The bilingual individual

FRANCOIS GROSJEAN

This article presents a general overview of the adult bilingual individual. First,
the bilingual is defined and discussed in terms of the complementary principle,
i.e. the fact that bilinguals acquire and use their languages for different pur-
poses, in different domains of life, with different people. Next, the various
language modes bilinguals find themselves in during their everyday interac-
tions are examined. These range from the monolingual mode when they are
communicating with monolinguals {and they have to deactivate all but one
language) to the bilingual rriode when they are interacting with other bilinguals
who share their two (or more) languages and with whom they can mix lan-
guages if they so wish (i.e. code-switch and borrow). The article ends with a
rapid survey of the psycholinguistics of bilingualism and, in particular, of how
bilinguals access their lexicon when perceiving mixed speech. The regular
bilingual Is compared to the interpreter bilingual whenever possible.

A number of myths surround bilingualism and bilinguals. For many people,

bilingualism is a rare phenomenon that is found only in bilingual or multilin-
gual countries such as Canada, Switzerland and Belgium. Supposedly bilin-
guals grow up speaking two (or more) languages, they have equal speaking
and writing fluency in their languages, they have accentless speech, and they
are excellent translators and interpreters. The reality is in fact quite different.
There are more bilinguals in the world than monolinguals, they are found in
every country of the world, in all classes of society and in all age groups, and
they usually acquire their languages at various times during their lives and not
just in early childhood. In addition, they are rarely equally fluent in all their
languages (many cannot read or write one of their languages), they often have
an accent in the language(s) they acquired late, and few bilinguals are
proficient translators and interpreters.

interpreting 2:1-2 ( 1997), 163- 187. DOl I0. 1075/intp.2.1-2.07gro

ISSN 1384-6647 I E-I SS N 1569-982X © John Benjamins Publishing Company
164 FRAN<;OIS GROSJEAN

Several aspects of the bilingual individual will be discussed in this paper.

The adult bilingual, as opposed to the child bilingual, will be the object of
study as will the stable bilingual, that is the person who is no longer in the
process of acquiring a second or third language. First the bilingual person will
be described in terms of language knowledge and language use. Then the
language modes bilinguals find themselves in when interacting with monolin-
guals and with other bilinguals in their everyday life will be examined.
Finally, a rapid survey of the psycholinguistics of bilingualism and in particu-
lar of how bilinguals undertake lexical access when perceiving mixed speech
will be presented. Throughout the paper an attempt will be made to compare
the "regular bilingual" with the distinct bilingual that the interpreter has
become (the latter will be called the "interpreter bilingual"). Student inter-
preters start off as regular bilinguals and slowly, through extensive training
and practice, become interpreter bilinguals. Emphasis here will be put on
regular bilinguals, not only because they have been studied in more depth but
also because they share many more traits with interpreter bilinguals than one
might think at first.

1. Describing the bilingual

1.1 Definition

Although a few researchers have defined bilinguals as those who have native-
like control of two or more languages (Bloomfield, 1933; Thiery, 1978), most
others agree that this position is not realistic. If one were to count as bilingual
only those people who pass as monolinguals in each of their languages, one
would be left with no label for the vast majority of people who use two or more
languages regularly but who do not have native-like fluency in each. This has
led researchers to propose other definitions of bilingualism, such as: the ability
to produce meaningful utterances in two (or more) languages, the command of
at least one language skill (reading, writing, speaking, listening) in another
language, the alternate use of several languages, etc. (Beatens-Beardsmore,
1986; Hakuta, 1986; Haugen, 1969; Romaine, 1995). In what follows, bilin-
guals will be defined as those people who use two (or more) languages (or
dialects) in their everyday lives. (For a discussion of notions such as domi-
nance, fluency, balance, etc., as applied to bilinguals, see Grosjean, 1982,
1985b).
 THE BILINGUAL INDIVIDUAL 165

Bilinguals differ from one another on a number of variables. In addition

to such factors as age, sex, socio-economic and educational status, one finds
the following: number and type of languages known and global competence
in these languages; language history (when and how the languages were
acquired and used); language stability (are one or several languages still being
acquired (or restructured) or has a certain language stability been reached)?;
competence in each of the four skill s (reading, writing, speaking, listening) in
each language; function of the languages (which languages are used, when,
with whom and for what reason?); language modes (how often and for how
long do the subjects find themselves at the various points along the language
mode continuum?); amount of code-switching and borrowing normally done,
etc. Despite the great diversity that exists between bilinguals, they all have
one thing in common: they lead their lives with two (or more) languages.
(Bilinguals who no longer use their different languages but who have retained
knowledge of them are termed "dormant bilinguals.") It is clear that the many
variables listed above need to be taken into account when assessing future
interpreters, and they have to be kept in mind during actual interpreter
training. They can explain in part the great variability found in the bilingual-
ism of student interpreters and, to some extent at least, they can account for
differential behavior when interpreters are working under various conditions
(from optimal to less than optimal).

1.2 The complementary principle

There are many reasons that bring languages into contact and hence foster
bilingualism: migrations of various kinds (economic, educational, political,
religious), nationalism and federalism, education and culture, trade and com-
merce, intermarriage, etc. These factors create various linguistic needs in
people who are in contact with two or more languages and who develop
competencies in their languages to the extent required by these needs. In
contact situations it is rare that all facets of life require the same language
(people would not be bilingual if that were so) or that they always demand
two languages (language A and B at work, at home, with friends, etc.). This
leads to what is called the complementary principle:
Bilinguals usually acquire and use their languages for different purposes, in
different domains of life, with different people. Different aspects of life
require different languages.
166 FRANc;OIS GROSJEAN

It is precisely because the needs and uses of the languages are usually quite
different that bilinguals rarely develop equal and total fluency in their lan-
guages. The level of fluency attained in a language (more precisely, in a
language skill) will depend on the need for that language and will be domain
specific. If reading and writing skills are not needed in a language, they will
not be developed. If a language is spoken with a limited number of people in
a reduced number of domains, it may be less fluent and more restricted than a
language used extensively. If a language is never used for a particular
purpose, it will not develop the linguistic properties needed for that purpose
(specialized vocabulary, stylistic variety, some linguistic rules, etc.).
In general, the failure to understand the complementary principle has
been a major obstacle to obtaining a clear picture of bilinguals and has had
many negative consequences (Grosjean, 1985b): bilinguals have been de-
scribed and evaluated in terms of the fluency and balance they have in their
two languages (when in fact they are rarely balanced); language skills in
bilinguals have almost always been appraised in terms of monolingual stan-
dards (but monolinguals use only one language for all domains or life whereas
bilinguals use two or more); research on bilingualism has in large part been
conducted in terms of the bilingual 's individual and separate languages (the
use of language A or of language B when in fact both languages are often used
simultaneously); and, finally , many bilinguals evaluate their language compe-
tencies as inadequate. Some criticize their mastery of language skills, others
strive their hardest to reach monolingual norms, others hide their knowledge
of their "weaker" language, and most do not perceive themselves as being
bilingual even though they use two (or more) languages in their everyday
lives.
The complementary principle can help us understand a number of phe-
nomena. First, it reflects the configuration of the bilingual 's language reper-
toire: what languages are known and to what extent, what they are used for,
with whom and when, why one language is less developed than another, etc.
Second, it helps to explain why the bilingual's language repertoire may
change over time: as the environment changes and the needs for particular
language skills also change, so will the bilingual's competence in his or her
variou.s language skills. New situations, new interlocutors and new language
functions will involve new linguistic needs and will therefore change the
language configuration of the person involved. Extreme cases of restructuring
are language forgetting .and a return to functional monolingualism, be it in the
 THE BILINGUAL INDIVIDUAL 167

person's first, second or third language. On this point, it should be noted that
interpreters must constantly strive to maintain their knowledge of, and their
fluency in, their various languages despite changes in their own personal lives
which may result in a particular language being used very little outside of
work. Third, an increasing understanding of the complementary principle has
changed researchers' view of bilinguals over the last few years. Bilinguals are
now seen not so much as the sum of two (or more) complete or incomplete
monolinguals but rather as specific and fully competent speakers-hearers who
have developed a communicative competence that is equal, but different in
nature, to that of monolinguals. This competence makes use of one language,
of the other, or of the two together (in the form of mixed speech, see below)
depending on the situation, the topic, the interlocutor, etc. Thus , it is perfectly
normal to find bilinguals who can only read and write one of their languages,
who have reduced speaking fluency in a language they use only with a limited
number of people, or who can speak only about a particular subject in one of
their languages. This in turn is leading to a redefinition of the procedure used
to evaluate the bilingual's competencies. Bilinguals are now being studied in
terms of their total language repertoire, and the domains of use and the
functions of the bilingual's various languages are now being taken into
account (see e.g. Romaine, 1995).
Finally, the complementary principle accounts for why regular bilinguals
are not usually very good translators and interpreters. Some may not know the
translation equivalents in the other language (words, phrases, set expressions,
etc.) which in turn will lead to perception and production problems. At the
level of words, for example, some of the bilingual's domains of life are
covered by the lexicon of only one language, others by the lexicon of the other
language only, and some by the two. Unless bilinguals acquired their second
language in manner which involves learning translation equivalents, many
will find themselves lacking vocabulary in various domains (work, religion,
politics, sports, etc.) even though they may appear to be perfectly fluent in
their two languages. Another reason that accounts for less than perfect
translation and interpreting (prior to training, of course) is that some bilin-
guals may not have the stylistic varieties needed in their two languages (they
simply do not need them in one or the other language). Yet another reason is
that some bilinguals may not have the cultural knowledge (pragmatic compe-
tence) required to understand an utterance in one of their languages (how
many are totally bicultural?). And a final reason is that most bilinguals have
168 FRAN<;:OIS GROSJEAN

not developed the necessary transfer skills needed for translation and inter-
pretation, not to mention the specific on-line processing and memory mecha-
nisms needed for the latter behavior. Taking into account the complementary
principle is a crucial prerequisite in interpreter training: interpreter bilinguals,
unlike regular bilinguals, will have to learn to use their languages (and the
underlying skills they have in them) for similar purposes, in similar domains
of life, with similar people. This is something regular bilinguals do not often
need to do.

2. The bilingual's language modes

In their everyday lives, bilinguals find themselves in various language modes

that correspond to points on a monolingual-bilingual mode continuum
(Grosjean, 1985b, 1997). A mode is a state of activation of the bilingual's
languages and language processing mechanisms. At one end of the con-
tinuum, bilinguals are in a totally monolingual language mode in that they are
interacting with monolinguals of one - or the other - of the languages they
know. One language is active and the other is deactivated. At the other end of
the continuum, bilinguals find themselves in a bilingual language mode in that
they are communicating with bilinguals who share their two (or more) lan-
guages and with whom they can mix languages (i.e. code-switch and borrow).
In this case, both languages are active but the one that is used as the main
language of communication (the base language) is more active than the other.
These are end points but bilinguals also find themselves at intermediary
points depending on such factors as interlocutor, situation, content of dis-
course and function of the interaction.
Figure 1 is a visual representation of the continuum. The languages (A
and B) are represented by a square located in the top and bottom parts of the
figure and their level of activation is depicted by the degree of darkness of the
square: black for a highly active language and white for a deactivated lan-
guage. In the figure, three hypothetical positions for the same speaker are
presented (discontinuous lines numbered from 1 to 3). In all positions, the
speaker is using language A as the main language of communication (the base
language) and it is therefore the most active language (black squares). In
position 1, the speaker is in a monolingual mode: language A is totally active
whereas language B is deactivated (Green, 1986, would even say that it is
 THE BILINGUAL INDIVIDUAL 169

LANGUAGE A
(base language)

• • •
I I

I
I

MONOLINGUAL 1 I 2 13 BILINGUAL
LANGUAGE LANGUAGE
MODE MODE

LANGUAGE B

Figure 1. Visual representation of the Language mode continuum. The speaker's positions
on the continuum are represented by the discontinuous vertical lines and the
level of language activation by the degree of darkness of the squares (black is
active and white is inactive).

inhibited). This mode arises when the person being spoken to (the interlocu-
tor) is monolingual (in this case, in language A), and/or the topic, the situation
and the purpose of interaction require that only one language be spoken to the
exclusion of the other(s). In position 2, the speaker is in an intermediary
mode. Language A is still the most active language (it is the language of
communication) but language B is also partly activated . This kind of mode
arises, for example, when a bilingual is speaking to another bilingual who
does not wish to use the other language (in this case, language B) or when a
bilingual is interacting with a person who has limited knowledge of the other
language. Any number of combinations of the factors listed above (interlocu-
tor, topic, situation, etc.) can lead to this intermediary position. In position 3,
the speaker is at the bilingual end of the continuum. Both languages are active
but language B is slightly less active than language A as it is not currently the
language of communication. This is the kind of mode bilinguals find them-
selves in when they are interacting with other bilinguals who share their two
(or more) languages and with whom they feel comfortable mixing languages.
170 FRAN<;OIS GROSJEAN

They usually first adopt a base language to use together (language A here,
hence its greater level of activation) but the other language, often referred to
as the guest language, is available in case it is needed for a word, a phrase or
a sentence (in the form of code-switches and borrowing; see below). Of
course, a change of topic or of situation may lead to a change of base
language. In our example, language B would become the most active (it
would be represented by a black square) and language A would be slightly
less active (the black square would contain white diagonal lines).
Because a particular mode corresponds to a state of activation of the
bilingual's languages and language processing mechanisms, it influences
both language production (maintenance or change of the base language,
amount and type of language mixing that takes place, etc.) and language
perception (speed of processing of a language, access to one or to both
lexicons, role of the less activated language, etc.). It should be noted also that
bilinguals differ regarding the extent to which they travel along the con-
tinuum; some rarely find themselves at the bilingual end whereas others
rarely leave this end (for example, bilinguals who live in communities where
the language norm is mixed language). The two end points of the continuum
will now be examined as will the bilingual's language behavior in the mono-
lingual and bilingual language modes . This will be followed by a discussion
of the manner in which the language mode continuum model has to be
adapted to take into account simultaneous interpreting.

2.1 The monolingual language mode

In this mode in more detail in Grosjean, 1982), bilinguals adopt

the language of the monolingual interlocutor(s) and deactivate their other
as completely as possible. Bilinguals who manage to do this
totally and, in addition, who speak the other language fluently and have no
foreign accent in it, will often "pass" as monolinguals. Although such cases
are relatively rare (many interpreters are among them), it is precisely these
cases that have led people to think that bilinguals are (or should be) two
monolinguals in one person. In fact, deactivation of the other language is
rarely total as is clearly seen in the interferences bilinguals produce (these are
also known as between-language deviations). An interference is a speaker-
specific deviation from the language being spoken due to the influence of the
other deactivated language. Interferences can occur at all levels of language
 THE BILINGUAL INDMDUAL 171

(phonological, lexical, syntactic, semantic, pragmatic) and in all modalities

(spoken, written or sign). They are of two kinds: static interferences which
reflect permanent traces of one language on the other (such as a permanent
accent, the meaning extensions of particular words, specific syntactic struc-
tures, etc.) and dynamic interferences (the ones interpreters know well) which
are the ephemeral intrusions of the other language as in the case of the
accidental slip on the stress pattern of a word due to the stress rules of the
other language, the momentary use of a syntactic structure taken from the
language not being spoken, etc. Following are some examples of interfer-
ences produced by a French person speaking English: At the phonetic level,
pronouncing Thank 'eaven for this instead of Thank heaven for this; at the
lexical level, using corns (from French comes) instead of horns in Look at the
corns on that animal! ; at the syntactic level, saying I saw this on the page five
(instead of on page five), and in writing, misspelling adress or appartment
(based on the French adresse and appartement).
In addition, if one of the bilingual's languages is mastered only to a certain
level of proficiency (this does not concern interpreters), deviations due to the
person's interlanguage (also known as within-language deviations) will also
occur. These include overgeneralizations (for example, taking irregular verbs
and treating them as if they were regular), simplifications (dropping pluraliza-
tion and tense markers, omitting functions words, simplifying the syntax, etc.)
as well as hypercorrections and the avoidance of certain words and expressions.
Between - and within - language deviations are clearly observable when
bilinguals are in a monolingual language mode but they also occur in the
bilingual language mode. In this latter mode, however, deviations such as
interferences are often difficult to distinguish from perfectly normal borrow-
ings. Unfortunately, this has resulted in grouping very different manifestations
of language contact under one heading such as interference or code-switching.
It should be noted finally that both types of deviations, although sometimes
quite apparent (such as a foreign accent), usually do not interfere with
communication. This is because bilinguals generally develop their languages
to the level of fluency required by the environment.

2.2 The bilingual language mode

In this mode, bilinguals are usually interacting with one another. First they
adopt a language to use together, that is a base language (also known as the
172 FRAN<;OIS GROSJEAN

"host" or " matrix" language). This process is called language choice and is
governed by a number of factors that are similar to those that play a role in
positioning the bilingual on the monolingual-bilingual mode continuum: the
interlocutors involved (the usual language of interaction, language profi-
ciency, language preference, socioeconomic status, age, sex, occupation,
education, kinship relation, attitude toward the languages, etc.); the situation
of the interaction (location, presence of monolinguals, degree of formality
and intimacy); the content of the discourse (topic, type of vocabulary needed)
and the function of the interaction (to communicate information, to create a
social distance between the speakers, to raise the status of one of the inter-
locutors, to exclude someone, to request something, etc.). Language choice
is a well-learned behavior (a bilingual rarely asks the conscious question,
"Which language should I be using with this person?") but it is also a very
complex phenomenon which only becomes apparent when it breaks down.
Usually, bilinguals go through their daily interactions with other bilinguals
quite unaware of the many psychological and sociolinguistic factors that
interact to help choose one language over another. It should be noted again
that the base language can change several times during a single conversation
if the situation, topic, interlocutor, etc. require it.
Once a base language has been chosen, bilinguals can bring in the other
language (the "guest" or "embedded" language) in various ways. One of
these ways is to code-switch, that is to shift completely to the other language
for a word, a phrase, a sentence. For example, Va chercher Marc and bribe
him avec un chocolat chaud with cream on top (Go get Marc and bribe him
with a hot chocolate with cream on top). Code-switching has long been
stigmatized (Baetens Beardsmore, 1986; Grosjean, 1982), and has been given
a number of pejorative names such as Franglais (the switching between
French and English) or Tex-Mex (the switching between English and Spanish
in the southwestern part of the United States). The consequence of this has
been that some bilinguals never switch while others restrict it to situations in
which they will not be stigmatized for doing so. Recently, code-switching has
received considerable attention from researchers (for an overview, see Milroy
& Muysken, 1995). For example, sociolinguists have concentrated on when
and why switching takes place in the social context. Reasons that have been
put forward are: to fill a linguistic need, to continue the last language used, to
quote someone, to specify the addressee, to exclude someone from the
conversation, to qualify a message, to specify speaker involvement, to mark
 THE BILINGUAL I NDIVIDUAL 173

group identity, to convey emotion, to change the role of the speaker, etc.
Linguists, on the other hand, have sought to study the types of code-switches
that occur (single words, phrases , clauses, sentences, etc.) as well as the
linguistic constraints that govern their appearance. Although there is still
considerable controversy over this latter aspect (are constraints universal or
language specific? how broad can a constraint be?), it is now clear that
switching is not simply a haphazard behavior due to some form of "semi-
lingualism" but that it is a well governed process used as a communicative
strategy to convey linguistic and social information.
The other way bilinguals can bring in the other, less activated, language
is to borrow a word or short expression from that language and to adapt it
morphologically (and often phonologically) into the base language. Thus,
unlike code-switching, which is the juxtaposition of two languages, borrow-
ing is the integration of one language into another. Most often both the form
and the content of a word are borrowed (to produce what has been called a
loanword or more simply a borrowing) as in the following examples taken
from French-English bilinguals: "Ca m 'etonnerait qu 'on ait code-switche
autant que fa" (I can't believe we code-switched as often as that) and
"Maman, tu peu.x me tier ltaje/ mes chaussures" (Mummy, can you tie my
shoes?). In these examples, the English words "code-switch" and "tie" have
been brought in and integrated into the French sentence. A second type of
borrowing, called a loanshift, consists in either taking a word in the base
language and extending its meaning to correspond to that of a word in the
other language, or rearranging words in the base language along a pattern
provided by the other language and thus creating a new meaning. An example
of the first kind of loanshift would be the use of humoroso by Portuguese-
Americans to mean 'humorous ' when the original meaning is 'capricious' .
An example of the second kind is the use of idiomatic expressions that are
translated literally from the other language, such as "/ put myself to think
about it" said by a Spanish-English bilingual, based on "Me puse a pensarlo".
It is important to distinguish idiosyncratic loans (also called "speech borrow-
ings" or " nonce borrowings") from words which have become part of a
language community's vocabulary and which monolinguals also use (called
"language borrowings" or "established loans"). Thus, in the following text,
every third or fourth word is an established loan from French which has now
become part of the English language: "The poet lived in the duke's manor.
That day, he painted, played music and wrote poems with his companions."
174 FRAN<;OIS GROSJEAN

Current research is examining, among other things, the differences and simi-
larities that exist between code-switches and borrowings (and within the
latter, between idiosyncratic borrowings and established borrowings), as well
as the impact of the two on language itself, such as first- and second-language
restructuring.

2.3 The case of simultaneous interpreting

What language mode are interpreters in when they are doing simultaneous
interpretation? Figure 2 is an attempt to answer this question. First, as can be
seen, the interpreter is in a bilingual mode and both languages are active.
However, one language is not more active than the other as is normally the
case in the bilingual mode. Both the source language and the target language
are active to the same extent as both are needed, for perception and produc-
tion respectively. Second, input and output components have been added to
each language and it is their level of activation that varies. The addition of

LANGUAGE A
(source language)

Input 1 Output
activated inhibited

MONOLINGUAL BILINGUAL
LANGUAGE LANGUAGE
MODE MODE

Input
activated
._._..Output
activated

LANGUAGE B
(target language)

Figure 2. Visual representation of the interpreter's position on the language mode

continuum when doing simultaneous interpreting. Both languages are active
(black squares) but they differ as to the level of activation of their input and
output mechanisms.
 THE BILINGUAL INDIVIDUAL 175

these components allows us to make a distinction between language (gram-

mar) and language processing mechanisms. The languages arc equally active
but the processing mechanisms are not. In this way, the interpreter will be
able to input the source language (and to a lesser extent the target language,
see below) and to output the target language only. Third, the input component
of the source and of the target language are both activated. No comments are
needed concerning the source language input except that it is probably more
strongly activated than its target language counterpart. As for the latter (the
target language input), at least three reasons require that it also be activated:
the interpreter must be able to monitor his/her overt speech (Levelt, 1989), the
client's occasional use of the target language must be processed, and a fellow
interpreter's cues must be heard or read. Fourth, the target language output
mechanism is activated and the source language output mechanism inhibited
(both for obvious reasons). In sum, the output mechanisms are in a monolin-
gual mode (only one language should normally be output) whereas the input
mechanisms are in a bilingual mode (input will take place in both the source
and the target language).
Of course, this is a very tentative view of interpreting in terms of the
monolingual-bilingual mode continuum model and it needs to be refined. One
concern is how strongly the source language output mechanism should be
inhibited. For example, it is well known that interpreters sometimes naturalize
a source-language term (Gile, 1995) by adapting it to the morphological and
phonological rules of the target language. (This is called a nonce borrowing in
the bilingualism literature). It is also common knowledge that interpreters
sometimes transcode a word or expression into the target language (called a
loanshift or loan translation in the field of bilingualism). Although the pro-
cessing mechanisms that account for these two phenomena are not clear, they
do not probably require the source language output mechanism to be acti-
vated. However, there remains the problem of what Gile (1995) refers to as
reproducing the sounds heard in the source-language speech (with or without
an explanation in the target language) which is simply code-switching in the
bilingualism terminology. Thus, although it is the target language that is being
output, the interpreter sometimes code-switches to the source language for a
word or ,Phrase. How this takes place in processing terms remains an unan-
swered question as the source language output mechanism is normally inhib-
ited. A momentary release of inhibition probably has to occur for a source
language utterance to be output. Future research should examine this issue as
well as how bilingual interpreters shift from being regular bilinguals in a
176 FRAN<;:OIS GROSJEAN

bilingual language mode when interacting with other bilinguals (they adopt a
base language and code-switch or borrow from time to time, etc.) to being
distinct bilinguals when interpreting; although they remain in the bilingual
mode, they can no longer mix their languages in this particular way.

3. The psycholinguistics of bilingualism

The psycholinguistics of bilingualism is aimed at studying the processes

involved in the production, perception, comprehension and memorization of
the bilingual's languages (spoken, written or signed) when used in a monolin-
gual or a bilingual language mode. Until recently the emphasis has been put
on th_e independence of the bilingual ' s languages (How does the bilingual
keep the two languages separate? Does the bilingual have one or two lexi-
cons?) to the de.triment of issues such as the on-line processing of language,
be it in a monolingual or in a bilingual language mode. Much research was
conducted in earlier years, for example, on the coordinate-compound-subor-
dinate distinction (see Grosjean, 1982, for a review). According to it, there are
three types of bilinguals: coordinate bilinguals who have two sets of meaning
units and two modes of expression, one for each language (this means that the
words of the two languages are totally separate entities); compound bilinguals
who have one set of meaning units and two modes of expression (equivalent
words in different langua.ges have the same meaning); and subordinate bilin-
guals who have the meaning units of the first language and two modes of
expression: that of the first language and that of the second, learned by means
of the first (here the bilingual processes words of the weaker language
through the words of the stronger language). Despite the inherent appeal of
this distinction, no amount of experimentation has brought conclusive evi-
dence for this trichotomy .
Another area of considerable investigation has been whether bilinguals
possess one or two internal lexicons. Proponents of the one-lexicon view state
that linguistic information is stored in a single semantic system. Words from
both languages are organized in one large lexicon, but each word is "tagged"
to indicate the language it belongs to (see, for example, Kolers, 1966;
Schwanenflugel & Rey, 1986). Other researchers have claimed that bilinguals
have two lexicons (e.g. Tulving & Colotla, 1970; Taylor, 1971), and that the
information acquired in one language is available in the other only through a
 THE BILINGUAL INDIVIDUAL 177

translation process. Despite a large number of studies, no clear-cut results

have been found . In fact, it has since been proposed that bilinguals have three
stores, one conceptual store corresponding to the bilingual's knowledge of the
world and two language stores, one for each language (Paradis, 1980). In
recent years, the question of lexical representation has been studied with
different paradigms and new models have been proposed (the word associa-
tion model, the concept mediation model, the asymmetrical model, etc.) but it
is still too early to say which one accounts for most of the data obtained
(Schreuder & Weltens, 1993; de Groot & Kroll, 1997). In fact, de Groot
( 1995) has recently proposed that the lexical memory of every bilingual
contains structures of various types (each one corresponding to a different
kind of model) and that these structures occur in different proportions across
bilinguals.
A third issue of interest has been the ability of bilinguals to keep their two
languages separate when in the monolingual mode. Researchers have postu-
lated the existence of a language switch which allows bilinguals to gate out
the other language (e.g. Macnamara, 1967; Macnamara & Kushnir, 1971),
and experimental studies have been conducted to find evidence for this
proposal. The results obtained have been inconclusive or, at the very least,
questionable (see Grosjean, 1982, for a review), and currently it is felt that no
switch, be it psycholinguistic or neurolinguistic, exits in bilinguals. Rather, it
has been proposed that bilinguals are probably using various activation and
deactivation procedures to maintain their languages separate in the monolin-
gual mode and to make them interact in the bilingual mode (Green, 1986;
Paradis, 1989; Grosjean, 1985b, 1997).
Now that it is more generally accepted that the bilingual is not two
monolinguals in one person, but a unique speaker-hearer using one language,
the other language, or both together depending on the interlocutor, situation,
topic, etc. (see above), current psycholinguistic research is trying to under-
stand the processing of language in the bilingual's different language modes.
Researchers are studying how bilinguals in the monolingual mode differ from
monolinguals in terms of perception and production processes, and they are
investigating the actual interaction of the two languages during processing in
the bilingual mode. Following is a review of some recent work on spoken
word recognition in the bilingual mode as it has some relevance to the type of
lexical access the interpreter has to engage in as he or she is interpreting
simultaneously.
178 FRANCOIS GROSJEAN

3.1 Lexical access in the bilingual language mode

A great deal of research has been conducted on lexical access in monolin-

guals. It is well known, for example, that certain properties of words affect
their recognition: their frequency of use, their length, their phonotactic con-
figuration, their uniqueness point, their lexical neighborhood. It is also known
that when words are presented in context, their lexical properties interact with
various sources of knowledge (linguistic rules, knowledge of the world,
discourse, etc.) to speed up or slow down the recognition process (see
Frauenfelder & Tyler, 1987, for a review). The exact nature of the "interac-
tion" between the properties of the words and the sources of knowledge
remains to be described adequately, and the controversy concerning the
moment at which "top-down" information enters the lexical access process
has yet to be resolved (Forster, 1976; Marslen-Wilson, 1987; McClelland &
Elman, 1986; Swinney, 1982). But one conclusion that emerges from this
research is that recognizing a word may not be a simple mapping between its
acoustic-phonetic properties and its entry in the mental lexicon (although see
Klatt, 1979). Instead, it may well be a rather complex process that involves
various narrowing-in and monitoring stages, correcting strategies, post-
access decision stages, and even look-ahead and look-back operations
(Grosjean, 1985a; Grosjean & Gee, 1987; McLelland & Elman, 1986).
Word recognition in bilinguals has received much less attention, espe-
cially as it concerns the access of code-switches and borrowings (often
grouped under the label "guest words"). The literature most closely related to
this latter question dates back a number of years and examines the perception
and production of language mixtures, most of them ungrammatical (Kolers,
1966; Macnamara & Kushnir, 1971 ; Neufeld, 1973; etc.). In a first explor-
atory study, Soares & Grosjean (1984) investigated the lexical access of base
language words and code-switched words by means of the Phoneme Trig-
gered Lexical Decision task (Blank, 1980). English-Portuguese bilingual
subjects were presented with sentences and were asked to listen for a word or
a nonword within them which began with a prespecified phoneme. Once this
word (or nonword) was found, the subjects had to indicate as quickly as
possible whether the item was a real word or not. English monolingual
subjects were run on English sentences only, whereas bilingual subjects were
tested on three separate sets of sentences (English, Portuguese and code-
switched). Before each set, every effort was made to induce the appropriate
 THE BILINGUAL INDIVIDUAL 179

language set: English, Portuguese or bilingual.

Two main findings emerged from this study. The first was that although
bilinguals accessed real words in English as quickly as English monolinguals,
they were substantially slower at responding to nonwords. This finding pro-
vided additional evidence for the residual activation of the other language
when the bilingual is processing only one language (Altenberg & Cairns,
1983; Obler & Albert, 1978). It was hypothesized that a non word triggers a
complete search of the base language lexicon (or an activation of the lexicon,
depending on the access theory one espouses), which is then immediately
followed by at least a partial search (or activation) of the other, less active,
lexicon. This occurs before the stimulus is classified as a nonword, hence the
longer reaction times. The second finding was that bilinguals took longer to
access code-switched words in the bilingual language mode than they did
base language words in the monolingual language mode. Although at first this
was accounted for by suggesting that bilinguals always search the base
language lexicon before the less activated "guest" lexicon, in a later paper
(Grosjean & Soares, 1986) it was suggested that a number of factors could
account for the delay, irrespective of the access strategy. These have been
grouped recently into four categories (Grosjean, 1997):
a. The listener. This category concerns the listener's fluency in the guest
language, the language mode the listener is in (one expects slower recognition
of guest words when the listener is not totally in the bilingual language mode),
the listener's attitude towards code-switching and borrowing (a negative
attitude will ·usually have an inhibitory effect on the guest lexicon) and the
listener's expectations for code-switches and borrowings based on the topic,
the speaker, the situation, etc. The higher the expectation, the easier will be
the guest word recognition. Although these various factors are probably
important, empirical evidence for the role they play during guest word recog-
nition is still lacking.
b. Base and guest language activation. This category is divided into two parts:
base language activation and guest language activation. As concerns base
language activation, more than a quarter of a century ago, Macnamara &
Kushnir (1971) showed that the bilingual listener has certain "expectations"
for strings of words and that one such expectation is that all words should be
in a single language. Although the terminology has changed a bit today (one
would speak in terms of activation and/or inhibition), there is now consider-
180 FRANc;Ois GROSJEAN

able evidence that the base language being spoken (which normally makes up
some 80% to 90% of a mixed utterance) has a strong effect on language
processing. It is more strongly activated and hence base language units
(phonemes, syllables, words) are favored over guest language units, at least
momentarily (for evidence, see Grosjean & Soares, 1986; Btirki-Cohen,
Grosjean & Miller, 1989; Grosjean, 1988). As concerns guest language
activation, evidence is emerging that the density of the code-switches (that is,
the number of code-switched words in a sentence) also influences their
recognition (Soares & Grosjean, 1984; Leuenberger, 1994).
c. Code-switching constraints. This category covers the higher order con-
straints (syntactic, semantic and pragmatic) that govern code-switching. These
have been studied extensively by linguists and sociolinguists (Romaine, 1995;
Myers-Scotton, 1993; Muysken, 1995; di Sciullo, Muysken, & Singh, 1986;
Poplack, Sankoff & Miller, 1988) but have not been the object of many word
recognition studies (see however, Leuenberger, 1994; Li, 1996).
d. Guest word properties. This final category concerns the properties of guest
words that affect their recognition. The first property is the word's phonotac-
tics, that is the sequential arrangements (or groupings) of the word' s units
such as consonant sequences, syllables, etc. It has been hypothesized (e.g.
Grosjean, 1988) that the more phonotactic cues there are for the word to
belong to the guest language, the easier it should be to recognize. The second
property concerns the actual phonetics of the word. If it contains sounds that
are specific to the guest language, if it is said clearly and fully in the phonetics
of the guest language (and not in that of the base language) and if it is said
with the prosody of the guest language, then all this should speed up its
recognition as the appropriate word will be activated more easily in the less
activated lexicon. One difference between code-switches and borrowings is
precisely their degree of phonetic integration in the base language; a code-
switch is not usually integrated in the base language (unless the speaker has
an accent in that language) whereas a borrowing is. One can therefore expect
differences in the recognition of code-switches and borrowings. Finally, the
presence of interlanguage neighbors (i.e. words that are phonologically simi-
lar in the base language) should affect the recognition of guest words. If the
guest word has a close homophone in the base language and, furthermore, if
this homophone is more frequent than the guest word, then the latter should be
recognized with more difficulty. Evidence for the role of some of these
 THE BILINGUAL INDIVIDUAL 181

factors (phonotactics, language phonetics, interlanguage neighbor proximity)

has been provided by Grosjean (1988), Li (1996), Btirki, Grosjean and Miller
( 1989), among others.
In summary, a number of studies show that the recognition of guest
words in bilingual mixed speech is a highly complex process governed by a
number of factors pertaining to the listener, the degree of activation of the two
languages, the linguistic constraints governing code-switching and borrow-
ing, and the properties of the guest words. A model that can account for some
of these factors will now be described.

3.2 BIMOLA: A bilingual model of lexical access

Grosjean (1988) proposed an interactive activation model of word recogni-

tion in bilinguals, which has since been named BIMOLA (Bilingual model of
lexical access; Uwy & Grosjean, in preparation). It is strongly inspired by
TRACE (McClelland & Elman, 1986) and is governed by two basic assump-
tions. First, it is assumed that bilinguals have two language networks (fea-
tures, phonemes, words, etc.) which are independent yet interconnected. They
are independent in the sense that they allow a bilingual to speak just one
language but they are also interconnected in that the monolingual speech of
bilinguals often shows the active interference of the other language, and in
that bilinguals can code-switch and borrow quite readily when they speak to
other bilinguals. This view has long been defended by Paradis ( 1986, 1989),
who proposes that both languages are stored in identical ways in a single
extended system, though elements of each language, because they normally
appear only in different contexts, form separate networks of connections, and
thus a subsystem within a larger system. According to this "subset hypoth-
esis", bilinguals have two subsets of neural connections, one for each lan-
guage (each can be activated or inhibited independently because of the strong
associations between elements), while at the same time they possess one
larger set from which they are able to draw elements of either language at any
time. The second assumption is that in the monolingual language mode, one
language network is strongly activated while the other is only very weakly
activated (the resting activation level of the units of this other network is
therefore very low) whereas in the bilingual language mode, both language
networks are activated but one more than the other (see Green, 1986; Myers-
Scotton, 1993, for a similar assumption).
182 GROSJEAN

BIMOLA
(Bilingual MQdel
of ,Lexical Access) Global language activation
Higher linguistic information

WORDS

Language
activation

PHONEMES

Language and
phonotactic
activation

FEATURES

ACOUSTIC WAVE

Figure 3. Visual representation of the BIMOLA model of lexical access in bilinguals. The
proximity of neighboring units (phonemes, words) is depicted by the degree of
darkness; darkly shaded units have close neighbors in the other language
whereas lightly shaded units do not. At the word level, word frequency is
represented by the size of the units; the larger the unit, the more frequent the
word.
 THE BILINGUAL INDIVIDUAL 183

Figure 3 presents a visual representation of the BIMOLA model as it

stands today. As can be seen, the feature level is common to both languages
but the next two levels - phonemes and words - are organized according to
the subset hypothesis, that is, both independently (each language - Lg. A and
Lg. B-are represented by a subset of units) but also interdependently (both
subsets are enclosed in a larger set). At both the word and phoneme levels,
units can have close or distant form neighbors, both within a language and
between languages. This is depicted by the degree of darkness of the units;
darkly shaded units have close neighbors in the other language whereas
lightly shaded units do not. At the word level, word frequency is represented
by the size of the units - the larger the unit, the more frequent the word.
Connections (mainly excitatory) are unidirectional between features and
phonemes and bidirectional between phonemes and words. Features activate
phonemes which in turn activate words. Descending connections bearing
information about the listener's base language and language mode, and
information from the higher linguistic levels (semantic, syntactic), serve to
activate words which in turn can activate phonemes. Language activation
(reflected by the overall activation of one language system over the other)
takes place through these descending connections but also through within
language connections at the phoneme and word levels. Finally, at the pho-
neme level, between phoneme connections within a language can allow for
phonotactic activation .
The model is currently being refined and implemented on computer
(Lewy & Grosjean, in preparation) and it appears to be able to account for a
number of effects found experimentally (see the previous section). Consider-
able time has been spent developing a combined French and English feature
system which will be used to activate phonemes of one or of both languages
depending on their similarity. Work is now being carried out on the higher
levels, interconnecting phonemes and words in the languages. Once the
model has been implemented, it will be assessed by comparing its behavior to
the data obtained experimentally and to that of the BIA model (Bilingual
Interactive Activation) model of visual word recognition (Grainger &
Dijkstra, 1992; van Heuven, Dijkstra, & Grainger, 1996). Although also an
interactive activation model, BIA has been developed for visual word recog-
nition and is different in a number of ways (e.g. presence of a language node,
many inhibitory connections, etc.). Finally, the BIMOLA model will need to
be adapted at some point to explain what takes place when the interpreter
184 FRAN<;OIS GROSJEAN

doing simultaneous interpretation accesses the words of the source language

(and sometimes of the target language) without letting the words of the target
language interfere too much with the process. The latter are highly active
(they are being used by the output mechanism) and yet they must not be too
involved in the recognition process. How the interpreter keeps the input and
output processes separate, at this level of processing but also at all other
levels, remains an intriguing question.

4. Conclusion

Despite what is already known about the bilingual individual, much more
research needs to be conducted on the topic. The emergence of a wholistic view
of bilingualism is encouraging researchers to move away from the monolingual
yardstick and develop a true linguistics and cognition of bilingualism. How-
ever, many issues require further study: the structure and organization of the
bilingual's different languages; the various processing operations involved in
the perception, production and memorization of language when the bilingual is
in a monolingual language mode or in a bilingual language mode; the linguistic
and psycholinguistic differences (and similarities) between code-switches,
borrowings and interferences; the organization of the bilingual brain; and
finally, cognitive processes in the bilingual individual.

Acknowledgments

Preparation of this paper was made possible in part by grants from the Swiss
National Science Foundation (12-45375.95; 32-49 106.96). The author would
like to thank Robert Hoffman and Dominic Massaro for their helpful com-
ments during the revision stage.

Universite de Neuchatel
 THE BILINGUAL I NDIVIDUAL 185

References

Altenberg, E., & Cairns, H. ( 1983 ). The effects of phonotactic constraints on lexical
processing in bilingual and monolingual subjects. Journal of Verbal Learning and
Verbal Behavior, 22 , 174-188.
Baetens Beardsmore, H. (1986). Bilingualism: Basic principles. Clevedon: Multilingual
Matters.
Blank, M. (1980). Measuring lexical access during sentence processing. Perception and
Psychophysics, 28, 1-8.
Bloomfield, L. (1933). Language. New York: Holt, Rinehart and Winston.
Biirk.i-Cohen, J., Grosjean, F., & Miller, J. (1989). Base language effects on word
identification in bilingual speech: Evidence from categorical perception experiments.
Language and Speech, 32(4), 355-371.
de Groot, A. (1995). Determinants of bilingual lexicosemantic organisation. Computer
Assisted Language Learning, 8(2-3), 151-180.
de Groot, A. , & Kroll, J. (1997). Tutorials in bilingualism: Psycholinguistic perspectives.
Majwah, NJ: LEA.
di Sciullo, A. , Muysken, P., & Singh, R. (1986). Government and code-mixing. Journal
of Linguistics, 22, 1-24.
Forster, K. (1976). Accessing the mental lexicon. In R. Wales & E. Walker (Eds.), New
approaches to language mechanism (pp. 257-287). Amsterdam: North-Holland.
Frauenfelder, U., & Tyler, L. (1987). Spoken word recognition. Cambridge, Mass.: MIT
Press.
Gile, D. (1995). Basic concepts and models for interpreter and translator training.
Amsterdam: John Benjamins.
Grainger, J., & Dijkstra, T. (1992). On the representation and use of language information
in bilinguals. In R. (Ed.), Cognitive processing in bilinguals (pp. 207-220).
New York: North-Holland.
Green, D. (1986). Control, activation, and resource: A framework and a model for the
control of speech in bilinguals. Brain and Language, 27, 210-223.
Grosjean, F. (1982). Life with two languages: An introduction to bilingualism. Cam-
bridge, Mass: Harvard University Press.
Grosjean, F. (1985a). The recognition of words after their acoustic offset: Evidence and
implications . .Perception and Psychophysics, 38, 299-310.
Grosjean, F. (1985b). The bilingual as a competent but specific speaker-hearer. Journal
of Multilingual and Multicultural Development, 6(6), 467-477 .
Grosjean, F. ( 1988). Exploring the recognition of guest words in bilingual speech.
Language and Cognitive Processes, 3(3), 233-274.
Grosjean, F. (1997). Processing mixed language: Issues, fmdings and models. In A. de
Groot & J. Kro ll (Eds.), Tutorials in bilingualism: Psycholinguistic perspectives (pp.
225-254). Majwah, NJ: LEA.
Grosjean, F., & Gee, J. (1987). Prosodic structure and spoken word recognition. Cogni-
tion, 25, 135-155.
Grosjean, F., & Soares, C. (1986). Processing mixed language: Some preliminary find-
ings. In J. Vaid (Ed.), Language processing in bilinguals: Psycholinguistic and
186 FRAN<;OIS GROSJEAN

neuropsychological perspectives (pp. 145-179). Hillsdale, NJ: Lawrence Erlbaum.

Hakuta, K. ( 1986). Mirror of language: The debate on bilingualism. New York: Basic
Books.
Haugen, E. (1969). The Norwegian language in America: A study in bilingual behavior.
Bloomington, IN: University of Indiana Press.
Klatt, D . (1979). Speech perception: A model of acoustic-phonetic analysis and lexical
access. Journal of Phonetics, 7, 279-312.
Kolers, P. (1966). Reading and talking bilingually. American Journal of Psychology, 3,
357-376.
Leuenberger, M. (1994). L'acces au lexique de code-switchs chez le bilingue: Effets de Ia
densite et du contexte. Travaux neucluitelois de linguistique (TRANEL), 21, 61-72.
Levelt, W. (1989). Speaking: From intention to articulation. Cambridge, Mass: MIT
Press.
Uwy, N. & Grosjean, F. A computational model of bilingual lexical access. In prepara-
tion.
Li, P. (1996). Spoken word recognition of code-switched words by Chinese-English
bilinguals. Journal of Memory and Language, 35, 757-774.
Macnamara, J. (1967). The bilingual's linguistic performance: A psychological over-
view. Journal of Social Issues, 23, 59-77.
Macnamara, J. , & Kushnir, S. (1971). Linguistic independence of bilinguals: The input
switch. Journal of Verbal Learning and Verbal Behavior, 10, 480-487 .
Marslen-Wilson, W. (1987). Functional parallelism in spoken word-recognition. Cogni-
tion, 25, 71 -102.
McClelland, J., & Elman, J. (1986). The TRACE model of speech perception. Cognitive
Psychology, 18, 1-86.
Milroy, L., & Muysken, P. (1995). One speaker, two languages: Cross-disciplinary
perspectives on code-switching. Cambridge: Cambridge University Press.
Muysken, P. (1995). Code-switching and grammatical theory. In L. Milroy & P. Muysken
(Eds.), One speaker, two languages (pp. 177-198). Cambridge: Cambridge University
Press.
Myers-Scotton, C. (1993). Duelling languages: Grammatical structure in codeswitching.
Oxford: Clarendon Press.
Neufeld, G. (1973). The bilingual's lexical store. Working Papers on Bilingualism, 1, 35-
65.
Ohler, L., & Albert, M. (1978). A monitor system for bilingual1anguage processing. In
M. Paradis (Ed.), Aspects of bilingualism (pp. 105-113). Columbia, SC: Hornbeam
Press.
Paradis, M. ( 1980). The language switch in bilinguals: Psycholinguistic and neuro-
linguistic perspectives. In P . Nelde (Ed.), Languages in contact and conflict (pp. 501-
506). Weisbaden: Franz Steiner Verlag.
Paradis, M. (1986). Bilingualism. In international encyclopedia of education (pp. 489-
493). Oxford: Pergamon Press.
Paradis, M . (1989). Bilingual and polyglot aphasia. In F. Boller & J. Grafman Eds),
Handbook of neuropsychology, vol. 2 (pp. 117-140). Amsterdam: Elsevier.
 THE Bll..INGUAL I NDIVIDUAL 187

Poplack, S., Sankoff, D., & Miller, C. (1988). The social correlates and linguistic
processes of lexical borrowing and assimilation. Linguistics, 26, 47-104.
Romaine, S. (1995). Bilingualism. London: Blackwell.
Schreuder, R., & Weltens, B. (1993). The bilingual lexicon. Amsterdam & Philadelphia:
John Benjamins.
Schwanenflugel, P., & Rey, M. (1986). Interlingual semantic facilitation : Evidence for a
common representational system in the bilingual lexicon. Journal of Memory and
Language, 25, 605-618.
Soares, C., & Grosjean, F. (1984). Bilinguals in a monolingual and a bilingual speech
mode: The effect on lexical access. Memory and Cognition, 12(4), 380-386.
Swinney, D. (1982). The structure and time-course of information interaction during
speech comprehension: Lexical segmentation, access and interpretation. In J. Mehler,
E. Walker, & M. Garrett (Eds.), Perspectives on mental representations (pp. 151-
167). Hillsdale, NJ: Lawrence Erlbaum Associates.
Taylor, I. (1971). How are words from two languages organized in bilinguals' memory?
Canadian Journal of Psychology, 25, 228-240.
Thiery, C. (1978). True bilingualism and second language learning. In D. Gerver and H.
Sinaiko (Eds.), Language, interpretation and communication (pp. 145-154). New
York: Plenum Press.
Tulving, E., & Colotla, V. (1970). Free recall of trilingual lists. Cognitive Psychology, 1,
86-98.
van Heuven, W., Dijkstra, T., & Grainger, J. (1996). Orthographic neighborhood effects
in bilingual word recognition. Unpublished manuscript, NICI, University of Nijmegen,
The Netherlands.