Caixeta 2003 studied this experimentally with two groups of Brazilian Portuguese–French bilinguals, one advanced and one intermediate in their knowledge of French. They were tested individually, in French, on a number of tasks by two experimenters, a French monolingual and a French–Portuguese bilingual. Caixeta found that the participants who had an intermediary level of French produced a greater percentage of guest elements than the advanced‐level participants.
Multilinguals
People who know and use three or more languages also find themselves in various language modes (see Dewaele, 2001). For example, trilinguals are in a monolingual mode when the people they are interacting with are monolingual in one of their three languages, or when they share only one language with another bilingual or multilingual. They can be in a bilingual mode if they share two of their interlocutor's languages (e.g., languages B and C) and they feel comfortable bringing one of the languages (e.g., language C) into the base language (language B). If they are with trilinguals with whom they share all their languages, then the mode can be trilingual, with one language being the most active, for some period of time at least. What is true of trilinguals is also true of quadrilinguals. For example, a quadrilingual can be in a language‐B monolingual mode where language B is being used (it is the base language) and languages A, C, and D are deactivated. This same person, in another situation, can be in a quadrilingual mode where, for example, language B is the base language and languages A, C, and D are also active.
Interpreters
To understand how interpreters undertake simultaneous interpretation, we have to call upon the languages involved but also upon their input and output mechanisms (Grosjean, 1997). Interpreters have to be in a bilingual mode where both languages are active. However, one language is not more active than the other as is normally the case in the bilingual mode. Here both the source language (the language being heard) and the target language (the language being spoken) are active to the same extent as both are needed, for perception and production respectively. This is relatively rare in normal bilingual communication. This said, the processing mechanisms differ according to the level of activation. The input mechanisms of both the source and the target language are active. The reason for the activation of the source mechanism is clear but why that of the target language? There are at least three reasons. Interpreters must be able to monitor their overt speech, the clients' occasional use of the target language must be processed, and the cues of fellow interpreters must be heard. As for the output mechanisms, only that of the target language is active; the source language's mechanism is not. The reason here is straightforward: Only one language has to be output—the target language.
Deaf Bilinguals
Like hearing bilinguals, deaf bilinguals find themselves in their everyday lives at various points along the language‐mode continuum (Grosjean, 2010). When they are communicating with monolinguals they restrict themselves to just one language and are therefore in a monolingual mode. They deactivate the other language and remain, as best they can, within the confines of the language being used (for example, a written form of the majority language). At other times, deaf bilinguals find themselves in a bilingual mode, that is with other bilinguals who share to some extent their two languages—sign language and the majority language—and with whom they can mix their languages. They choose a base language—usually a form of sign language (the natural sign language of the community or a signed version of the spoken language). Then, according to various momentary needs, and by means of signing, finger spelling, mouthing, and so forth, they bring in the other language in the form of code switches or borrowings.
Language Mode in Research
Several research issues are related to language mode.
Language Mode as a Confounding Variable
Since language mode is a cognitive phenomenon that has its roots in human interaction, it is present in many research projects, but mostly in a covert way. The consequence is that the data obtained are variable due to the fact that participants are probably situated at various points along the language‐mode continuum. In addition, the data can be ambiguous given the frequent confound between language mode and the variable under study. A few examples are examined below.
It is rare that researchers working on interferences/transfers put their bilingual participants in a strictly monolingual mode when they obtain language samples. This is unfortunate as they invariably obtain other contact phenomena such as borrowings and code switches which may not be of any interest in the study. For example, Marian and Kaushanskaya 2007 examined a database obtained in the study of autobiographical memories in bilinguals in order to observe crosslinguistic transfer and borrowing. The first author, herself also bilingual in Russian and English, interviewed all participants individually, in English in one session, and in Russian in the other. The participants were thus, de facto, in an intermediate language mode (they knew the experimenter was bilingual) and they brought in various types of contact phenomena. The types of phenomena would have been different and the number much less had participants been interviewed by monolinguals of the two languages.
A much researched psycholinguistic issue concerns the presence or absence of language‐selective processing in bilinguals, that is whether bilinguals call on two (or more) languages when listening to, or reading, one language only. Beauvillain and Grainger 1987, for example, found evidence for nonselective lexical access when bilinguals were shown interlexical homographs. The problem, however, is that the bilingual participants in their experiment had to be in a bilingual mode to complete the task: They had to read a context word in one language and then decide whether the next word, always in the other language, was a word or not in that language. It is no surprise, therefore, that a result indicating nonselective processing was obtained. Many other studies which have failed to control for language mode sufficiently well have been carried out since then and there is now a growing myth that processing is nonselective (see, e.g., Dijkstra & van Hell, 2003, and its discussion in Grosjean, 2008). A close examination of the research situations, the methodologies, and the stimuli used in these studies leads one to conclude that most of the time the other language was being activated either by top‐down or by bottom‐up factors. Hence the nonselective processing found in experiments.
Finally, in the bilingual‐language‐development literature, it has been proposed by some that children who acquire two languages simultaneously go through an early fusion stage in which the languages are in fact one system (one lexicon, one grammar, etc.). They then slowly differentiate their languages, by separating first their lexicons and then their grammars. Evidence for this has come from the observation of language mixing in very young bilingual children and from the fact that there is a gradual reduction of mixing as the child grows older. However, according to researchers such as Genesee 1989, many of these children are in a bilingual mode when recorded, that is, the caretakers are usually bilingual themselves and they are probably overheard using both languages, if not actually mixing their languages (see Goodz, 1989). Thus, as with the other studies mentioned above, language mode is a confounding factor that impinges on the results obtained.
Language Mode as a Control Variable
Early attempts to control for language mode used two approaches, both of them inappropriate. The first was to put the participants in a “language set” (called by some a “language mode”) by giving them instructions in one language, getting them to carry out preliminary tasks in that language, occasionally presenting reminders in that language, and so forth. What this does is activate a particular base language (the variable depicted on the vertical axis in Figure 1), but it does not guarantee a particular position on the monolingual‐bilingual‐mode continuum. A second inappropriate approach, which has been used a lot with bilingual children, second language learners, and aphasic or