Available online at www.sciencedirect.

com

Acta Psychologica 128 (2008) 416–430

www.elsevier.com/locate/actpsy

Language selection in bilingual speech: Evidence for

inhibitory processes
Judith F. Kroll a,*, Susan C. Bobb a, Maya Misra b, Taomei Guo c
a
Department of Psychology, Pennsylvania State University, University Park, PA 16802, USA
b
Department of Communication Sciences and Disorders, Pennsylvania State University, PA 16802, USA
c
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, PR China

Received 21 December 2007; received in revised form 31 January 2008; accepted 5 February 2008
Available online 20 March 2008

Abstract

Although bilinguals rarely make random errors of language when they speak, research on spoken production provides compelling
evidence to suggest that both languages are active when only one language is spoken (e.g., [Poulisse, N. (1999). Slips of the tongue: Speech
errors in first and second language production. Amsterdam/Philadelphia: John Benjamins]). Moreover, the parallel activation of the two
languages appears to characterize the planning of speech for highly proficient bilinguals as well as second language learners. In this
paper, we first review the evidence for cross-language activity during single word production and then consider the two major alternative
models of how the intended language is eventually selected. According to language-specific selection models, both languages may be
active but bilinguals develop the ability to selectively attend to candidates in the intended language. The alternative model, that candi-
dates from both languages compete for selection, requires that cross-language activity be modulated to allow selection to occur. On the
latter view, the selection mechanism may require that candidates in the nontarget language be inhibited. We consider the evidence for
such an inhibitory mechanism in a series of recent behavioral and neuroimaging studies.
Ó 2008 Elsevier B.V. All rights reserved.

PyscINFO classification: 2300; 2440; 2720; 2530

Keywords: Psycholinguistics; Bilingual language processing; Language production

1. Introduction Unlike the bottom–up processing that characterizes word

recognition (e.g., Dijkstra & Van Heuven, 2002), speaking
When bilinguals perform even the simplest production is initiated by a conceptually-driven process that takes a
task, such as speaking the name of a familiar object in one thought and maps it on to available lexical information.
of their two languages, there is evidence that both languages In theory, the conceptual nature of spoken production
are active and influence performance (e.g., Colomé, 2001; should allow the language to be selected early in speech
Costa, Miozzo, & Caramazza, 1999; Hermans, Bongaerts, planning as one aspect of the thought to be expressed.
De Bot, & Schreuder, 1998; Kroll, Bobb, & Wodniekca, Although some instances of early language selection may
2006). Although there is abundant evidence for parallel be possible, for example, when bilinguals who have a stron-
activity of the bilingual’s two languages in comprehension ger first (L1) than second (L2) language speak in their L1
tasks (e.g., Spivey & Marian, 1999; Van Heuven, Dijkstra, (e.g., Bloem & La Heij, 2003; La Heij, 2005), most recent
& Grainger, 1998), finding that the unintended language is studies of bilingual word production have shown that the
available during spoken production remains surprising. intention to speak one language only does not suffice to limit
activation to alternatives in that language alone.
*
Corresponding author. Tel.: +1 814 863 0126; fax: +1 814 863 7002. A striking aspect of bilingual speech is that proficient
E-mail address: jfk7@psu.edu (J.F. Kroll). bilinguals do not make random errors of language. At the

0001-6918/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.actpsy.2008.02.001
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 417

same time, they are able to code switch with ease with others influence because L1 is more skilled than L2 and the rapid
who are similarly bilingual (e.g., Muysken, 2000; Myers– time course of speech planning in L1 may not provide an
Scotton, 2002). Although one might argue that fluency at opportunity for L2 to come into play (e.g., Bloem & La Heij,
the level of sentence or discourse production is supported 2003; Kroll, Dijkstra, Janssen, & Schriefers, in preparation).
by a range of mechanisms that might be unavailable in In contrast, when bilinguals speak in the L2, particularly
decontextualized word production, the fact remains that when they are more dominant in L1, there may be multiple
bilingual spoken production is better than what might be influences of L1 on L2 (e.g., Costa, Caramazza, & Sebas-
expected if we assume that both languages are available in tian-Galles, 2000, 2006; Hermans et al., 1998; Hoshino &
parallel and potentially compete for selection. That observa- Kroll, 2008). Notably, these cross-language interactions
tion has led some to propose that bilinguals possess an exqui- in production appear to function between lexical and
site mechanism of cognitive control that develops as they sub-lexical levels (Costa, La Heij, & Navarrete, 2006), to
gain skill in the L2 (e.g., Green, 1998) and that has conse- extend to feed-back as well as feed-forward interactions
quences more generally for executive control processes (Kroll et al., in preparation), and to late specification of
(e.g., Bialystok, Craik, Klein, & Viswanathan, 2004) and for the phonetic properties of realized speech (e.g., Engstler
their neural representation (e.g., Abutalebi & Green, 2007). & Goldrick, 2007; Gerfen, Jacobs, & Kroll, 2005).
Understanding the way in which spoken word produc- Although there is not complete agreement about the
tion is accomplished in bilinguals when two or more extent to which each result in the past studies uniquely
alternatives are available requires that the nature of cross- demonstrates the presence and the locus of cross-language
language activity be characterized and that a mechanism activation in the planning of words in each of the bilin-
of selection be specified. A number of previous studies have gual’s two languages (see Costa et al., 2006), taken
considered the first of these questions in detail (see Costa, together, the evidence is quite compelling. If alternatives
2005, and Kroll et al., 2006, for recent reviews). The avail- are active in the two languages, how is the correct word
able evidence provides support for a number of different selected? Two types of selection mechanisms have been
loci of cross-language activation during the planning of a contrasted. According to a language-specific selection
single word utterance. Fig. 1 is a representative model of model (e.g., Costa et al., 1999), information about words
bilingual word production adapted from previous work in the unintended language may be activated but those
by Hermans (2000) and Poulisse and Bongaerts (1994). words are not candidates themselves for selection. Note
The general assumption in models of lexical production that the presence of cross-language activation itself rules
(e.g., De Bot & Schreuder, 1993; Levelt, 1989) is that at out an extreme language-specific model in which one of
least three component processes must be engaged prior to the two languages is effectively switched off or inhibited
articulation. A concept and its closest lexical representation in advance to enable the bilingual to function as a monolin-
must be selected and the phonology that corresponds to gual speaker (and see Wang, Xue, Chen, Xue, & Dong,
that lexical representation must be specified. For bilinguals, 2007, for recent neuroimaging evidence suggesting that
because there are multiple alternatives in each language, there is no brain area uniquely associated with a language
there can be activation of abstract candidates at the lemma switch). The proposed language-specific model is function-
level or among phonological competitors.1 Note that the ally a ‘‘mental firewall” such that the language cue effec-
model illustrated in Fig. 1 assumes that a language cue rep- tively signals the activated alternatives that are on the
resents the intention to name the object in one of the two right side of the wall. A threshold version of the lan-
languages. As we will discuss later, the representation of guage-specific model assumes that the language cue acts
the intention to speak one language alone may be influ- to set the activation level higher for candidates in the target
enced by the relative dominance of the two languages for language, thereby avoiding potential competition between
the bilingual, by the context in which spoken production them at the point when selection occurs. Finkbeiner,
occurs, and by features of the two languages themselves. Gollan, and Caramazza (2006) proposed the threshold
Kroll et al. (2006) argued that cross-language alternatives model to be a mechanism to avoid what they consider to
may be active at any of the loci shown in the model. The be the ‘‘hard” problem of lexical competition. In contrast,
degree to which there is sustained activity of the nontarget the non-specific language model assumes that words in
language will depend on a variety of factors, including the both languages are potential candidates for selection. The
language of production, proficiency in the L2, the task that non-specific language model allows competition for selec-
initiates speech planning, and the degree to which specific tion such that candidates within and across languages
lexical alternatives are primed. As noted above, when pro- actively compete with alternatives in the unintended
duction occurs in L1, there may be little evidence of L2 language which are eventually inhibited to allow accurate
production to proceed (e.g., Green, 1998). Costa and
1
Santesteban (2004) recently proposed a reconciliation of
For the purpose of this discussion we assume that concepts and
these alternatives by arguing that more proficient bilinguals
conceptual features are largely shared across languages although the same
concept may give rise to different patterns of lexical activation across the
have acquired the skills to avoid the ‘‘hard” problem,
bilingual’s two languages (e.g., De Groot, 1992; Francis, 2005; Tokowicz whereas L2 learners and less proficient bilinguals may be
& Kroll, 2007). more likely to face cross-language competition that
418 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

Fig. 1. A model of bilingual spoken word production (adapted from Hermans, 2000 and Poulisse and Bongaerts, 1994).

requires subsequent inhibition. On this view, the models duction. The first approach is to use a variant of the Stroop
are both correct but describe different states of bilingualism. task (Stroop, 1935) in which pictures are named (e.g., Her-
In the remainder of this paper we consider the selection mans et al., 1998) or words are translated (e.g., LaHeij
models in more detail, reviewing existing findings as well as et al., 1990). During speech planning, a distractor word,
new evidence that we believe provides a more compelling related to the name of the picture or word to be translated,
case for the need for an inhibitory mechanism for even pro- is presented visually or auditorily. The general prediction is
ficient bilinguals. In the course of our review, we consider that at the point in planning when the language of speaking
the role that the production tasks used may have contrib- is selected, there should no longer be any effect of distrac-
uted to this debate and the evidence that bilinguals can tors in the nontarget language.
exploit language-specific cues when they are present. Most The second approach involves mixing and/or switch-
critically, we consider the emerging literature that examines ing the languages of production to examine the conse-
the neural basis of language selection using event related quence of having to prepare alternatives in both
potentials (ERPs) and functional magnetic resonance languages. If both language alternatives are normally
imaging (fMRI) to examine the time course and localiza- active during the planning of a single word utterance
tion of bilingual speech planning processes. in one language alone, then mixing the language of pro-
duction should have little consequence for performance
2. Testing accounts of language selection using behavioral relative to blocked language naming conditions (e.g.,
evidence Kroll et al., in preparation). In a sense, forcing the lan-
guages to be mixed potentially disrupts the mechanism of
Three approaches have been adopted to test the lan- selection that would ordinarily be adopted under blocked
guage-specific vs. non-specific models of spoken word pro- conditions. Identifying the way in which that disruption
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 419

is manifest provides insight into the selection mechanism 2.1. Picture-word interference
itself (e.g., Meuter & Allport, 1999). Likewise, when bil-
inguals are required to switch between their two lan- One of the claims for cross-language competition in
guages, we can examine the magnitude of switch costs bilingual word production comes from findings using the
and the relative impact for the L1 vs. the L2. If one lan- picture-word interference (PWI) paradigm. As noted ear-
guage must be inhibited to produce the other, these dif- lier, in the standard version of the task, a picture is pre-
ferential inhibitory demands should be revealed in the sented followed by a distractor word at a variable SOA
pattern of processing costs observed following a language with respect to the presentation of the picture. The objec-
switch. tive of the task is to name the picture while ignoring the
A third approach to examining language selection pro- distractor. The typical finding in PWI is that words that
cesses is to exploit the presence of shared cross-language are semantically related to the picture cause interference,
features. The logic of this approach is similar to the one while words that are phonologically and/orthographically
that has been used extensively in the literature on bilin- related facilitate picture naming (La Heij, Van der Heijden,
gual word recognition in which performance on language & Schreuder, 1985; Lupker, 1979, 1982). In the bilingual
ambiguous words, such as cognates, interlingual homo- version of the task, distractor words from the nontarget
graphs, and interlingual neighbors can be compared with language that are semantically or phonologically related
performance on language unambiguous words (e.g., Dijk- to the picture’s name also produce interference and facilita-
stra, 2005). Because languages often share aspects of their tion, respectively (e.g., Costa et al., 1999; Hermans et al.,
lexical and/or sub-lexical representations, it is possible to 1998). Hermans et al. (1998) also reported an inhibitory
have bilinguals perform a task in one language alone and effect for distractors that were phonological neighbors of
to ask whether their performance in producing words the translation. The ‘‘phono-translation” effect followed a
with shared-language features is similar to words that time course similar to effects observed for semantic distrac-
are unique to one language alone. If both languages are tors which led Hermans et al. to conclude that the nontar-
active when even a single language is required, then these get language was active to the level of the lemma. In and of
language ambiguous materials should give rise to a differ- themselves, these effects of the nontarget language distrac-
ent pattern of performance than language unique materi- tors suggest that speech planning is open to the influence of
als. Furthermore, bilinguals and monolinguals would be the unintended language.
expected to perform similarly on the language unique Most critically, pictures paired with distractors that are
words but only bilinguals would be predicted to respond the translation of the target word itself also facilitate pic-
differentially to the language ambiguous words. In previ- ture naming (Costa et al., 1999; Hermans, 2004). Costa
ous production studies, the effect of cross-language cog- et al. argued that if lexical items compete for selection, then
nate status has been examined as a means of a translation distractor should cause the most interference
determining whether the phonology of the unintended of all since it directly activates the competing lexical alter-
language is active during the planning of the target utter- native. They took the translation facilitation effect as sup-
ance (e.g., Christoffels, De Groot, & Kroll, 2006; Costa port for a language-specific mechanism that does not
et al., 2000; Kroll et al., in preparation). Although the consider lexical nodes in the competing language for selec-
presence of cross-language activity does not, in and of tion. An important feature of the translation facilitation
itself, reveal the nature of the selection mechanism, a effect is that it is short lived relative to the effects of identi-
comparison of these effects for language pairs with differ- cal distractors (i.e., the picture’s name in the target lan-
ent properties provides a means to determine whether lan- guage). The larger and more extended facilitation effect
guage selection is sensitive to language-specific features for the identity condition was taken as further support
(e.g., Hoshino & Kroll, 2008). for language-specific access since it suggests that lexical
In the next section of the paper, we review briefly the activation of the unintended language had less of an effect
evidence from investigations that have adopted each of on the production of the target name.
these approaches and consider the implications for each Hermans (2004) argued that the differential time course
of the models of language selection. We summarize the and magnitude of the identity condition versus the transla-
findings of a set of experiments that we believe provide sup- tion condition could be accommodated in either a lan-
port for the competition-for-selection alternative and the guage-non-specific model that assumes competition or a
need for an inhibitory process. Here, we describe the cen- language-specific model of lexical selection that does not.
tral findings in past research using each of the three pri- Under the former logic, cross-language activation reduces
mary empirical approaches used to examine bilingual the amount of facilitation of the conceptual, lexical, and
word production: picture-word interference, language phonological activation. Under the latter model, other fac-
switching, and the effects of cognate status on picture nam- tors such as a delay in activation from the translation dis-
ing. In the section that follows, we then consider how very tractor or stronger activation by identity distractors could
recent ERP and fMRI studies on bilingual speech planning change the time course and effect magnitude, respectively.
might allow us to test the alternative selection models more The debate concerning the interpretation of translation
sensitively. facilitation also raises an important issue regarding the
420 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

processing of the distractor word in PWI. If bottom–up was replaced by a dot pattern in which participants named
activation of the distractor word intrudes into processing the number of dots only in the L1, they again found no evi-
that would not otherwise occur during the planning of dence of switch costs. These results led them to conclude
the picture’s name, it becomes difficult to tell whether the that Meuter and Allport’s language switching results sup-
observed effects reveal the locus of cross-language activa- port neither the suppression of the unintended language
tion and language selection during speech planning or a as a whole nor of the specific lexical item in the unintended
complex interaction between the processing of the word language. In a further experiment, they manipulated the
and the picture. Following this line of argument, it speed of response and found that the speed of response
becomes difficult to use PWI data alone to adjudicate availability, and not language identity per se, appears to
clearly between the two selection models. be crucial in determining the pattern of switch cost.
The role of valence and speed of response in accounting
2.2. Language switching for the pattern of switch costs suggests that switch costs
may not provide a simple or direct means to reveal cross-
In a seminal study investigating control in bilingual pro- language competition. Although the Finkbeiner et al.
duction, Meuter and Allport (1999) empirically tested the (2006) results are also potentially problematic for an inhib-
proposal of cross-language competition and subsequent itory account, they suggest that the symmetry or asymme-
suppression/inhibition using a switching paradigm. Central try of switch costs in and of itself may not reveal the means
to the logic of their experiment is the idea of the Task Set of lexical selection.
Inertia hypothesis developed first in non-linguistic task- A recent study by Gollan and Ferreira (2007) reached
switching studies (Allport, Styles, & Hsieh, 1994). In these the same conclusion regarding the difficulty in relying on
tasks, one observes what seems to be a paradoxical effect the symmetry of switch costs to adjudicate between alterna-
where, following a task switch, the dominant task shows tive models of bilingual speech planning. They asked
a greater switching cost than the non-dominant task. Meu- bilinguals to switch between their two languages in simple
ter and Allport explain this phenomenon in terms of sup- picture naming. However, unlike previous studies, the deci-
pression, such that the dominant task must be actively sion to switch was under the person’s control so that he or
suppressed in order to prevent interference with the non- she could use whichever language was preferred to name a
dominant task. When switching into the dominant task, given picture. Under these conditions, Gollan and Ferreira
the active suppression of the dominant task on the preced- found that even L1 dominant bilinguals who typically pro-
ing non-dominant trial persists to disrupt processing on the duce an asymmetric pattern of switch costs when switches
subsequent trial. To investigate whether language switch- are required, produced symmetric switch costs when
ing reveals a similar pattern, Meuter and Allport asked switches were under their own control. Critically, even
bilingual participants to rapidly name Arabic numbers in under these voluntary switching conditions with symmetric
either their L1 or L2, cued by the background color on switch costs, there was an overall inhibitory effect for the
which the number appeared. Results supported the Task more dominant language, suggesting that language mixing
Set Inertia hypothesis, with asymmetric switch costs requires that the dominant language be inhibited. The sym-
observed such that switch costs were greater for naming metric pattern of switch costs observed in this study also
in the L1 than in the L2. Meuter and Allport (1999) pro- shows that valence per se is not the critical factor in deter-
posed a mechanism of suppression whereby the stronger mining the pattern of switch costs. In the context of volun-
L1 is inhibited to allow production in the weaker L2. tary switching, items are inherently bivalent and
More recently, Finkbeiner, Almeida, Janssen, and yet allowing bilinguals to control the pattern of switching
Caramazza (2006) challenged the original interpretation eliminated asymmetric switch costs.
of Meuter and Allport’s (1999) results due to what they A final study took a different approach but also reached
argued was a confound in the nature of the design. Their the conclusion that the symmetry of switch costs may not
claim is based on the idea of valence. Bivalent stimuli that be the most reliable index of the presence of inhibition of
are paired with two distinct responses typically elicit the the nontarget language. Wodniecka, Bobb, Kroll, and
characteristic asymmetric pattern of switch costs. Univa- Green (in preparation) used a competitor priming para-
lent stimuli, however, do not typically show switch costs. digm to determine whether the presence of asymmetric
In Meuter and Allport’s data, valence was confounded switch costs was uniquely correlated with the presence of
with language switching since digits were named in both inhibitory processing. The logic of this study was to
L1 and L2. To tease apart valence from language switch- develop a new means to induce competition that might
ing, Finkbeiner et al. used a bivalent digit-naming task sim- require inhibition and to then ascertain whether only those
ilar to Meuter and Allport. They then interspersed a bilinguals who were likely to produce asymmetric switch
picture naming task in which pictures were only named costs would also reveal evidence for inhibition. Costa and
in the participants’ L1. These stimuli were therefore univa- Santesteban (2004) demonstrated that L2 learners, like
lent. Their results replicated the asymmetrical switch costs the L1 dominant bilinguals in the original Meuter and
for bivalent digit naming, but there were no switch costs for Allport (1999) study, showed an asymmetric pattern of
univalent picture naming. When the picture naming task switch costs, with larger costs when switching into the L1
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 421

than into the L2. Highly proficient and balanced bilinguals 2.3. The effect of language-specific properties
in that study produced a symmetric pattern of switch costs,
but they also produced longer naming latencies for the L1 A third source of evidence on bilingual production
than the L2, a result that suggested the presence of L1 inhi- comes from studies that exploit language-specific proper-
bition. A later study (Costa, Santesteban, & Ivanova, 2006) ties to determine whether the nontarget language is active
replicated the symmetrical pattern for proficient bilinguals and available for selection. A number of studies have taken
but also showed that proficiency alone was sufficient to this approach by examining the effects of cross-language
produce symmetry; it was not necessary that they also be cognates on production. As previously noted, cognates
early bilinguals. are translation pairs that possess shared lexical features
Wodniecka et al. (in preparation) first demonstrated that across languages (e.g., phonology and/or orthography).
the asymmetric vs. symmetric pattern of switch costs in pic- Picture naming studies have shown that bilinguals name
ture naming could be replicated in two groups of late biling- pictures whose names are cognates faster than pictures
uals who differed in whether they were dominant in the L1 whose names are non-cognates (e.g., Costa et al., 2000;
or relatively balanced across the two languages. They then Hoshino & Kroll, 2008; Kroll et al., in preparation). These
showed that both groups, regardless of the symmetry of results suggest that the phonology of lexical candidates in
the switch costs produced, were subject to inhibition in the unintended language is active during the planning of
the context of a competitor priming paradigm. Briefly, the speech so that cognate items receive activation from two
design of the competitor priming task involved a study sources – from both the L1 and the L2. Critically, only bil-
phase and a test phase. At study, participants named pic- inguals show these effects; performance for monolinguals is
tures in separate blocks in Spanish and in English. At test, similar for both types of pictures, suggesting that this is an
they again named pictures in Spanish and English but some effect of the activation of the nontarget language. Costa
of the pictures at test were new pictures, not seen at study, et al. found that the magnitude of the cognate effect in pic-
and others were old pictures, previously named at study. ture naming was also larger in the bilingual’s L2 than in the
Of the old pictures, some were named in the same language L1, suggesting that the more dominant language is more
at study and test (i.e., were congruent) and others were likely to influence the less dominant language than the
named in different languages at study and test (i.e., were reverse. These results support not only parallel activation
incongruent). If in the process of planning a spoken word of a bilingual’s languages but also suggest that parallel acti-
there is competition for selection among alternatives within vation to the level of phonology interacts and competes in
and across languages, then previously naming a picture in the selection of the name of a picture.
the nontarget language should increase cross-language com- Hoshino and Kroll (2008) replicated the cognate facili-
petition. If there is not active competition for selection, then tation for picture naming reported by Costa et al. (2000)
naming the same picture that had been named previously for Spanish–English bilinguals and then extended the
but in a different language should produce facilitation results to Japanese–English bilinguals whose languages
attributable to repetition priming of the picture itself and do not share the same-script. Their results, in addition to
its concept. That is, a language-specific selection mechanism highlighting the extent to which both languages of a bilin-
would not be affected by the mismatch of language other gual are active (i.e., to the level of phonology), also demon-
than to reduce the facilitation predicted in the congruent strate that bilinguals cannot easily exploit all possible cues
condition when picture, concept, and word are repeated that might be available to achieve early language selection.
identically. The critical result in the Wodniecka et al. study Although the script itself is not present in a simple picture
was that there is significant facilitation only in the congruent naming experiment, other research suggests that resonance
condition. Because the incongruent condition included the among orthographic and phonological codes as a conse-
repetition of the picture and concept, factors that have been quence of fluent literacy can be observed during spoken
shown in previous research to produce facilitatory priming word recognition and production (e.g., Damian & Bowers,
(e.g., Francis, Augustini, & Saenz, 2003; Hernandez & 2003; Ziegler, Muneaux, & Grainger, 2003). In the Hoshi-
Reyes, 2002; Sholl, Sankaranarayanan, & Kroll, 1995; no and Kroll experiment, the life experience of proficient
Van Turennout, Bielamowicz, & Martin, 2003), these Japanese–English bilinguals in reading different-script lan-
results suggest that an inhibitory component must have also guages did not function as a cue to mitigate cross-language
been present to counteract that facilitation. Statistically, the activation during picture naming. If inhibitory processes
pattern of competitor priming was identical for both bal- are not necessary to modulate cross-language activity, then
anced and L1 dominant bilinguals although these two bilinguals must be able to exploit available cues to direct
groups differed in whether they produced symmetric or attention to the intended language and/or to raise activa-
asymmetric language switch costs. tion of lexical items in the intended language above the
Taken together, the results of each of these studies, activation threshold of lexical alternatives in the unin-
using different converging tasks, suggest that inhibition is tended language. These results contribute to a growing
required to overcome the activation of competitors from body of evidence showing that it is difficult to identify
the nontarget language when bilinguals produce words in how a language cue would function if it is used at all
one of their two languages. (e.g., Emmorey, Borinstein, & Thompson, 2005; Schwartz
422 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

& Kroll, 2006.) Guo and Peng (2006) also recently reported was present for L2 only at a 0 ms SOA; by 500 ms there
an ERP study showing that Chinese–English bilinguals was no facilitation for naming a picture in L2 whose name
produce the same translation facilitation effect in a variant shared phonology with its translation equivalent in L1. In
of picture-word interference as same-script bilinguals. By contrast to the results of past studies and the results of the
all accounts, bilinguals do not appear to use what might blocked conditions, there was robust cognate facilitation
seem to be obvious choices for cues to language selection for L1 under mixed conditions that extended throughout
such as script, sentence context, or language modality.2 the entire time course, from 0 to 1000 ms. Taken together
Kroll et al. (in preparation) used a delayed picture nam- with the slower RTs for L1 in the mixed condition, the
ing task in which relatively proficient Dutch–English bil- absence of a cognate effect for L2 under just those condi-
inguals were cued as to the language in which to name tions when the cost to L1 processing was greatest, suggests
the picture when they heard an auditorily presented tone. that L1 was actively inhibited while L2 was prepared.
In mixed conditions, one of two tones signaled the produc- Although L1 is the more dominant language, under these
tion of one of the bilingual’s two languages. In blocked conditions it no longer influenced the processing of L2.
conditions, one tone signaled naming in the target language Taken together, the results of these studies suggest that
and the other tone required a ‘‘no” response. In both mixed inhibition may be required to overcome the activation of
and blocked conditions, the tone was presented at a vari- competitors from the nontarget language when bilinguals
able delay following the onset of the picture. By comparing produce words in one of their two languages. We now turn
the time to name pictures whose names were cognates or to a set of very recent studies that have adopted electro-
not under these conditions of mixed and blocked naming, physiological and neuroimaging methods to examine the
it was possible to track the time course of parallel activa- time course of cross-language activation and language
tion of the phonology associated with both languages. selection. As we will see, the evidence on the neural basis
The results of the Kroll et al. (in preparation) study of language selection also converges on the conclusion that
revealed clear evidence for an asymmetry that resembles inhibitory control is required to enable the bilinguals to
the asymmetry observed in language switching (e.g., Meu- speak one language alone.
ter & Allport, 1999). Picture naming in L1 (Dutch) was
slower under mixed conditions when L2 (English) was 3. ERP and fMRI studies of bilingual word production
required to be active than when it was optional under
blocked conditions. In contrast, picture naming in L2 In the search for a locus of selection effects in bilingual
was relatively unaffected by the requirement to have L1 word processing, we and others have turned to methods
active, suggesting that even under the blocked naming con- that can better elucidate the time course and locus of cog-
ditions in L2, L1 was also active. Perhaps the most striking nitive processes involved in selecting and producing words
result, however, was that picture naming was slower in L1 in both the languages. The different accounts of language
than in L2 under the mixed language conditions. This find- selection that we have already discussed would seem to
ing, like the results of the previously reviewed language make contrasting hypotheses about the time at which these
switching experiments, suggests that the mixed language effects should reveal themselves. Accounts which claim that
conditions potentially impose a requirement to inhibit the language cues, operating at a conceptual level, can guide
L1. the selection process might suggest that effects of language
To test the hypothesis that the longer latencies for L1 selection should be seen early in the time course of process-
relative to L2 in the mixed conditions reflect active inhibi- ing, at least in a language production task. In contrast,
tion of the L1, Kroll et al. (in preparation) examined the accounts that propose an inhibitory process in response
effects of cognate status on L1 and L2 production. In pre- to competition from alternatives in both languages would
vious studies of simple picture naming (e.g., Costa et al., be more consistent with a later locus of selection following
2000; Hoshino & Kroll, 2008), reliable facilitation has been activation and competition among within and between lan-
observed when bilinguals name cognate pictures in their guage alternatives. While behavioral methods have been
L2. That facilitation has been attributed to the parallel acti- used to examine time course issues by varying stimulus
vation of the L1 phonology of the picture’s name. The past onset asynchronies and manipulating task constraints,
studies have also reported facilitation for naming L1 cog- behavioral approaches have the distinct disadvantage of
nates, but the effects in these studies have been smaller relying on a discrete measure which may reflect the com-
and less reliable for L1 than for the L2. In the Kroll bined result of many stages and loci of processing.
et al. study, the blocked conditions replicated the past find- Response times on the order of hundreds of milliseconds
ings. There were significant effects of cognate facilitation may obscure the fine-grained series of events which under-
for naming in L2 but no effects of cognate status on L1. lie fluent language processing.
In contrast, in the mixed conditions, cognate facilitation Unlike response time measures, event-related potentials
(ERPs) can allow for evaluation of neurocognitive pro-
2
In other cross-script experiments, Guo and Peng (2005) and Hoshino cesses with millisecond resolution. This sensitivity to time
(2006) showed that script can modulate cross-language activity to some course, coupled with the fact that the ERP method is
degree but only when it is perceptually present. non-invasive, relatively inexpensive, and well-suited for
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 423

use with a variety of populations, makes it the cognitive of cross-language phonology. Despite the somewhat con-
neuroscience method of choice when questions of time flicting results, these ERP studies show consistent evidence
course of processing are at issue.3 The ERP technique pro- that both languages are activated during speech planning
vides an invaluable opportunity to ‘‘observe” on-line pro- and that activation of the nontarget language may even
cessing of stimuli without requiring overt responses or spread to the phonological level at least for certain tasks
additional decision processes needed for most behavioral or language pairings. Thus, a further concern of current
measures. However, ERP recording can also be performed ERP studies is how bilinguals can select the correct words
concurrently with many behavioral measures to allow for a in the correct language and whether they have to inhibit
direct comparison. This feature of the method is particu- activation of the nontarget language.
larly useful, since recent evidence has suggested that ERPs One approach in the literature has been to evaluate the
may reveal aspects of L2 acquisition that are obscured in role of executive function as a possible locus to inhibit the
behavioral measures (e.g., McLaughlin, Osterhout, & activation of the nontarget language. These studies have
Kim, 2004; Tokowicz & MacWhinney, 2005). primarily used the switching paradigm or ‘‘language mix-
In the section below, we focus on the literature pertain- ing” to examine this issue. The main finding of these studies
ing to tasks involving language switching or mixing, pri- has been that modulations of the N2 component, observed
marily in contexts that should require processing to the to be maximal over the frontal and central scalp, may
phonological level, either because language production is reflect the cognitive control system in bilingual speech pro-
required or because a task involves a decision based on duction. Effects on the N2 component have been inter-
the name of the stimulus. Thus, we will not review the preted as evidence for inhibitory effects in these tasks,
growing ERP literature examining reading in the L2 (e.g., since this component has also been found to be sensitive
Alvarez, Holcomb, & Grainger, 2003; De Bruijn, Dijkstra, to response inhibition processes required for the perfor-
Chwilla, & Schriefers, 2001; Kotz, 2001; Kotz & Elston- mance of go/no-go tasks (e.g., Schmitt, Rodriguez-Forn-
Güttler, 2004; Kotz & Hernandez, 2004; Weber-Fox & ells, Kutas, & Munte, 2001). Jackson, Swainson,
Neville, 1996) or reading of code-switched sentences Cunnington, and Jackson (2001) investigated executive
(Moreno, Federmeier, & Kutas, 2002; Proverbio, Cok, & control during language switching by recording ERPs dur-
Zani, 2004). In addition, we have focused on studies in ing a visually cued naming task in which bilinguals named
which the stimuli themselves should not cue the language digits in either L1 or L2. Switch-related modulation of
choice (i.e., numerals or pictures, rather than words), but ERP components was observed on the N2 component,
where task demands have been made explicit by the exper- around 310 ms after stimulus onset over the parietal and
imenter through external language cues. These studies most frontal cortices. As illustrated in Fig. 2, switch trials were
closely mirror the logic described in the behavioral observed to increase this negative ERP component com-
approaches described above. pared to non-switch trials. Importantly, this effect persists
Previous ERP studies have suggested that both lan- throughout the recording epoch, suggesting that it is not
guages are activated even when bilinguals intended to a transient effect, but rather reflects a process that remains
speak only one of their languages, and that the time course active throughout the process of lexical selection. This
and magnitude of nontarget language activation might be effect over the frontal scalp was significant when switching
modulated by the relative proficiency of their two lan-
guages (Guo & Peng, 2006). However, results concerning
how far into processing both languages are active have
been somewhat inconsistent. Similar to behavioral studies,
ERP studies using cognates (Christoffels, Firk, & Schiller,
2007; Rodriguez-Fornells et al., 2005) have found evidence
that phonological information of the nontarget language
was activated in tasks which involved overt or tacit picture
naming. In contrast, when using non-cognates in a picture-
word interference paradigm, Guo and Peng (2005) did not
obtain significant activation of the L1 phonology when
Chinese–English bilinguals spoke words in L2 although
Guo and Peng (2006) reported significant activation of
the L1 translation in L2 production. It is not clear how
the different-scripts for Chinese and English may account
for the observed differences with respect to the activation
Fig. 2. Adapted data from Jackson et al. (2001) demonstrating the ERP
language switching effect observed in their digit-naming task. Bars on the
3
While the spatial resolution of ERP signals is relatively poor, since x-axis indicate 100 ms intervals, and the component peaking just after
different combinations of neural generators could propagate similar 300 ms is described by these authors as the ‘‘N2”. Note that these data are
patterns, this need not be a limitation when research questions focus on collapsed across L1 and L2 and collapsed across sensors in the left fronto-
time course. central quadrant of a 128-channel geodesic sensor net.
424 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

from L1 to L2 but not when switching from L2 to L1, sug- waveforms than switch trials. Finally, both their behavioral
gesting that switching into the non-dominant language and ERP data showed that the mixed language context had
from the dominant language required greater allocation a strong effect on L1 and L2, as compared to the blocked
of resources than making switches in the opposite direc- language context. Specifically, both languages showed a
tion. These results are consistent with claims that speaking greater negativity for non-switch trials in the mixed naming
in the L2 may require active inhibition of the L1. However, context as compared to trials in the blocked naming con-
switches to the dominant L1 should not require such a text in the earlier epoch, while in the later epoch a reversal
demanding process. of this effect was found for L1, but not for L2. Thus,
Verhoef, Roelofs, and Chwilla (2006) also examined blocked naming trials showed an enhanced negativity in
switch costs using the ERP technique. In their study, highly the L1 in the later epoch. Taken together, Christoffels
proficient, but unbalanced, Dutch–English bilinguals were et al. argued that their results suggest that language control
asked to perform a cued picture naming task. They manip- takes place via global inhibition of languages which acts
ulated the time available for preparing the picture’s name, specifically to change the availability of the L1.
allowing either a short interval of 500 ms or a long interval In recent experiments (Guo, Misra, & Bobb, 2007; Mis-
of 1250 ms. The behavioral data revealed a larger switch ra, Guo, Bobb, & Kroll, 2007), we have evaluated the time
cost for L1 than for L2 at short intervals. However, for course of lexical activation and the interaction of a bilin-
long intervals, switch costs were symmetrical for both lan- gual’s languages during speech production ERPs. In two
guages. They argued that these results challenged the pro- experiments unbalanced Chinese–English bilinguals named
ficiency hypothesis proposed by Costa and Santesteban pictures while ERPs were recorded. Picture names were
(2004) who attributed symmetry differences in switching untrained and repetitions of each picture were carefully
costs to language proficiency, since symmetrical switch minimized and controlled. Participants named pictures in
costs could be observed in the very same unbalanced biling- Chinese or English, depending on the picture’s background
uals when longer preparation intervals were provided. Fur- color. In addition, pictures were named at both short
thermore, ERP data in this experiment also showed (250 ms) and long (1000 ms) delays, with ERPs evaluated
evidence for differential switching costs by preparation only at the long delays to minimize artifact. The short delay
interval, reflected in modulations of the N2. However, naming trials were included to ensure the early preparation
these authors interpret the N2 to reflect attentional control of responses and to allow for evaluation of immediate
mechanisms rather than response inhibition and suggest behavioral responses (using logic similar to that described
that the observed pattern reveals that more attentional by Jackson et al. (2001)). In one experiment, pictures to
resources were engaged for the long preparation trials, be named in Chinese and English alternated in a predict-
but this engagement was not fully maintained during long able fashion in a mixed naming paradigm. In another
non-switch trials in the L1, thus contributing to the experiment, participants named pictures in one language
observed switch cost patterns. in the first block and then named the same pictures in the
Christoffels et al. (2007) recently examined bilingual lan- other language in the second block. The effects of switching
guage control using a language switching task. ERPs and from one language to another were evaluated for each
naming latencies were recorded while unbalanced Ger- experiment, and mixed naming was compared to blocked
man–Dutch bilinguals named pictures. The bilinguals naming between experiments. Results suggest that there is
attended university in their L2 context and commonly a processing cost associated with forcing both languages
switched between their languages in daily life. Picture to be active, reflected in effects on the P200, N300 (consis-
names were trained in advance, and participants named tent in latency with the ‘‘N2” described in other ERP lan-
pictures in both blocked and mixed language conditions. guage switching paradigms), and N400 (see Fig. 2).
Additionally, Christoffels et al. manipulated cognate status However, in contrast to expectations based on the behav-
between translation equivalents to examine phonological ioral literature, in which costs to the first language are typ-
activation of the nontarget language. Both behavioral ically greater, processing costs were similar for both
results and ERP results revealed a cognate facilitation languages in most conditions. Also, similar to results from
effect in both languages and for both blocked and mixed Jackson et al. (2001), the effects of language mixing and
language naming, suggesting that phonological informa- language switching began early, but persisted throughout
tion from the nontarget language was activated. However, the recording epoch. Representative results from this para-
in contrast to previous studies, equal switch costs were digm are presented in Fig. 3.
observed for both languages behaviorally, which the Although there is little doubt that executive control is
authors attribute to participants’ experience of commonly involved in tasks where people have to change frequently
switching between languages in their daily life. In addition, from one language to the other, there remains a question
a small switching effect in the ERP data was obtained for as to whether bilinguals have to inhibit the nontarget lan-
L1 but not for L2 during two windows interpreted to be guage when speaking in only one of their two languages.
consistent with the N2 (275–375 ms and 375–475 ms). Rodriguez-Fornells et al. (2005) examined ERP and neuro-
However, in contrast to previous ERP studies on language imaging evidence for interference of phonological informa-
switching, non-switch trials elicited more negative ERP tion from the bilinguals’ nontarget language and inhibition
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 425

Fig. 3. Top panel: Grand average ERP waveforms at electrode site Fz for blocked and mixed picture naming trials for L1 and L2. Bottom panel: Grand
average ERP waveforms at electrode site Fz for non-switch and switch picture naming trials for L1 and L2 (adapted from Misra et al., 2007 and Guo et al.,
2007).

of this interference by the frontal cortex in a task where to monolinguals: the left dorso-lateral prefrontal cortex
responses could be based on access to only one of a bilin- (DLPFC) and the supplementary motor area (SMA).
gual’s two languages. In their study, in order to avoid These neural areas have been associated with executive
vocalization artifacts during EEG and fMRI data acquisi- function in a variety of other tasks, suggesting that biling-
tion, a variant of the go/no-go picture naming task was uals recruit ‘‘typical ‘executive function’ brain areas”
employed. German–Spanish bilinguals were required to (Rodriguez-Fornells et al., 2005, p. 427).
respond when the name of the picture began with a conso- In a recent review of the literature, Rodriguez-Fornells,
nant and to withhold a response for words starting with a De Diego Balaguer, and Munte (2006) further suggested
vowel. The target language was changed on every block, that cognitive control executed by the left dorso-lateral pre-
but responses within a block did not require switching frontal cortex is required in bilinguals, and the degree of
between languages. Stimuli were selected such that on half activation of this mechanism might be related to the simi-
of the trials the names in both languages (Spanish and Ger- larity of languages in use at the lexical, grammatical, and
man) would lead to the same response (coincidence condi- phonological levels. Abutalebi and Green (2007) also
tion, e.g., vowel coincidence Esel – asno ‘‘donkey” or reviewed fMRI evidence on bilingual language production
consonant coincidence Spritze – jeringuilla ‘‘syringe”), and claimed that there is a single network mediating the
whereas on the other half responses were different for the representation of a person’s L1 and L2 and that cortical
two languages (noncoincidence condition, e.g., Erdbeere- and subcortical structures generally associated with execu-
fresa ‘‘strawberry”). Interference was evident behaviorally tive function such as LPFC (left prefrontal cortex) and
by slower response times (RTs) for incongruent than con- ACC (anterior cingulate cortex) are engaged by bilinguals
gruent trials for bilinguals as compared to monolingual to inhibit lexical competition between languages in order
controls. For the ERPs, an enhanced negativity with a to successfully select the intended language. The implica-
frontal maximum was found between 300 and 600 ms (sim- tion is that as a bilingual’s proficiency in L2 is increased,
ilar to the N2 described elsewhere) for incongruent as com- a reduction in prefrontal activity should be observed due
pared to congruent trials. These results provide evidence to changes in the internal structure that will mediate the
for cross-language interference at the phonological level way in which control mechanisms are used.
in bilinguals. In addition, the results of fMRI data col- A recent fMRI study by Wang et al. (2007) reported that
lected in the same paradigm showed two regions associated both the frontal gyrus and the ACC are involved in lan-
with the noncoincidence effect in bilinguals when compared guage switching, providing further evidence for the neural
426 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

mechanism of the inhibition in bilingual word production. bilinguals can indeed utilize information that might func-
Another recent fMRI study (Abutalebi et al., in press) tionally enable a language cue to direct attention to alter-
investigated whether the neural network underlying lan- natives in the target language alone. The available
guage control in bilinguals differs from that involved in evidence on this issue is mixed at best, and again, very
general executive functions that control switching between few studies have directly investigated the nature of lan-
competing tasks within language. They found that lan- guage cues. In the studies we have reviewed, there is a sug-
guage selection processes engaged in contexts during which gestion that cross-language script differences sometimes
both languages must remain active recruited the left cau- affect the degree of observed cross-language activation dur-
date and the ACC in a manner that could be distinguished ing speech planning (e.g., Guo & Peng, 2005; Hoshino,
from areas engaged in within-language selection decisions. 2006) and sometimes do not (e.g., Hoshino & Kroll, 2008).
Taken together, the evidence from both ERPs and other Studies addressing other levels of language processing
neuroimaging methods support a view in which brain areas have also demonstrated the degree to which parallel activa-
associated with inhibitory processing function to aid biling- tion of the two languages proceeds even in the face of clear
uals in selecting the appropriate language alternative. evidence for the presence of one language rather than the
other. For example, a number of experiments have asked
4. Conclusions: interpreting the evidence on language whether the language of a sentence context constrains lex-
selection ical access for language ambiguous words. Schwartz and
Kroll (2006) examined the performance of Spanish–English
The behavioral and neuroimaging studies we have bilinguals naming words that were cognates or controls
reviewed suggest that the problem of language selection is when they were embedded within a sentence context that
indeed a hard problem, contrary to the suggestion that it appeared in one language alone. The sentences were highly
may be possible to bypass processes that negotiate compe- constrained semantically so that the upcoming target word
tition and potential inhibition across the bilingual’s two to be named was highly predictable from the context. The
languages (Finkbeiner et al., 2006). The available evidence cognates used in context had already been shown to reveal
is not conclusive, as this is a relatively young area of inves- the effects of cross-language activation when named out of
tigation and like other research on bilingualism, there are a context. Schwartz and Kroll found that the cognate facili-
host of factors whose influence is not fully understood (e.g. tation observed out of the context was eliminated in con-
proficiency in the L2, context of language use, age of acqui- text when sentences were highly semantically constrained.
sition, similarity of the two languages). Despite that, there Most critically, under conditions of low semantic con-
is a strong suggestion in the material that we have reviewed straint, the magnitude of cognate facilitation was identical
that not only are the bilingual’s two languages activated in to what was observed out of context. That is, the language
parallel, but that they compete for selection during spoken of the sentence context itself, in the absence of high seman-
production. In this final section, we consider a set of tic constraints, was not able to be exploited to direct access
remaining issues that relate to this conclusion. to the target language alternative (see Van Hell, 1998, for
similar results). Although it is difficult to know how far
4.1. Language cues we might generalize these findings, they suggest that obvi-
ous cues to the target language are not necessarily func-
As noted earlier in the discussion, a language-specific tional cues to language selection. Duyck, Van Assche,
selection model has to assume that bilinguals are able to Drieghe, and Hartsuiker (2007) recently reported a similar
effectively represent the intention to speak one language result in an eye tracking study, suggesting that the failure
alone. The evidence on cross-language activation suggests to overcome the parallel activation of the two languages
that the intention to speak one language is not sufficient in context can be observed even under experimental condi-
to restrict activation to that language. The model of bilin- tions that are more ecologically valid.
gual language production shown in Fig. 1 assumes that
there is a representation of a language cue. The language 4.2. The weak link hypothesis
cue potentially provides a means to represent the intention
to use one of the bilingual’s two language and also to weigh Another alternative to the selection-by-competition
the influence of information that should bias production model is the weak link hypothesis (e.g., Gollan & Acenas,
towards one of the languages. For a model to enable lan- 2004; Gollan, Montoya, Cera, & Sandoval, 2008; Gollan,
guage-specific selection, it must be able to exploit available Montoya, Fennema-Notestine, & Morris, 2005). The claim
language cues. Those cues may take different forms, here is that because bilinguals are likely to use each of their
depending on the linguistic context in which the two lan- two languages less often than a monolingual uses one lan-
guages are used and the specific goals of the task that is guage, the words in each language will be functionally
performed. They may also be related to aspects of the lar- lower frequency and therefore less available than they are
ger cultural and perceptual environment in which the two for monolingual speakers. The focus of this work has been
languages are used. An important consideration in testing on the performance of bilingual vs. monolingual speakers,
the language-specific alternative is to determine whether showing that bilinguals experience more tip-of-the-tongue
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 427

(TOT) states than monolingual speakers and are generally ipants were native English speakers who were not highly
slower to name pictures, even in the language that is nom- proficient in Spanish as the L2, and may therefore reflect
inally their L1 than their monolingual counterparts. On the stage of L2 development that Costa and Santesteban
this view, slower spoken production and increased TOTs (2004) identified as requiring inhibition, they demonstrate
do not indicate cross-language competition that must be a phenomenon that is clearly inhibitory and that like other
resolved but rather weaker links between the semantics evidence reviewed here, affects the more dominant L1.
and the phonology. It remains to be seen whether it will Related findings have been reported by Linck, Kroll,
be possible to adjudicate between this alternative and those and Sunderman (in review) in a study of L2 learners
that are the focus of the current paper, but there are at least immersed in the L2 environment during study abroad.
two features of this work that make it seem unlikely that it Linck et al. compared the performance of a group of
will provide a comprehensive account of bilingual produc- immersed learners with group of classroom learners
tion. At an empirical level, the bilingual participants in matched on length of L2 study and working memory span.
most of the studies on which the weak link alternative is Participants performed a translation recognition task in
based have been heritage speakers of Spanish who, for which distractors to be rejected included pairs that were
the most part, had Spanish as their L1 but were educated related to the target translation by similarity to lexical form
almost entirely in English and for whom English has (e.g., man-mano, which means hand in Spanish) or by sim-
become the dominant language. In contexts in which it is ilarity to the lexical form of the translation (e.g., man-ham-
likely that there is language attrition or the failure to fully bre, which means hunger in Spanish), or to the meaning of
acquire the two language equally and in similar contexts, it the translation (e.g., man-mujer, which means woman in
seems likely that there will be processing consequences that Spanish). Classroom learners were slower to reject these
may be quite different than those encountered by bilinguals false pairs that were lexically and semantically related than
for whom the two languages are used more equally or in completely unrelated control pairs. In contrast, immersed
similar contexts (but see Ivanova & Costa, 2008, for evi- learners appeared to be immune to the effects of lexical
dence that the same pattern may be observed for early bil- interference, even when the form similarity of the lexical
inguals who actively use both languages). At a theoretical pairs was easily perceived. In a verbal fluency task in which
level, it is very difficult to see how the weak link hypothesis they were asked to generate as many members of a seman-
can account for the benefits to cognitive performance in the tic category as they could think in 30 s, the immersed learn-
realm of executive function that have been reported in a ers generated a larger number of Spanish exemplars than
series of recent papers (e.g., Bialystok et al., 2004; Costa, the classroom learners. But most critically, the immersed
Hernandez, & Sebastián-Gallés, 2008). Although the rela- learners generated fewer exemplars in English, their domi-
tion between the resolution of cross-language competition nant language, even when the English task was blocked for
and the acquisition of expertise in the realm of executive retrieval. The overall pattern of results suggest that in the
control is indirect at best, there is clear evidence that bilin- L2 immersion context, the L1 is actively inhibited. How
gualism confers benefits more generally to just those con- the inhibition observed in a brief immersion experience
trol skills that affect the ability to resolve conflicting later maps on to enduring consequences for cognitive con-
information. The selection-by-competition model of lan- trol is a rich topic for future research. For present pur-
guage choice could easily have the consequence of confer- poses, the findings simply underline the need for inhibition.
ring such expertise as bilinguals negotiate cross-language To summarize, we have reviewed a range of empirical
competition in multiple contexts. It is difficult to see how studies that address the question of how bilinguals select
the weak links alternative would have any way of account- the language they intend to speak.4 Although the findings
ing for these positive cognitive consequences of bilingual- in any particular study may be constrained by aspects of
ism. To the contrary, it supports a deficit view of the methodology, the overall picture, including evidence
bilingualism that would be unlikely to confer these benefits. from both behavioral and ERP experiments, suggests that
there is cross-language activation during the planning of
4.3. Other evidence for inhibition speech that potentially extends quite far along into the time
course of processing. The available evidence also suggests
A number of recent papers have proposed that inhibi- that although some of these effects may be larger when bil-
tion of the L1 may be required to modulate the activity inguals are less proficient in the L2, providing greater oppor-
of the L2. Levy, McVeigh, Marful, and Anderson (2007) tunity for L1 to influence performance in L2, they are
used the retrieval-induced forgetting paradigm that has present and characterize the speech of even highly proficient
been studied primarily in the domain of memory research bilinguals. Taken together with the available evidence on the
(Anderson, Bjork, & Bjork, 1994), to investigate the degree
to which the phonology of the L1 is inhibited when the L2 4
Given space constraints, we have not reviewed the literature on
is spoken. They demonstrated that increasing practice in
bilingual speech errors although that area of research also provides
retrieving the Spanish name of a picture has the effect of support for the presence of cross-language interactions that sometime
suppressing retrieval of the phonology associated with result in errors that reveal the activation of the nontarget language (see
the English name of the same picture. Although the partic- Poulisse, 1997, 1999, for reviews of that work).
428 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

degree to which language cues are difficult to exploit, the Costa, A., Caramazza, A., & Sebastian-Galles, N. (2000). The cognate
data demonstrate that the ‘‘hard problem” of language facilitation effect: Implications for the model of lexical access. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 26,
selection does not go away with increasing bilingual exper- 1283–1296.
tise. They suggest that bilinguals become skilled in negotiat- Costa, A., Hernandez, M., & Sebastián-Gallés, N. (2008). Bilingualism
ing the existing cross-language competition rather than in aids conflict resolution: Evidence from the ANT task. Cognition, 106,
learning to avoid that competition from the start. That con- 59–86.
clusion is compatible with the claims that bilingualism more Costa, A., La Heij, W., & Navarrete, E. (2006). The dynamics of bilingual
lexical access. Bilingualism: Language and Cognition, 9, 137–151.
generally confers enhanced cognitive control (e.g., Bialystok Costa, A., Miozzo, M., & Caramazza, A. (1999). Lexical selection in
et al., 2004). Mapping the connections more directly bilinguals: Do words in the bilingual’s two lexicons compete for
between mechanisms of language control and cognitive skill selection? Journal of Memory and Language, 41, 365–397.
will be a rich topic for future research. Costa, A., Roelstraete, B., & Hartsuiker, R. J. (2006). The lexical bias
effect in bilingual speech production: Evidence for feedback between
lexical and sublexical levels across languages. Psychonomic Bulletin &
Acknowledgements Review, 13, 972–977.
Costa, A., & Santesteban, M. (2004). Lexical access in bilingual speech
The writing of this article was supported in part by NSF production: Evidence from language switching in highly proficient
Grant BCS-0418071 and NIH Grant F33HD055003 to Ju- bilinguals and L2 learners. Journal of Memory and Language, 50,
dith F. Kroll, by NIH Grant R56-HD053146 to Judith F. 491–511.
Costa, A., Santesteban, M., & Ivanova, I. (2006). How do highly
Kroll, Maya Misra, Taomei Guo, and Chip Gerfen, and by proficient bilinguals control their lexicalization process? Inhibitory and
the Open Project Grant at State Key Laboratory for Cog- language-specific selection mechanisms are both functional. Journal of
nitive Neuroscience and Learning, Beijing Normal Univer- Experimental Psychology: Learning, Memory, and Cognition, 32,
sity to Judith F. Kroll, Maya Misra, and Taomei Guo, and 1057–1074.
by the National Natural Science Foundation of China Damian, M. F., & Bowers, J. S. (2003). Effects of orthography on speech
production in a form-preparation paradigm. Journal of Memory and
(30600179) to Taomei Guo. Language, 49, 119–132.
De Bot, K., & Schreuder, R. (1993). Word production and the bilingual
References lexicon. In R. Schreuder & B. Weltens (Eds.), The bilingual lexicon
(pp. 191–214). Amsterdam: John Benjamin.
Abutalebi, J., Annoni, J. M., Zimine, I., Pegna, A. J., Seghier, M. L., Lee- De Bruijn, E. R. A., Dijkstra, A., Chwilla, D. J., & Schriefers, H. J. (2001).
Jahnke, H. et al. (in press). Language control and lexical competition Language context effects on interlingual homograph recognition:
in bilinguals: An event-related FMRI study. Cerebral Cortex. Evidence from event-related potentials and response times in semantic
Abutalebi, J., & Green, D. (2007). Bilingual language production: The priming. Bilingualism: Language and Cognition, 4, 155–168.
neurocognition of language representation and control. Journal of De Groot, A. M. B. (1992). Determinants of word translation. Journal of
Neurolinguistics, 20, 242–275. Experimental Psychology: Learning, Memory, and Cognition, 18,
Allport, A., Styles, E. A., & Hsieh, S. (1994). Shifting intentional set: 1001–1018.
Exploring the dynamic control of tasks. In C. Umilta & M. Dijkstra, T. (2005). Bilingual word recognition and lexical access. In J. F.
Moscovitch (Eds.), Attention and performance XV: Conscious and Kroll & A. M. B. De Groot (Eds.), Handbook of bilingualism:
nonconscious information processing (pp. 421–452). Hillsdale, NJ: Psycholinguistic approaches (pp. 179–201). New York: Oxford Uni-
Erlbaum. versity Press.
Alvarez, R. P., Holcomb, P. J., & Grainger, J. (2003). Accessing word Dijkstra, A., & Van Heuven, W. J. B. (2002). The architecture of the
meaning in two languages: An event-related brain potential study of bilingual word recognition system: From identification to decision.
beginning bilinguals. Brain and Language, 87, 290–304. Bilingualism: Language and Cognition, 5, 175–197.
Anderson, M. C., Bjork, R. A., & Bjork, E. L. (1994). Remembering can Duyck, W., Van Assche, E., Drieghe, D., & Hartsuiker, R. J. (2007).
cause forgetting: Retrieval dynamics in long-term memory. Journal of Visual word recognitions by bilinguals in a sentence context: Evidence
Experimental Psychology: Learning, Memory and Cognition, 20, for nonselective access. Journal of Experimental Psychology: Learning,
1063–1087. Memory, and Cognition, 33, 663–679.
Bialystok, E., Craik, F. I. M., Klein, R., & Viswanathan, M. (2004). Emmorey, K., Borinstein, H. B., & Thompson, R. (2005). Bimodal
Bilingualism, aging, and cognitive control: Evidence from the Simon bilingualism: Code-blending between spoken English and American
task. Psychology and Aging, 19, 290–303. sign language. In J. Cohen, K. T. McAlister, K. Rolstad, & J.
Bloem, I., & La Heij, W. (2003). Semantic facilitation and semantic MacSwan (Eds.), Proceedings of the 4th international symposium on
interference in word translation: Implications for models of lexical bilingualism (pp. 663–673). Somerville, MA: Cascadilla Press.
access in language production. Journal of Memory and Language, 48, Engstler, C., & Goldrick, M. (2007). Lexically conditioned variation
468–488. across languages. In Poster presented at the 4th international workshop
Christoffels, I. K., De Groot, A. M. B., & Kroll, J. F. (2006). Memory and on language production.
language skill in simultaneous interpreting: The role of expertise and Finkbeiner, M., Almeida, J., Janssen, N., & Caramazza, A. (2006). Lexical
language proficiency. Journal of Memory and Language, 54, 324–345. selection in bilingual speech production does not involve language
Christoffels, I. K., Firk, C., & Schiller, N. O. (2007). Bilingual language suppression. Journal of Experimental Psychology: Learning, Memory,
control: An event-related brain potential study. Brain Research, 1147, and Cognition, 32, 1075–1089.
192–208. Finkbeiner, M., Gollan, T., & Caramazza, A. (2006). Bilingual lexical
Colomé, A. (2001). Lexical activation in bilinguals’ speech production: access: What’s the (hard) problem? Bilingualism: Language and
Language-specific or language-independent? Journal of Memory and Cognition, 9, 153–166.
Language, 45, 721–736. Francis, W. S. (2005). Bilingual semantic and conceptual representation.
Costa, A. (2005). Lexical access in bilingual production. In J. F. Kroll & In J. F. Kroll & A. M. B. De Groot (Eds.), Handbook of bilingualism:
A. M. B. De Groot (Eds.), Handbook of bilingualism: Psycholinguistic Psycholinguistic approaches (pp. 251–267). New York: Oxford Uni-
approaches (pp. 308–325). New York: Oxford University Press. versity Press.
 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430 429

Francis, W. S., Augustini, B. K., & Saenz, S. P. (2003). Repetition priming Kroll, J. F., Bobb, S. C., & Wodniekca, Z. (2006). Language selectivity is
in picture naming and translation depends on shared processes and the exception, not the rule: Arguments against a fixed locus of
their difficulty. Journal of Experimental Psychology: Learning, Mem- language selection in bilingual speech. Bilingualism: Language and
ory, and Cognition, 29, 1283–1297. Cognition, 9, 119–135.
Gerfen, C., Jacobs, A., & Kroll, J. F. (2005). Inhibiting first language Kroll, J. F., Dijkstra, A., Janssen, N., & Schriefers, H. (in preparation).
phonology in planning and producing speech in a second language. In Selecting the language in which to speak: Experiments on lexical access
Presented at the European society for cognitive psychology. in bilingual production.
Gollan, T. H., & Acenas, L. A. (2004). What is a TOT? Cognate and La Heij, W. (2005). Selection processes in monolingual and bilingual
translation effects on tip-of-the-tongue states in Spanish–English and lexical access. In J. F. Kroll & A. M. B. De Groot (Eds.), Handbook of
Tagalog–English bilinguals. Journal of Experimental Psychology: bilingualism: Psycholinguistic approaches (pp. 289–307). New York:
Learning, Memory, and Cognition, 30, 246–269. Oxford University Press.
Gollan, T. H., & Ferreira, V. S. (2007). Natural language switching: LaHeij, W., De Bruyn, E., Elens, E., Hartsuiker, R., Helaha, D., & Van
Expanding the role of inhibitory control. In Poster presented at the Schelven, L. (1990). Orthographic facilitation and categorical inter-
48th annual meeting of the psychonomic society. ference in a word-translation variant of the Stroop task. Canadian
Gollan, T. H., Montoya, R. I., Cera, C., & Sandoval, T. C. (2008). More Journal of Psychology, 44, 76–83.
use almost always means a smaller frequency effect: Aging, bilingual- La Heij, W., Van der Heijden, A. H. C., & Schreuder, R. (1985). Semantic
ism, and the weaker links hypothesis. Journal of Memory and priming and Stroop-like interference in word-naming tasks. Journal of
Language, 58, 787–814. Experimental psychology: Human Perception and Performance, 11,
Gollan, T. H., Montoya, R. I., Fennema-Notestine, C., & Morris, S. K. 62–80.
(2005). Bilingualism affects picture naming but not picture classifica- Levelt, W. J. M. (1989). Speaking: From intention to articulation.
tion. Memory and Cognition, 33, 1220–1234. Cambridge: MIT Press.
Green, D. (1998). Mental control of the bilingual lexico-semantic system. Levy, B. J., McVeigh, N. D., Marful, A., & Anderson, M. C. (2007).
Bilingualism: Language and Cognition, 1, 67–81. Inhibiting your native language: The role of retrieval-induced forget-
Guo, T., & Peng, D. (2005). Was the native language activated during the ting during second language acquisition. Psychological Science, 18,
production of the second language: An ERP study with unskilled 29–34.
Chinese–English bilinguals. In Poster presented at the annual meeting of Linck, J. A., & Kroll, J. F., & Sunderman, G. (in review). Learning a
the cognitive neuroscience society. second language during immersion: Evidence for increased inhibitory
Guo, T., Misra, M., Bobb, S. C., & Kroll J. F. (2007). Behavioral and control.
ERP measures of competition in bilingual spoken word production. In Lupker, S. J. (1979). The semantic nature of response competition in the
Paper presented at the 6th international symposium on bilingualism. picture–word interference task. Memory & Cognition, 7, 485–495.
Guo, T., & Peng, D. (2006). ERP evidence for parallel activation of two Lupker, S. J. (1982). The role of phonetic and orthographic similarity in
languages in bilingual speech production. NeuroReport, 17, 1757–1760. picture–word interference. Canadian Journal of Psychology, 36,
Hermans, D. (2000). Word production in a foreign language. Unpublished 349–367.
doctoral dissertation, University of Nijmegen, Nijmegen, The McLaughlin, J., Osterhout, L., & Kim, A. (2004). Neural correlates of
Netherlands. second-language word learning: Minimal instruction produces rapid
Hermans, D. (2004). Between-language identity effects in picture-word change. Nature Neuroscience, 7, 703–704.
interference tasks: A challenge for language-nonspecific or language- Meuter, R. F. I., & Allport, A. (1999). Bilingual language switching in
specific models of lexical access? International Journal of Bilingualism, naming: Asymmetrical costs of language selection. Journal of Memory
8, 115–125. and Language, 40, 25–40.
Hermans, D., Bongaerts, T., De Bot, K., & Schreuder, R. (1998). Misra, M., Guo, T., Bobb, S. C., & Kroll, J. F. (2007). Electrophysio-
Producing words in a foreign language: Can speakers prevent logical correlates of bilingual word production. In Poster presented at
interference from their first language? Bilingualism: Language and the annual meeting of the cognitive neuroscience society.
Cognition, 1, 213–229. Moreno, E. M., Federmeier, K. D., & Kutas, M. (2002). Switching
Hernandez, A. E., & Reyes, I. (2002). Within- and between-language languages, switching palabras (words): An electrophysiological study
priming differ: Evidence from repetition of pictures in Spanish–English of code switching. Brain and Language, 80, 188–207.
bilinguals. Journal of Experimental Psychology: Learning, Memory, Muysken, P. (2000). Bilingual speech: A typology of code-mixing. Cam-
and Cognition, 28, 726–734. bridge, UK: Cambridge University Press.
Hoshino, N. (2006). A psycholinguistic study of native language constraints Myers–Scotton, C. (2002). Contact linguistics: Bilingual encounters and
on speaking words in a second language. Unpublished doctoral grammatical outcomes. Oxford, UK: Oxford University Press.
dissertation, The Pennsylvania State University, University Park, PA. Poulisse, N. (1997). Language production in bilinguals. In A. M. B. De
Hoshino, N., & Kroll, J. F. (2008). Cognate effects in picture naming: Groot & J. F. Kroll (Eds.), Tutorials in bilingualism: Psycholinguistic
Does cross-language activation survive a change of script? Cognition, perspectives (pp. 201–224). Mahwah, NJ: Erlbaum.
106, 501–511. Poulisse, N. (1999). Slips of the tongue: Speech errors in first and second
Ivanova, I., & Costa, A. (2008). Does bilingualism hamper lexical access in language production. Amsterdam/Philadelphia: John Benjamins.
speech production? Acta Psychologica, 127, 277–288. Poulisse, N., & Bongaerts, T. (1994). First language use in second
Jackson, G. M., Swainson, R., Cunnington, R., & Jackson, S. R. (2001). language production. Applied Linguistics, 15, 36–57.
ERP correlates of executive control during repeated language switch- Proverbio, A. M., Cok, B., & Zani, A. (2004). Electrophysiological
ing. Bilingualism: Language and Cognition, 4, 169–178. measures of language processing in bilinguals. Journal of Cognitive
Kotz, S. A. (2001). Neurolinguistic evidence for bilingual memory Neuroscience, 14, 994–1017.
representation: A comparison of reaction times and event-related Rodriguez-Fornells, A., De Diego Balaguer, R., & Munte, T. F. (2006).
brain potentials. Bilingualism: Language and Cognition, 4, 143–154. Executive control in bilingual language processing. Language Learn-
Kotz, S. A., & Elston-Güttler, K. (2004). The role of proficiency on ing, 56, 133–190.
processing categorical and associative information in the L2 as Rodriguez-Fornells, A., van der Lugt, A., Rotte, M., Britti, B., Heinze, H.
revealed by reaction times and event-related brain potentials. Journal J., & Munte, T. F. (2005). Second language interferes with word
of Neurolinguistics, 17, 215–235. production in fluent bilinguals: Brain potential and functional imaging
Kotz, S. A., & Hernandez, A. (2004). An ERP study on L2 concreteness evidence. Journal of Cognitive Neuroscience, 17, 422–433.
effects: The role of proficiency and cognate status. In Paper presented Schmitt, B. M., Rodriguez-Fornells, A., Kutas, M., & Munte, T. F.
at the annual meeting of the cognitive neuroscience society. (2001). Electrophysiological estimates of semantic and syntactic
430 J.F. Kroll et al. / Acta Psychologica 128 (2008) 416–430

information access during tacit picture naming and listening to words. Van Heuven, W. J. B., Dijkstra, T., & Grainger, J. (1998). Orthographic
Neuroscience Research, 41, 293–298. neighborhood effects in bilingual word recognition. Journal of Memory
Schwartz, A. I., & Kroll, J. F. (2006). Bilingual lexical activation in and Language, 39, 458–483.
sentence context. Journal of Memory and Language, 55, 197– Van Turennout, M., Bielamowicz, L., & Martin, A. (2003). Modulation of
212. neural activity during object naming: Effects of time and practice.
Sholl, A., Sankaranarayanan, A., & Kroll, J. F. (1995). Transfer between Cerebral Cortex, 13, 381–391.
picture naming and translation: A test of asymmetries in bilingual Verhoef, K. M. W., Roelofs, A., & Chwilla, D. J. (2006). Dynamics of
memory. Psychological Science, 6, 45–49. language switching: Evidence from event-related potentials in overt
Spivey, M. J., & Marian, V. (1999). Cross talk between native and second picture naming. In Poster presented at the annual meeting of the
languages: Partial activation of an irrelevant lexicon. Psychological cognitive neuroscience society.
Science, 10, 281–284. Wang, Y., Xue, G., Chen, C. S., Xue, F., & Dong, Q. (2007). Neural bases
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. of asymmetric language switching in second-language learners: An
Journal of Experimental Psychology, 18, 643–662. ER-fMRI study. NeuroImage, 35, 862–870.
Tokowicz, N., & Kroll, J. F. (2007). Number of meanings and concrete- Weber-Fox, C. M., & Neville, H. J. (1996). Maturational constraints on
ness: Consequences of ambiguity within and across languages. functional specializations for language processing: ERP and behav-
Language and Cognitive Processes, 22, 727–779. ioral evidence in bilingual speakers. Journal of Cognitive Neuroscience,
Tokowicz, N., & MacWhinney, B. (2005). Implicit and explicit measures 8, 231–256.
of sensitivity to violations in second language grammar: An event- Wodniecka, Z., Bobb, S. C., Kroll, J. F., & Green, D. W. (in preparation).
related potential investigation. Studies in Second Language Acquisition, Selection processes in bilingual word production: Evidence for
27, 173–204. inhibition.
Van Hell, J. G. (1998). Cross-language processing and bilingual memory Ziegler, J. C., Muneaux, M., & Grainger, J. (2003). Neighborhood effects
organization. Unpublished dissertation, University of Amsterdam, in auditory word recognition: Phonological competition and ortho-
Amsterdam, The Netherlands. graphic facilitation. Journal of Memory and Language, 48, 779–793.