend to study, in a free fluency task, the eventual interlingual effect that would support the existence of a compound lexicon. Results will be presented and discussed in terms of a quantitative (number of words,ratio words/time) as well as a qualitative analysis of productions in the monolingual and the bilingual groups.Particular attention will be given to the ratio word/time in the preferred versus non-preferred language in the bilingual group.

¾ ¾ ¾


Abstract do Paper apresentado no 5th International Congress of the International Society of Applied Psycholinguistics, Porto, de 25 a 27 de Junho de 1997


Verbal Information Processing in Bilinguals ( Portuguese/French) in a Dichotic Listening Task






Different methods have been used to study verbal processing, namely for decoding. Since dichotic listening is a method suited to throw light onto processes of perception and lateralization for verbal stimuli, it may contribute towards the study of information processing differences between mono and bilingual subjects.

In this study, Portuguese and French monolinguals were given a dichotic listening task in their respective native languages. Bilingual participants performed three different tasks, the ones given to each group of monolinguals and an original one (mixed Portuguese/French) all made specifically for this purpose. In the mixed task, a word in one language was always paired with another word of the other language.

Thus interference between dichotic stimuli also occurred at the level of the language itself.

The results provided no evidence for global differences between monolinguals and bilinguals. However, differences in information processing between monolinguals and bilinguals will be discussed on the basis of the analysis of the type and relative frequency of errors.



* ESEV-Viseu-Portugal

** FPCE-Univ. Porto-Portugal

*** ICBAS-Univ. Porto-Portugal