Fig. 1 - Aix Marseille Université
Fig. 1 - Aix Marseille Université
Fig. 1 - Aix Marseille Université
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
U NIVERSITE D’AIX-MARSEILLE I - U NIVERSITE DE P ROVENCE<br />
UFR DE P SYCHOLOGIE, S CIENCES DE L’EDUCATION<br />
THÈSE<br />
en vue de l'obtention du grade de<br />
DOCTEUR DE L'UNIVERSITÉ D'AIX-MARSEILLE I<br />
Formation doctorale : Psychologie<br />
Présentée et soutenue publiquement par<br />
K ATHERINE M IDGLEY<br />
le 13 février 2009<br />
L E LEXIQUE BILINGUE<br />
ETUDES ELECTROPHYSIOLOGIQUES<br />
DES INTERACTIONS INTER- LANGUES<br />
sous la direction de J ONATHAN G RAINGER<br />
membres du jury :<br />
Jonathan Grainger, Directeur de Recherche, CNRS<br />
Sonja Kotz, Professeur, Institut Max Planck, rapporteur<br />
Jean-Marc Lavaur, Professeur, Université Paul Valéry<br />
Janet van Hell, Professeur, Université de Nimègue, rapporteur<br />
Johannes Ziegler, Directeur de Recherche, CNRS
à mes bilingues
Remerciements<br />
Il y a un nombre de personnes sans qui ce travail ne serait pas ce qu'il est.<br />
There are many who have helped shape this work. But first I must thank the<br />
Je remercie en premier lieu les participants qui se sont livrés avec joie et bonne<br />
participants who lent their brains to my questioning. Without them science<br />
humeur à mes expérimentations. Sans eux la science n’existerait pas.<br />
cannot exist. To all who have helped me along the way, I thank you.<br />
Et à tous ceux qui m’ont assisté d’une façon ou d’autre - je vous remercie.<br />
You know who you are.
Sommaire<br />
1 | Introduction 9<br />
Les Modèles 11<br />
Potentiels évoqués 14<br />
Etudes bilingues en potentiels évoqués 18<br />
Difficultés en études bilingues 22<br />
Les études présentées 24<br />
2 | Les effets de langue chez des apprenants en langue seconde et des 33<br />
bilingues confirmés grâce à la technique des potentiels évoqués<br />
Résumé pagination du journal - 1 35<br />
Introduction 2 36<br />
1° Expérience 4 38<br />
Méthodes 4 38<br />
Résultats 7 41<br />
Discussion 8 42<br />
2° Expérience 9 43<br />
Méthodes 9 43<br />
Résultats 10 44<br />
Discussion 12 46<br />
3° Expérience 13 47<br />
Méthodes 13 47<br />
Résultats 13 47<br />
Discussion 15 49<br />
Discussion générale 15 49<br />
Références 18 52<br />
3 | Une étude électrophysiologique des effets de voisinage 55<br />
orthographique inter-langues<br />
Résumé pagination du journal - 123 57<br />
Introduction 123 57
Résultats 125 59<br />
Discussion générale 130 64<br />
Méthodes 131 65<br />
Références 135 69<br />
4 | Les effets de cognats sur la compréhension des mots chez les 71<br />
apprenants en langue seconde : une étude en potentiels évoqués<br />
Résumé 74<br />
Introduction 75<br />
Méthodes 81<br />
Résultats 86<br />
Discussion 92<br />
Références 97<br />
5 | L'amorçage masqué de répétition et par équivalent de traduction 101<br />
chez des apprenants en langue seconde : un regard sur le décours<br />
temporel de l'activation de la forme et du sens<br />
Résumé 104<br />
Introduction 105<br />
1° Expérience 111<br />
Méthodes 111<br />
Résultats 117<br />
Discussion 122<br />
2° Expérience 123<br />
Méthodes 124<br />
Résultats 124<br />
Discussion 130<br />
Discussion générale 131<br />
Références 136<br />
6 | Discussion générale 139<br />
| Bibliographie 151
Introduction<br />
Chapitre 1<br />
Introduction<br />
P | 9
Introduction<br />
A true universal account of language processing will require a<br />
detailed understanding of both monolingual comprehension<br />
and bilingual comprehension and the representations and<br />
processes involved.<br />
This statement, loosely quoted from de Groot and Kroll’s introduction of Tutorials<br />
in Bilingualism (1997), is partly the result of at least two important observations. One -<br />
bilingualism is prevalent. Grosjean (1982) estimates that more than half the world’s<br />
population has some experience in a language other than a native language. Two - many<br />
of the existing accounts of lexical processing that have been elaborated to date describe<br />
monolingual processing in great detail with rare contingencies for describing bilingual<br />
processing. However, we can point to two models that directly tackle the question of the<br />
bilingual lexicon, its organization and processes.<br />
The Models<br />
One of the earliest and most cited models describing the specifics of the bilingual<br />
lexicon is the Revised Hierarchical Model of Kroll and Stewart (1994). The model<br />
assumes a separation of lexical and conceptual representations of the lexicon with links as<br />
the provision for mapping form to meaning. The RHM posits a single conceptual level<br />
that can be accessed via links from lexical items from both a bilingual’s first language<br />
(L1) and second language (L2). In this model the conceptual links are stronger between<br />
well established L1 lexical entries and conceptual representations than they are between<br />
P | 11
Introduction<br />
lexical entries of a newly acquired L2 and conceptual representations. (See figure 1.) The<br />
RHM also describes an interconnectivity between the lexical representations of a<br />
bilingual’s two languages with direct lexical links that are stronger from L2 lexical<br />
representations to L1 lexical representations than they are from L1 lexical representation<br />
to L2 lexical representations.<br />
<strong>Fig</strong>ure 1. The revised Hierarchical Model (RHM)<br />
This model provides an architecture around which it is possible to make hypotheses<br />
about the interactions of a bilingual’s two languages. This model also leaves room to<br />
describe the changes that occur over time as a bilingual’s, or a learner’s experience with<br />
both L1 and L2 evolves. As a learner becomes more proficient in his L2 conceptual links<br />
between the L2 lexical representations and concepts will strengthen while links between<br />
L2 lexical representations and L1 lexical representations will weaken.<br />
Strategies on the part of the learner, such as the express study of vocabulary and<br />
grammar as is the case with classroom learning, lead to explicit knowledge. This explicit<br />
learning is often the method for late learners of an L2. The strong lexical links from L2 to<br />
L1 reflect a type of explicit learning where L2 items rely on their L1 translations to<br />
P | 12
Introduction<br />
access meaning. This asymmetry is supported by translation studies reporting better<br />
performance from L2 to L1 than vice-versa and by studies showing better performance in<br />
translation as compared to picture naming in L2 (Kroll & Curley, 1988; Chen & Leung,<br />
1989; de Groot, 1992).<br />
While the RHM proposes separate stores for L1 and L2 lexical items, providing for<br />
the possibility of interaction between the two stores, another bilingual lexicon model, the<br />
Bilingual Interactive Activation Model or BIA model (Grainger & Dijkstra, 1992; van<br />
Heuven, Dijkstra, & Grainger, 1998) posits a completely integrated and highly interactive<br />
bilingual lexicon as does its successor the BIA+ model (Dijkstra & van Heuven, 2002 - See<br />
figure 2.) The BIA model is an implemented connectionist model in the line of<br />
McClelland and Rumelhart’s Interactive Activation (IA) model (1981). This model adds<br />
on to the monolingual IA model. It adds an integrated lexicon of two languages and an<br />
extra representational layer containing two language nodes. These language nodes<br />
connect to all the items in both lexica. The model posits non-selective bottom-up<br />
processing in which words from both languages are activated along with languagespecific<br />
top-down processing in which the language nodes selectively inhibit activity in<br />
words of the non-appropriate language. At the word level all words inhibit each other.<br />
Activated word nodes from the same language send activation to the corresponding<br />
language node. The activated language node sends inhibitory feedback to all word nodes<br />
in the other language. Language nodes collect activation from words in the language that<br />
they represent and inhibit active words of the other language. This model addresses<br />
questions of language selectivity as it is possible within this model’s framework to obtain<br />
results that show both language selectivity and language non-selectivity as does the<br />
bilingual lexicon literature.<br />
P | 13
Introduction<br />
<strong>Fig</strong>ure 2. The Bilingual Interactive Activation Model (BIA)<br />
Throughout the chapters of this work we will refer to these two models. The RHM<br />
is a descriptive model that captures the asymmetries of a bilingual’s two lexicons and can<br />
also be used to describe the changes over time of a second language learner. The BIA<br />
model is an instantiated computational model that assumes certain proficiency. Many<br />
phenomena related to bilingual word recognition can be simulated with the BIA model.<br />
Event-Related Potentials - ERPs<br />
The ERP technique involves measuring the brain's electrical activity at the scalp<br />
(electroencephalogram or EEG) and time-locking this activity to specific stimulus events.<br />
The individual EEG waveforms to the stimulus events are then averaged across the same<br />
or experimentally similar items to yield a waveform characterizing the measurable part of<br />
the brain's electrical response to the stimulus category of interest. The important thing to<br />
keep in mind is that it is the averaging process that extracts the voltage signature of a<br />
particular stimulus condition from the background noise. The waveforms resulting from<br />
P | 14
Introduction<br />
the averaging, the ERPs, are a series of peaks and valleys that are identified as ERP<br />
components. Researchers have linked the various components in the ERP waveform to a<br />
number of cognitive processes (Coles & Rugg, 1995; Lau, Phillips & Poeppel, 2008 for<br />
reviews).<br />
One ERP component that is of particular interest for the presented studies is the<br />
N400. The N400 is a negative-going wave that usually starts around 250-300 ms and<br />
peaks at approximately 400 ms after the onset of a stimulus. Numerous studies have<br />
found the N400 to be involved in some aspect of semantic processing. Kutas and Hillyard<br />
(1980, 1984) were the first to report on this component. They found it in a sentential<br />
context to be larger in amplitude in response to sentence final words that are semantically<br />
anomalous (e.g., "He takes his coffee with cream and dog.) and that it is greatly reduced<br />
or even absent to high probability congruent sentence endings (e.g., "He takes coffee with<br />
cream and sugar."). Based on findings such as these and others it has been argued that the<br />
N400 reflects the process of semantic integration (Brown & Hagoort, 1993; Holcomb,<br />
1993). Larger N400s are taken as being indicative of a more effortful or involved<br />
integration process.<br />
Subsequent work has shown that the N400 is also elicited to words outside of a<br />
sentence context. For example, in one early study Bentin, McCarthy & Wood (1985)<br />
demonstrated N400s to word pairs in a semantic priming task. Target words preceded by<br />
semantically related prime words (e.g., doctor-NURSE) produced smaller N400s than<br />
target words preceded by unrelated words (e.g., truck-NURSE). Other studies using<br />
words in lists with no sentence or semantic context have also reported effects on the<br />
N400. For example, Münte, Wieringa, Weyerts, Szentkuti, Matzke and Johannes (2001)<br />
showed that the N400 is sensitive to word frequency with low frequency words having a<br />
larger N400 amplitude than high frequency words. The general conclusion from this and<br />
P | 15
Introduction<br />
other studies is that virtually any word processed for meaning will produce an N400<br />
response (Holcomb, 1993).<br />
An important example of a study using single word presentation was one by<br />
Holcomb, O’Rourke and Grainger, (2002). In this study the authors demonstrated that<br />
N400 amplitude is sensitive to the size of a word’s orthographic neighborhood. The N400<br />
was larger to items that were similar to many other words (e.g., words like “time”) than to<br />
items that resembled only a few or no other word (e.g., “yacht”). They interpreted this<br />
result as supporting a view of the N400 whereby it reflects activity at the interface of<br />
form and meaning during word processing.<br />
More recently research using masked priming paradigms has elaborated on the<br />
timing of visual word recognition. Holcomb and Grainger (2006) describe the modulation<br />
of the N400 and earlier components, including the N250, that reflect the sequential<br />
overlapping steps in the processing of printed words. The N250 component is also of<br />
interest for the work presented here. Holcomb and Grainger compared targets that were<br />
full repetitions of, partial repetitions of, or unrelated to their masked primes. They found<br />
strong modulation of the N400 and N250 for full repetitions as compared to unrelated<br />
targets (see <strong>Fig</strong>ure 3). The amplitude of these components in the case of partial repetitions<br />
compared to unrelated items was between that of full repetitions and unrelated primes.<br />
This led Holcomb and Grainger to propose that the N250 reflected the mapping of<br />
sublexical form representations (e.g., letters and letter combinations) onto lexical<br />
representations.<br />
P | 16
Introduction<br />
<strong>Fig</strong>ure 3. From Holcomb and Grainger 2006, fig. 3b. ERPs time locked to target onset in the three<br />
repetition conditions at CP1 site. Negative voltages are plotted upward.<br />
Throughout this work we will present research conducted using ERPs.<br />
Electrophysiological measures were preferred here because they more directly reflect the<br />
processing of items, being an on-line measure of brain activity rather than, as in the case<br />
with behavioral measures such as reaction times, just one data point after processing has<br />
been completed. Moreover, ERPs are multidimensional in that not only are they a<br />
measure of “voltage” but they also they contain precise time-course information and scalp<br />
distribution information. This allows us to draw conclusions not only about the time<br />
course of word processing but also concerning possible differences in the nature of this<br />
processing. For example, one effect observed using behavioral methods could be shown<br />
to have an effect on two different ERP components, which by inference would implicate<br />
two different perceptual or cognitive processes. Similarly, differences in time-course of<br />
effects could point to different loci for these effects. Furthermore since much research<br />
designed to probe the bilingual lexicon has been conducted using behavioral measures it<br />
P | 17
Introduction<br />
is our position that the added multidimensional sensitivity of this electrophysiological<br />
data can only enrich our collective knowledge of the subject.<br />
ERP studies of bilingual visual word processing<br />
Yet ERPs have been used to study issues of bilingual language comprehension. A<br />
recent review by Moreno, Rodríguez-Fornells & Laine (2008) listed over 30 publications<br />
where ERPs were used to characterize some aspect of bilingual language processing, and<br />
this was by no means an exhaustive list (we found 45 studies published since 1996).<br />
Of particular relevance to the experiments in this dissertation is the much smaller<br />
set of studies that have used ERPs to examine visual word processing in second language<br />
learners. In one of the first such studies Kotz (2001) set out to examine semantic priming<br />
in early fluent Spanish-English bilinguals. ERPs were recorded while participants made<br />
lexical decisions on target items in either Spanish or English. The results demonstrated<br />
that both associative (e.g., girl-BOY or chica-CHICO) and categorical (junior-BOY or<br />
joven-CHICO) priming produced the typical reduction in N400 amplitude in both L1 and<br />
in L2. This pattern is consistent with the view that, at least in proficient bilinguals, there<br />
is not an asymmetry in lexical-semantic connections in L1 and L2. A more recent study<br />
by this same group (Kotz & Elston-Guttler, 2004) used the same ERP priming method but<br />
with native German late learners of English. Participants were assigned to two<br />
proficiency groups. Unlike the early proficient participants in Kotz (2001) neither group<br />
of late learners showed a robust categorical priming effect on the N400 in their L2,<br />
although both groups showed effects of associative priming. This pattern of results was<br />
taken to indicate that categorical relationships require an early exposure to a second<br />
language, while associative relationships can be acquired later.<br />
P | 18
Introduction<br />
In a study similar to that of Kotz (2001), Phillips, Segalozitw, O’Brien & Yamasaki,<br />
(2004) investigated individual differences in second language (L2) proficiency by looking<br />
at the efficiency of semantic priming. Twenty-nine L1 English speakers varying in L2<br />
(French) proficiency read lists of words in English and French blocked by language.<br />
Critical target words were either primed by a semantic associate in the preceding trial<br />
(e.g. adult-CHILD) or were unprimed (e.g. rabbit-CHILD). High proficiency participants<br />
showed an N400 priming effect in both L1 and L2 (i.e., smaller N400s for targets<br />
following related primes than unrelated primes), but low proficiency participants showed<br />
only a priming effect in L1. Moreover, in the high proficiency group the N400 effect was<br />
delayed by 50 ms in L2 compared to L1.<br />
Two ERP studies have been conducted looking at interlingual-homographs in<br />
Dutch-English bilinguals (de Bruijn, Dijkstra, Chwilla & Schriefers, 2001; Kerkhofs,<br />
Dijkstra, Chwilla & de Bruijn, 2006). Interlingual homographs are words that are spelled<br />
the same in both languages (i.e., they have the same orthographic form) but have different<br />
meanings in the two languages. In the first study, de Bruijn et al. (2001) had bilinguals<br />
perform a generalized lexical decision task on triplets of items, responding with “yes” if<br />
all three items were correct Dutch and/or English words, and with “no” if one or more of<br />
the items was not a word in either language. The critical manipulation was that sometimes<br />
the second item in a triplet was an interlingual homograph whose English meaning was<br />
semantically related to the third item of the triplet (e.g., HOUSE - ANGEL - HEAVEN,<br />
where ANGEL means “sting” in Dutch). In such cases, the first item was either an<br />
exclusively English (HOUSE) or an exclusively Dutch (ZAAK) word. The ERP results<br />
showed clear N400 priming effects on the third item and these were not affected by the<br />
language of the first item in the triplets. In other words the N400 to ANGEL-HEAVEN<br />
was attenuated compared to an unrelated pair of words even when the ANGEL was<br />
P | 19
Introduction<br />
preceded by a Dutch word which presumably puts the reader in a Dutch mode of<br />
processing for the subsequent word ANGEL (sting), which is not related to the word<br />
HEAVEN. The authors claimed this finding was evidence in support of a strong bottomup<br />
role with respect to bilingual word recognition.<br />
A second study (Kerkhofs et al., 2006) also used a semantic priming paradigm and<br />
interlingual homographs. This study examined the effects of semantic and lexical–<br />
orthographic context in Dutch–English bilinguals who performed an English lexical<br />
decision task in which homograph target words like STEM (meaning “voice” in Dutch)<br />
were preceded by primes like ROOT (a related word in English) or FOOL (and unrelated<br />
word in English). N400 effects were found for homograph targets preceded by related<br />
(ROOT-STEM) compared to unrelated primes (FOOL-STEM) a result consistent with the<br />
findings of de Bruijn et al. (2001). The amplitude of the N400 effect was also modulated<br />
by the word frequency of the targets in both their Dutch and English reading, although in<br />
the opposite direction (i.e., smaller N400s for high frequency English and larger N400s<br />
for high frequency Dutch readings). These data would also appear to argue for a strong<br />
bottom-up role with respect to bilingual word recognition.<br />
A seemingly contradictory finding comes from Rodríguez-Fornells, Rotte, Heinze,<br />
Nosselt & Münte, (2002) who recorded ERPs and fMRI from bilingual Spanish/Catalan<br />
and monolingual Spanish participants. Participants were instructed to press a button when<br />
presented with words in one language, while ignoring words and pseudowords in the<br />
other language. The ERPs of bilingual participants to words of the non-target language<br />
were not sensitive to word frequency in that language suggesting that participants were<br />
able to reject these items at an early stage before semantic analysis. However, one<br />
difference between this study and the Dutch studies is that the Spanish/Catalan speakers<br />
were highly proficient in both languages and acquired both languages at an early age. The<br />
P | 20
Introduction<br />
Dutch/English bilinguals were presumably less proficient in L2 and learned it later. The<br />
ability to reject items from the other language then might reflect differences in the age of<br />
acquisition or proficiency.<br />
A 2004 study by McLaughlin employed a novel approach for using ERPs in the<br />
study of second language learning. In their study ERPs were recorded in a lexical<br />
decision semantic priming paradigm in L1 English speakers (similar to the Kotz 2001<br />
study). However, they only included L2 (French) stimuli and tracked changes in the ERPs<br />
as participants progressed through their first university French course. In the first session<br />
participants who had had only 14 hours of instruction in French showed significant<br />
differences between word and pseudoword N400 responses. The results were similar to<br />
monolingual studies which have consistently shown larger N400s for pseudowords than<br />
real words (e.g., Holcomb & Neville, 1990). A control group of monolingual English<br />
speakers (with no French experience) showed no difference between real French words<br />
and French-like pseudowords. Moreover, these effects of word-pseudoword N400s<br />
continued to increase as participants received more instruction in French from 14 to 63 to<br />
138 hours. Moreover, with more instruction semantic priming effects on the N400 began<br />
to emerge.<br />
A recent study by Thierry and Wu (2007) illustrates another way ERPs have been<br />
used to augment behavioral measures of bilingual processing. Similar to the Dutch studies<br />
using interlingual homographs Thierry and Wu sought to determine if the native language<br />
of bilingual individuals is active during second-language comprehension. Chinese–<br />
English bilinguals were required to decide whether English (L2) words presented in pairs<br />
were related in meaning or not (i.e., semantic priming). However, rather than present L1<br />
words that participants could read along with the L2 items (a procedure that arguably<br />
alerts participants to the bilingual nature of the task), Thierry and Wu unbeknownst to<br />
P | 21
Introduction<br />
their participants, included L2 words which concealed (masked) a Chinese word that was<br />
a translation of the L2 item. Interestingly, there was no evidence of the masked L1 items<br />
affecting behavioral performance, but it did significantly modulate the N400 in expected<br />
direction (smaller N400s for translations compared to unrelated words), establishing that<br />
English words were automatically and unconsciously translated into Chinese. Critically,<br />
the same modulation was found in Chinese monolinguals reading the same words in<br />
Chinese. These findings demonstrate that native-language activation is an unconscious<br />
correlate of second-language comprehension.<br />
These ERP studies make important contributions to the literature on bilingual visual<br />
word processing. However, even though these specific pieces have been added it is clear<br />
that many aspects of bilingual word processing still remain unexplored with<br />
electrophysiological methods. For example, although there have been two studies using<br />
interlingual homophones to probe bilingual word processing no studies using ERPs have<br />
been conducted using cognates. No studies examining the effects of orthographic<br />
neighborhood and its effects in a bilingual context have been reported. More surprisingly,<br />
there are no reported studies that directly compare processing between an L1 and an L2.<br />
It is our position that electrophysiological methods can bring a wealth of information to<br />
what is known about bilingual visual word processing. This dissertation presents four<br />
articles, three of which are published or accepted for publication, that we hope will add<br />
significantly to the field of bilingual visual word recognition.<br />
Thorny issues in bilingual studies<br />
Bilingualism may be the norm for many of the world’s populations but its forms<br />
certainly cannot be generalized. The language experience of bilinguals and second<br />
language learners varies enormously. State agendas regarding the obligation, the amount<br />
P | 22
Introduction<br />
and the starting age of second language study in schools vary wildly from country to<br />
country. Populations where balanced bilingualism is rampant can be found (e.g. the<br />
Basque country, Catalonia, San Diego County) as can populations where a majority<br />
language is not the dominant L1 of a great number of speakers (e.g. New York City). The<br />
emphasis on teaching English as a second language has been on the rise in many<br />
countries across the globe. These quick observations illustrate the fact that there are<br />
varying levels of bilingualism, various bilingual situations and varying attitudes towards<br />
bilingualism and second language learning that could very well affect its study.<br />
Furthermore we can imagine that there are important differences in the lexicons of<br />
bilinguals that share very similar languages like Spanish and French when compared to<br />
bilinguals that come from very different language groups like Finnish and French and that<br />
a difference in script between a bilingual’s two languages like between Hebrew and<br />
French would result in its own specificity. These factors tend to complicate the<br />
generalization of studies that may use comparisons between any two language<br />
combinations difficult.<br />
Another thorny issue is the question of proficiency and how to measure it in order to<br />
compare results across studies. Furthermore in the context of studying learners it is clear<br />
that a learner’s competence in L2 changes over time. What isn’t clear is how to select a<br />
homogenous group of L2 learners and how to compare them to other populations of L2<br />
learners.<br />
The bottom line in bilingual studies, given all the possible variation due to different<br />
language pairs, proficiency, homogeneity of population, is that precaution must always be<br />
exercised when comparing results of studies that have been conducted in different<br />
populations.<br />
P | 23
Introduction<br />
We sought solutions to these issues for the research presented in this dissertation.<br />
We consistently studied learners and bilinguals from the same language combination, i.e.,<br />
French and English. Our L2 learner participants were students at Tufts University in the<br />
Boston area who were engaged in studying French and students from the University of<br />
Provence in the <strong>Aix</strong>-en-Provence area who were studying English. One solution was to<br />
keep our populations as homogeneous as possible by selecting participants from<br />
university learners at the same level of study. However, we did note differences between<br />
the group of American students and the group of French students. These differences were<br />
presumably due to the difference of commitment between the two groups. The French<br />
university students were studying only English while the Tufts students had French in<br />
addition to many other subjects. Moreover, we conducted our studies in two ERP<br />
laboratories, one at Tufts University and one in <strong>Marseille</strong> at the LPC (Laboratoire de<br />
Psychologie Cognitive) that were designed to be as identical as possible allowing us to<br />
compare our results.<br />
The presented studies<br />
The studies that are the starting point for this dissertation do not address the<br />
complex question of language interactivity but are a simple interrogation of the<br />
processing differences between an L1 and an L2. In chapter two we present research that<br />
looks at the differences in processing using electrophysiological methods and observing<br />
language learners and proficient bilinguals. Across three experiments we compared ERPs<br />
to words passively read for meaning in bilingual participants of different proficiency<br />
levels and from different L1 language populations who were all adult learners of their L2.<br />
The first two experiments in chapter two investigate mechanisms underlying word<br />
recognition in second language learners. Are the mechanisms involved in word<br />
P | 24
Introduction<br />
recognition in L1 and L2 basically the same, that is, are L2 words integrated into a<br />
common set of lexical representations much like newly acquired words in L1? Or is there<br />
a different mechanism involved in L2 word recognition in adult learners? Are L2 words<br />
somehow associated with L1 equivalents as the RHM proposes and as explicit adult<br />
learning may induce?<br />
An integrated lexicon view of how languages are represented in the bilingual mind<br />
would appear to predict that L2 words should be processed much in the same way as L1<br />
words. According to this account, words in L2 would be influenced by the same factors<br />
that are known to influence L1 processing. By a pure frequency account of word<br />
processing, words in a second language learner’s L2 should behave like low-frequency<br />
words in L1 (Van Petten & Kutas, 1990; Münte et al., 2001). Much prior behavioral<br />
research shows that performance in L2 is slower and less accurate than performance in<br />
L1. This could be accounted for by the relative frequency characteristics of the words in<br />
each language with L2 items having a lower subjective frequency. Age-of-acquisition<br />
(AoA) is a factor that has been seen to influence word processing in research with<br />
monolinguals (Morrison & Ellis, 1995, 2000). In the case of adult learners of an L2 AoA<br />
could be a factor driving differences in performance to L1 and L2 words. In this case we<br />
would expect to see ERP language effects that are similar to AoA effects in<br />
monolinguals. Alternatively L2 words might be less interconnected to other L2 words,<br />
having smaller orthographic neighborhoods or fewer semantic associations than L1 words<br />
and these differences could be another factor that accounts for differences in<br />
performance. Our language effects in this case would mimic known monolingual<br />
neighborhood effects (Holcomb et al., 2002).<br />
On the other hand, any observed language effects could be the result of mechanisms<br />
that are specific to the developing bilingual lexicon such as the RHM would predict, i.e.,<br />
P | 25
Introduction<br />
lexical level links between L2 and L1 items that favor an L1 based processing of L2<br />
words. Effects due to this type of mechanism may have a previously unobserved ERP<br />
signature.<br />
In the third experiment in chapter three the same language processing comparisons<br />
are made in a population of proficient bilinguals. The comparison of results between<br />
learners and proficient bilinguals permits us to observe the evolution of language effects<br />
over time and structure questions about how L2 processing mechanisms may change as a<br />
learner becomes bilingual. We call on the RHM and the BIA model to describe the<br />
language effects of our different populations. Throughout this work we are not interested<br />
in adjudicating between these two models. We choose to accept their different strengths<br />
and to reconcile them with the observations that the BIA model better reflects lexical<br />
processing in relatively proficient bilinguals, while the RHM is a better model for<br />
describing lexical processing in evolving bilinguals.<br />
Having addressed the question of processing specificities of a bilingual’s two<br />
languages in chapter two we move on to a significant debate in the literature on bilingual<br />
language comprehension in chapter three. This debate opposes proponents of languageselective<br />
processing to those who posit non-selective access to a set of representations<br />
that is shared by both languages.<br />
According to one version of the language-selective hypothesis a switching<br />
mechanism guides the linguistic input to the appropriate set of language-specific lexical<br />
representations (Macnamara, 1967; Macnamara & Kushnir, 1972). When a bilingual is<br />
operating in a monolingual context, according to this hypothesis, the language of the<br />
incoming information is completely predictable so there should be no cross-language<br />
interference and the input will be treated by the appropriate set of language<br />
representations. In contrast, in the non-selective access hypothesis word representations<br />
P | 26
Introduction<br />
from both languages are activated by orthographic or phonological similarities with the<br />
input resulting in competition between form-related words both within and across<br />
languages (van Heuven et al., 1998).<br />
Previous experiments demonstrating cross-language interference have provided<br />
evidence that bilinguals cannot block interference from the irrelevant language. There<br />
exists the possibility, however, that the mere presence of words in both languages will<br />
prevent bilinguals from processing in a pure “monolingual” mode (Grosjean, 1988). In<br />
order to provide more convincing evidence in favor of non-selective access, crosslanguage<br />
interference must be demonstrated in conditions where there is no explicit<br />
activation of the both languages simultaneously. The research presented in chapter three<br />
is based on a study by van Heuven et al. (1998) that avoided this mixing of a bilingual’s<br />
two languages. Van Heuven et al. did not explicitly manipulate the presence or absence of<br />
stimuli that belong to both languages such as inter-lingual homophones or cognates.<br />
Rather, they manipulated the presence of potential cross-language interference in the<br />
form of phonologically or orthographically similar words from the other language.<br />
They, as did we, manipulated the number of orthographic neighbors of items in the<br />
non-task relevant (non-presented) language in order to observe any cross-language<br />
interference. We observed the effects of cross-language neighborhood size on the N250<br />
and the N400 components in two experiments with proficient French-English bilinguals<br />
and English monolinguals and presented the results of a simulation using the BIA+<br />
computational model. The non-selective hypothesis proposes that the initial feed-forward<br />
sweep of information from the linguistic input can make contact with lexical<br />
representations from both. This is a central hypothesis of the Bilingual Interactive-<br />
Activation model (Grainger & Dijkstra, 1992; van Heuven et al, 1998). We discuss the<br />
results of the bilingual study and the simulation within the framework of the BIA+ model.<br />
P | 27
Introduction<br />
The interactivity of a bilingual’s word representations is touched upon in chapter<br />
three and we continue to explore interactivity in the next chapter. Chapter four presents a<br />
study that exploits an oft used device for bilingual research, the cognate. Cognates are<br />
favored in bilingual research because they are a special case of interactivity, where form<br />
and meaning coincide across languages. Cognates are words like “table” that in French<br />
and English share orthography, much phonology and complete semantic overlap. Because<br />
of their shared form with L1 items, cognates, during L2 acquisition, could be a learner’s<br />
first foothold into the new lexicon. Presumably in the early stages of acquisition this<br />
would result in different patterns of processing for cognates and non-cognates while<br />
processing in L2. In the case of L1 processing, if cognates showed different patterns of<br />
processing when compared to non-cognates this would be evidence of L1 changing as a<br />
function of learning an L2 and also point to an integrated, interactive lexicon.<br />
In many behavioral studies cognate items have been shown to elicit different RT<br />
patterns than non-cognate items; they are recognized more rapidly than non-cognates,<br />
they have been shown to be translated more quickly than non-cognates, and stronger<br />
priming has been found for cognates than for non-cognates (Dijkstra, Grainger & van<br />
Heuven, 1999; Lemhöfer & Dijkstra, 2004; Lemhöfer, Dijkstra & Michel, 2004; de<br />
Groot, 1992; Sánchez-Casas, Davis, & García-Albea, 1992; van Hell & de Groot,1998; de<br />
Groot, Delmaar & Lupker, 2000; van Hell & Dijkstra, 2002; Cristoffanini, Kirsner, &<br />
Milech, 1986; Lalor & Kirsner, 2001; de Groot & Nas, 1991; Gollan, Forster, & Frost,<br />
1997; Voga & Grainger, 2007). The majority of previous behavioral studies that observed<br />
a cognate advantage did so in paradigms that tested processing in L2. What about cognate<br />
processing in L1? Studies in this direction have given mixed results (Caramazza &<br />
Brones, 1979; Gerard & Scarborough, 1989; van Hell & de Groot,1998; de Groot et al.,<br />
2000; van Hell & Dijkstra, 2002) making it appear that word recognition in L2 benefits<br />
P | 28
Introduction<br />
from cognate status, whereas word recognition in L1 may be somewhat immune to such<br />
influences. However effects of cognate status on word recognition in L1 have been<br />
documented. Van Hell and Dijkstra (2002) showed clear effects of cognate status during<br />
L1 processing and this while testing 2 groups of bilinguals of different proficiencies.<br />
They observed that only more fluent bilinguals showed clear effects of an L3 on L1<br />
processing while the less proficient bilinguals showed only a trend for these effects. They<br />
concluded that a certain level of proficiency is necessary in the bilinguals’ non-target<br />
language relative to their target language in order to observe effects on processing in the<br />
target language. In other words a bilingual must have enough fluency in L2 for L2 word<br />
status to influence L1 processing.<br />
We used electrophysiological measures once again to improve our chances of<br />
observing effects in our population of L2 learners. We sought evidence for mechanisms<br />
that would be the basis of the observed behavioral advantage in processing cognate words<br />
compared with non-cognates and this during processing in both languages. We again<br />
observed the N400 component that reflects processing of a word and can inform us of the<br />
relative difficulty of settling on a unique form-meaning interpretation of a word.<br />
Chapters two, three and four explore the bilingual lexicon in proficient bilinguals<br />
and second language learners primarily seeking insights about the similarities and<br />
differences of processing in L1 and L2, language selectivity and the level of integration<br />
and interactivity of the bilingual lexicon. Yet on a broader perspective, bilinguals and<br />
second-language learners provide an ideal testing ground for general theories of how<br />
form and meaning representations interact during language comprehension in general.<br />
The study presented in chapter five examines masked within-language repetition priming<br />
in L1 and L2 and masked translation repetition priming in order to examine the relative<br />
P | 29
Introduction<br />
contributions of form and meaning based representations in bilingual word processing.<br />
However, on a larger scale the results may speak to word recognition in general.<br />
The question of when semantic information becomes available during visual word<br />
recognition and the nature of the form-level processing that is necessary for that to occur<br />
during L1 processing and during L2 processing is addressed in this chapter.<br />
Non-cognate translation equivalents (e.g., the English word “tree” and its French<br />
translation “arbre”) arguably provide the closest possible semantic relation between two<br />
distinct word forms. They therefore provide an ideal testing ground for the interplay<br />
between form-level and semantic-level processes during visual word recognition.<br />
Studies investigating masked non-cognate translation priming effects have produced<br />
mixed results. Effects of masked translation priming have been consistently found in the<br />
direction of L1 priming L2 (De Groot & Nas, 1991; Gollan et al, 1997; Sanchez-Casas et<br />
al., 1992). However observing similar priming in the L2 to L1 direction has not been<br />
conclusive. Many studies failed to observe masked translation priming in this direction<br />
(Altarriba, 1992; Fox, 1996; Keatley & de Gelder, 1992; Keatley, Spinks, & de Gelder,<br />
1994; Kroll & Sholl, 1992). Robust masked translation priming has been reported for<br />
languages of different scripts in both directions (Finkbeiner, Forster, Nicol, & Nakamura,<br />
2004; Gollan et al., 1997; Voga & Grainger, 2007) but also found in a same script context<br />
by Grainger and Frenck-Mestre (1998) in proficient English-French bilinguals in the L2-<br />
L1 direction.<br />
In recent research, the masked priming paradigm has been used with ERP<br />
recordings to map out the time-course of component processes in visual word recognition<br />
(e.g., Grainger, Kiyonaga & Holcomb, 2006; Holcomb & Grainger, 2006; Kiyonaga,<br />
Grainger, Midgley & Holcomb et al., 2007). Holcomb and Grainger (2006) described a<br />
P | 30
Introduction<br />
cascade of ERP components elicited in masked priming paradigms. In chapter four we are<br />
interested in two of these components the N250 and the N400. The N250 is a negativegoing<br />
component which peaks near 250 ms. Holcomb and Grainger propose that the<br />
N250 reflects processes in visual word recognition where sublexical form representations<br />
(letters and letter combinations) are mapped onto the lexical system. The N400 has been<br />
observed in a host of word processing studies and described as reflecting some aspect of<br />
semantic processing (e.g., Kutas & Hillyard, 1980, 1984; Kounios & Holcomb, 1992,<br />
1994). In masked priming paradigms both the N250 and the N400 have been found to be<br />
modulated by the degree of form overlap across prime and target stimuli, with unrelated<br />
primes generating more negative waveforms than primes that are the same word as<br />
targets, or orthographically similar words.<br />
The study presented in chapter four compares within-language repetition priming<br />
and translation priming in order to obtain an improved picture of the time-course of form<br />
and meaning activation both within and between the lexical and semantic systems of<br />
second language learners. The N250 and N400 ERP components will be used to infer<br />
form-level and semantic-level influences on processing. Even if these two components<br />
reflect to some extent a combination of form and semantic influences, we expect formlevel<br />
influences to be greater on the N250, and semantic level influences to be greater on<br />
the N400.<br />
This then allows us to test the predictions of the RHM (Kroll & Stewart, 1994) and<br />
BIA model (Grainger & Dijkstra, 1992). In the RHM there are stronger links from L2<br />
lexical representations to L1 lexical representations than from L1 lexical representations<br />
to L2 lexical representations, and weaker links between L2 lexical representations and<br />
concepts than between L1 lexical representations and concepts. L2 primes should<br />
therefore affect form-level processing of the upcoming translation in L1, modulating the<br />
P | 31
Introduction<br />
N250 component, and in consequence the N400. L1 primes, on the other hand, should<br />
mostly affect semantic level processing of upcoming L2 translates, and therefore only<br />
modulate the N400 component. According to the BIA model, translation priming is<br />
always semantically mediated (i.e., there are no excitatory connections between lexical<br />
form representations of translation equivalents), hence most of the effects should be<br />
evident in the N400. Some priming effects are nevertheless expected on the N250<br />
component via feedback from semantics to lexical representations (Voga & Grainger,<br />
2007), and these should be most evident with L1 primes and L2 targets, simply because it<br />
is assumed that L1 words are processed more rapidly and efficiently than L2 words. If<br />
this is indeed the case, we should also observe smaller and later effects in L2-L2<br />
repetition priming than L1-L1 repetition priming.<br />
The four articles in this dissertation, while not a complete picture of bilingual word<br />
recognition, present studies using electrophysiological measures that inform the existing<br />
body of research. The following chapters address important questions about similarities<br />
and differences of processing in L1 and L2, language selectivity and the level of<br />
integration and interactivity of the bilingual lexicon and provide insight into the timecourse<br />
of bilingual lexical processing and the more general question of how form and<br />
meaning representations interact during language comprehension.<br />
P | 32
Effets de langage<br />
Chapitre 2<br />
Les effets de langue chez des<br />
apprenants en langue seconde et des bilingues<br />
confirmés grâce à la technique des potentiels évoqués*<br />
Dans cette étude nous avons examiné les effets de langue chez des apprenants en langue<br />
seconde avec la méthode des potentiels évoqués. Dans trois expériences nos sujets ont été<br />
confrontés à des listes constituées de mots appartenant à leur langue première (L1) et de<br />
mots appartenant à leur langue seconde (L2). Dans la première expérience, les<br />
participants dont la première langue était l’anglais et qui suivaient à l’Université des<br />
cours de français, ont lu deux listes de mots: l’une entièrement composée de mots anglais,<br />
l’autre de mots français. Nous avons observé un important effet de langue sur l’amplitude<br />
de la composante N400, effet qui commençait à se manifester dès 150 ms après la<br />
présentation du stimulus. Un pattern similaire a été mis en évidence au cours de<br />
l’expérience suivante dans laquelle les apprenants avaient pour L1 le français et pour L2<br />
l’anglais. La similitude des effets dans ces deux premières expériences nous portent à<br />
croire que les effets observés sont dus à des dominances linguistiques et non à la langue<br />
elle-même. Dans la troisième expérience, les participants étaient des bilingues<br />
compétents. Ils ont produit un pattern différent concernant les effets de langue, ces<br />
derniers semblent-ils modulés par le niveau de compétence linguistique du sujet. Ces<br />
résultats renforcent l’hypothèse selon laquelle des mécanismes spécifiques seraient<br />
impliqués dans la reconnaissance des mots au cours des phases précoces d’acquisition de<br />
la L2 chez des apprenants tardifs en comparaison à la reconnaissance des mots en L1.<br />
Mots-clefs : traitement des mots visuel, bilinguisme, N400<br />
________________________<br />
* Article in press, Journal of Neurolinguistics, August 2008<br />
P | 33
ARTICLE IN PRESS<br />
+ MODEL<br />
Journal of Neurolinguistics xx (2008) 1e20<br />
www.elsevier.com/locate/jneuroling<br />
Language effects in second language learners and<br />
proficient bilinguals investigated with event-related<br />
potentials *<br />
Katherine J. Midgley a,b, *, Phillip J. Holcomb a , Jonathan Grainger c<br />
a Department of Psychology, Tufts University, Medford, MA, USA<br />
b Université de Provence - <strong>Aix</strong>-<strong>Marseille</strong> I, <strong>Marseille</strong>, France<br />
c CNRS, <strong>Marseille</strong>, France<br />
Received 30 March 2008; received in revised form 6 August 2008; accepted 11 August 2008<br />
Abstract<br />
The present study examines language effects in second language learners. In three experiments<br />
participants monitored a stream of words for occasional probes from one semantic category and ERPs<br />
were recorded to non-probe critical items. In Experiment 1 L1 English participants who were university<br />
learners of French saw two lists of words blocked by language, one in French and one in English. We<br />
observed a large effect of language that mostly affected amplitudes of the N400 component, but starting as<br />
early as 150 ms post-stimulus onset. A similar pattern was found in Experiment 2 with L1 French and L2<br />
English, showing that the effect is due to language dominance and not language per se. Experiment 3<br />
found that proficient French/English bilinguals exhibited a different pattern of language effects showing<br />
that these effects are modulated by proficiency. These results lend further support to the hypothesis that<br />
word recognition during the early phases of L2 acquisition in late learners of L2 involves a specific set of<br />
mechanisms compared with recognition of L1 words.<br />
Ó 2008 Elsevier Ltd. All rights reserved.<br />
Keywords: Visual word processing; Bilingualism; N400<br />
* This research was supported by grant numbers HD043251 & HD25889.<br />
* Corresponding author. Department of Psychology, Tufts University, 490 Boston Avenue, Medford, MA 02155, USA.<br />
Tel.: þ1 617 627 3521; fax: þ1 617 627 3181.<br />
E-mail address: kj.midgley@tufts.edu (K.J. Midgley).<br />
0911-6044/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.<br />
doi:10.1016/j.jneuroling.2008.08.001<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
2 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
What are the basic mechanisms underlying word recognition in a second language (L2) in<br />
late learners of a foreign language, and to what extent do they overlap with the mechanisms<br />
involved in word recognition in the first language (L1)? One possibility is that the mechanisms<br />
are basically the same, and L2 words are integrated into a common set of lexical representations<br />
much like newly acquired words in L1. This would appear to be an unlikely possibility for late<br />
learners of a second language, at least in the early phases of L2 acquisition, for several reasons.<br />
First, anecdotally, many L2 learners report using translation into L1 as a general heuristic for<br />
processing L2 words. Second, given the relatively large number of cognates (words that share<br />
form and meaning in the two languages) in languages such as English and French, it would be<br />
more economical to establish L2eL1 formeform associations rather than creating new<br />
formemeaning associations. This is the position adopted by the revised hierarchical model<br />
(RHM) of word recognition in bilinguals proposed by Kroll and Stewart (1994) (<strong>Fig</strong>. 1).<br />
Nevertheless, the alternative hypothesis, that L2 words are basically processed in the same<br />
way as L1 words, would fit with an account of lexical representation in bilinguals according to<br />
which words from both languages are stored together. Indeed, there is an abundance of<br />
behavioral evidence in favor of a language non-selective, integrated lexicon view of written<br />
word comprehension in bilinguals, at least when the two languages share the same alphabet<br />
(e.g., Beauvillain & Grainger, 1987; Brysbaert, Van Dyck, & Van de Poel, 1999; De Groot,<br />
Delmaar, & Lupker, 2000; Dijkstra, Grainger, & van Heuven, 1999; Dijkstra, van Heuven, &<br />
Grainger, 1998; Dijkstra, Timmermans, & Schriefers, 2000; Dyer, 1973; Guttentag, Haith,<br />
Goodman, & Hauch, 1984; Lemhöfer et al., 2008; Nas, 1983). According to this view, bottomup<br />
processing of a printed word proceeds independently of the language to which that word<br />
belongs, up to the level of whole-word representations stored in a word-form lexicon that is<br />
common to both languages, and possibly beyond. Perhaps the strongest evidence in favor of this<br />
non-selective approach to bilingual lexical access was provided by van Heuven, Dijkstra, and<br />
Grainger (1998) who showed that word recognition in one language was affected by the target<br />
word’s orthographic neighbors in the other language. The major conclusion from this and<br />
<strong>Fig</strong>. 1. The revised hierarchical model (Kroll & Stewart, 1994).<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
3<br />
related behavioral research is that bilinguals cannot completely stop interference from the nontarget<br />
language (see Midgley, Holcomb, van Heuven, & Grainger (in press), for further<br />
evidence from ERPs). These results counter the predictions of the input-switch mechanism first<br />
proposed by Macnamara and Kushnir (1972), according to which the words of the bilingual’s<br />
two languages are kept apart (in separate stores) and the incoming sensory input is directed to<br />
the appropriate set of words as a function of context.<br />
The non-selective access/integrated lexicon view of how languages are represented in the<br />
bilingual mind would appear to predict that L2 words should be processed much in the same<br />
way as L1 words. Therefore, according to this account, words in a second language learner’s L2<br />
should behave like low-frequency words in the L1. This therefore contrasts with the predictions<br />
of the RHM (Kroll & Stewart, 1994) that assigns very different mechanisms for processing<br />
words in L2 compared with L1. Of course, behavioral results show that performance in L2 is<br />
slower and less accurate than performance in L1, but this could be attributed to the frequency<br />
characteristics of the words in each language. Exposure to L2 words is generally much lower<br />
than exposure to words in L1, especially for beginning bilinguals. Therefore L2 words will have<br />
lower subjective frequencies than L1 words, and these differences in subjective frequency could<br />
be driving the behavioral effects. Furthermore, L2 words are learned after the majority of L1<br />
words have already been learned, so age-of-acquisition (AoA) could be another factor driving<br />
observed differences in performance to L1 and L2 words. Finally, L2 words might be less<br />
interconnected to other L2 words (e.g., have smaller orthographic neighborhoods) than L1<br />
words and these differences might also account for differences in performance.<br />
Therefore, one key question that emerges from the above review of the behavioral literature<br />
on word recognition in bilinguals is whether L2 words show any processing specificities<br />
compared with L1 words that are quantitatively or qualitatively different than those attributable<br />
to subjective frequency, AoA or neighborhood density. In order to address this issue, the present<br />
study compared ERPs to words in L1 and L2 in lists that are blocked by language, hence<br />
avoiding issues of predictability and language-switching (see Chauncey, Grainger, & Holcomb,<br />
2008, for a recent ERP study of language-switching). A number of previous studies have<br />
presented words in participants’ L1 and L2 and compared various ERP effects such as priming<br />
or anomaly detection between languages (e.g., De Bruijn, Dijkstra, Chwilla, & Schriefers,<br />
2001; Hahne & Friederici, 2001; Kerkhofs, Dijkstra, Chwilla, & de Bruijn, 2006; Kotz, 2001;<br />
Kotz & Elston-Guttler, 2004; Moreno & Kutas, 2005; Phillips, Klein, Mercier, & de Boysson,<br />
2006; Phillips, Segalowitz, O’Brien, & Yamasaki, 2004; Weber-Fox & Neville, 1996).<br />
Surprisingly though, none of these studies has systematically compared ERPs to words in L1<br />
and L2. ERPs would appear to be ideally suited for discerning word level differences between<br />
languages, both because of their excellent temporal resolution as well as their ability to<br />
differentiate multiple sensory and cognitive influences in a single experiment.<br />
There are also comparatively few ERP studies that have examined second language learners<br />
in the process of formal classroom instruction. One notable exception is a study by<br />
McLaughlin, Osterhout, and Kim (2004). These authors showed that the N400 component to<br />
visually presented items in L2 (French) differentiated between words and pseudowords after<br />
only a very brief period of L2 learning (14 h) though semantic priming effects were not<br />
observed at this point in L2 learning. This result suggests that ERPs are sensitive to lexical<br />
representations laid down in the very initial phases of L2 learning. McLaughlin et al. did not<br />
report direct comparisons between L2 and L1, but a study by Alvarez, Grainger, and Holcomb<br />
(2003) did directly compare ERPs recorded to Spanish (L2) and English (L1) words in native<br />
English speakers enrolled in intermediate university Spanish classes. The finding of most<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
interest here was that starting as early as 150 ms and continuing on as late as 700 ms there were<br />
differences in the time course of ERP effects between L1 and L2. However, because their<br />
design included a repetition factor and all of their reported effects of language interacted with<br />
this factor, it is not clear whether there were pure differences between languages when<br />
participants read words in each language for the first time (i.e., prior to a repetition). Examination<br />
of their <strong>Fig</strong>s. 2 and 3 does, however, suggest that L1 words produced somewhat larger<br />
N400s than L2 words.<br />
In the current study we examined differences in the ERPs generated by L1 and L2 words for<br />
American students of French (Experiment 1), French students of English (Experiment 2) and<br />
proficient FrencheEnglish bilinguals (Experiment 3). Table 1 summarizes the different level of<br />
expertise in L1 and L2 of these three groups of participants.<br />
1. Experiment 1 e learners of French<br />
ARTICLE IN PRESS<br />
+ MODEL<br />
4 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
In this experiment ERP language effects for L1 and L2 items were measured during<br />
a passive reading for meaning task. Participants were native speakers of American English<br />
selected from beginning and intermediate French courses at university and none had had<br />
a significant immersion experience. To better match these participants to the native French<br />
participants learning English in Experiment 2, an additional criterion for selection was<br />
a relatively early (by US educational standards) and ongoing exposure to French.<br />
1.1. Methods<br />
1.1.1. Participants<br />
Twenty-two participants from Tufts University (17 female, mean age ¼ 19.8, SD ¼ 1.7) who<br />
were enrolled in French courses were paid for their participation. All were right handed (Edinburgh<br />
Handedness Inventory e Oldfield, 1971) and had normal or corrected-to-normal visual acuity with<br />
no history of neurological insult or language disability. English was reported to be the first language<br />
learned by all participants (L1) and French their primary second language (L2). Participants began<br />
their study of French on average at the age of 12.1 years (range 5e18 years, SD ¼ 2.5).<br />
<strong>Fig</strong>. 2. A critical trial in the L1 block.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
5<br />
<strong>Fig</strong>. 3. Electrode montage and electrode analysis sites (connected by lines).<br />
Participants’ auto-evaluation of English and French language skills were surveyed by<br />
questionnaire. On a seven-point Likert scale (1 ¼ unable; 7 ¼ expert) participants reported their<br />
abilities to read, speak and comprehend English and French as well as how frequently they read<br />
in both languages (1 ¼ rarely; 7 ¼ very frequently). The overall average of self-reported<br />
language skills in English was 7.0 (SD ¼ 0.0) and in French was 4.2 (SD ¼ 0.9). Our participants<br />
reported their average frequency of reading in English as 6.4 (SD ¼ 0.9) and in French as<br />
3.6 (SD ¼ 1.1). See Table 1 for a comparison of participants across the 3 experiments.<br />
1.1.2. Stimuli<br />
The critical stimuli for this experiment were 80 four to seven letter non-cognate morphemically<br />
simple English words and their translations into French. The English items had a mean<br />
Table 1<br />
Average self-evaluation on a seven-point scale of overall language skills and reading frequency of the three groups of<br />
participants in Experiments 1e3<br />
Experiment 1 Experiment 2 Experiment 3<br />
L1 e E L2 e F L1 e F L2 e E L1 e F L2 e E<br />
Language skills 7.0 (0.0) 4.2 (0.9) 6.8 (0.4) 3.9 (1.1) 6.9 (0.3) 5.7 (1.0)<br />
Reading 6.4 (0.9) 3.6 (1.1) 6.4 (1.2) 2.7 (1.3) 6.3 (1.1) 5.8 (1.5)<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
6 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
log frequency (CELEX, 1993) of 1.73 (SD ¼ 0.63, range 0.00e3.09). The French items had<br />
a mean log frequency (New, Pallier, Ferrand, & Matos, 2001) of 1.63 (SD ¼ 0.53, range 0.24e<br />
2.65). The log frequencies of the items in the two languages were not found to be statistically<br />
different (t(79) ¼ 1.16, p ¼ 0.25). The log frequencies of the translation equivalents correlated<br />
highly (r ¼ 0.71, p < 0.01). The average length of the English items was 4.7 letters (SD ¼ 0.97)<br />
while the average length of the French items was 5.5 (SD ¼ 1.09).<br />
Forty animal names were selected as probe items and an additional 200 non-critical words<br />
were used as fillers for each language block. These non-critical items and the probe items were,<br />
like the critical items, four to seven letters in length (mean length: 5.5 letters for English items<br />
(SD ¼ 1.00), 5.8 letters for French items (SD ¼ 0.99)). These items had slightly lower log<br />
frequencies than the critical items (average log frequency for English items: 1.23 (SD ¼ 0.52),<br />
French: 1.17 (SD ¼ 0.50)).<br />
The 80 English items were divided into two lists of 40 items and each participant saw one of<br />
the two lists. The same was done with the French items such that there was no repetition of<br />
translation equivalents. That is if a participant saw ‘‘tree’’ in the English block they would not<br />
see ‘‘arbre’’ in the French block. Both an English block and a French block were presented to<br />
each participant in a counterbalanced fashion across participants. The two blocks were thus<br />
comprised of 40 animal probes, 40 critical items and 200 fillers. All items in a block were in<br />
one language and there was no repetition of translation equivalents across blocks.<br />
1.1.3. Procedure<br />
Animal names served as probe items in a go/no-go semantic categorization task in which<br />
participants were instructed to rapidly press a single button whenever they detected an animal<br />
name. Participants were told to read all other words passively without responding (i.e., critical<br />
stimuli did not require an overt response). Stimulus lists were a pseudorandom mixture of<br />
critical trials, fillers and probe trails. A practice session was administered before the main<br />
experiment to familiarize the participant with the task.<br />
The visual stimuli were presented on a 19 00 monitor located directly in front of the participant<br />
at a distance of approximately 150 cm. Stimuli were displayed at high contrast as white<br />
letters on a black background in the Verdana font (letter matrix 20 pixels wide 40 pixels tall).<br />
Each trial began with a fixation cross followed by a blank screen and then an item. The item<br />
was on screen for 300 ms, followed by 1000 ms of blank screen and then a 2500 ms blink<br />
symbol (see <strong>Fig</strong>. 2). The participants were instructed to blink only during this blink symbol.<br />
This symbol was followed by 500 ms of blank screen after which the next trial began with<br />
a fixation cross.<br />
1.1.4. EEG recording procedure<br />
Participants were seated in a comfortable chair in a sound attenuated darkened room. The<br />
electroencephalogram (EEG) was recorded from 29 active tin electrodes held in place on the<br />
scalp by an elastic cap (Electrode-Cap International e see <strong>Fig</strong>. 3). In addition to the 2 scalp<br />
sites, additional electrodes were attached to below the left eye (LE, to monitor for vertical eye<br />
movement/blinks), to the right of the right eye (VE, to monitor for horizontal eye movements),<br />
over the left mastoid bone (A1, reference) and over the right mastoid bone (recorded actively to<br />
monitor for differential mastoid activity). All EEG electrode impedances were maintained<br />
below 5 kU (impedance for eye electrodes was less than 10 kU and for the references electrodes<br />
less than 2 kU). The EEG was amplified by an SA Bioamplifier with a bandpass of 0.01 and<br />
40 Hz and the EEG was continuously sampled at a rate of 200 Hz throughout the experiment.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
7<br />
1.1.5. Data analysis<br />
Averaged ERPs were formed off-line from trials free of ocular and muscular artifact (9% of<br />
trials rejected for artifact) and were lowpass filtered at 15 Hz. The approach to data analysis<br />
involved the selection of a subset of the 29 scalp sites (see <strong>Fig</strong>. 3). Average waveforms were<br />
formed for the two levels of LANGUAGE (L1 vs. L2), three levels of POSTERIOReANTERIOR<br />
(posterior, central and anterior) and three levels of LATERALITY (right, midline and left). The<br />
main analysis approach involved measuring mean amplitudes in three temporal epochs;<br />
150e300 ms, 300e500 ms, 600e800 ms capturing activity prior, during and after the typical<br />
N400. Separate repeated measures analyses of variance (ANOVAs) were used to analyze the<br />
data in each of these three epochs. 1 The Geisser and Greenhouse (1959) correction was applied<br />
to all repeated measures with more than one degree of freedom in the numerator.<br />
1.2. Results<br />
1.2.1. Visual inspection of ERPs<br />
The ERPs time locked to critical target items are plotted in <strong>Fig</strong>. 4. Plotted in <strong>Fig</strong>. 5 are the<br />
voltage maps resulting from subtracting L2 from L1 ERPs (these plots reveal the scalp<br />
distribution of differences between languages). As can be seen in <strong>Fig</strong>. 4, early in the waveforms<br />
there is a small negativity (sharpest at anterior sites) peaking around 100 ms (N1), this is<br />
followed by a prominent positivity peaking near 200 ms (P2). The P2 is visible at all sites, but is<br />
larger at more anterior electrodes. Up to this point the ERPs for the two languages are quite<br />
similar. However, starting just after the peak of the P2 there are clear differences in the ERPs<br />
for the two languages. These can be seen most clearly in the voltage maps in <strong>Fig</strong>. 5. At posterior<br />
sites there are differences on the negativity that starts at about 250 ms and which peaks at about<br />
400 ms (N400). Here the difference appears to be due to a prolonged attenuation of the N400<br />
for L2 compared to L1 items starting as early as 200 ms and lasting through 500 ms. Also<br />
evident is a reduction and/or delay in the negativity which peaks at about 300 ms at anterior<br />
sites in L1 and at about 450 ms in L2 and a subsequent larger anterior sustained negativity for<br />
L2 compared to L1 words.<br />
1.2.2. Analyses of ERP data<br />
1.2.2.1. 150e300 ms epoch. In this epoch there was a main effect of LANGUAGE (F(1,21) ¼ 8.56,<br />
p ¼ 0.008) with L1 items being more negative-going than L2 items.<br />
1.2.2.2. 300e500 ms epoch. In the traditional N400 epoch there were again differences between<br />
the languages, however, these effects differed as a function of scalp site (LANGUAGE POSTERIORe<br />
ANTERIOR, F(2,42) ¼ 15.75, p ¼ 0.0002; LANGUAGE LATERALITY, F(2,42) ¼ 5.76, p ¼ 0.009).<br />
Examination of <strong>Fig</strong>. 4 suggests that the LANGUAGE POSTERIOReANTERIOR interaction reflects that<br />
at anterior sites L2 items are slightly only more negative-going than L1 items, but at posterior<br />
sites L1 items are substantially more negative-going than L2 items. The LANGUAGE LATERALITY<br />
interaction indicates that L1 was quite a bit more negative than L2 at midline and right hemisphere<br />
sites compared to left hemisphere sites.<br />
1 We performed a first pass analysis including the factor of order to test for differential effects of which target<br />
language block occurred first for all three epochs and for all three Experiments. There were no interactions involving the<br />
order and language (all Fs < 2). In all of the analyses reported we collapsed across this factor.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
8 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
<strong>Fig</strong>. 4. ERP language effects from nine sites (see <strong>Fig</strong>. 3 for electrode locations) for native English speakers reading<br />
critical words in their L1 (English) and L2 (French) in Experiment 1.<br />
1.2.2.3. 600e800 ms epoch. In the final epoch there were again language effects that differed<br />
as a function of scalp site (LANGUAGE POSTERIOReANTERIOR, F(2,42) ¼ 11.64, p ¼ 0.001). <strong>Fig</strong>. 4<br />
suggests that this interaction reflects that L2 was more negative-going than L1 at anterior sites<br />
while at posterior sites L1 was still slightly more negative-going than L2.<br />
1.2.3. Behavioral data<br />
Participants detected on average 98.2% (SD ¼ 2.7%) of probes in the L1 block. In the L2 block<br />
the participants detected 78.9% of probes (SD ¼ 10.5%). Participants produced false alarms on an<br />
average of 0.5 items (SD ¼ 1.1) in L1 (0.2%) and on 7.6 items (SD ¼ 7.8) in L2 (3.8%).<br />
1.3. Discussion<br />
Experiment 1 tested learners of French still at a relatively early point in acquiring their<br />
second language. ERPs were time locked to passively read words in L1 (English) and L2<br />
(French). Both languages elicited a similar pattern of early ERP components. However, in the<br />
time frame of the N400 component there were clear effects of language dominance. At<br />
posterior electrode sites L1 items were associated with larger N400-like negativities starting as<br />
early as 200 ms and continuing on through 500 ms. At anterior sites L1 items started off more<br />
negative-going than L2 items (between 200 and 400 ms), but after 400 ms L2 items became<br />
more negative than L1 items. This latter effect suggests that the anterior negativity produced by<br />
words in both languages is delayed by some 150 ms in L2.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
9<br />
<strong>Fig</strong>. 5. Voltage maps computed by subtracting ERPs recorded to L2 words (French) from ERPs recorded from L1 words<br />
(English) in Experiment 1. Note that computing the subtraction in this direction (L1eL2) results in the larger early<br />
posterior negativity from L1 ERPs (see <strong>Fig</strong>. 4) producing a larger effect at posterior sites (blue areas at 300 and 400 ms)<br />
and the larger late anterior negativity from L2 ERPs (see <strong>Fig</strong>. 4) producing an effect at anterior sites (red areas at 600,<br />
700 and 800 ms).<br />
2. Experiment 2<br />
In Experiment 1, L2 speakers of French showed smaller posterior negativities and delayed/<br />
prolonged anterior negativities to words read in their L2 compared to their L1. It seems likely<br />
that these effects are due to these participants less competent language status in L2, although it<br />
is possible that they are due entirely or in part to inherent differences between French and<br />
English words. Neville et al. have demonstrated that there are both similarities and differences<br />
in ERP effects for users of different languages (English and ASL) and that some of these effects<br />
can be attributed to language competence but that some effects reflect basic differences in the<br />
languages themselves (Neville et al., 1997; Neville, Mills, & Lawson, 1992). In order to test if<br />
the language effects seen in Experiment 1 are due to specific characteristics of English and<br />
French or reflect the different level of participants’ competence in English and French, in<br />
Experiment 2 we tested native speakers of French that are learners of English in a similar<br />
paradigm with different items.<br />
2.1. Methods<br />
2.1.1. Participants<br />
Eighteen participants (14 female, mean age ¼ 22.1, SD ¼ 4.7) were recruited from the<br />
University of Provence and paid for their participation. All were right handed (Edinburgh<br />
Handedness Inventory e Oldfield, 1971) and had normal or corrected-to-normal visual acuity<br />
with no history of neurological insult or language disability. French was reported to be the first<br />
language learned by all participants (L1) and English their primary second language (L2).<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
+ MODEL<br />
10 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
Participants began their study of English in their sixth year of primary school at approximately<br />
the age of 12 years, as is customary in the French school system.<br />
Participants’ auto-evaluation of English and French language skills were surveyed by<br />
questionnaire. On a seven-point Likert scale (1 ¼ unable; 7 ¼ expert) participants reported their<br />
abilities to read, speak and comprehend English and French as well as how frequently they read<br />
in both languages (1 ¼ rarely; 7 ¼ very frequently). The overall average of self-reported<br />
language skills in French was 6.8 (SD ¼ 0.4) and in English was 3.9 (SD ¼ 1.1). Our participants<br />
reported their average frequency of reading in French as 6.4 (SD ¼ 1.2) and in English as<br />
2.7 (SD ¼ 1.3). See Table 1 for a comparison of participants across the 3 experiments.<br />
2.1.2. Stimuli<br />
The critical stimuli for this experiment were 74 four and five letter English words and 74<br />
four and five letter French words. The English items had a mean log frequency (CELEX, 1993)<br />
of 1.08 (SD ¼ 0.489, range 0.301e2.158). The French items had a mean log frequency (New<br />
et al., 2001) of 1.16 (SD ¼ 0.611, range 0.000e2.615). These log frequencies were not found to<br />
be statistically different (t(73) ¼ 0.654, p ¼ 0.51). The average length of the English items was<br />
4.34 letters (SD ¼ 0.48) while the average length of the French items was 4.43 (SD ¼ 0.50).<br />
The critical stimuli were morphemically simple items. Stimulus lists were formed by mixing<br />
the above critical items with probe words which were all members of the semantic category of<br />
‘‘body parts’’ (20% of trials). These probe items were, like the critical items, four and five<br />
letters in length (mean length: 4.5 letters (SD ¼ 0.51) for both English and French items). These<br />
items had slightly higher log frequencies than the critical items (average log frequency for<br />
English items: 1.53 (SD ¼ 0.80), French: 1.69 (SD ¼ 0.61)).<br />
2.1.3. Procedure<br />
The procedure was the same as Experiment 1. Two blocks were presented in counterbalanced<br />
order. In the L1 block only French items were presented and in the L2 block only<br />
English items were presented.<br />
The laboratory in France was designed to be as similar as possible to the laboratory at Tufts<br />
University. The same EEG recording system is used and the same software is used in Experiments<br />
1e3 as well as the same experimenters.<br />
The visual stimuli were presented on a 15 00 monitor located directly in front of the participant<br />
at a distance of approximately 150 cm. Stimuli were displayed at high contrast as white<br />
letters on a black background in the Verdana font (letter matrix 30 pixels wide 60 pixels tall).<br />
All else was as in Experiment 1. See <strong>Fig</strong>. 2 for a typical L2 trial.<br />
2.1.4. Data analysis<br />
Data analysis was identical to Experiment 1. Trials containing ocular and muscular artifact<br />
were excluded from analysis (13% of trials rejected due to artifact).<br />
2.2. Results<br />
ARTICLE IN PRESS<br />
2.2.1. Visual inspection of ERPs<br />
The ERPs time locked to critical target items are plotted in <strong>Fig</strong>. 6. Plotted in <strong>Fig</strong>. 7 are the<br />
voltage maps resulting from subtracting L2 from L1. As can be seen, and similar to Experiment 1,<br />
early in the waveforms there is a small negativity (sharpest at anterior sites) peaking around<br />
100 ms (N1). This is followed by a prominent positivity peaking near 200 ms (P2). The P2 is<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
11<br />
<strong>Fig</strong>. 6. ERP language effects from nine sites (see <strong>Fig</strong>. 2 for electrode locations) for native French speakers reading<br />
critical words in their L1 (French) and L2 (English) in Experiment 2.<br />
visible at all sites, but is larger at more anterior electrodes. Again, as in Experiment 1, up to the<br />
peak of the P2 the ERPs for the two languages are quite similar. However, starting near the peak of<br />
the P2 there are clear differences in the ERPs for the two languages. Most evident across the scalp,<br />
but most notable at posterior sites there are differences on the negativity starting as early as<br />
200 ms and peaking at about 450 ms (N400). This difference appears to be due to a prolonged<br />
attenuation of the N400 for L2 compared to L1 items (see the large central/posterior blue area in<br />
<strong>Fig</strong>. 7 between 200 and 600 ms). Another difference is a reduction in the anterior negativity in L2<br />
compared to L1 especially after 400 ms (the red area in <strong>Fig</strong>. 7 between 500 and 800 ms).<br />
2.2.2. Analyses of ERP data<br />
2.2.2.1. 150e300 ms epoch. In this epoch there was a main effect of LANGUAGE (F(1,21) ¼ 7.42,<br />
p ¼ 0.014) with L1 items being more negative-going than L2 items.<br />
2.2.2.2. 300e500 ms epoch. In this epoch there were differences between the languages<br />
(F(1,17) ¼ 17.83, p ¼ 0.001) as well as an interaction between LANGUAGE POSTERIOReANTE-<br />
RIOR LATERALITY (F(4,68) ¼ 5.45, p ¼ 0.002). <strong>Fig</strong>. 6 suggests that this interaction reflects that<br />
the difference between languages (L1 more negative than L2) tended to be larger at more<br />
posterior and midline sites.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
12 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
<strong>Fig</strong>. 7. Voltage maps computed by subtracting ERPs recorded to L2 words (English) from ERPs recorded from L1 words<br />
(French) in Experiment 2. Note that as in <strong>Fig</strong>. 4 there is a large early central/posterior effect resulting from L1 ERPs<br />
being more negative-going (blue areas at 200, 300, 400 and 500 ms) and a large late anterior effect resulting from L2<br />
ERPs being more negative at anterior sites (red areas at 500, 600, 700 and 800 ms).<br />
2.2.2.3. 600e800 ms epoch. In the final epoch there were again language effects that differed<br />
as a function of scalp site (LANGUAGE POSTERIOReANTERIOR LATERALITY (F(4,68) ¼ 8.60,<br />
p ¼ 0.0002)). <strong>Fig</strong>. 6 suggests that at anterior sites, especially along the midline and over the<br />
right hemisphere, L2 was more negative-going than L1, while at posterior sites, especially<br />
along the midline, L1 was still more negative than L2.<br />
2.2.3. Behavioral data<br />
Participants detected on average 87.0% (SD ¼ 10.9%) of probes in the L1 block. In the L2<br />
block the participants detected 52.9% of probes (SD ¼ 21.7%). Participants produced false<br />
alarms on an average of 1.4 items (SD ¼ 1.1) in L1 (1.9%) and on 1.0 items (SD ¼ 1.0) in L2<br />
(1.4%).<br />
2.3. Discussion<br />
Experiment 2 tested French learners of English still at a relatively early point in acquiring<br />
their second language. As in Experiment 1, ERPs were time locked to passively read words but<br />
now French is the participants’ L1 and English their L2. Again, both languages elicited<br />
a similar pattern of early ERP components. Just after the peak of the P2, at the beginning of the<br />
time frame of the N400 component there were again effects of language dominance. At central<br />
and posterior electrode sites L1 items were associated with larger N400-like negativities<br />
starting as early as 200 ms and continuing on through 700 ms. At anterior sites L1 items started<br />
off more negative-going than L2 items (between 200 and 500 ms) but after 500 ms L2 items<br />
became more negative than L1 items especially at midline and right hemisphere sites. This<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
13<br />
latter effect is consistent with the possibility that the anterior negativity produced by words in<br />
both languages is delayed by some 150 ms in L2. The overall pattern of effects in Experiments<br />
1 and 2 was quite similar suggesting that the observed differences in L1 and L2 are not due to<br />
specific attributes of either language, but rather are more general effects of language competence<br />
in L1 and L2. This hypothesis is further tested in Experiment 3 with more balanced<br />
FrencheEnglish bilinguals.<br />
3. Experiment 3<br />
In order to provide a further test of the hypothesis that the language effects obtained in<br />
Experiments 1 and 2 are due to different levels of proficiency in each language, Experiment 3<br />
tests a group of proficient FrencheEnglish bilinguals in the same paradigm as Experiment 2.<br />
3.1. Methods<br />
3.1.1. Participants<br />
Twenty participants (13 female, mean age ¼ 23.0, SD ¼ 4.7) were recruited as proficient<br />
bilinguals and paid for their participation. All were right handed (Edinburgh Handedness<br />
Inventory e Oldfield, 1971) and had normal or corrected-to-normal visual acuity with no<br />
history of neurological insult or language disability. French was reported to be the first language<br />
learned by all participants (L1) and English their primary second language (L2). Participants<br />
began their study of English in their sixth year of primary school at approximately the age of 12<br />
years, as is customary in the French school system.<br />
Participants’ auto-evaluation of French and English language skills were surveyed by<br />
questionnaire. On a seven-point Likert scale (1 ¼ unable; 7 ¼ expert) participants reported their<br />
abilities to read, speak and comprehend English and French as well as how frequently they read<br />
in both languages (1 ¼ rarely; 7 ¼ very frequently). The overall average of self-reported<br />
language skills in French was 6.9 (SD ¼ 0.3) and in English was 5.7 (SD ¼ 1.0). Our participants<br />
reported their average frequency of reading in French as 6.3 (SD ¼ 1.1) and in English as<br />
5.8 (SD ¼ 1.5). See Table 1 for a comparison of participants across the 3 experiments.<br />
3.1.2. Stimuli and procedure<br />
The stimuli and procedure for this experiment were the same as in Experiment 2.<br />
3.1.3. Data analysis<br />
Data analysis was identical to Experiment 1. Trials containing ocular and muscular artifact<br />
were excluded from analysis (7% of trials rejected).<br />
3.2. Results<br />
3.2.1. Visual inspection of ERPs<br />
The ERPs time locked to critical target items are plotted in <strong>Fig</strong>. 8. Plotted in <strong>Fig</strong>. 9 are the<br />
voltage maps resulting from subtracting L2 (English) from L1 (French) ERPs. As can be seen,<br />
and similar to Experiments 1 and 2, early in the waveforms there is a small negativity (sharpest<br />
at anterior sites) peaking around 100 ms (N1). This is followed by a prominent positivity<br />
peaking near 200 ms (P2). In Experiments 1 and 2 there were prominent differences between<br />
languages that appeared to begin just after the P2 (<strong>Fig</strong>s. 4 and 6). These effects are not so<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
14 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
<strong>Fig</strong>. 8. ERP language effects from nine sites (see <strong>Fig</strong>. 3 for electrode locations) for native French speakers reading<br />
critical words in their L1 (French) and L2 (English) in Experiment 3.<br />
apparent in Experiment 3 (<strong>Fig</strong>. 8). Still evident, however, is the delay in the anterior negativity<br />
which peaks after 350 ms at the frontal sites in L1 and just after 450 ms in L2. Moreover, this<br />
negativity appears to now be larger in L2 than L1. Another difference between the ERPs in<br />
Experiment 3 and those in Experiments 1 and 2 is that in Experiment 3 there are no longer large<br />
differences on the negativity that peaks between 400 and 500 ms (N400) at posterior sites. Here<br />
the peak of the negativity is roughly equivalent for the two languages. It is only in the epoch<br />
following the N400 where there are small differences between languages (L2 more negative<br />
than L1).<br />
3.2.2. Analyses of ERP data<br />
3.2.2.1. 150e300 ms epoch. Consistent with the visual observation of <strong>Fig</strong>. 8 there were no<br />
significant effects of LANGUAGE in this epoch.<br />
3.2.2.2. 300e500 ms epoch. In this epoch there were differences between the languages as<br />
a function of scalp site (LANGUAGE POSTERIOReANTERIOR (F(2,38) ¼ 5.21, p ¼ 0.022)). Examination<br />
of <strong>Fig</strong>. 8 suggests that this interaction was sensitive to the difference between languages<br />
from the front to the back of the head.<br />
3.2.2.3. 600e800 ms epoch. In the final epoch there were again language effects that differed<br />
as a function of scalp site (LANGUAGE LATERALITY (F(2,38) ¼ 4.20, p ¼ 0.031)). <strong>Fig</strong>. 8 suggests<br />
that differences between the languages were greater over midline and right hemisphere sites.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
15<br />
<strong>Fig</strong>. 9. Voltage maps computed by subtracting ERPs recorded to L2 words (English) from ERPs recorded from L1 words<br />
(French) in Experiment 3. Note that the early posterior is much smaller than in <strong>Fig</strong>s. 5 and 7 reflecting the fact that L1<br />
ERPs are only slightly more negative-going than L2 ERPs (pale blue area at 400 ms), but that there is still a large late<br />
anterior effect resulting from L2 ERPs being more negative at anterior sites (red areas at 500, 600, 700 and 800 ms).<br />
3.2.3. Behavioral data<br />
Participants detected on average 92.8% (SD ¼ 5.7%) of probes in the L1 block. In the L2 block<br />
the participants detected 93.0% of probes (SD ¼ 10.7%). Participants produced false alarms on an<br />
average of 1.8 items (SD ¼ 1.1) in L1 (2.4%) and on 2.2 items (SD ¼ 3.9) in L2 (2.9%).<br />
3.3. Discussion<br />
This experiment examined proficient bilinguals in both their L1 (French) and L2 (English).<br />
As in Experiments 1 and 2, ERPs were time locked to passively read words in both languages.<br />
Both languages elicited a similar pattern of early ERP components and this similarity in early<br />
effects extended through 300 ms. It was not until the middle portion of the N400 (around<br />
350 ms) that the somewhat larger posterior N400 for L1 items could be seen in this Experiment.<br />
At anterior sites the negativity which peaked around 350 ms for L1 was clearly delayed in L2,<br />
peaking around 50e100 ms later. Also the late negativity that follows the N400 especially at<br />
anterior sites was larger in L2 than in L1. However, the latency shift in anterior N400 was<br />
certainly smaller than seen in the less proficient participants of Experiments 1 and 2.<br />
4. General discussion<br />
Across three experiments we compared ERPs to words passively read for meaning in bilingual<br />
participants who were comparatively late learners of their L2 (after age 12). In Experiment 1,<br />
participants were native speakers of English (L1 English) and were in the process of learning<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
16 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
French (L2). In Experiment 2 participants were native speakers of French (L1 French) and were in<br />
the process of learning English (L2). In Experiment 3 participants were again native speakers of<br />
French (L1 French) but were also competent speakers of English (L2).<br />
In all three experiments words in L1 and L2 produced a series of early ERP components (N1<br />
and P2) prior to 200 ms that were quite similar in morphology (shape) scalp distribution and<br />
amplitude (see <strong>Fig</strong>s. 4, 6 and 8). This is not surprising because these early exogenous ERP<br />
components are believed to reflect sensory and perceptual characteristics of the eliciting<br />
stimulus (Luck, 2005) and in the current study we carefully matched these characteristics<br />
across the words used in the two languages (e.g., no French words with accents were used).<br />
The most prominent feature of ERPs to passively read content words after 200 ms is the<br />
N400 (Osterhout & Holcomb, 1995). The N400 in single word paradigms is believed to be<br />
sensitive to word processing as early as the word-form/meaning interface (e.g., Holcomb,<br />
O’Rourke, & Grainger, 2002) and therefore is likely to be the first ERP component that would<br />
carry any language effects in a passive reading paradigm. As expected words presented in L1<br />
were associated with a large broadly distributed N400, and this was true for both English and<br />
French L1 speakers. Turning first to the two less competent L2 groups (Experiment 1 and 2),<br />
perhaps most notable is the observation that both experiments produced a very similar pattern<br />
of L1eL2 ERP effects. Because the language of L1 and L2 were reversed across the experiments,<br />
this suggests that language-specific differences were not at the root of the effects<br />
observed. In both experiments the L2 N400 was attenuated relative to L1 especially at centroposterior<br />
sites. However, at anterior sites while the N400 started out smaller in L2 than L1, this<br />
difference appeared to be due in part to the anterior N400 being somewhat delayed in L2<br />
compared to L1. In fact, looking at the epoch just after the traditional N400 at anterior sites, L2<br />
words are more negative-going than L1 words and this difference continued all the way to the<br />
end of the recording epoch. Moreno et al (2005) also found delayed N400s as well as<br />
a posterior N400 negativity in competent L2 speakers in a sentence reading task. They found<br />
that this late negativity was also associated with language dominance, with ERPs to the less<br />
dominant language having the larger late negativity.<br />
Experiment 3 examined the impact of L2 proficiency level on these language differences by<br />
testing participants with roughly comparable proficiency in their L1 and L2 (in Experiments 1<br />
and 2 participants were significantly less competent in L2), but who had began learning their L2<br />
at a similarly late age. In this experiment the difference between L1 and L2 in centro-posterior<br />
N400 amplitude that was seen previously in Experiments 1 and 2 was virtually non-existent.<br />
The L2 words in Experiment 3 generated a centro-posterior N400 of similar size to words in L1.<br />
This pattern of similar N400 amplitude for L1 and L2 has been reported in previous studies of<br />
competent L2 speakers in other language paradigms (e.g., Moreno et al., 2005; Neville et al.,<br />
1992) and is consistent with the hypothesis that the posterior N400 effect reflects competence in<br />
L2, less competent users producing smaller N400s. The anterior N400 latency shift seen clearly<br />
in Experiments 1 and 2 was still evident in Experiment 3, but the latency difference was much<br />
smaller. This would suggest that the anterior N400 latency shift reflects competence in L2, but<br />
contrary to the posterior N400 effect continues to reflect differences in L1 and L2 processing<br />
even in relatively proficient bilinguals.<br />
These large differences in the ERPs generated by words in L1 and L2 and their modulation<br />
by proficiency in L2 are in line with the hypothesis that word recognition in L2 involves distinct<br />
mechanisms compared with the first language, at least in the relatively early phases of L2<br />
acquisition in late learners of L2. One specific model, the RHM (Kroll & Stewart, 1994),<br />
predicted such differences in L1 and L2 word recognition over and above possible differences<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
17<br />
due to subjective frequency, AoA, and other correlated factors. In the RHM, it is the process of<br />
translation from an L2 word-form to its equivalent in L1 that is the key distinguishing characteristic<br />
of L2 vocabulary acquisition in beginning bilinguals. That is, word recognition in L2<br />
is achieved mainly by translation of the L2 word-form into its equivalent word-form in L1,<br />
which then leads to the appropriate semantic activation. The delayed anterior N400 latency to<br />
L2 words might well reflect this specific process of L2 word recognition, that is inevitably<br />
delayed compared with L1 word recognition for which the translation route is non-dominant<br />
(see <strong>Fig</strong>. 1). The fact that an albeit smaller latency shift was still observed in the proficient<br />
bilinguals of Experiment 3 suggests that the translation route is still operational in these<br />
participants, but to as lesser degree.<br />
Do these results therefore force us to reject an integrated lexicon account of bilingual lexical<br />
representation, such as implemented in the bilingual interactive-activation model (BIA-model,<br />
Grainger & Dijkstra, 1992; van Heuven et al., 1998)? One way of saving the BIA-model is to<br />
assume that it better reflects processing in relatively proficient bilinguals, while the RHM is a better<br />
model of lexical processing in beginning bilinguals. The idea would be that during L2 acquisition,<br />
the L2eL1 translation route of the RHM would be gradually replaced by L2 lexical representations<br />
that become part of an integrated network along with L1 representations. This could be achieved by<br />
gradually weakening the L2eL1 connections between translation equivalents, and gradually<br />
strengthening the connections between L2 word-forms and semantics. Thus as proficiency<br />
develops in L2, lexical processing in L2 becomes more and more akin to lexical processing in L1.<br />
But why is the posterior N400 smaller for L2 items in less competent speakers? Indeed, one<br />
might have expected a greater N400 amplitude to L2 words compared with L1 words, given<br />
that N400 amplitude is often associated with processing effort or cost. One possibility is that in<br />
less competent speakers exposure to L2 words is generally much lower than exposure to words<br />
in L1. Therefore L2 words will have lower subjective frequencies than L1 words. Differences in<br />
subjective frequency could be the source of the smaller posterior N400 for L2 items. However,<br />
the effects of word frequency on the N400 have typically been reported to be just the opposite<br />
of those obtained here (i.e., larger N400s for less frequent words e Van Petten & Kutas, 1990),<br />
so a mechanism like subjective word frequency is an unlikely source.<br />
Another possible explanation for these effects is that they reflect differential activity in<br />
a mechanism similar to the age-of-acquisition effect for words in L1. That is, that words<br />
acquired younger in life are afforded some special privilege (Zevin & Seidenberg, 2004).<br />
Plotted in <strong>Fig</strong>. 10 are ERPs recorded to words in the same passive reading paradigm of the<br />
current study in adult monolingual speakers of English. These waves have been segregated<br />
according to whether they were learned before or after the age of eight years. As can be seen,<br />
the very small effects seen at anterior sites for the two word categories are in the opposite<br />
direction to those predicted by the late negative effects seen for words in L2 in <strong>Fig</strong>s. 4, 6 and 8.<br />
So AoA appears to be an unlikely source of the posterior N400 language effect.<br />
Finally, another possibility is that L2 words, especially in a less competent speaker may be<br />
less interconnected to other L2 words e that is, on average, they would tend to have smaller<br />
orthographic neighborhoods. At least one previous study has shown that words from dense<br />
orthographic neighborhoods have larger N400s than words from sparse neighborhoods<br />
(Holcomb et al., 2002). Therefore, the larger posterior N400 amplitude to L1 words might be<br />
a reflection of the greater number of other words that are co-activated during target word<br />
recognition. A similar argument could be made at the level of semantic representations, with<br />
the semantic representations of L1 words being more richly interconnected within the semantic<br />
network than L2 words. Such an analysis would fit with one account of the effects of<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
18 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
<strong>Fig</strong>. 10. ERPs in the same passive reading task and at the same 9 scalp sites as the current study. ERPs are based on 250<br />
words divided at the median on age-of-acquisition ratings.<br />
concreteness on N400 amplitude (concrete words generate larger N400 amplitudes than abstract<br />
words, e.g., Kounios & Holcomb, 1994) according to which this would be due to the greater<br />
semantic interconnectivity for concrete words.<br />
5. Conclusions<br />
The present study provides evidence that the processing of L1 and L2 words in late L2<br />
learners diverge in two distinct ways as reflected in the ERP waveforms generated by these<br />
words during silent reading for meaning. An anterior part of the N400 component showed<br />
a distinct latency shift with L2 amplitudes peaking later than L1 amplitudes, and although the<br />
latency shift was still present in proficient bilinguals, it was smaller in magnitude. This<br />
latency shift might well reflect differences in processing difficulty associated with words in L1<br />
and L2. The posterior part of the N400 revealed larger amplitudes to L1 compared with L2<br />
words in our low-proficiency participants, and very little difference in the proficient<br />
bilinguals. It is suggested that these amplitude differences might reflect a lower level of<br />
interconnectivity of L2 lexical and semantic representations in beginning bilinguals, that<br />
disappears with increasing competence in L2 and greater integration of L2 words in<br />
a common lexical-semantic network.<br />
References<br />
Alvarez, R., Grainger, J., & Holcomb, P. J. (2003). Accessing word meaning in primary and secondary languages:<br />
a comparison using event-related brain potentials. Brain & Language, 87, 290e304.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
19<br />
Beauvillain, C., & Grainger, J. (1987). Accessing interlexical homographs: some limitations of a language-selective<br />
access. Journal of Memory and Language, 26(6), 658e672.<br />
Brysbaert, M., Van Dyck, G., & Van de Poel, M. (1999). Visual word recognition in bilinguals: evidence from masked<br />
phonological priming. Journal of Experimental Psychology: Human Perception and Performance, 25(1), 137e148.<br />
CELEX English database (Release E25). (1993). Available from. http://www.mpi.nl/world/celex.<br />
Chauncey, K., Grainger, J., & Holcomb, P. J. (2008). Code-switching effects in bilingual word recognition: a masked<br />
priming study with event-related potentials. Brain & Language.<br />
De Bruijn, E. R. A., Dijkstra, T., Chwilla, D. J., & Schriefers, H. J. (2001). Language context effects on interlingual<br />
homograph recognition: evidence from event-related potentials and response times in semantic priming. Bilingualism:<br />
Language and Cognition, 4, 155e168.<br />
De Groot, A. M., Delmaar, P., & Lupker, S. J. (2000). The processing of interlexical homographs in translation<br />
recognition and lexical decision: support for non-selective access to bilingual memory. Quarterly Journal of<br />
Experimental Psychology A: Human Experimental Psychology, 53A(2), 397e428.<br />
Dijkstra, T., Grainger, J., & van Heuven, W. J. B. (1999). Recognition of cognates and interlingual homographs: the<br />
neglected role of phonology. Journal of Memory and Language, 41, 496e518.<br />
Dijkstra, T., van Heuven, W. J. B., & Grainger, J. (1998). Simulating cross-language competition with the bilingual<br />
interactive activation model. Psychologica Belgica, 38(3e4), 177e196.<br />
Dijkstra, T., Timmermans, M., & Schriefers, H. (2000). On being blinded by your other language: effects of task<br />
demands on interlingual homograph recognition. Journal of Memory and Language, 42(4), 445e464.<br />
Dyer, F. N. (1973). The Stroop phenomenon and its use in the study of perceptual, cognitive, and response processes.<br />
Memory & Cognition, 1(2), 106e120.<br />
Geisser, S., & Greenhouse, S. (1959). On methods in the analysis of profile data. Psychometrica, 24, 95e112.<br />
Grainger, J., & Dijkstra, T. (1992). On the representation and use of language information in bilinguals. In R. J. Harris<br />
(Ed.), Cognitive processing in bilinguals (pp. 207e220). Amsterdam: North-Holland.<br />
Guttentag, R. E., Haith, M. M., Goodman, G. S., & Hauch, J. (1984). Semantic processing of unattended words by<br />
bilinguals: a test of the input switch mechanism. Journal of Verbal Learning & Verbal Behavior, 23(2), 178e188.<br />
Hahne, A., & Friederici, A. D. (2001). Processing a second language: late learners’ comprehension mechanisms as<br />
revealed by event-related brain potentials. Bilingualism: Language and Cognition, 4, 123e141.<br />
van Heuven, W. J. B., Dijkstra, T., & Grainger, J. (1998). Orthographic neighborhood effects in bilingual word<br />
recognition. Journal of Memory and Language, 39(3), 458e483.<br />
Holcomb, P. J., O’Rourke, T., & Grainger, J. (2002). An event-related brain potential study of orthographic similarity.<br />
Journal of Cognitive Neuroscience, 14, 938e950.<br />
Kerkhofs, R., Dijkstra, T., Chwilla, D. J., & de Bruijn, E. R. A. (2006). Testing a model for bilingual semantic priming<br />
with interlingual homographs: RT and N400 effects. Brain Research, 1068, 170e183.<br />
Kotz, S. A. (2001). Neurolinguistic evidence for bilingual language representation: a comparison of reaction times and<br />
event-related brain potentials. Bilingualism: Language and Cognition, 4, 143e154.<br />
Kotz, S. A., & Elston-Guttler, E. (2004). The role of proficiency on processing categorical and associative information<br />
in the L2 as revealed by reaction times and event-related brain potentials. Journal of Neurolinguistics, 17, 215e235.<br />
Kounios, J., & Holcomb, P. J. (1994). Concreteness effects in semantic processing: event-related brain potential evidence<br />
supporting dual-coding theory. Journal of Experimental Psychology: Learning, Memory and Cognition, 20, 804e823.<br />
Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture naming: evidence for asymmetric<br />
connections between bilingual memory representations. Journal of Memory and Language, 33, 149e174.<br />
Lemhöfer, K., Dijkstra, T., Schriefers, H., Baayen, H., Grainger, J., & Zwitserlood, P. (2008). Native language influences<br />
on word recognition in a second language: a mega-study. Journal of Experimental Psychology: Learning,<br />
Memory, and Cognition, 34(1), 12e31.<br />
Luck, S. J. (2005). An introduction to the event-related potential technique. Cambridge, MA: MIT Press.<br />
Macnamara, J., & Kushnir, S. L. (1972). Linguistic independence of bilinguals: the input switch. Journal of Verbal<br />
Learning & Verbal Behavior, 10(5).<br />
McLaughlin, J., Osterhout, L., & Kim, A. (2004). Neural correlates of second-language word learning: minimal<br />
instruction produces rapid change. Nature Neuroscience, 7(7), 703e704.<br />
Midgley, K. J., Holcomb, P. J., van Heuven, W. J. B., & Grainger, J. An electrophysiological investigation of crosslanguage<br />
effects of orthographic neighborhood. Brain Research, in press.<br />
Moreno, E., & Kutas, M. (2005). Processing semantic anomalies in two languages: an electrophysiological exploration<br />
in both languages of SpanisheEnglish bilinguals. Cognitive Brain Research, 22, 205e220.<br />
Nas, G. (1983). Visual word recognition in bilinguals: evidence for a cooperation between visual and sound based codes<br />
during access to a common lexical store. Journal of Verbal Learning & Verbal Behavior, 22(5), 526e534.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
ARTICLE IN PRESS<br />
+ MODEL<br />
20 K.J. Midgley et al. / Journal of Neurolinguistics xx (2008) 1e20<br />
Neville, H. J., Coffey, S. A., Lawson, D. S., & Fischer, A. (1997). Neural systems mediating American sign language:<br />
effects of sensory experience and age of acquisition. Brain & Language, 57(3), 285e308.<br />
Neville, H. J., Mills, D. L., & Lawson, D. S. (1992). Fractionating language: different neural subsystems with different<br />
sensitive periods. Cerebral Cortex, 2(3), 244e258.<br />
New, B., Pallier, C., Ferrand, L., & Matos, R. (2001). Une base de données lexicales du français contemporain sur<br />
internet: Lexique. L’Année Psychologique, 101, 447e462.<br />
Oldfield, R. C. (1971). The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia, 9(1),<br />
97e113.<br />
Osterhout, L., & Holcomb, P. J. (1995). The electrophysiology of language comprehension. In M. D. Rugg, &<br />
M. G. H. Coles (Eds.), Electrophysiology of mind (pp. 171e215). Oxford University Press.<br />
Phillips, N. A., Klein, D., Mercier, J., & de Boysson, C. (2006). ERP measures of auditory word repetition and<br />
translation priming in bilinguals. Brain Research, 1125, 116e131.<br />
Phillips, N. A., Segalowitz, N., O’Brien, I., & Yamasaki, N. (2004). Semantic priming in a first and second language:<br />
evidence from reaction time variability and event-related brain potentials. Journal of Neurolinguistics, 17, 237e262.<br />
Van Petten, C., & Kutas, M. (1990). Interactions between sentence context and word frequency in event-related brain<br />
potentials. Memory & Cognition, 18, 380e393.<br />
Weber-Fox, C. M., & Neville, H. J. (1996). Maturational constraints on functional specializations for language processing:<br />
ERP and behavioral evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8, 231e256.<br />
Zevin, J. D., & Seidenberg, M. S. (2004). Age-of-acquisition effects in reading aloud: test of cumulative frequency and<br />
frequency trajectory. Memory & Cognition, 32, 31e38.<br />
Please cite this article in press as: Katherine J. Midgley et al., Language effects in second language learners and<br />
proficient bilinguals investigated with event-related potentials, Journal of Neurolinguistics (2008), doi:10.1016/<br />
j.jneuroling.2008.08.001
Voisinage orthographique inter-langues<br />
Chapitre 3<br />
Une investigation électrophysiologique<br />
des effets de voisinage orthographique inter-langues*<br />
Au cours d'une expérience où des sujets bilingues français-anglais lisaient des listes de<br />
mots anglais et français, des potentiels évoqués ont été enregistrés. Les mots se<br />
distinguaient en termes de nombre de voisins orthographiques qu'ils possèdent dans<br />
l'autre langue ; peu ou beaucoup. C'est le nombre de voisins français pour des mots cibles<br />
anglais et vice-versa qui est manipulé. Les résultats ont montré des effets de taille de<br />
voisinage inter-langue sur la composante ERP N400. Celle-ci survenait plus précocement<br />
et se distribuait plus largement pour des mots cibles anglais (L2) que pour des mots cibles<br />
français (L1). Dans une expérience contrôle destinée à démontrer que ces effets n'étaient<br />
dus à aucun facteur parasite, des participants monolingues anglais ne lisaient que la liste<br />
des mots cibles anglais variant par le nombre de voisins français (langue inconnue d'eux).<br />
Ces sujets ont produit un pattern d'effets totalement différent de celui des bilingues. Ces<br />
résultats témoignent de la perméabilité inter-langue impliquée dans la reconnaissance des<br />
mots dans un contexte bilingue, phénomène dont la prédiction et la simulation sont<br />
correctement prises en charge par le modèle BIA+ (Modèle d'Activation Interactive<br />
Bilingue, version plus).<br />
Mots-clefs : potentiels évoqués, bilinguisme, voisinage orthographique, N400<br />
________________________<br />
* Article in Brain Research vol. 1246, pp. 123-135, December 2008<br />
P | 55
BRAIN RESEARCH 1246 (2008) 123– 135<br />
available at www.sciencedirect.com<br />
www.elsevier.com/locate/brainres<br />
Research Report<br />
An electrophysiological investigation of cross-language effects<br />
of orthographic neighborhood<br />
Katherine J. Midgley a,b, ⁎, Phillip J. Holcomb a , Walter J.B. vanHeuven c , Jonathan Grainger b<br />
a Tufts University, Medford, MA, USA<br />
b CNRS and University of Provence, <strong>Marseille</strong>, France<br />
c University of Nottingham, Nottingham, UK<br />
ARTICLE INFO<br />
Article history:<br />
Accepted 17 September 2008<br />
Available online 9 October 2008<br />
Keywords:<br />
ERP<br />
Bilingualism<br />
Orthographic neighborhood<br />
N400<br />
ABSTRACT<br />
In Experiment 1 ERPs were recorded while French–English bilinguals read pure language lists<br />
of French and English words that differed in terms of the number of orthographic neighbors<br />
(many or few) they had in the other language. That is the number of French neighbors for<br />
English target words was varied and the number of English neighbors for French target words<br />
was varied. These participants showed effects of cross-language neighborhood size in the<br />
N400 ERP component that arose earlier and were more widely distributed for English (L2)<br />
target words than French (L1) targets. In a control experiment that served to demonstrate<br />
that these effects were not due to any other uncontrolled for item effects, monolingual L1<br />
English participants read only the list of English targets that varied in the number of French<br />
(an unknown L) neighbors. These participants showed a very different pattern of effects of<br />
cross-language neighbors. These results provide further crucial evidence showing crosslanguage<br />
permeability in bilingual word recognition, a phenomena that was predicted and<br />
correctly simulated by the bilingual interactive-activation model (BIA+).<br />
© 2008 Elsevier B.V. All rights reserved.<br />
1. Introduction<br />
A long-standing debate in the literature on bilingual language<br />
comprehension concerns the relative permeability of the<br />
representations dedicated to processing each language. Traditionally,<br />
this debate has opposed proponents of early languageselective<br />
processing with proponents of a non-selective access<br />
to a set of representations shared by both languages. The<br />
language-selective hypothesis is typically associated with the<br />
notion of a switching mechanism that guides the linguistic<br />
input to the appropriate set of language-specific lexical representations<br />
(Macnamara, 1967). According to this hypothesis,<br />
there should be no cross-language interference when the<br />
language of the incoming information is completely predictable<br />
(i.e., in a monolingual context). When this is the case, information<br />
extracted from the stimulus is sent directly to the<br />
appropriate set of language-specific representations.<br />
The non-selective access hypothesis proposes, on the other<br />
hand, that the initial feed-forward sweep of information from<br />
the linguistic input can make contact with lexical representations<br />
from both languages as a function of their orthographic<br />
or phonological overlap with the input. This is the central<br />
hypothesis of the Bilingual Interactive-Activation model<br />
(Grainger & Dijkstra, 1992; van Heuven et al., 1998), and its<br />
successor the BIA+ model (Dijkstra & van Heuven, 2002). As a<br />
consequence, word representations from both languages are<br />
activated and they compete with each other due to lateral<br />
inhibition at the word level. Therefore the model predicts not<br />
⁎ Corresponding author. Tufts University, 490 Boston Avenue, Medford, MA 02155, USA. Fax: +1 617 627 3181.<br />
E-mail address: kj.midgley@tufts.edu (K.J. Midgley).<br />
0006-8993/$ – see front matter © 2008 Elsevier B.V. All rights reserved.<br />
doi:10.1016/j.brainres.2008.09.078
124 BRAIN RESEARCH 1246 (2008) 123– 135<br />
only within-language interference but also cross-language<br />
interference effects.<br />
Evidence in favor of language-selective access to languagespecific<br />
representations, was first provided by language<br />
switching experiments (Macnamara & Kushnir, 1972; Soares<br />
& Grosjean, 1984; Beavillain & Grainger, 1987; Thomas &<br />
Allport, 2000; Alvarez et al., 2003). All of these studies have<br />
shown that switching languages incurs a processing cost<br />
compared to a situation where there is no language switch.<br />
Thus, for example, in Grainger's and Beauvillain's (1987) study,<br />
lexical decision responses to words in one language were<br />
slower when the word on the preceding trial was from the<br />
other language compared with a word from the same<br />
language. Although switch costs have traditionally been<br />
taken as evidence for language-selective access, Grainger<br />
and Dijkstra (1992) provided an interpretation within the<br />
framework of a non-selective access model. Language switch<br />
costs are therefore not necessarily diagnostic of languageselective<br />
access.<br />
Evidence in favor of non-selective access to a common set<br />
of representations was provided by experiments demonstrating<br />
cross-language interference using bilingual versions of the<br />
Stroop task (Dyer, 1973), the flanker task (Guttentag et al.,<br />
1984), experiments showing evidence for co-activation of nontarget<br />
language representations during the processing of<br />
cross-language homographs (Beauvillain & Grainger, 1987;<br />
De Groot et al., 2000; Dijkstra et al., 1999, 2000; Jared & Szucs,<br />
2002; van Heuven et al., 2008) and cross-language homophones<br />
(Brysbaert et al., 1999; Nas, 1983; Dijkstra et al., 1999),<br />
and experiments showing differential processing of cognate<br />
words compared with non-cognate words (e.g., de Groot &<br />
Nas, 1991; van Hell & de Groot, 1998; van Hell & Dijskstra,<br />
2002). These cross-language influences have generally been<br />
interpreted as showing that bilinguals cannot block interference<br />
from the irrelevant language. However, proponents of<br />
selective access have argued that the mere presence of words<br />
in the irrelevant language (as is the case in Stroop and Flanker<br />
interference experiments) is enough to prevent processing in a<br />
pure “monolingual” mode (e.g., Grosjean, 1988). The same<br />
critique can be leveled against research examining processing<br />
of cross-language homographs, homophones, and identical<br />
cognates, since these stimuli are also words in the other<br />
language. In order to provide more convincing evidence in<br />
favor of non-selective access, cross-language interference<br />
must be demonstrated in conditions where there is no explicit<br />
activation of the irrelevant language.<br />
These conditions were respected in two studies, one<br />
investigating spoken word recognition (Marian & Spivey,<br />
2003), and the other investigating visual word recognition<br />
(van Heuven et al., 1998). Critically, and contrary to all prior<br />
research, these studies did not explicitly manipulate the<br />
presence or absence of other language stimuli. Rather, they<br />
manipulated the presence of potential cross-language interference<br />
in the form of phonologically or orthographically<br />
similar words from the other language. To do so, Marian and<br />
Spivey (2003) applied the visual world paradigm (see Tanenhaus<br />
et al., 2000, for a description of this technique). In one<br />
version of this paradigm, participants are requested to pick up<br />
one of four objects placed in front of them. The instructions<br />
are delivered auditorily (e.g., “pick up the candle”) and eye<br />
movements are recorded. The standard finding is that a<br />
significant proportion of eye movements are made to objects<br />
whose name is phonologically similar to the target (e.g.,<br />
“candy”), suggesting at least partial access to the distracter's<br />
lexical representation during target word processing. In<br />
Marian and Spivey's (2003) study, the phonological similarity<br />
of targets and distracters was manipulated within and<br />
between languages in bilingual participants. As well as the<br />
standard within-language effect, they also found a significant<br />
percentage of eye movements to distracter objects in the<br />
cross-language condition, but only for targets in L2. Thus, in<br />
the absence of any overt presentation of L1 words, comprehension<br />
of words in L2 would appear to be influenced by<br />
implicit activation of phonologically similar L1 words.<br />
Most relevant for the present study is van Heuven et al.'s<br />
(1998) investigation of cross-language neighborhood effects in<br />
bilinguals (applying Coltheart et al., 1977, definition of an<br />
orthographic neighbor). Prior work has shown that withinlanguage<br />
manipulations of this variable significantly affects<br />
performance in standard word recognition tasks (e.g.,<br />
Andrews, 1989; Carreiras et al., 1997; Grainger, 1990; Grainger<br />
et al., 1989). Van Heuven et al. (1998) found a significant effect<br />
of number of orthographic neighbors both within languages<br />
and across languages in bilingual participants (see also<br />
Grainger & Dijkstra, 1992). Most important, the crosslanguage<br />
neighborhood effects disappeared in an experiment<br />
testing monolingual participants with the same materials.<br />
Therefore, as predicted by the BIA model, the cross-language<br />
neighborhood effect found in bilingual participants suggests<br />
that the processing of a given word (among a list of words<br />
from one language only) generates activation in orthographically<br />
similar words not only within that language but also in<br />
the other language.<br />
There is, however, some variability in the effects of<br />
orthographic neighborhood reported in monolingual studies<br />
using behavioral measures, with some studies showing<br />
facilitatory effects (i.e., faster responses and/or lower error<br />
rates to words with large numbers of neighbors compared<br />
with words with few neighbors), and others showing inhibitory<br />
effects (see Andrews, 1997; Grainger & Jacobs, 1996, for<br />
review and discussion of possible mechanisms). These<br />
discrepancies in prior behavioral research were the primary<br />
motivation for Holcomb et al. (2002) study. These authors<br />
investigated the effects of orthographic neighborhood density<br />
in English using event-related potentials (ERPs). In one<br />
experiment, participants had to read words presented in<br />
isolation and press a response button whenever they saw an<br />
animal name (randomly appearing in 19.5% of trials). The<br />
amplitude of the N400 ERP component, a negative going<br />
waveform that peaks around 400 ms post-target onset, was<br />
found to vary significantly with the neighborhood density of<br />
target words. Words with large numbers of orthographic<br />
neighbors generated greater N400 amplitudes (i.e., more<br />
negative-going waveforms in the 300–500 ms time window).<br />
Most critically, and unlike prior behavioral findings, these<br />
effects of orthographic neighbor on ERP amplitudes did not<br />
depend on the task that participants had to perform (semantic<br />
categorization or lexical decision).<br />
One ERP study has examined neighborhood effects in<br />
bilingual participants. Rüschemeyer et al. (2008) examined
BRAIN RESEARCH 1246 (2008) 123– 135<br />
125<br />
the effects of phonological neighborhood during L2 processing.<br />
They found ERP effects in the same direction as Holcomb<br />
et al. (i.e., items from larger neighborhoods elicit greater N400<br />
amplitudes).<br />
Holcomb and Grainger (2007) have proposed a tentative<br />
mapping of ERP components onto underlying processes<br />
involved in visual word recognition, couched with the framework<br />
of a generic interactive-activation model. Based primarily<br />
from evidence obtained with the masked priming<br />
paradigm, these authors suggested that much of the mapping<br />
of form onto meaning arises as early as 200 ms post-target<br />
onset (the beginning of the N250 component found in masked<br />
priming) and culminating in the N400. This processing would<br />
initially involve the mapping of prelexical form representations<br />
onto whole-word representations, with the N400 reflecting<br />
the mapping of whole-word form representations onto<br />
semantics. In the interactive-activation framework adopted<br />
by Holcomb and Grainger (2007), competition between wholeword<br />
form representations is thought to be the primary cause<br />
of inhibitory effects of orthographic and phonological neighbors.<br />
Thus the greater negativity to words with many<br />
orthographic neighbors reported by Holcomb et al. (2002),<br />
would reflect inhibition operating across lexical representations<br />
leading to increased difficulty in settling on a unique<br />
form-meaning association. The bilingual version of interactive-activation<br />
(the BIA model and its successor the BIA+<br />
model) predicts similar effects of cross-language neighbors<br />
due to lexical competition operating within an integrated<br />
lexicon of word forms from both languages.<br />
1.1. Experiment 1<br />
The present study provides a further investigation of crosslanguage<br />
neighborhood effects using ERP recordings. We<br />
combine the basic manipulation of cross-language neighborhood<br />
in the van Heuven et al. (1998) study with the procedure<br />
used in the Holcomb et al. (2002) study. In Experiment 1<br />
bilingual participants saw pure lists of French and English<br />
words that varied in terms of the number of orthographic<br />
neighbors in the other language (many or few). Given the<br />
number of items per condition required for an ERP study we<br />
did not manipulate within-language neighborhood, although<br />
this was equated (see Holcomb et al., 2002, for a withinlanguage<br />
ERP investigation of neighborhood effects). The<br />
participants tested in Experiment 1 were L1 French and had<br />
a relatively high level of proficiency in their L2 (English). They<br />
were tested in an L1 context, that is, in France although these<br />
participants reported using their L2 on a daily basis for work<br />
or study. On the basis of the non-selective access hypothesis<br />
and prior ERP effects of neighborhood density found in<br />
monolinguals, it was predicted that the N400 would be<br />
sensitive to the number of neighbors in the non-presented<br />
language, with larger amplitudes for items with many otherlanguage<br />
neighbors compared with items with few otherlanguage<br />
neighbors.<br />
1.2. Experiment 2<br />
Although within language neighborhood size was carefully<br />
controlled in Experiment 1, our words with many and few<br />
neighbors could differ by chance on some other withinlanguage<br />
dimension. In order to be absolutely sure that it is<br />
non-target language activation that is driving the crosslanguage<br />
neighborhood effects, it is important to show that<br />
these effects are not a result of some other uncontrolled for<br />
property of the words. Monolingual L1 English participants<br />
with no or little exposure to French as an L2 should show no<br />
effect of French orthographic neighborhood size during the<br />
processing of L1 (English) target words. If, on the other hand, it<br />
is an uncontrolled L1 variable that is driving the effect, then<br />
the results should resemble the pattern found in the bilingual<br />
participants tested with English words in Experiment 1. This<br />
prediction was tested in Experiment 2.<br />
1.3. Simulation study<br />
Finally, a simulation study was run on the BIA+ model<br />
(Dijkstra & van Heuven, 2002). The model was tested with<br />
exactly the same stimuli as used in Experiments 1 and 2 in<br />
order to evaluate its ability to account for the precise pattern of<br />
cross-language neighborhood effects found in the present<br />
experiments.<br />
2. Results<br />
2.1. Experiment 1 results<br />
2.1.1. Visual inspection of ERPs<br />
The ERP grand mean waveforms for English targets for 12 scalp<br />
sites are plotted in <strong>Fig</strong>. 1A while the grand mean waveforms<br />
for French targets are plotted in <strong>Fig</strong>. 2A. <strong>Fig</strong>s. 1B (L2) and 2B (L1)<br />
contain voltage maps computed by subtracting ERPs for items<br />
with few cross-language neighbors from ERPs for items with<br />
many cross-language neighbors. We included these to better<br />
visualize the scalp distribution of neighborhood size effects at<br />
three points in time. The first includes voltages at the center of<br />
the early analysis window (275 ms), while the second and third<br />
are centered at early (350 ms) and later (450 ms) in the N400<br />
window. As can be seen in these figures, for all ERPs anterior to<br />
the occipital sites the first visible component was a negativegoing<br />
deflection between 90 and 150 ms after stimulus onset<br />
(N1). This was followed by a positive deflection occurring at<br />
approximately 200 ms (P2). A negativity followed the P2<br />
peaking around 400 ms (N400). At occipital sites the first<br />
observable component is the P1, which peaked near 100 ms<br />
and was followed by the N1 at 190 ms and a broad P2 between<br />
250 and 300 ms. The P2 was followed by the N400 between 400<br />
and 600 ms.<br />
2.1.2. Analyses of ERP data<br />
2.1.2.1. 175–275 ms epoch. As can be seen in <strong>Fig</strong>. 1<br />
differences due to cross-neighborhood-size began to emerge<br />
in this epoch. The omnibus ANOVA on the mean amplitude<br />
values revealed a marginal main effect of cross-neighborhood-size<br />
(F(1,19)=3.78, p=.067), and a significant language by<br />
cross-neighborhood-size interaction (F(1,19)=4.41, p=.049),<br />
the latter indicating a difference in the cross-neighborhoodsize<br />
effect for the two languages.
126 BRAIN RESEARCH 1246 (2008) 123– 135<br />
<strong>Fig</strong>. 1 – (A) Results of bilinguals reading English (L2) targets with either many orthographic neighbors in French (L1) or few<br />
orthographic neighbors in French. (B) Scalp voltage maps at three time points between English words with few French<br />
neighbors and many French neighbors (units are in microvolts).
BRAIN RESEARCH 1246 (2008) 123– 135<br />
127<br />
<strong>Fig</strong>. 2 – (A) Results of bilinguals reading French (L1) targets with either many orthographic neighbors in English (L2) or few<br />
orthographic neighbors in English. (B) Scalp voltage maps at three time points between French words with few English<br />
neighbors and many English neighbors.
128 BRAIN RESEARCH 1246 (2008) 123– 135<br />
Follow-up analyses examining the effects of crossneighborhood-size<br />
separately for the two target languages<br />
revealed that English (L2) words with many French orthographic<br />
neighbors were more negative-going than English<br />
words with few orthographic neighbors in French (main<br />
effect of cross-neighborhood-size: F(1,19)=8.08, p=.01). Moreover,<br />
this cross-neighborhood-size effect tended to be larger<br />
over the left hemisphere and midline electrode sites than<br />
over right hemisphere sites (cross-neighborhood-size×laterality<br />
interaction: F(2,38)=4.49, p=.033 — see <strong>Fig</strong>. 1B, left).<br />
There was however, no evidence that French (L1) words<br />
were affected by the number of English neighbors in this<br />
epoch (all Fsb1.0 involving cross-neighborhood-size — see<br />
<strong>Fig</strong>. 2).<br />
2.1.2.2. 300–500 ms epoch. As can be seen in <strong>Fig</strong>s. 1 and 2<br />
differences due to cross-neighborhood-size continued into<br />
this epoch. The omnibus ANOVA produced marginal effects<br />
for cross-neighborhood-size (F(1,19)=3.24, p =.089) and a<br />
cross-neighborhood-size×language interaction (F(1,19)=4.13,<br />
p=.056) and importantly a significant cross-neighborhoodsize×language×electrode<br />
site interaction (F(3,57)=3.85,<br />
p=.037). This latter interaction indicated differences in the<br />
scalp distribution of the cross-neighborhood-size effect for the<br />
two languages.<br />
Follow-up analyses examining the effects of cross-neighborhood-size<br />
separately for the two target languages demonstrated<br />
that English words (L2) with many French (L1)<br />
orthographic neighbors were again more negative-going<br />
than English words with few French neighbors (main effect<br />
of cross-neighborhood-size: F(1,19)=8.63, p=.008). However,<br />
unlike the earlier epoch where the cross-neighborhood-size<br />
effect was larger over left and midline sites, in this epoch the<br />
effect was more widespread across the head and was<br />
bilaterally more symmetrical (see <strong>Fig</strong>. 1B middle and right).<br />
Also different from the earlier epoch where there was no<br />
evidence of significant cross-neighborhood-size effects for<br />
French words, in this window French words with many<br />
English orthographic neighbors did produce evidence of<br />
more negative-going ERPs than French words with few English<br />
neighbors (although the main effect of cross-neighborhoodsize<br />
was not significant, pN.778). This was revealed in a<br />
significant cross-neighborhood-size ×electrode site interaction<br />
(F(3,57)=4.5, p=.029). As can be seen in <strong>Fig</strong>s. 2A and B,<br />
these effects were not widespread across the scalp and were<br />
significant only at the three most posterior sites (occipital<br />
cross-neighborhood-size F(1,19)=5.74, p=.027).<br />
2.1.3. Experiment 1 behavioral results<br />
Participants averaged 17 (SD=0.94) out of 18 hits in their L1<br />
(95%) and 14 (SD=1.77) out of 18 hits in their L2 (79%) for probe<br />
words. This difference was significant (t(19)=7.78, p=.001).<br />
Participants produced false alarms on an average of 1.8 items<br />
(SD=1.01) in L1 (2.4%) and on 1.9 items (SD=2.87) in L2 (2.5%).<br />
This difference between languages was not significant (pN.9).<br />
In a post translation task participants were asked to translate<br />
all 74 L2 target words that they had seen in the experiment.<br />
The mean number of correct translations was 53 (SD=9.2) or<br />
71%. The mean number of correct translations of probe items<br />
was 15 (SD=1.88) or 81%.<br />
2.2. Discussion of Experiment 1<br />
The results of Experiment 1 show effects of cross-language<br />
orthographic neighborhood density in the ERP waveforms<br />
generated during the processing of words in L1 and L2. These<br />
cross-language neighborhood effects had an earlier onset and<br />
were more widely distributed when the targets were in L2.<br />
This is important evidence in favor of initial non-selective<br />
access processes in bilingual word recognition, as assumed in<br />
the BIA+ model (Dijkstra & van Heuven, 2002). Participants in<br />
this study read words in one language only (and knew that<br />
they would only receive words in one language in a given list),<br />
yet the orthographic characteristics of the words in the nonpresented<br />
language influenced the way our participants<br />
reacted to these stimuli. These results can be explained by a<br />
combination of non-selective access (a string of letters<br />
activates compatible whole-word orthographic representations<br />
in both of a bilingual's languages) and lateral inhibition<br />
across word representations in an integrated lexicon. A given<br />
stimulus word generates activation in all whole-word orthographic<br />
representations that are partly compatible with the<br />
stimulus, and these co-activated word representations inhibit<br />
processing of the target word itself. The increased difficulty in<br />
target word processing is reflected in the greater ERP<br />
negativities between 200 and 500 ms, a result similar to that<br />
previously reported for neighborhood density in a monolingual<br />
context (Holcomb et al., 2002), and compatible with the<br />
time-course of component processes in visual word recognition<br />
proposed by Holcomb and Grainger (2006, 2007).<br />
2.3. Experiment 2 results<br />
2.3.1. Analyses of ERP data<br />
2.3.1.1. 175–275 ms epoch. As can be seen in <strong>Fig</strong>. 3 there<br />
does not appear to be much of a cross-neighborhood-size<br />
effect in this epoch (cross-neighborhood-size main effect:<br />
F(1,19)=2.85, p=.108; cross-neighborhood-size×electrode site<br />
pN0.2) and this small marginal main effect is actually in the<br />
opposite direction compared to the results for English targets<br />
in Experiment 1.<br />
2.3.1.2. 300–500 ms epoch. As can be seen in <strong>Fig</strong>. 3, there is<br />
only a small effect of cross-neighborhood-size in this epoch (F<br />
(1,19)=3.00, p=.100; cross-neighborhood-size×electrode site<br />
pN0.2) and importantly this marginal main effect is in the<br />
opposite direction compared to the results for English targets<br />
in Experiment 1 as in the previous epoch.<br />
2.3.2. Experiment 2 behavioral results<br />
Participants averaged 84% (SD=9.5%) hit rate for probe words.<br />
2.3.3. Combined analysis of Experiments 1 and 2<br />
A between participants analysis was carried out, comparing<br />
the English list of the L1 French participants in Experiment 1<br />
with the English list of the monolingual participants in<br />
Experiment 2 to insure that our effects, showing the influence<br />
of cross-language neighborhood on word recognition, are not<br />
due to any properties of the English items that we may not<br />
have effectively controlled. While there was no main effect of
BRAIN RESEARCH 1246 (2008) 123– 135<br />
129<br />
<strong>Fig</strong>. 3 – (A) Results of the monolinguals reading English targets with either many orthographic neighbors in French or few<br />
orthographic neighbors in French. (B) Scalp voltage maps showing the difference at three time points between English words<br />
with few French neighbors and many French neighbors.
130 BRAIN RESEARCH 1246 (2008) 123– 135<br />
cross-neighborhood-size in either epoch (both pN.260), there<br />
was a significant interaction of cross-neighborhood-size by<br />
experiment in both epochs (early: F(1,38)=10.69, p=.002; late:<br />
F(1, 38)=11.26, p=.002).<br />
2.4. Discussion of Experiment 2<br />
The results of Experiment 2 demonstrate that the pattern of<br />
effects seen in these monolinguals is not similar to those of<br />
bilinguals when processing a list of English words in which<br />
cross-language orthographic neighborhood size varies. This<br />
implies that the effects of cross-language neighborhood size<br />
that was found in the bilinguals in Experiment 1 are not due to<br />
some confound or uncontrolled property of the English list. If<br />
that had been the case, we should have seen very similar<br />
effects of this variable in Experiment 2.<br />
Our results clearly indicate that when bilinguals read lists<br />
of words in one of their languages, the brain's reaction to these<br />
word stimuli is influenced by the orthographic characteristics<br />
of the words in the other (non-presented) language. The BIA+<br />
model (Dijkstra & van Heuven, 2002) accounts for such crosslanguage<br />
influences in terms of non-selective activation of<br />
word representations in both of the bilingual's languages. In a<br />
simulation study we put the BIA+ model to test with the<br />
stimuli used in the present experiments.<br />
2.5. Simulation study results<br />
The mean number of cycles to reach the identification threshold<br />
for the different experimental conditions in Experiments 1<br />
and 2 is presented in Table 1. We conducted separate analyses<br />
for Experiment 1 (bilinguals) and Experiment 2 (English<br />
monolinguals). The data of the bilinguals revealed a significant<br />
effect of Language (F(1,144)=23.42, pb.001), Cross-Neighborhood-Size<br />
(F(1,144)=4.41, pb.05), and importantly a significant<br />
interaction between these factors (F(1,144)=5.21, pb.05). This<br />
interaction is due to a significant inhibition effect (0.7 cycles) of<br />
cross-language neighborhood size for English words (F(1,72)=<br />
8.51, pb.01) and not for French words (F(1,72)b1). As expected,<br />
Table 1 – Simulated response times of French–English<br />
bilinguals and English monolinguals in the BIA+ model<br />
for the different experimental conditions tested in<br />
Experiments 1 (bilingual participants) and 2 (monolingual<br />
participants)<br />
Neighborhood Bilinguals Monolinguals<br />
size<br />
English targets<br />
French targets<br />
High French 21.1 20.1<br />
Low French 20.4 20.0<br />
difference 0.7 0.1<br />
High English 20.0<br />
Low English 20.1<br />
difference −0.1<br />
Average number of cycles to reach the word identification threshold<br />
for English words with many French neighbors (High French) or<br />
few French neighbors (Low French) and French words with many<br />
English neighbors (High English) or few English neighbors (Low<br />
English).<br />
the data of the English monolinguals did not show any effect of<br />
cross-language neighborhood size (F(1,72)b1).<br />
2.6. Simulation study discussion<br />
The results of the simulation study show that the BIA+ model<br />
correctly predicts an influence of cross-language orthographic<br />
neighborhood size. The effect of cross-language neighbors<br />
was significant in the simulation of bilinguals recognizing L2<br />
words. As expected, no effect of cross-language neighborhood<br />
size was found in the simulation of English monolinguals. The<br />
BIA+ model therefore simulates the pattern of cross-language<br />
neighborhood effects for French–English bilinguals and English<br />
monolinguals that, overall, mimics the effects found in<br />
our ERP experiments. The simulation study revealed a<br />
significant interaction between cross-language neighborhood<br />
size and language, thus correctly accounting for the stronger<br />
effects that were found in L2 than in L1 in Experiment 1.<br />
3. General discussion<br />
In the present study French–English bilinguals were shown<br />
pure-language lists of words that varied in terms of the<br />
number of orthographic neighbors they had in the other<br />
language (the number of cross-language neighbors). French<br />
native speakers who were relatively proficient in English were<br />
found to be sensitive to the cross-language neighborhood<br />
density of words in both their L1 (French) and their L2<br />
(English). Words with many cross-language neighbors generated<br />
a more negative-going ERP waveform in the region of the<br />
N400 than words with few cross-language neighbors. This<br />
cross-language neighborhood effect appeared earlier (in the<br />
175–275 ms epoch) and was more widely distributed across the<br />
scalp when the target words were in English (L2) and the<br />
neighbors in L1. Effects of cross-language neighborhood on<br />
French (L1) words only appeared in the 300–500 ms epoch and<br />
were limited to the most posterior electrode sites. The strong<br />
effects of L1 (French) neighbors on processing L2 (English)<br />
words cannot be attributed to any properties of the English<br />
items apart from their French neighborhood size because<br />
English monolingual participants did not show the same<br />
pattern of ERPs to these stimuli.<br />
These results provide considerable support for the nonselective<br />
access hypothesis embodied in the BIA+-model<br />
(Dijkstra & van Heuven, 2002), and contradict the notion of<br />
early language-specific selection in bilingual language comprehension.<br />
Our participants saw lists of words in one<br />
language only and were therefore in appropriate conditions<br />
for using language-specific selection processes. The results<br />
clearly indicate that such selection processes were not<br />
effective in blocking the activation of word representations<br />
in the irrelevant language. We found evidence for early<br />
activation of non-target language representations that influenced<br />
the processing of target words. The more negativegoing<br />
waveforms found for words with large numbers of crosslanguage<br />
neighbors is interpreted as reflecting a greater<br />
difficulty in settling on a single form-meaning interpretation<br />
of the stimulus (Holcomb et al., 2002). Words with more crosslanguage<br />
neighbors suffer from the co-activation of the lexical
BRAIN RESEARCH 1246 (2008) 123– 135<br />
131<br />
representations of these neighbors, as reflected in the<br />
typically longer RTs found to these stimuli in behavioral<br />
studies (Grainger & Dijkstra, 1992; van Heuven et al., 1998).<br />
The results of Experiment 1 show that L2 neighbors have a<br />
later and less widely distributed effect on L1 target processing<br />
than L1 neighbors have on L2 target processing. This is<br />
perfectly in line with one major principle implemented in all<br />
connectionist models of language processing — that frequency<br />
of exposure determines connection strength. Because<br />
none of our participants in Experiment 1 were early balanced<br />
bilinguals we can assume that exposure to L2 words is overall<br />
much lower than exposure to L1 words. This exposure<br />
difference is thus reflected in word frequency differences<br />
between L1 and L2. Therefore, L2 word representations will on<br />
average be more weakly activated by a stimulus than L1 word<br />
representations, and this imbalance will be exaggerated in a<br />
competitive network where the dominant representation<br />
inhibits all others. This therefore accounts for why the effects<br />
of L2 neighbors are weaker, less widely distributed (since more<br />
time is required for propagation), and appear later than the<br />
effects of L1 neighbors. Simulations run on the BIA+ model<br />
show effectively that cross-language neighborhood effects are<br />
stronger when targets are in L2 compared with targets in L1<br />
(see Table 1).<br />
Experiment 2 of the present study tested monolingual<br />
English participants with the same list of English words<br />
presented to the French–English bilinguals of Experiment 1.<br />
Since these monolingual participants did not show that same<br />
pattern of effects of cross-language neighborhood (i.e., of<br />
French language neighbors) as the bilinguals, this allows us to<br />
reject uncontrolled for within-language variables as the<br />
source of the cross-language neighborhood effect found in<br />
Experiment 1. Therefore, the present study adds to the<br />
behavioral literature on effects of cross-language orthographic<br />
and phonological similarity (Marian & Spivey, 2003; van<br />
Heuven et al., 1998) showing that the process of word<br />
comprehension in bilingual participants presented with<br />
words in one of their languages is influenced by the similarity<br />
of these words to words in the non-presented language.<br />
The present study provides important information concerning<br />
the time-course of cross-language neighborhood<br />
effects. The results of Experiment 1 show relatively early<br />
influences of L1 orthographic neighbors on the processing of<br />
L2 words, emerging as early as 200 ms post-target onset (see<br />
<strong>Fig</strong>. 1). Such early influences were not found in the withinlanguage<br />
neighborhood manipulation of Holcomb et al. (2002).<br />
This can be explained by differences in the relative frequency<br />
of target words and their orthographic neighbors in the<br />
Holcomb et al. and the present study. Orthographic neighbors<br />
will tend to have higher subjective frequencies when these<br />
neighbors are L1 words and the target a word in L2 (compared<br />
to L1 neighbors of L1 words), and the more frequent the<br />
orthographic neighbors are relative to the target word, the<br />
more rapidly they can influence target word processing.<br />
Furthermore, as processing develops and word recognition is<br />
in its final stages (i.e., a stable form-meaning association is<br />
established), activation of the target word itself will dominate<br />
processing and neighborhood effects disappear.<br />
Furthermore, the precise timing of the effects found in the<br />
present study is in line with the time-course analysis of visual<br />
word recognition proposed by Holcomb and Grainger (2006,<br />
2007). According to their analysis, form-level (orthographic<br />
and phonological) processing of printed words initiates<br />
around 200 ms post-target onset with the mapping of<br />
prelexical representations onto whole-word forms, and culminates<br />
at around 400 ms (the peak of the N400) with the<br />
mapping of lexical form onto meaning. Within the generic<br />
interactive-activation model adopted by Holcomb and Grainger,<br />
effects of orthographic neighborhood are generated by<br />
competition arising between co-activated whole-word representations.<br />
This lexical-level competition already affects the<br />
early mapping of prelexical form representations onto wholeword<br />
form representations and further influences processing<br />
upstream, increasing the difficulty of mapping whole-word<br />
forms onto semantics. Given that orthographic neighborhood<br />
correlates highly with phonological neighborhood (e.g., Grainger<br />
et al., 2005), it is likely that part of the effects of<br />
orthographic neighborhood are being driven by competition<br />
between phonologically similar words. However, this possibility<br />
is greatly reduced in a cross-language neighborhood<br />
manipulation as used in the present study, given the lower<br />
levels of phonological overlap between orthographically<br />
similar words from different languages.<br />
In conclusion, the present study provides further evidence<br />
for cross-language permeability in bilingual word<br />
recognition, in particularly stringent testing conditions.<br />
First, following the behavioral study of van Heuven et al.<br />
(1998) participants saw words of one language only in a<br />
given list, and cross-language interference was evaluated by<br />
a manipulation of the number of orthographic neighbors in<br />
the non-presented language. Second, our participants had<br />
to silently read words for meaning and respond (on noncritical<br />
trials) whenever a body part appeared, a procedure<br />
that minimizes contamination by decision-related processes.<br />
The ERPs generated by target words on critical trials<br />
were found to be sensitive to the number of orthographic<br />
neighbors of that word in the other language of our bilingual<br />
participants. This constitutes perhaps the strongest evidence<br />
to date in favor of initial parallel access to representations<br />
in both languages when bilinguals are reading in one<br />
language.<br />
4. Experimental procedures<br />
4.1. Experiment 1<br />
4.1.1. Participants<br />
Twenty-two participants were recruited and compensated for<br />
their time. The data from two participants was not used due to<br />
excessive noise in their ERP data. Of the remaining 20, thirteen<br />
were women (mean age=23 years, SD=4.7), all were right<br />
handed (Edinburgh Handedness Inventory — Oldfield, 1971)<br />
and had normal or corrected-to-normal visual acuity with no<br />
history of neurological insult or language disability.<br />
French was reported to be the first language learned by all<br />
participants (L1) and English their primary second language<br />
(L2). All participants began their study of English in their sixth<br />
year of primary school at approximately the age of 12 years,<br />
as is customary in the French school system. Participants'
132 BRAIN RESEARCH 1246 (2008) 123– 135<br />
daily use of English, auto-evaluation of English and French<br />
language skills and a history of study and of immersion in<br />
English were surveyed by questionnaire. Participants reported<br />
daily use of English to be on average 42% (SD=26.8%) of their<br />
total language use. On a seven point scale (1=unable;<br />
7=expert) participants reported their abilities to read, speak<br />
and comprehend English and French as well as how<br />
frequently they read in both languages (1=rarely; 7=very<br />
frequently). The overall average of self-reported languages<br />
skills in French was 6.9 (SD=0.32) and in English was 5.7<br />
(SD=0.95). Our participants reported their average frequency<br />
of reading in French as 6.3 (SD=1.05) and in English as 5.8<br />
(SD=1.47).<br />
4.1.2. Stimuli<br />
For the selection of stimuli a French lexicon was extracted<br />
from the Lexique database (New et al., 2004), and an English<br />
lexicon from the CELEX database (Baayen et al., 1995). These<br />
lexicons contained only 4 and 5-letter, monosyllabic and bisyllabic<br />
words with at least 1 occurrence per million, and were<br />
used to calculate the number of orthographic neighbors of<br />
words within and across languages. The final set of stimuli for<br />
the study were 74 English and 74 French words between four<br />
and five letters in length with half of the items in each<br />
language having many orthographic neighbors in the other<br />
language and other half having few neighbors in the other<br />
language (an orthographic neighbor is defined as a word of the<br />
same length having all but one letter in common respecting<br />
letter position (Coltheart et al., 1977).<br />
The English items from large French neighborhoods had a<br />
mean number of French neighbors of 5.9 (range=4–13,<br />
SD=2.2). For the English items with few French neighbors<br />
the mean number of French neighbors was 1.1 (range=0–3,<br />
SD=1.1). These means were significantly different (t(72)=4.58,<br />
p=0.036). The means of within language neighbors for these<br />
two groups of English words were 6.5 (SD=3.4) for items with<br />
many French neighbors and 6.9 (SD=3.6) for items with few<br />
French neighbors. These means were not significantly different<br />
(t(72)=1.29, p=0.26). The mean frequency per million, of<br />
English words with many French neighbors was 12.9 (SD=13.9)<br />
while for items with few French neighbors the mean<br />
frequency was 12.8 (SD=13.0). These means were not significantly<br />
different (t(72)=0.04, p=0.97). The mean number of<br />
letters for English items was not significantly different for the<br />
two conditions (see Table 2 for mean lengths, t(72)=0.73,<br />
p=0.47).<br />
Table 2 – Stimulus characteristics for the two languages<br />
and the two conditions<br />
English<br />
targets<br />
French<br />
targets<br />
Mean<br />
number of<br />
cross<br />
language<br />
neighbors<br />
Mean<br />
number of<br />
within<br />
language<br />
neighbors<br />
Mean<br />
frequency<br />
count per<br />
million<br />
Mean<br />
length<br />
Many 5.9 (2.2) 6.5 (3.4) 12.9 (13.9) 4.4 (0.5)<br />
Few 1.1 (1.1) 6.9 (3.6) 12.8 (13.0) 4.3 (0.5)<br />
Many 7.8 (3.6) 6.6 (3.1) 15.2 (15.0) 4.4 (0.5)<br />
Few 0.7 (1.0) 4.7 (3.3) 14.2 (11.9) 4.5 (0.5)<br />
The French items from large English neighborhoods had a<br />
mean number of English neighbors of 7.8 (=5–19, SD=3.6). For<br />
the French items with few English neighbors the mean<br />
number of English neighbors was 0.7 (range=0–5, SD=1.0).<br />
These means were significantly different (t(72)= 25.62,<br />
pb0.001). The means of within language neighbors for these<br />
two groups of French words were 6.6 (SD=3.1) for items with<br />
many English neighbors and 4.7 (SD=3.3) for items with few<br />
English neighbors. These means were not significantly<br />
different (t(72)=0.50, p=0.48). The mean frequency per million,<br />
of French words with many English neighbors was 15.2<br />
(SD=15.0) while for items with few English neighbors the<br />
mean frequency was 14.2 (SD=11.9). These means were not<br />
significantly different (t(72)=0.30, p=0.76). The mean number<br />
of letters for French items was not significantly<br />
different for the two conditions (, t(72) =1.41, p =0.16).<br />
Mean lengths can be seen in Table 2.<br />
Two lists were formed, one with the 74 English words in a<br />
pseudorandom order and one with the 74 French words in a<br />
pseudorandom order. Intermixed in each list was a second<br />
group of 18 probe words which were all members of the<br />
semantic category of “body parts” (probes were English words<br />
in the English list and French words in the French list). The<br />
order of the list, blocked by language was counter-balanced<br />
across participants.<br />
4.1.3. Procedure<br />
The word stimuli in each list were presented as white letters<br />
centered vertically and horizontally on a black background on<br />
a 15 in. color monitor (Toshiba Tekbright). Presentation of all<br />
visual stimuli and digitizing of the EEG was synchronized<br />
with the vertical retrace interval (60 Hz refresh rate) of the<br />
stimulus PCs video card (ATI Radeon) to assure precise time<br />
marking of ERP data. The participants were seated so that<br />
their eyes were at a distance of approximately 1.5 m from the<br />
screen. The maximum height and width of the stimuli were<br />
such that no saccades would be required during reading of<br />
the single word stimuli (i.e., the width of the word filled less<br />
than 2 degrees of the participant's visual field). Participant<br />
responses were made using a button box held in the lap<br />
throughout the experiment. A go/no-go semantic categorization<br />
task was used in which participants were instructed to<br />
read all words but to press a button whenever they saw a<br />
word referring to a body part. Eighteen trials in each language<br />
block were body part words (19.5% of all trials). As can be seen<br />
in <strong>Fig</strong>. 4, each trial began with the onset of a fixation cross<br />
which remained on screen for 200 ms and was followed by<br />
300 ms of blank screen. A target word then appeared for a<br />
duration of 300 ms was followed by 1000 ms of blank screen.<br />
Each trial ended with a screen indicating that it was<br />
permissible to move or blink the eyes [( - - )]. This screen<br />
had a duration of 2500 ms. The next trial began after 500 ms<br />
of blank screen with the fixation cross.<br />
4.1.4. EEG recording<br />
Participants were seated in a comfortable chair in a sound<br />
attenuating room and were fitted with an elastic cap equipped<br />
with 29 tin electrodes (Electro-cap International — see <strong>Fig</strong>. 5<br />
for the location of electrodes). Two additional electrodes were<br />
used to monitor for eye-related artifact (blinks and vertical or
BRAIN RESEARCH 1246 (2008) 123– 135<br />
133<br />
<strong>Fig</strong>. 4 – Schematic of two trials in the English block, one with a target word (grape) and another with a probe word requiring a<br />
button pressing response (foot).<br />
horizontal eye movement); one below the left eye (VE) and one<br />
horizontally next to the right eye (HE). All electrodes were<br />
referenced to an electrode placed over the left mastoid (A1). A<br />
final electrode was placed over the right mastoid (A2 — used to<br />
determine if there was any asymmetry between the mastoids;<br />
none was observed). The 32 channels of electrophysiological<br />
data were amplified using an SA Instruments Bio-amplifier<br />
system with 6db cutoffs set at .01 and 40 Hz. The output of the<br />
bio-amplifier was continuously digitized at 200 Hz throughout<br />
the experiment.<br />
After electrode placement instructions for the experimental<br />
task were given in French then a short practice list (in the<br />
<strong>Fig</strong>. 5 – Electrode Montage and analysis sites (in grey).
134 BRAIN RESEARCH 1246 (2008) 123– 135<br />
language of the first block) was presented to assure good<br />
performance during experimental runs and to accustom the<br />
participant to the coming language. A practice list was also<br />
run before the second block in the language of the second<br />
block. The order of language blocks was counterbalanced<br />
across participants. There were three pauses within each<br />
block; the length of these pauses was determined by the<br />
participant. Each language block typically required 15 min. At<br />
the end of the ERP experiment participants were asked to give<br />
a translation of the 92 English words that they had seen during<br />
the experiment (74 critical items and 18 body parts). These<br />
post-translations were graded for accuracy and reported as<br />
behavioral results.<br />
4.1.5. Data analysis<br />
ERPs were averaged separately for English target words that<br />
had many or few orthographic neighbors in French, and<br />
French target words that had many or few orthographic<br />
neighbors in English. Only trials contaminated by eye movement<br />
activity were rejected prior to averaging (7.1% of trials).<br />
Because we did not assume that translation performance is<br />
equivalent to L2 word representation all French items were<br />
averaged regardless of post-translation results. All target<br />
items were baselined to the average of activity in the 40 ms<br />
pre-target period and were lowpass filtered at 15 Hz. 1 The<br />
ERPS were then quantified by measuring the mean amplitude<br />
in two latency windows: 175–275 ms to capture pre-N400<br />
activity, and 300–500 ms to capture the N400 itself. In order<br />
to analyze the scalp distribution of the various ERP components<br />
omnibus repeated measures analyses of variance<br />
(ANOVAs) were carried out for 12 electrode sites from<br />
representative frontal (FC1, Fz and FC2) middle (C3, Cz and<br />
C4), parietal (CP1, Pz and CP2) and occipital (O1, Oz, and O2)<br />
locations. This arrangement allowed for a single omnibus<br />
ANOVA with factors of language (French vs. English), crossneighborhood-size<br />
(many vs. few), electrode-site (F vs. C vs.<br />
CP vs. O) and laterality (left vs. medial vs. right). Significant<br />
interactions in the omnibus analyses involving language and<br />
cross-neighborhood-size were decomposed with followed-up<br />
ANOVAs looking at each LANGUAGE (French/English) separately.<br />
The Geisser-Greenhouse (1959) correction was applied<br />
to repeated measures with more than one degree of freedom<br />
in the numerator. 2<br />
4.2. Experiment 2<br />
4.2.1. Participants<br />
Twenty participants (11 women) were recruited and compensated<br />
for their time (mean age=20 years, SD=1.3). All were<br />
right handed (Edinburgh Handedness Inventory — Oldfield,<br />
1971) with normal or corrected-to-normal visual acuity and no<br />
history of neurological insult or language disability. All<br />
1 A 40ms baseline was chosen because using the more traditional<br />
100 pre-stimulus baseline resulted in a substantial difference<br />
between conditions in the first 50ms after target word onset.<br />
2 We performed a first pass analysis including the factor of<br />
order to test for differential effects of which target language block<br />
occurred first. There were no interactions involving the order and<br />
language or cross-neighborhood-size factors (all Fs b 2). In all of<br />
the analyses reported we collapsed across this factor.<br />
participants reported to be monolingual native English speakers<br />
and to have had no classroom exposure to French as an L2.<br />
4.2.2. Stimuli and procedure<br />
Materials and experimental task were the same as in Experiment<br />
1 but only the English list was presented to these<br />
participants.<br />
4.2.3. Data analysis<br />
Trials contaminated by eye movement activity were rejected<br />
prior to averaging (8.9%). ERPs were averaged for English<br />
target words that had many or few orthographic neighbors in<br />
French. All target items were baselined to the average of<br />
activity in the 40 ms pre-target period and were lowpass<br />
filtered at 15 Hz. The ERPs were then quantified, as in<br />
Experiment 1, by measuring the mean amplitude in two<br />
latency windows: 175–275 ms and 300–500 ms. The analysis<br />
approach was identical to Experiment 1, but the factor of<br />
target language was eliminated as these monolingual participants<br />
only read words in their L1.<br />
4.3. Simulation study<br />
The model was implemented with a 4 and 5-letter French<br />
word lexicon extracted from the Lexique database (New et al.,<br />
2004), and a 4 and 5-letter English word lexicon from the CELEX<br />
database (Baayen et al., 1995). Only words with at least 2<br />
occurrences per million (opm) were included in the lexicons.<br />
The bilinguals of Experiment 1 were not perfectly balanced<br />
bilinguals, therefore word frequencies (implemented as resting-level<br />
activations of word nodes in the BIA+ model) were<br />
adjusted to simulate such unbalanced high proficiency<br />
bilinguals (see Dijkstra & van Heuven, 1998). The restinglevel<br />
activations of French (L1) were scaled between the<br />
default word node resting-level activation values of the<br />
Interactive Activation (IA) model (McClelland & Rumelhart,<br />
1981). Thus, the resting-level activation of the most frequent<br />
word was set to 0 and the resting-level activation of the least<br />
frequent words (2 opm) was set to −0.92. Other word nodes<br />
were assigned resting-level values between −0.92 and −0.01<br />
based on their word frequency. The resting-level activations of<br />
the English (L2) words were scaled for the bilinguals between<br />
−1.20 and 0. To simulate the English monolingual data of<br />
Experiment 2 we conducted simulations with only English<br />
words with resting-level activations between −0.92 and 0.<br />
Parameters of the BIA+ model with 4-letter words were<br />
identical to those of the IA model. Parameters for the<br />
simulations with the 5-letter word lexicons were identical to<br />
the simulation of the 4-letter words except for the letter-toword<br />
excitation parameter, which was reduced from 0.07 to<br />
0.06 as in the simulations of Grainger and Jacobs (1996). Target<br />
words were presented to the model until the target word<br />
reached the word identification threshold of 0.70.<br />
Acknowledgments<br />
This research was supported by grant numbers HD25889 and<br />
HD043251. The authors would like to thank Courtney Brown<br />
for her help in data collection.
BRAIN RESEARCH 1246 (2008) 123– 135<br />
135<br />
REFERENCES<br />
Alvarez, R.P., Holcomb, P.J., Grainger, J., 2003. Accessing word<br />
meaning in two languages: an event-related brain potential<br />
study of beginning bilinguals. Brain Lang. 87 (2), 290–304.<br />
Andrews, S., 1989. Frequency and neighborhood effects on lexical<br />
access: activation or search? J. Exper. Psychol., Learn., Mem.,<br />
Cogn. 15 (5), 802–814.<br />
Andrews, S., 1997. The effect of orthographic similarity on lexical<br />
retrieval: resolving neighborhood conflicts. Psychon. Bull. Rev.<br />
4 (4), 439–461.<br />
Baayen, R.H., Piepenbrock, R., Gulikers, L., 1995. The Celex Lexical<br />
Database. Linguistic Data Consortium, Philadelphia, PA.<br />
Beauvillain, C., Grainger, J., 1987. Accessing interlexical<br />
homographs: some limitations of a language-selective access.<br />
J. Mem. Lang. 26 (6), 658–672.<br />
Brysbaert, M., Van Dyck, G., Van de Poel, M., 1999. Visual word<br />
recognition in bilinguals: evidence from masked phonological<br />
priming. J. Exp. Psychol. Hum. Percept. Perform. 25 (1), 137–148.<br />
Carreiras, M., Perea, M., Grainger, J., 1997. Effects of the<br />
orthographic neighborhood in visual word recognition:<br />
cross-task comparisons. J. Exper. Psychol., Learn., Mem.,<br />
Cogn. 23 (4), 857–871.<br />
Coltheart, M., Davelaar, E., Jonasson, J.T., Besner, D., 1977. Access<br />
to the internal lexicon. In: Dornic, S. (Ed.), Attention and<br />
Performance VI. Academic Press, London, pp. 535–555.<br />
de Groot, A.M., Nas, G.L., 1991. Lexical representation of cognates<br />
and noncognates in compound bilinguals. J. Mem. Lang. 30 (1),<br />
90–123.<br />
de Groot, A.M., Delmaar, P., Lupker, S.J., 2000. The processing of<br />
interlexical homographs in translation recognition and lexical<br />
decision: support for non-selective access to bilingual memory.<br />
Q. J. Exp. Psychol. A Human Exp. Psychol. 53A (2), 397–428.<br />
Dijkstra, T., van Heuven, W.J.B., 1998. The BIA model and bilingual<br />
word recognition. In: Grainger, J., Jacobs, A.M. (Eds.), Localist<br />
Connectionist Approaches to Human Cognition. Lawrence<br />
Erlbaum Associates, Mahwah, NJ, pp. 189–225.<br />
Dijkstra, T., Van Heuven, W.J.B., 2002. Modeling bilingual word<br />
recognition: past, present and future: reply. Bilingualism Lang.<br />
Cogn. 5 (3), 219–224.<br />
Dijkstra, T., Grainger, J., van Heuven, W.J.B., 1999. Recognition of<br />
cognates and interlingual homographs: the neglected role of<br />
phonology. J. Mem. Lang. 41, 496–518.<br />
Dijkstra, T., Timmermans, M., Schriefers, H., 2000. On being blinded<br />
by your other language: effects of task demands on interlingual<br />
homograph recognition. J. Mem. Lang. 42 (4), 445–464.<br />
Dyer, F.N., 1973. The stroop phenomenon and its use in the study<br />
of perceptual, cognitive, and response processes. Mem. Cogn.<br />
1 (2), 106–120.<br />
Grainger, J., 1990. Word frequency and neighborhood frequency<br />
effects in lexical decision and naming. J. Mem. Lang. 29 (2),<br />
228–244.<br />
Grainger, J., Beauvillain, C., 1987. Language blocking and lexical<br />
access in bilinguals. Q. J. Exp. Psychol. 39A, 295–319.<br />
Grainger, J., Dijkstra, T., 1992. On the representation and use of<br />
language information in bilinguals. In: Harris, R.J. (Ed.),<br />
Cognitive Processing in Bilinguals. North-Holland, Amsterdam,<br />
pp. 207–220.<br />
Grainger, J., Jacobs, A.M., 1996. Orthographic processing in visual<br />
word recognition: a multiple read-out model. Psychol. Rev.<br />
103 (3), 518–565.<br />
Grainger, J., O'Regan, J., Jacobs, A.M., Segui, J., 1989. On the role of<br />
competing word units in visual word recognition: the neighborhood<br />
frequency effect. Percept. Psychophys. 45 (3), 189–195.<br />
Grainger, J., Muneaux, M., Farioli, F., Ziegler, J., 2005. Effects of<br />
phonological and orthographic neighborhood density interact<br />
in visual word recognition. Q. J. Exp. Psychol. 58A, 981–998.<br />
Greenhouse, S.W., Geisser, S., 1959. On methods in the analysis of<br />
profile data. Psychometrika 24, 95–112.<br />
Grosjean, F., 1988. Exploring the recognition of guest words in<br />
bilingual speech. Lang. Cogn. Processes 3 (3), 233–274.<br />
Guttentag, R.E., Haith, M.M., Goodman, G.S., Hauch, J., 1984.<br />
Semantic processing of unattended words by bilinguals: a test<br />
of the input switch mechanism. J. Verbal Learn. Verbal Behav.<br />
23 (2), 178–188.<br />
Holcomb, P.J., Grainger, J., 2006. On the time-course of visual word<br />
recognition: an ERP investigation using masked repetition<br />
priming. J. Cogn. Neurosci. 18 (10), 1631–1643.<br />
Holcomb, P.J., Grainger, J., 2007. Exploring the temporal dynamics<br />
of visual word recognition in the masked repetition priming<br />
paradigm using event-related potentials. Brain Res. 1180,<br />
39–58.<br />
Holcomb, P.J., Grainger, J., O'Rourke, T., 2002. An electrophysiological<br />
study of the effects of orthographic neighborhood size on printed<br />
word perception. J. Cogn. Neurosci. 14 (6), 938–950.<br />
Jared, D., Szucs, C., 2002. Phonological activation in bilinguals:<br />
evidence from interlingual homograph naming. Bilingualism<br />
Lang. Cogn. 5, 225–239.<br />
Macnamara, J., 1967. The linguistic independence of bilinguals.<br />
J. Verbal Learn. Verbal Behav. 6 (5), 729–736.<br />
Macnamara, J., Kushnir, S.L., 1972. Linguistic independence of<br />
bilinguals: the input switch. J. Verbal Learn. Verbal Behav. 10 (5).<br />
McClelland, J.L., Rumelhart, D.E., 1981. An interactive model of<br />
context effects in letter perception: Part i. An account of basic<br />
findings. Psychol. Rev. 88 (5), 375–407.<br />
Marian, V., Spivey, M., 2003. Competing activation in bilingual<br />
language processing: within- and between-language competition.<br />
Bilingualism Lang. Cogn. 6, 97–115.<br />
Nas, G., 1983. Visual word recognition in bilinguals: evidence for a<br />
cooperation between visual and sound based codes during<br />
access to a common lexical store. J. Verbal Learn. Verbal Behav.<br />
22 (5), 526–534.<br />
New, B., Pallier, C., Brysbaert, M., Ferrand, L., 2004. Lexique 2: a new<br />
French lexical database. Behav. Res. Meth. Instrum. Comput.<br />
36 (3), 516–524.<br />
Oldfield, R.C., 1971. The assessment and analysis of<br />
handedness: the Edinburgh inventory. Neuropsychologia 9 (1),<br />
97–113.<br />
Ruschemeyer, S., Nojack, A., Limbach, M., 2008. A mouse with a<br />
roof? Effects of phonological neighbors on processing of words<br />
in sentences in a non-native language. Brain Lang. 104 (2),<br />
113–121.<br />
Soares, C., Grosjean, F., 1984. Bilinguals in a monolingual and a<br />
bilingual speech mode: the effect on lexical access. Mem. Cogn.<br />
12 (4), 380–386.<br />
Tanenhaus, M.K., Magnuson, J.S., Dahan, D., Chambers, C., 2000.<br />
Eye movements and lexical access in spoken-language<br />
comprehension: evaluating a linking hypothesis between<br />
fixations and linguistic processing. J. Psycholinguist. Res. 29 (6),<br />
557–580.<br />
Thomas, M.S., Allport, A., 2000. Language switching costs in<br />
bilingual visual word recognition. J. Mem. Lang. 43 (1), 44–66.<br />
van Hell, J.G., De Groot, A.M.B., 1998. Conceptual representation<br />
in bilingual memory: effects of concreteness and cognate<br />
status in word association. Bilingualism Lang. Cogn. 1 (3),<br />
193–211.<br />
van Hell, J.G., Dijkstra, T., 2002. Foreign language knowledge can<br />
influence native language performance in exclusively native<br />
contexts. Psychon. Bull. Rev. 9 (4), 780–789.<br />
van Heuven, W.J., Dijkstra, T., Grainger, J., 1998. Orthographic<br />
neighborhood effects in bilingual word recognition. J. Mem.<br />
Lang. 39 (3), 458–483.<br />
van Heuven, W.J.B., Schriefers, H., Dijkstra, T., Hagoort, P., 2008.<br />
Language conflict in the bilingual brain. Cereb. Cortex 39 (11),<br />
2706–2716.
Effets de cognats<br />
Chapitre 4<br />
Les effets de cognats sur<br />
la compréhension des mots chez les apprenants<br />
en langue seconde : une étude en potentiels évoqués*<br />
La technique des potentiels évoqués a permis d’examiner différents patterns concernant le<br />
traitement des cognats et des non-cognats chez des apprenants en langue seconde. Des<br />
apprenants du français (L2), dont la première langue (L1) est l’anglais, ont lu des listes de<br />
mots en L1 et L2. Les potentiels évoqués ont été enregistrés par rapport aux deux<br />
conditions : cognats et non-cognats. Des comparaisons entre cognats et non-cognats ont<br />
été réalisées pour les deux langues. Dans les deux langues, les cognats ont produit des<br />
amplitudes réduites dans la région de la N400 par rapport aux non-cognats. Néanmoins,<br />
les différences entre les cognats et les non-cognats pendant le traitement en L1 se<br />
manifestaient dans la première partie de la composante N400 alors que ces mêmes<br />
différences pour le traitement en L2 n'ont été observées que sur une partie plus tardive de<br />
cette même composante. La discussion abordera les effets des cognats sur la<br />
compréhension des mots chez des apprenants en langue seconde.<br />
Mots-clefs : Bilinguisme, Cognats, Acquisition en langue seconde, Reconnaissance<br />
visuelle des mots, Potentiels évoqués<br />
________________________<br />
* Article submitted, Journal of Cognitive Neuroscience, January 2009<br />
P | 71
Effets de cognats<br />
Effects of cognate status on word comprehension in second language learners:<br />
an ERP investigation<br />
Katherine J. Midgley<br />
Tufts University, Medford, Ma USA & CNRS & Université d'<strong>Aix</strong>-<strong>Marseille</strong>,<br />
<strong>Marseille</strong>, France<br />
Phillip J. Holcomb<br />
Tufts University, Medford, Ma USA<br />
Jonathan Grainger<br />
CNRS & Université d'<strong>Aix</strong>-<strong>Marseille</strong>, <strong>Marseille</strong>, France<br />
Address for correspondence:<br />
Katherine J Midgley<br />
Department of Psychology<br />
Tufts University<br />
Medford, MA 02155 USA<br />
kj.midgley@tufts.edu<br />
Voice: (617) 627-3521<br />
Fax: (617) 627-3181<br />
*This Research was supported by grant numbers HD043251 & HD25889.<br />
P | 73
Effets de cognats<br />
Abstract<br />
ERPs were used to explore the different patterns of processing of cognate and noncognate<br />
words in the first (L1) and second language (L2) of a population of second<br />
language learners. L1 English students of French were presented with blocked lists of L1<br />
and L2 words, and ERPs to cognates and non-cognates were compared within each<br />
language block. For both languages, cognates had smaller amplitudes in the N400<br />
component when compared to non-cognates. L1 items that were cognates showed early<br />
differences in amplitude in the N400 epoch when compared to non-cognates. L2 items<br />
showed later differences between cognates and non-cognates than L1 items. The results<br />
are discussed in terms of how cognate status affects word recognition in second language<br />
learners.<br />
Keywords: Bilingualism, Cognates, Second language acquisition, ERPs, Visual word<br />
recognition,<br />
P | 74
Effets de cognats<br />
The 1066 victory of William the conqueror at the battle of Hastings and the<br />
subsequent centuries of Norman rule had enormous impact on the England of the middle<br />
ages. One of the legacies of this period can be found in the relationship of the French and<br />
English languages. The imposition of French on English during that time resulted in the<br />
incorporation of many tokens of French into English, a language already accepting of<br />
many borrowed forms. This shared history leaves the two languages today with a<br />
multitude of shared words like “table”. This type of non-accidental overlap of form in<br />
translation equivalents is what defines cognate words. Many English–French cognates<br />
have complete written overlap like “fruit” or near complete overlap like “mask” and<br />
“masque”. In the study presented here, we examined the processing of cognate words<br />
such as these as compared to non-cognate words by English-French beginning bilinguals.<br />
Cognates’ membership in both of a bilingual’s lexicons make them a special case. It<br />
is this special status that will be exploited in the present study in order to address different<br />
questions of bilingual lexical access. Because of their shared form with L1 items,<br />
cognates, during L2 acquisition, could be a learner’s first foothold into the new lexicon.<br />
Presumably in the early stages of acquisition this would result in different patterns of<br />
processing for cognates and non-cognates while processing L2. In the case of L1<br />
processing, if cognates showed different patterns of processing when compared to noncognates<br />
this would be evidence of the L1 changing as a function of learning an L2, and<br />
also point to an integrated lexicon with non-selective access for the two languages.<br />
P | 75
Effets de cognats<br />
In behavioral studies cognate items have been shown to elicit different response<br />
patterns compared to non-cognate items. Cognates are more rapidly recognized than noncognates<br />
in isolated word recognition tasks such as lexical decision (Dijkstra, Grainger &<br />
van Heuven, 1999; Lemhöfer & Dijkstra, 2004; Lemhöfer, Dijkstra & Michel, 2004).<br />
Cognate items have been shown to be translated more quickly than non-cognate items (de<br />
Groot, 1992; Sánchez-Casas, Davis, & García-Albea, 1992). Both Dijkstra et al. (1999)<br />
and Lemhöfer and Dijkstra (2004) tested cognates mixed with non-cognate words from<br />
L2. However, studies testing cognate processing in an L1 context have given mixed<br />
results. Some authors failed to observe a difference between cognate and non-cognate<br />
words in an L1 context (Caramazza & Brones, 1979; Gerard & Scarborough, 1989) while<br />
others did find effects (van Hell & de Groot,1998; de Groot, Delmaar, & Lupker, 2000;<br />
van Hell & Dijkstra, 2002). Thus, in general, it would appear that word recognition in L2<br />
benefits from cognate status, whereas word recognition in L1 is more impervious to such<br />
influences. This is in line with other observed asymmetries in bilingual lexical<br />
processing, such as the greater strength of translation priming from L1-L2 compared with<br />
L2-L1 (see Midgley, Holcomb, & Grainger, in press-b, for a recent example).<br />
Effects of cognate status on word recognition in L1 have, however, been<br />
documented. Van Hell and Dijkstra (2002) showed clear effects of cognate status during<br />
L1 processing and this while testing 2 groups of trilinguals of different proficiencies in<br />
their L3. In experiments 2 and 3 they used a lexical decision task in L1 (Dutch) to<br />
examine the influence of cognates from both L2 (English) and L3 (French). They found a<br />
similar advantage for L2 cognates for both groups. However a significant advantage for<br />
L3 cognate items was observed only for the groups having more proficiency in L3. The<br />
group with less proficiency in L3 showed only a weak trend toward a cognate advantage.<br />
They concluded that a certain level of proficiency is necessary in the bilinguals’ non-<br />
P | 76
Effets de cognats<br />
target language relative to their target language in order to observe effects on processing<br />
in the target language. In other words a bilingual must have enough fluency in an L2 or<br />
L3 for cognate status to influence L1 processing.<br />
Strong cognate effects have also been reported in a variety of priming paradigms<br />
(Bowers, Mimouni, & Arguin, 2000; Cristoffanini, Kirsner, & Milech, 1986; Lalor &<br />
Kirsner, 2001), including masked priming (e.g., de Groot & Nas, 1991; Gollan, Forster, &<br />
Frost, 1997; Sánchez-Casas et al., 1992) where the contribution of strategic factors are<br />
less likely to have influenced the findings. Voga and Grainger (2007) found priming for<br />
both cognates and non-cognates across different scripts (Greek and French). When<br />
cognates and non-cognates were compared to unrelated primes cognates showed greater<br />
priming, but when cognates and non-cognates were compared to phonologically related<br />
primes this advantage disappeared. Voga and Grainger concluded that this difference in<br />
priming size as a function of baseline comparison was evidence that the cognate<br />
advantage is simply due to the additional form overlap of cognates and not some special<br />
status of cognate words. Together this literature points towards an advantage in<br />
processing for words from a bilingual’s two languages when the items share both form<br />
and meaning.<br />
The nature and locus of the cognate advantage<br />
In behavioral studies the locus of the cognate effect is difficult to establish. In<br />
laboratory tasks such as lexical decision, demand characteristics of an experiment could<br />
exert their influence on performance in ways that are relatively uninformative about the<br />
word recognition system per se. Furthermore the observation that a certain level of L2<br />
proficiency is needed in order to observe behavioral effects on L1 processing raises the<br />
possibility that effects with lower levels of proficiency were not observed because of poor<br />
P | 77
Effets de cognats<br />
measurement sensitivity. Perhaps a more sensitive measure could be used to observe L2<br />
effects on L1 processing even at relatively early stages of second language acquisition.<br />
Consistent with this view van Hell and Dijkstra showed a non-significant trend in less<br />
proficient bilinguals in the same direction as the significant effect they reported for the<br />
more proficient participants. Of interest here is whether with a more sensitive measure<br />
they would have found stronger evidence for an effect of cognate status on L1 even in<br />
relatively non-proficient learners of a second language. Such an effect would be strong<br />
evidence that even at early stages of becoming bilingual that there are profound changes<br />
in the L1 as a function of learning a new language.<br />
What mechanisms could be at the basis of the observed behavioral advantage in<br />
processing cognate words compared with non-cognates in bilinguals and L2 learners? The<br />
most straightforward interpretation is in terms of increased exposure to the same<br />
orthographic and/or phonological patterns, and most critically, the same association<br />
between a given form representation and its corresponding meaning. This would account<br />
for why increased proficiency in the non-native language causes an increase in the effects<br />
of cognate status when processing L1 words (van Hell & Dijkstra, 2002). The strong<br />
cognate advantage found when processing L2 words would be due to the cognate words<br />
benefiting from pre-existing form-meaning associations in the L1. However, this specific<br />
processing advantage must be evaluated against a general background of overall<br />
differences in processing L1 and L2 words related to a multitude of other factors<br />
(Midgley, Holcomb & Grainger, in press-a).<br />
ERPs and bilingual word recognition<br />
The present experiment examined the nature and time course of cognate processing<br />
as it compares to non-cognate processing in a bilingual’s two languages by using<br />
P | 78
Effets de cognats<br />
electrophysiological measures. Electrophysiological measures are used because they more<br />
directly reflect the processing of items, being an on-line measure of brain activity rather,<br />
as in the case with RTs, just one data point after processing has been completed. Eventrelated<br />
potentials (ERPs) are measures of the brain’s electrical activity recorded at the<br />
scalp and obtained by averaging time-locked responses to stimuli onset thus extracting<br />
the voltage signature of the processing of the items of interest from the background<br />
electroencephalogram. ERPs are multidimensional in that they contain time-course<br />
information, scalp distribution information along with the voltage measures. This<br />
multidimensionality is informative not only about the time course of word processing but<br />
also about differences in the nature of this processing.<br />
Of particular interest to studies of word processing using ERPs is a negative going<br />
component that starts around 250 ms post word onset and continues on until about 600<br />
ms. This ERP component, called the N400, has been shown to reflect lexical and<br />
semantic processes associated with word recognition, being larger whenever a word is<br />
more difficult to process or integrate into its surrounding context (Lau, Phillips &<br />
Poeppel, 2008; Holcomb et al., 1993). Numerous studies have shown that the amplitude<br />
of the N400 is sensitive to a host of linguistic variables including word frequency (larger<br />
to low frequency words than high, e.g., Van Petten & Kutas, 1990; Münte et al., 2001)<br />
and orthographic neighborhood density (larger to words with more dense neighborhoods,<br />
Holcomb, Grainger & O’Rourke, 2002; Midgley, Holcomb, van Heuven & Grainger,<br />
2008). These results, plus the results of many other studies using single word stimuli<br />
(e.g., Holcomb & Grainger, 2006; 2007) all suggest that an increased amplitude in the<br />
N400 component reflects an increased difficulty in processing the target word, and more<br />
specifically in terms of settling on a unique form-meaning interpretation.<br />
P | 79
Effets de cognats<br />
The present study<br />
In the present experiment, we sought electrophysiological evidence for the cognate<br />
advantage reported in previous behavioral experiments, and investigated whether this<br />
cognate effect differs while processing in L2 and L1. Participants read lists of words for<br />
meaning while making occasional button presses to probes from a specific semantic<br />
category. The critical items were words that were either cognates or non-cognates. In one<br />
block of trials all items were in L1, and in the other block all items were in L2. This<br />
design allowed us to directly compare ERPs to cognates and non-cognates in the two<br />
language blocks. Based on previous ERP work investigating single word recognition and<br />
prior behavioral research on cognate processing we expected to see reduced negativities<br />
to cognate words compared with non-cognate words in the N400 time window.<br />
Furthermore, on the hypothesis that the cognate advantage reflects an accumulation of the<br />
benefits of exposure to a given form-meaning association across two languages, then we<br />
would expect to see stronger effects when processing L2 words than when processing L1<br />
words. However, with respect to the relative time-course of visual word recognition in L1<br />
and L2 (L1 being arguably faster than L2), we expect to see earlier effects of cognate<br />
status in L1 compared with L2 words. This is because the mapping between form and<br />
meaning, the hypothesized locus of the cognate advantage, will occur earlier in the<br />
processing of an L1 word than an L2 word. Here it is assumed that cognates are processed<br />
differently in L1 and L2, due to a global influence of language context that tunes the<br />
lexical processor to be prepared for words from one or the other language (Grainger,<br />
Midgley & Holcomb, in preparation).<br />
P | 80
Effets de cognats<br />
Methods<br />
Participants<br />
Fifty participants were recruited and compensated for their time. The data from<br />
eight participants was not used due to excessive artifacts in their data or incomplete scalp<br />
recordings. Of the remaining 42, 33 were women. The mean age of participants was 20<br />
years (SD = 1.7), all reported to be right handed and had normal or corrected-to-normal<br />
visual acuity with no history of neurological insult or language disability. English was<br />
reported to be the first language learned by all participants (L1) and French their primary<br />
second language (L2). All participants were undergraduate students at Tufts University<br />
who were either currently enrolled in a French class or had previously studied French at<br />
Tufts.<br />
Stimuli<br />
One hundred and sixty items were chosen that were cognates in English and French.<br />
These items were cognates with complete overlap of form (eg. “table” in both English<br />
and French) and very close cognates (“victim” and “victime”). The mean orthographic<br />
overlap of the items was 89.0% (SD=14.7%). One hundred and sixty items were chosen<br />
(80 English items and 80 French items) that were non-cognates between English and<br />
French. That is, they had no obvious form overlap (e.g., “apple” and “pomme”). The<br />
mean orthographic overlap of these non-cognate items was 7.2% (SD=10.5%). All items<br />
in all conditions and for both languages were between four and seven letters in length.<br />
The English cognates had a mean frequency per million of 31.48 (SD = 49.94) (CELEX<br />
database, 1993; Baayen, Piepenbrock, & Gulikers, 1995) while the French cognates had a<br />
mean frequency per million of 26.50 (SD = 30.91) (Lexique database; New, Pallier,<br />
Brysbaert, & Ferrand, 2004). These means were not statistically different (t(318) = 1.19,<br />
P | 81
Effets de cognats<br />
p = 0.24). The English non-cognates had a mean frequency per million of 38.33 (SD =<br />
43.16) while the French non-cognates had a mean frequency per million of 28.17 (SD =<br />
24.92). These means were not statistically different (t(158) = 1.82, p = 0.07). Furthermore<br />
English cognate mean frequency compared to English non-cognate mean frequency did<br />
not differ significantly (t(238) = 1.16, p = 0.25) nor did French cognate mean frequency<br />
differ from French non-cognate mean frequency (t(238) = 0.42, p = 0.68).<br />
Two lists were formed, each list being composed of two blocks; an English block<br />
and a French block. Each list contained 80 English cognates and English 80 non-cognates<br />
for the English block and 80 French cognates and 80 French non-cognates. The items in<br />
these two lists were counterbalanced to avoid repetition of the cognate items across<br />
language. That is to say that no one participant saw both an English cognate and its<br />
French equivalent. The items in each language block were in a pseudorandom order.<br />
Intermixed in each list was a second group of 40 probe items which were all members of<br />
the semantic category of “animal names” (probes were English animal names in the<br />
English block and French animal names in the French block). All participants saw the<br />
same animal names. The animal names varied in cognate status similar to the critical<br />
items (i.e., a mix of complete and close cognates and non-cognates) and were also four to<br />
seven letters in length. The order of the language blocks was counter-balanced across<br />
participants.<br />
Procedure<br />
The word stimuli in each list were presented as white letters centered vertically and<br />
horizontally on a black background on a 19 inch color CRT monitor. Presentation of all<br />
visual stimuli and digitizing of the EEG was synchronized with the vertical retrace<br />
interval (60 Hz refresh rate) of the stimulus PCs video card (ATI Radeon) to assure<br />
P | 82
Effets de cognats<br />
precise time marking of ERP data. The participants were seated so that their eyes were at<br />
a distance of approximately 1.5 meters from the screen. The maximum height and width<br />
of the stimuli were such that no saccades would be required during reading of the single<br />
word stimuli. Participant responses were made using a button box held in the lap<br />
throughout the experiment. A go/no-go semantic categorization task was used in which<br />
participants were instructed to read all words for meaning and to press a button whenever<br />
they saw a word referring to an animal name. Forty trials in each language block were<br />
animal names (12% of all trials – see <strong>Fig</strong>ure 1 for a typical series of trials). As can be<br />
seen, each trial began with the presentation of an item for a duration of 300 ms followed<br />
by a blank screen for a duration of 1000 ms. Each trial ended with a stimulus indicating<br />
that it was permissible to move or blink the eyes. This blink stimulus [“( - - )”] had a<br />
duration of 2000 ms followed by 500 ms of blank screen before the next item appeared.<br />
<strong>Fig</strong>ure 1. Schematic of three trials in the English block, a cognate, a probe word and a non-cognate.<br />
Only the probe word (buffalo) requires a button pressing response.<br />
P | 83
Effets de cognats<br />
After electrode placement instructions for the experimental task were given in<br />
English, the L1 of the participants, then a short practice list in the language of the first<br />
block was presented to assure good performance during experimental runs and to<br />
accustom the participant to the coming language. A practice list was also run before the<br />
second block in the language of that block. There were four pauses within each block; the<br />
length of these pauses was determined by the participant. Each language block typically<br />
required 15 minutes. Participants were asked to press a button on the response box every<br />
time they saw an animal name. At the end of the ERP experiment participants were asked<br />
to give a translation of the English words that they had seen during the experiment. These<br />
post-translations were graded for accuracy.<br />
EEG recording<br />
Participants were seated in a comfortable chair in a sound attenuating room and<br />
were fitted with an elastic cap equipped with 29 tin electrodes (Electro-cap International -<br />
- see <strong>Fig</strong>ure 2 for the location of electrodes). Two additional electrodes were used to<br />
monitor for eye-related artifact (blinks and vertical or horizontal eye movement); one<br />
below the left eye (VE) and one horizontally next to the right eye (HE). All electrodes<br />
were referenced to an electrode placed over the left mastoid (A1). A final electrode was<br />
placed over the right mastoid (A2 - used to determine if there was any asymmetry<br />
between the mastoids; none was observed). The 32 channels of electrophysiological data<br />
were amplified using an SA Instruments Bio-amplifier system with 6db cutoffs set at .01<br />
and 40Hz. The output of the bio-amplifier was continuously digitized at 200 Hz<br />
throughout the experiment.<br />
P | 84
Effets de cognats<br />
<strong>Fig</strong>ure 2. Electrode Montage.<br />
Data analysis<br />
Averaged ERPs time-locked to item onset of words in each category were formed<br />
off-line from trials free of ocular and muscular artifact (less than 8.1 % of trials). ERPs<br />
were averaged separately for English cognates, English non-cognates, French cognates<br />
and French non-cognates providing factors of LANGUAGE and COGNATE-STATUS IN A 2X2<br />
DESIGN. All items were baselined to the average of activity in the 100 ms pre-target<br />
period and were lowpass filtered at 15 Hz. The ERPs were then quantified by measuring<br />
the mean amplitude in three latency windows: 200 -300 ms to capture pre-N400 activity,<br />
and 300-500 ms to capture the N400 itself and 500 – 800 to capture late cognate effects.<br />
In order to thoroughly analyze the full montage of 29 scalp sites we employed an<br />
approach to data analysis that we have successfully applied in a number of previous<br />
studies (e.g., Holcomb, Reder, Misra & Grainger, 2005). In this scheme the 29 channel<br />
electrode montage is divided up into seven separate parasagittal columns along the<br />
antero-posterior axis of the head (see <strong>Fig</strong>ure 2). The electrodes in each of three pairs of<br />
lateral columns and one midline column are analyzed in four separate ANOVAs. Three of<br />
P | 85
Effets de cognats<br />
these analyses (referred to as column 1 or c1, column 2 or c2 and column 3 or c3)<br />
involved an anterior/posterior ELECTRODE-SITE factor with either three, four or five levels,<br />
as well as a HEMISPHERE factor (left vs. right). The forth “midline” analysis included a<br />
single anterior/posterior ELECTRODE-SITE factor with five levels.<br />
Significant interactions in the omnibus analyses involving factors of language and<br />
cognate-status were decomposed with followed-up ANOVAs looking at each language<br />
(English and French) separately. The Geisser-Greenhouse (1959) correction was applied<br />
to repeated measures with more than one degree of freedom in the numerator. Finally, to<br />
more carefully track the temporal properties of cognate effects we also performed timecourse<br />
analyses (TCAs) comparing the cognate and non-cognate ERPs at each of five the<br />
midline electrodes and for the two languages in eight consecutive 100 ms windows<br />
between 0 and 800 ms.<br />
Results<br />
Visual inspection of ERPs<br />
Plotted in <strong>Fig</strong>ure 3 are the grand mean ERP waveforms for cognate and non-cognate<br />
L1 (English) items. Contained in <strong>Fig</strong>ure 3a are ERPs from all 29 scalp sites, in <strong>Fig</strong>ure 3b<br />
enlarged plots of three midline sites <strong>Fig</strong>ure 4 contains the same plots for the L2 (French)<br />
block of trials. <strong>Fig</strong>ure 5a shows voltage maps at six points in time for L1 items, <strong>Fig</strong>ure 5b<br />
for L2 items. The voltage maps are a subtraction of ERPs for items that are cognates from<br />
ERPs for items that are non-cognates. Accompanying time-course analyses are presented<br />
in Table 1. As can be seen in these figures 3 and 4 for ERPs anterior to the occipital sites<br />
the first visible component was a negative-going deflection between 90 and 150 ms after<br />
stimulus onset (N1). This was followed by a positive deflection occurring at<br />
approximately 200 ms (P2). A negativity followed the P2 peaking around 400 ms (N400).<br />
P | 86
Effets de cognats<br />
At occipital sites the first observable component is the P1, which peaked near 100 ms and<br />
was followed by the N1 at 190 ms and a broad P2 between 250 and 300 ms. The P2 was<br />
followed by the N400 between 400 and 600 ms.<br />
Analyses of ERP data<br />
200 - 300 ms epoch.<br />
An omnibus ANOVA on the mean amplitude values in this epoch revealed<br />
significant main effects of LANGUAGE at all columns (midline: F(1,41) = 7.64, p = 0.009,<br />
c1: F(1,41) = 9.14, p = 0.004, c2: F(1,41) = 12.69, p = 0.001, c3: F(1,41) = 7.59, p =<br />
0.009). There was also a main effect of COGNATE-STATUS, but it reached significance only<br />
at column 1 (F(1,41) = 9.14, p = 0.004). The two-way interaction between LANGUAGE and<br />
COGNATE-STATUS was reliable at all columns (midline: F(1,41) = 9.93, p = 0.003, c1:<br />
F(1,41) = 11.76, p = 0.001, c2: F(1,41) = 10.23, p = 0.003, c3: F(1,41) = 9.23, p = 0.003).<br />
Follow-up analyses examining the effects of COGNATE-STATUS separately for the<br />
two languages in this epoch revealed effects at all columns for L1 (midline: F(1,41) =<br />
14.14, p = 0.001, c1: F(1,41) = 16.03, p < 0.001, c2: F(1,41) = 13.04, p = 0.001, c3:<br />
F(1,41) = 11.01, p = 0.002) with non-cognates being more negative than cognates. For L2<br />
no effect of COGNATE-STATUS was observed in this epoch.<br />
300 - 500 ms epoch.<br />
An omnibus ANOVA on the mean amplitude values in this epoch revealed<br />
significant main effects of COGNATE-STATUS at all columns (midline: F(1,41) = 8.86, p =<br />
0.005, c1: F(1,41) = 9.53, p = 0.004, c2: F(1,41) = 11.21, p = 0.002, c3: F(1,41) = 13.46,<br />
p = 0.001) as well as three-way interactions of LANGUAGE x COGNATE-STATUS x<br />
P | 87
Effets de cognats<br />
ELECTRODE-SITE at all columns (midline: F(4,164) = 7.34, p < 0.001, c1: F(2,82) = 9.36, p<br />
= 0.001, c2: F(3,123) = 9.56, p < 0.001, c3: F(4,164) = 4.39, p = 0.021).<br />
Follow-up analyses examining the effects of COGNATE-STATUS separately for the<br />
two languages revealed effects of COGNATE-STATUS at all columns for L1 (midline:<br />
F(1,41) = 23.44, p < 0.001, c1: F(1,41) = 16.18, p < 0.001, c2: F(1,41) = 15.48, p < 0.001,<br />
c3: F(1,41) = 14.70, p < 0.001). These analyses suggest that during the English task noncognate<br />
words tended to produce more negative-going ERPs in this epoch than cognate<br />
words. For the L2 task (French target words) there were no main effects of COGNATE-<br />
STATUS, however, there were significant interactions of COGNATE-STATUS x ELECTRODE-<br />
SITE (midline: F(4,164) = 12.21, p < 0.001, c1: F(2,82) = 9.86, p = 0.001, c2: F(3,123) =<br />
14.59, p < 0.001, c3: F(4,1641) = 9.53, p = 0.001). As can be seen at Pz in <strong>Fig</strong>ure 4b<br />
while French non-cognates tended to be more negative-going at anterior and central sites,<br />
at the more posterior sites cognates tended to be more negative-going than non-cognates.<br />
500 - 800 ms epoch.<br />
An omnibus ANOVA on the mean amplitude values in this epoch revealed<br />
significant main effects of COGNATE-STATUS in all four columnar analyses (midline:<br />
F(1,41) = 4.41, p = 0.042, c1: F(1,41) = 6.17, p = 0.017, c2: F(1,41) = 6.94, p = 0.012, c3:<br />
F(1,41) = 4.39, p = 0.042). There were also two-way interactions of LANGUAGE x<br />
ELECTRODE-SITE (midline: F(4,164) = 3.75, p = 0.026, c1: F(2,82) = 4.47, p = 0.021, c2:<br />
F(3,123) = 4.07, p = 0.034) and a three-way interaction of LANGUAGE x COGNATE-STATUS<br />
x ELECTRODE-SITE (midline: F(4,164) = 7.39, p = 0.001, c1: F(2,82) = 3.96, p = 0.039, c2:<br />
F(3,123) = 6.32, p = 0.004, c3: F(4,164) = 4.00, p = 0.027).<br />
P | 88
Effets de cognats<br />
<strong>Fig</strong>ure 3a. Results of L2 learners reading L1 (English) items that are cognates or non-cognates<br />
<strong>Fig</strong>ure 3b. Enlargement of five sites from 3a.<br />
P | 89
Effets de cognats<br />
<strong>Fig</strong>ure 4a. Results of L2 learners reading L2 (French) items that are cognates or non-cognates.<br />
<strong>Fig</strong>ure 4b. Enlargement of five sites from 4a.<br />
P | 90
Effets de cognats<br />
<strong>Fig</strong>ure 5a. Scalp voltage maps at six time points showing the difference in voltage between L1 noncognate<br />
items and L1 cognate items (units are in microvolts)<br />
<strong>Fig</strong>ure 5b. Scalp voltage maps at six time points showing the difference in voltage between L2 noncognate<br />
items and L2 cognate items (units are in microvolts).<br />
Follow-up analyses examining the effects of COGNATE-STATUS separately for the<br />
two languages revealed no effects at any column for L1 (all p > .150). For L2, COGNATE-<br />
STATUS was marginally significant at midline and column 1 (F(1,41) = 3.83, p = 0.057,<br />
F(1,41) = 3.39, p = 0.073) and reached significance at columns 2 and 3 (F(1,41) = 5.78, p<br />
= 0.021, F(1,41) = 7.82, p = 0.008). As can be seen in <strong>Fig</strong>ure 4 non-cognates tended to be<br />
more negative-going than cognates at the more lateral sites in this epoch.<br />
Time-course analysis<br />
Table 1. Time-course analysis of the Cognate effect in 100 ms epochs at five midline sites.<br />
0-100ms 100-200ms 200-300ms 300-400ms 400-500ms 500-600ms 600-700ms 700-800ms<br />
L1 : FPz - - N~C N>>C - - C>>N C>>N<br />
Fz N>C N>>>C N~C - C>N -<br />
Cz N>>>C N>>>C N>C - - -<br />
Pz - N>C N>>>C N>>>C N>>C N>C N~C N>>C<br />
Oz - N>>C - - - N>>C<br />
L2 : FPz - - - - N>>>C N>C N~C -<br />
Fz - N>>>C N>>C N~C N~C<br />
Cz - N>C N>C - -<br />
Pz - - - C~N - N>C - -<br />
Oz C~N C>>>N - - - C~N<br />
C= cognates, N = non-cognates, letter on the left is more negative than the one on the right<br />
- ns, ~ p = 0.1, > p = 0.05, >> p = 0.01, >>> p = 0.001<br />
P | 91
Effets de cognats<br />
Behavioral results<br />
Participants averaged 39.4 (SD = 1.7) out of 40 hits in their L1 (98.4%) and 34.2<br />
(SD = 4.3) out of 40 hits in their L2 (85.5%) for the animal probe words. Participants<br />
produced false alarms on an average of 0.8 items (SD = 1.1) in L1 (0.3%) and on 15.8<br />
items (SD = 5.7) in L2 (2.4%). In a post translation task participants were asked to<br />
translate all 240 L2 items that they had seen in the experiment. The mean number of<br />
correct translations was 175 (SD = 25.0) or 73%.<br />
Discussion<br />
In this experiment, testing a group of second language learners, we sought<br />
electrophysiological evidence for effects of cognate status reported in prior behavioral<br />
research. We recorded and compared ERPs to cognate and non-cognate words while<br />
participants were processing blocked lists of words in their L1 and their L2. ERP<br />
negativities in the region of the N400 component were found to be sensitive to cognate<br />
status in both language blocks. As in a number of previous behavioral studies (Dijkstra et<br />
al., 1999; Lemhöfer & Dijkstra, 2004; Lemhöfer et al., 2004; de Groot, 1992; Sánchez-<br />
Casas et al., 1992) there were robust effects of cognate status when participants were<br />
processing words in L2. In the present experiment ERPs were more negative for noncognates<br />
(i.e., larger N400s) than cognates, although this effect did not start until around<br />
300 ms and did not become widespread across the scalp until after 550 ms. Perhaps more<br />
interesting, because there have been fewer behavioral studies showing these effects, there<br />
were also cognate effects in the L1 block. Like L2, non-cognates were more negative than<br />
cognates, but this difference started earlier, in a 200-300 ms window. These effects were<br />
widespread across the scalp and continued through the traditional 300-500 ms N400<br />
window.<br />
P | 92
Effets de cognats<br />
The principle effect of cognate status in the present experiment was therefore a<br />
reduced negativity (smaller N400 amplitude) to cognate words compared with noncognate<br />
words in both L1 and L2. This fits with the general hypothesis that the mapping<br />
of form to meaning is facilitated in cognate words. Other examples of an interpretation of<br />
reduced N400 amplitude as reflecting greater ease in mapping form onto meaning in<br />
single word recognition are effects of word frequency (Van Petten & Kutas, 1990; Münte<br />
et al., 2001) effects of orthographic neighborhood (Midgley et al., 2008; Holcomb et al.,<br />
2002), and effects of masked primes (Holcomb & Grainger, 2006; 2007).<br />
However, perhaps the central finding of the present experiment is that of an<br />
influence of cognate status on word recognition in the first language (L1). Prior<br />
behavioral research had provided mixed findings on this particular issue. Several previous<br />
studies had not found cognate effects in L1 (Caramazza & Brones, 1979; Gerard &<br />
Scarborough, 1989) and others found an influence of cognate status on L1 items only in<br />
relatively proficient participants (van Hell & Dijkstra, 2002). Here, using<br />
electrophysiological measures, we were able to observe cognate effects in relatively low<br />
proficiency language learners processing words in their L1.<br />
The ERP data were also informative about the timing of cognate effects. These<br />
effects began to emerge at about 200 ms in L1, but not until about 400 ms in L2. In the<br />
introduction we hypothesized that the cognate advantage seen in behavioral research<br />
would reflect an accumulation of the benefits of exposure to a given form-meaning<br />
association across two languages. On this basis we not only expected to see effects of<br />
cognate status on processing L1 words, but we also were led to predict earlier effects of<br />
cognate status in L1 compared with L2 words. This is because the mapping between form<br />
and meaning, the hypothesized locus of the cognate advantage, should arise earlier in the<br />
processing of an L1 word than an L2 word. The results of the present experiment are in<br />
P | 93
Effets de cognats<br />
line with this prediction, and therefore lend support to an account of cognate effects in<br />
terms of increased exposure to form-meaning associations. Again, as mentioned in the<br />
introduction, interpreting cognate effects as effects of cumulative exposure to the same<br />
(or similar) form-meaning associations does not imply that these words will be processed<br />
identically in each language. Language context is thought to have a global influence on<br />
lexical processing in L1 and L2, such that a cognate word will be processed like a word in<br />
L1 or L2 as function of the context.<br />
There is, however, an alternative interpretation of the difference in timing of<br />
cognate effects in L1 and L2 found in the present experiment. This interpretation is<br />
described within the framework of the revised hierarchical model (RHM) of second<br />
language vocabulary acquisition (Kroll & Stewart, 1994). In the RHM, beginning<br />
bilinguals first learn to map new word forms in the L2 onto their translation equivalents<br />
in L1. Hence meaning access from an L2 word is initially achieved via a translation route<br />
from the L2 word to its L1 translate, and from there, to the word meaning. Increased<br />
exposure to the L2 eventually allows the development of direct associations between L2<br />
word forms and meaning. Within this theoretical framework, one could argue that<br />
cognate words are particularly easy to process in the early phases of L2 acquisition given<br />
that their translations equivalents have the same or very similar orthographic forms. Noncognate<br />
words would be harder to process because of the increased difficulty in mapping<br />
the L2 form onto its translation equivalent in L1. This would therefore explain why the<br />
cognate effect arises quite late in processing L2 words, but it does not explain why there<br />
is an effect in L1. Extra exposure to form-meaning associations while learning an L2<br />
therefore remains one viable account of the early effects of cognate status found in L1.<br />
The pattern of effects found in L2 also differed with respect to the L1 pattern in one<br />
other notable way. That is the reversed cognate effect found in posterior sites at around<br />
P | 94
Effets de cognats<br />
300 ms post-stimulus onset in L2. This might well reflect a conflict in the mapping of<br />
orthography to phonology that is exaggerated in the case of cognate words, since the<br />
same orthographic pattern maps onto two distinct pronunciations (e.g., the different<br />
pronunciations of the word “table” in French and English). The fact that this is only seen<br />
in L2 can be explained by the relative dominance of the L1 pronunciation over the L2<br />
pronunciation of cognate words. Furthermore, the timing of this putative phonological<br />
effect is in line with estimates of phonological influences on visual word recognition in<br />
monolinguals (e.g., Bentin, Mouchetant-Rostaing, Giard, Echallier, & Pernier, 1999;<br />
Grainger, Kiyonaga, & Holcomb, 2006).<br />
Are cognates special in any way that would not be predicted by a simple<br />
combination of shared form and meaning across languages? On the basis of the results of<br />
their masked priming study, Voga and Grainger (2007) suggested not. On the other hand,<br />
Van Hell and Dijkstra and others (e.g., De Groot et al., 2000; Dijkstra et al., 1998, 1999)<br />
have argued that cognates have a special type of representation in the mental lexicon, and<br />
cognate effects are not just cumulative frequency effects (i.e., that it is not just the sum of<br />
the frequencies of the cognate items across languages that is driving the cognate effect).<br />
They point to effects of “faux amis” or inter-lingual homographs as evidence against a<br />
cumulative frequency interpretation of cognate effects. Unlike cognates, inter-lingual<br />
homographs, whose accidental overlap of form with an absence of overlap of meaning<br />
(“four” in French means “oven”) show a disadvantage in processing, with slower RTs<br />
than control words. According to a cumulative frequency explanation of the cognate<br />
advantage, inter-lingual homographs should also show a processing advantage and they<br />
do not. However, the critical difference between cross-language homographs and cognate<br />
words is that homographs map onto different meanings in each language. Therefore, if as<br />
we hypothesize, it is the frequency of form-meaning mappings that underlies the cognate<br />
P | 95
Effets de cognats<br />
advantage, one would not expect to see such an advantage for cross-language<br />
homographs.<br />
Nevertheless, there is one remaining problem with the form-meaning frequency<br />
account of cognate effects that we propose. This is that one would expect to see a much<br />
larger cognate effect in L2 than in L1, because L2 words should benefit from much larger<br />
amounts of exposure to L1 words than vice versa. There was no evidence for this in the<br />
present experiment, with effects of comparable magnitude in L1 and L2. As already<br />
suggested above, the fact that relatively low-proficiency bilinguals were tested in the<br />
present experiment might point to differences in ease of translation as the source of the<br />
cognate effect in L2. Thus, it could be that cognates do have a special status during the<br />
earliest phases of second language learning, and that this special status is gradually lost as<br />
proficiency in L1 increases. In order to investigate this possibility, future work should<br />
compare cognate effects in participants with different levels of proficiency in their L2.<br />
It is clear that more empirical work must be done to address questions like these, but<br />
it is always interesting to provide parsimonious explanations for language effects. In a<br />
bilingual context, parsimony could be achieved by turning to well documented, well<br />
understood and well modeled L1 lexical phenomena to explain L2 phenomena rather than<br />
developing a long list of particulars for L2 processing. A true universal account of<br />
language processing must parsimoniously describe both monolingual processing and<br />
bilingual processing and the representations and processes involved.<br />
P | 96
Effets de cognats<br />
REFERENCES<br />
Baayen, R. H., Piepenbrock, R. & Gulikers, L. (1995). The CELEX Lexical Database<br />
Philadelphia: University of Pennsylvania, Linguistic Data Consortium.<br />
Bentin, S., Mouchetant-Rostaing, Y., Giard, M.H., Echallier, J.F. & Pernier, J. (1999).<br />
ERP manifestations of processing printed words at different psycholinguistic<br />
levels: time course and scalp distribution. Journal of Cognitive Neuroscience,<br />
11(3), 35–60.<br />
Bowers, J. S., Mimouni, Z. & Arguin, M. (2000). Orthography plays a critical role in<br />
cognate priming: Evidence from French/English and Arabic/French cognates.<br />
Memory & Cognition, 28, 1289-1296.<br />
Caramazza, A. & Brones, I. (1979). Lexical access in bilinguals. Bulletin of the<br />
Psychonomic Society, 13, 212-214.<br />
CELEX English database (Release E25). (1993). Available: www.mpi.nl/world/celex<br />
Cristoffanini, P., Kirsner, K. & Milech, D. (1986). Bilingual lexical representation: The<br />
status of Spanish–English cognates. Quarterly Journal of Experimental<br />
Psychology, 38A, 367-393.<br />
De Groot, A. M. B. (1992). Determinants of word translation. Journal of Experimental<br />
Psychology: Learning, Memory, and Cognition, 18, 1001-1018.<br />
De Groot, A. M. B., Delmaar, P. & Lupker, S. J. (2000). The processing of interlexical<br />
homographs in translation recognition and lexical decision: Support for<br />
nonselective access to bilingual memory. Quarterly Journal of Experimental<br />
Psychology, 53A, 397-428.<br />
De Groot, A. M. B. & Nas, G.L. (1991). Lexical representation of cognates and noncognates<br />
in compound bilinguals. Journal of Memory & Language, 30, 90-123.<br />
Dijkstra, A., van Jaarsveld, H. & Ten Brinke, S. (1998). Interlingual homograph<br />
recognition: Effects of task demands and language intermixing. Bilingualism:<br />
Language & Cognition, 1(1), 51-66.<br />
Dijkstra, T., Grainger, J. & van Heuven, W.J.B. (1999). Recognition of cognates and<br />
interlingual homographs: The neglected role of phonology. Journal of Memory<br />
and Language, 41, 496-518.<br />
Gerard, L. D. & Scarborough, D. L. (1989). Language-specific lexical access of<br />
homographs by bilinguals. Journal of Experimental Psychology, 15(2), 305-315.<br />
Gollan, T. H., Forster, K. I. & Frost, R. (1997). Translation priming with different scripts:<br />
Masked priming with cognates and non-cognates in Hebrew-English bilinguals.<br />
Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(5),<br />
1122-1139.<br />
Grainger, J. Kiyonaga, K. & Holcomb, P.J. (2006). The time-course of orthographic and<br />
phonological code activation. Psychological Science, 17, 1021-1026.<br />
Grainger, J., Midgley, K.J. & Holcomb P.J. (in preparation). Identical Words are Read<br />
Differently in a Bilingual’s Two Languages.<br />
P | 97
Effets de cognats<br />
Greenhouse, S.W. & Geisser, S. (1959). On methods in the analysis of profile data.<br />
Psychometrica, 24, 95-112.<br />
Holcomb, P.J. (1993). Semantic priming and stimulus degradation: Implications for the<br />
role of the N400 in language processing. Psychophysiology, 30(1), 47-61.<br />
Holcomb, P.J. & Grainger, J. (2006). On the time course of visual word recognition: An<br />
event-related potential investigation using masked repetition priming. Journal of<br />
Cognitive Neuroscience, 18(10), 1631-1643.<br />
Holcomb, P.J. & Grainger, J. (2007). Exploring the temporal dynamics of visual word<br />
recognition in the masked repetition priming paradigm using event-related<br />
potentials. Brain Research, 1180, 39–58.<br />
Holcomb, P.J., Grainger, J. & O'Rourke, T. (2002). An electrophysiological study of the<br />
effects of orthographic neighborhood size on printed word perception. Journal of<br />
Cognitive Neuroscience, 14 (6), 938–950.<br />
Holcomb, P.J., Reder, L., Misra, M. & Grainger, J. (2005). Masked priming: An eventrelated<br />
brain potential study of repetition and semantic effects. Cognitive Brain<br />
Research, 24(1), 155-172.<br />
Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture<br />
naming: evidence for asymmetric connections between bilingual memory<br />
representations. Journal of Memory and Language, 33, 149-174.<br />
Lalor, E. & Kirsner, K. (2001). The representation of “false cognates” in the bilingual<br />
lexicon. Psychonomic Bulletin & Review, 8, 552-559.<br />
Lau, E.F., Phillips, C. & Poeppel, D. (2008). A cortical network for semantics:<br />
(de)constructing the N400. Nature Neuroscience. 9, 920-933.<br />
Lemhöfer, K. & Dijkstra, T. (2004). Recognizing cognates and interlingual homographs:<br />
Effects of code similarity in language-specific and generalized lexical decision.<br />
Memory & Cognition, 32(4), 533-550.<br />
Lemhöfer, K., Dijkstra, T. & Michel, M.C. (2004). Three languages, one ECHO: Cognate<br />
effects in trilingual word recognition. Language & Cognitive Processes, 19, 585-<br />
611.<br />
Midgley, K.J., Holcomb P.J. & Grainger, J. (in press-a). Language effects in second<br />
language learners and proficient bilinguals investigated with event-related<br />
potentials. Journal of Neurolinguistics.<br />
Midgley, K.J., Holcomb P.J. & Grainger, J. (in press-b). Masked repetition and<br />
translation priming in second language learners: a window on the time-course of<br />
form and meaning activation using ERPs. Psychophysiology.<br />
Midgley, K.J., Holcomb P.J., van Heuven, W.J.B. & Grainger, J. (2008). An<br />
Electrophysiological investigation of cross-language effects of orthographic<br />
neighborhood. Brain Research, 1246, 123-135<br />
Münte, T., Wieringa, B.M., Weyerts, H., Szentkuti, A., Matzke, M. & Johannes, S.<br />
(2001). Differences in brain potentials to open and closed class words: class and<br />
frequency effects, Neuropsychologia, 39(1), 91-102.<br />
P | 98
Effets de cognats<br />
New, B., Pallier, C., Ferrand L. & Matos, R. (2001). Une base de données lexicales du<br />
français contemporain sur internet: Lexique. L'Année Psychologique, 101, 447-<br />
462.<br />
Sanchez-Casas, R. M., Davis, C. W. & Garcia-Albea, J. E. (1992). Bilingual lexical<br />
processing: Exploring the cognate/non-cognate distinction. European Journal of<br />
Cognitive Psychology, I, 293—310.<br />
van Hell, J. G. & De Groot, A. M. B. (1998). Conceptual representation in bilingual<br />
memory: Effects of concreteness and cognate status in word association.<br />
Bilingualism: Language & Cognition, 1(3), 193-211.<br />
van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can influence native<br />
language performance in exclusively native contexts. Psychonomic Bulletin &<br />
Review, 9(4), 780-789.<br />
Van Petten, C. & Kutas, M. (1990). Interactions between sentence context and word<br />
frequency in event-related brain potentials. Memory & Cognition, 18, 380-393.<br />
Voga, M. & Grainger, J. (2007). Cognate status and cross-script translation priming.<br />
Memory & Cognition, 35, 938-952.<br />
P | 99
L'amorçage masqué par équivalent de traduction<br />
Chapitre 5<br />
L'amorçage masqué de répétition et par équivalent de<br />
traduction chez des apprenants en langue seconde : un regard<br />
sur le décours temporel de l'activation de la forme et du sens*<br />
La technique des potentiels évoqués et celle de l'amorçage masqué par équivalent de<br />
traduction ont permis d'examiner le décours temporel de l'activation de la forme et du<br />
sens pendant la reconnaissance de mots chez des apprenants en langue seconde. Les<br />
cibles étaient soit identiques et dans la même langue que l'amorce, soit des traductions de<br />
l'amorce, soit des items non reliés à l'amorce. Lors de la première expérience, les cibles<br />
étaient choisies dans la L2 des participants et, dans la deuxième, elles l'étaient dans la L1.<br />
En ce qui concerne le traitement en L2, les conditions d'amorçage de répétition et<br />
d'équivalent de traduction ont produit des effets sur les composantes N250 et N400. Pour<br />
le traitement en L1, seule la répétition a modulé la N250 tandis que les deux conditions<br />
d'amorçage ont eu des effets sur la N400. Ces résultats suggèrent une activation rapide<br />
des représentations sémantiques lors du traitement de la forme des mots écrits et<br />
permettent de supposer l'absence de connexions facilitatrices entre les représentations des<br />
traductions chez les apprenants en L2. Leurs implications dans les théories du traitement<br />
des mots chez les bilingues seront abordées.<br />
Mots-clefs : Bilinguisme, Acquisition en langue seconde, Reconnaissance visuelle des<br />
mots, N400, N250, Amorçage masqué<br />
________________________<br />
* Article in press, Psychophysiology, August 2008<br />
P | 101
L'amorçage masqué par équivalent de traduction<br />
Masked Repetition and Translation Priming in Second Language Learners:<br />
A Window on the Time-Course of Form and Meaning Activation using ERPs<br />
Katherine J. Midgley<br />
Tufts University, Medford, Ma USA & CNRS & Université d'<strong>Aix</strong>-<strong>Marseille</strong>,<br />
<strong>Marseille</strong>, France<br />
Phillip J. Holcomb<br />
Tufts University, Medford, Ma USA<br />
Jonathan Grainger<br />
CNRS & Université d'<strong>Aix</strong>-<strong>Marseille</strong>, <strong>Marseille</strong>, France<br />
Address for correspondence:<br />
Katherine J Midgley<br />
Department of Psychology<br />
Tufts University<br />
Medford, MA 02155 USA<br />
kj.midgley@tufts.edu<br />
Voice: (617) 627-3521<br />
Fax: (617) 627-3181<br />
*This Research was supported by grant numbers HD043251 & HD25889.<br />
P | 103
L'amorçage masqué par équivalent de traduction<br />
Abstract<br />
ERPs and masked translation priming served to examine the time-course of form and<br />
meaning activation during word recognition in second language learners. Targets were<br />
repetitions of, translations of, or were unrelated to the immediately preceding prime. In<br />
Experiment 1 all targets were in the participants’ L2. In Experiment 2 all targets were in<br />
the participants’ L1. In Exp 1 both within-language repetition and L1-L2 translation<br />
priming produced effects on the N250 component and the N400 component. In Exp 2<br />
only within-language repetition produced N250 effects while both types of priming<br />
produced N400 effects. These results suggest rapid involvement of semantic<br />
representations during on-going form-level processing of printed words, and an absence<br />
of facilitatory connections between the form representations of non-cognate translation<br />
equivalents in L2 learners. The implications for bilingual theories of word processing are<br />
discussed.<br />
Keywords: Bilingualism, second language acquisition, visual word recognition, N400,<br />
N250, masked priming<br />
P | 104
L'amorçage masqué par équivalent de traduction<br />
Bilinguals and second-language learners provide an ideal testing ground for general<br />
theories of how form and meaning representations interact during language<br />
comprehension. They also represent a fascinating example of the versatility with which<br />
humans manipulate abstract symbols in order to communicate complex information.<br />
Learning a second language (once a first language has been acquired) involves acquiring<br />
a new set of arbitrary forms to re-represent a pre-established set of concepts (although<br />
some new concepts will of course be acquired with the new language). This likely<br />
involves at least a partial restructuring of the form representations in the first language, as<br />
proposed by two prominent models of bilingual word comprehension and second<br />
language learning – the Revised Hierarchical Model (RHM, Kroll & Stewart, 1994) and<br />
the Bilingual Interactive Activation (BIA) model (Grainger & Dijkstra, 1992; van<br />
Heuven, Dijkstra, & Grainger, 1998). According to the RHM, newly acquired lexical<br />
form representations in L2 are connected to their equivalent lexical form representations<br />
in L1 in order to facilitate access to semantics. According to the BIA model, the newly<br />
acquired lexical form representations are gradually integrated into a common network of<br />
lexical form representations for both languages. One specific goal of the present study is<br />
to test the predictions of these two models with respect to effects of non-cognate<br />
translation primes (i.e., translation equivalents having minimal form overlap).<br />
A more general goal of the present study is to investigate the nature of formmeaning<br />
interactions at the level of individual words. Part of that general endeavor<br />
involves describing exactly when semantic information becomes available during visual<br />
P | 105
L'amorçage masqué par équivalent de traduction<br />
word recognition, and the nature of the form-level processing that is necessary for that to<br />
occur. Demonstrations of early involvement of semantics during visual word recognition<br />
are few and far between. Standard semantic priming effects (e.g., bread-butter) are easily<br />
obtained when primes are clearly visible and targets immediately follow primes (e.g.,<br />
Meyer & Schvaneveldt, 1971). However, when prime exposure duration is reduced such<br />
that primes are barely visible, then the standard result is no semantic priming, but<br />
significant form priming effects (e.g., teble-table). This is true for behavioral studies (e.g.,<br />
Rastle, Davis, Marslen-Wilson, & Tyler 2000), and for electrophysiological studies (e.g.,<br />
Holcomb et al., 2005; Holcomb & Grainger, submitted). 1 This pattern therefore fits with<br />
models of visual word recognition according to which form-level processing must be<br />
complete (or close to completion) before any semantic-level information can be accessed<br />
(e.g., Forster, 1976). In cascaded activation models, on the other hand, semantic-level<br />
processing ought to follow form-level processing very rapidly (McClelland, 1979;<br />
McClelland & Rumelhart, 1981). According to this account, as soon as form<br />
representations are even partially activated by a given stimulus, activation immediately<br />
starts to spread to higher levels.<br />
The rather weak evidence for any early semantic activation obtained using the<br />
standard semantic priming paradigm might, however, be due to the relatively weak<br />
manipulation involved, and the fact that there is no general consensus as to how to<br />
measure semantic relatedness. Non-cognate translation equivalents (e.g., the English<br />
word “tree” and its French translation “arbre”) arguably provide the closest possible<br />
1 There is some controversy surrounding the presence/absence of N400 masked semantic priming effects.<br />
While Deacon et al. (2000), Kiefer (2002) and Grossi (2006) have reported significant N400 effects in<br />
masked semantic priming experiments, Holcomb et al. (2005) have shown that masked semantic priming<br />
N400 effects are predicted by conscious awareness of the prime. This suggests that monolingual masked<br />
semantic priming is unreliable at best and is most likely due to conscious processing of the primes.<br />
P | 106
L'amorçage masqué par équivalent de traduction<br />
semantic relation between two distinct word forms. They therefore provide an ideal<br />
testing ground for the interplay between form-level and semantic-level processes during<br />
visual word recognition.<br />
However, in line with the rather weak evidence for masked within-language<br />
semantic priming effects, behavioral studies investigating masked non-cognate translation<br />
priming effects have also produced mixed results. One standard finding in this area is that<br />
close-cognates (e.g., the English word “chair” and its French translation “chaise”)<br />
generate stronger priming effects than non-cognates (De Groot & Nas, 1991; Gollan,<br />
Forster, & Frost, 1997; Sanchez-Casas et al., 1992). Evidence for significant effects of<br />
non-cognate translation primes has mostly been obtained in language pairs with different<br />
scripts, such as Japanese and English, Hebrew and English or Greek and French<br />
(Finkbeiner, Forster, Nicol, & Nakamura, 2004; Gollan et al., 1997; Voga & Grainger,<br />
2007). The change in script across prime and target would allow improved processing of<br />
masked primes by providing the lexical processor with a distinct cue as to which<br />
language the word belongs to. However, robust non-cognate priming was reported by<br />
Grainger and Frenck-Mestre (1998) in same-script conditions (English-French<br />
bilinguals), but the effect was only robust when participants had to perform a semantic<br />
categorization task on target words (see Finkbeiner et al. 2004, for a replication with<br />
Japanese-English bilinguals). In the Grainger and Frenck-Mestre study, non-cognate<br />
translation primes did not significantly facilitate lexical decision responses to target<br />
words. The presence of translation priming in a semantic categorization task and not in<br />
the lexical decision task suggests that the effect is indeed semantically mediated, and not<br />
the result of direct form-level connections between translation equivalents as postulated<br />
in the RHM of Kroll and Stewart (1994).<br />
P | 107
L'amorçage masqué par équivalent de traduction<br />
The present study combines masked within-language and between-language (i.e.,<br />
non-cognate translation) priming with ERP recordings in order to: a) examine the relative<br />
contributions of form and meaning based representations in bilingual word processing, b)<br />
have a potentially overall more sensitive measure of priming effects, and c) provide finer<br />
grained information about the time-course of priming effects both within and betweenlanguages.<br />
Two prior ERP studies have examined unmasked non-cognate translation<br />
priming effects. (Alvarez et al., 2003; Phillips et al, 2006). 2 Of particular relevance to the<br />
current study is one by Alvarez et al. Using visually presented words these authors found<br />
evidence for translation priming effects in the N400 component that were larger and<br />
started earlier when primes were in L2 and targets in L1. This result is in line with the<br />
predictions of the RHM (Kroll & Stewart, 1994). According to this model, L2 words<br />
should automatically activate their L1 translate, and should do so to a greater extent and<br />
more rapidly than L1 words activate their L2 translate. However, there is an alternative<br />
interpretation of the Alvarez et al findings, Because the SOA in their study was 2700 ms<br />
and the primes were unmasked translation priming might have been due to participants<br />
using an overt translation strategy. If this strategy was employed it would presumably be<br />
used primarily in one direction to aid in L2 comprehension. In other words the L2 items<br />
during the long SOA would be overtly translated into L1 in order to facilitate the semantic<br />
categorization decision required on each trial and the resulting priming on subsequent L1<br />
targets would actually be more like L1-L1 priming. In the current study we rectify the<br />
problems of the Alvarez et al study by using masked priming and by blocking by<br />
language. This combination should minimize the possibility of any strategic influences on<br />
2 Unlike the Alvarez et al. study and the current study the bilinguals in the Phillips et al. study were nearly<br />
equally competent in both languages and the stimuli were auditory which makes comparisons between<br />
studies difficult. However, Phillips et al. did report asymmetrical N400 effects for within and betweenlanguage<br />
priming on the N400.<br />
P | 108
L'amorçage masqué par équivalent de traduction<br />
processing as participants should be unaware of the occurrence of primes in the non-target<br />
language and therefore should not engage in overt prime translation (Forster et al., 2002).<br />
Blocking by language for target words should also minimize the bilingual nature of the<br />
study because participants should only be aware of words in a single language in each<br />
block. This is important because it has been suggested (Grosjean & Miller, 1994) that<br />
bilinguals may have different processing modes; one for situations where both the<br />
bilinguals languages are required and one for monolingual situations. In the current<br />
experiment we were interested in biasing the participants towards monolingual processing<br />
in order to more clearly explore the more automatic aspects of interactivity between<br />
languages, that is, those not under strategic control.<br />
In recent research, the masked priming paradigm has been combined with ERP<br />
recordings to successfully map out the time-course of component processes in visual<br />
word recognition (e.g., Grainger et al., 2006; Holcomb & Grainger, 2006; Kiyonaga et al.,<br />
2007). Holcomb and Grainger (2006) described a cascade of ERP components found to<br />
be sensitive to their repetition priming manipulation. The first of these relevant to the<br />
current study is the N250, a negative-going component which peaks near 250 ms.<br />
Holcomb and Grainger reported that the N250 was more negative and had a slightly<br />
earlier peak latency to target words that were not full repetitions of or that had no overlap<br />
with their primes. Full repetitions produced the least N250 activity. Holcomb and<br />
Grainger postulate that the N250 reflects processes in visual word recognition where<br />
sublexical form representations (letters and letter combinations) are mapped onto the<br />
lexical system. The second component of relevance here was the N400, a negativity that<br />
starts around 350 ms and peaks between 400 and 600 ms. The N400 was more negative<br />
for unrelated items than for items that were partial repetitions and was least negative for<br />
fully repeated items. The results of a host of studies are consistent with the hypothesis<br />
P | 109
L'amorçage masqué par équivalent de traduction<br />
that the N400 reflects some aspect of semantic processing (e.g., Kutas & Hillyard, 1980;<br />
Kounios & Holcomb, 1992, 1994). Given the influence of partially overlapping nonword<br />
primes (e.g., teble-table), Holcomb and Grainger (2006) argued however, that the N400<br />
may also be sensitive to processing at the interface between whole-word form<br />
representations and semantics.<br />
In order to obtain an improved picture of the time-course of form and meaning<br />
activation both within and between the lexical and semantic systems of second language<br />
learners, the present study compares within-language repetition priming and non-cognate<br />
translation priming using the same paradigm as Holcomb and Grainger (2006). This<br />
methodology should allow us to observe and compare the time course of processing<br />
during L1 and L2 repetition priming as well as translation priming in both directions (L1-<br />
L2, L2-L1). The N250 and N400 ERP components will be used to infer form-level and<br />
semantic-level influences on processing. Even if these two components reflect to some<br />
extent a combination of form and semantic influences, we expect form-level influences to<br />
be greater on the N250, and semantic level influences to be greater on the N400. This<br />
then allows us to test the predictions of the RHM (Kroll & Stewart, 1994) and BIA model<br />
(Grainger & Dijkstra, 1992). In the RHM there are stronger links from L2 to L1 than from<br />
L1 to L2 lexical form representations, and weaker links between L2 lexical<br />
representations and meaning than between L1 lexical representations and meaning. L2<br />
primes should therefore affect form-level processing of the upcoming translate in L1,<br />
modulating the N250 component, and in consequence the N400. L1 primes, on the other<br />
hand, should mostly affect semantic level processing of upcoming L2 translates, and<br />
therefore only modulate the N400 component. According to the BIA model, translation<br />
priming is always semantically mediated (i.e., there are no excitatory connections<br />
between lexical form representations of translation equivalents), hence most of the effects<br />
P | 110
L'amorçage masqué par équivalent de traduction<br />
should be evident in the N400. Some priming effects are nevertheless expected on the<br />
N250 component via feedback from semantics to lexical representations (Voga &<br />
Grainger, 2007), and these should be most evident with L1 primes and L2 targets, simply<br />
because it is assumed that L1 words are processed more rapidly and efficiently than L2<br />
words. If this is indeed the case, we should also observe smaller and later effects in L2-L2<br />
repetition priming than L1-L1 repetition priming.<br />
Experiment 1<br />
In Experiment 1, ERP masked repetition priming effects of both within and crosslanguage<br />
primes (L2 – L2 repetition priming and L1 to L2 translation priming) on L2<br />
targets were measured in a 67ms SOA prime-target paradigm designed to decompose the<br />
different components elicited by priming. In this experiment all visible items are targets<br />
in participants’ L2. Masked primes are either L1 or L2 words.<br />
Methods<br />
Participants. Thirty-six participants (32 female, mean age = 20.3 SD = 1.2) were<br />
recruited during their second year of the English studies at the Université de Provence in<br />
<strong>Aix</strong>-en-Provence, France and paid for their participation. All were right handed<br />
(Edinburgh Handedness Inventory - Oldfield, 1971) and had normal or corrected-tonormal<br />
visual acuity with no history of neurological insult or language disability. French<br />
was reported to be the first language learned by all participants (L1) and English their<br />
primary second language (L2). All participants began their study of English in their sixth<br />
year of primary school at approximately the age of 12 years, as is customary in the French<br />
school system.<br />
P | 111
L'amorçage masqué par équivalent de traduction<br />
Participants’ auto-evaluation of English and French language skills were surveyed<br />
by questionnaire. On a seven point Likert scale (1 = unable; 7 = expert) participants<br />
reported their abilities to read, speak and comprehend English and French as well as how<br />
frequently they read in both languages (1 = rarely; 7= very frequently). The overall<br />
average of self-reported languages skills in French was 6.8 (SD = 0.1) and in English was<br />
4.9 (SD = 0.1). Our participants reported their average frequency of reading in French as<br />
6.4 (SD = 0.9) and in English as 5.4 (SD = 1.3). After the experiment participants were<br />
asked to translate all of the L2 target words that they saw into their L1. The mean score<br />
on this post-test was 82.4% (SD = 9.0, range 63.9% to 97.7%.).<br />
Stimuli The critical stimuli for this experiment were 400 four to eight letter English<br />
words and their translations into French. The English items had a mean CELEX<br />
(http://www.ru.nl/celex) log frequency of 1.74 (SD = 0.61, range 0.00 - 3.51). The French<br />
items had a mean Lexique (New et al., 2001) log frequency of 1.65 (SD = 0.63, range<br />
0.00 – 3.14). The correlation of the log frequencies of the English and French items was<br />
0.74 (p < .001). In selecting items care was taken to avoid any cross language<br />
homophones (e.g., lasse-lace) as well as cross language homographs (e.g., "coin"<br />
meaning corner in French). Overly polysemous words were also avoided (e.g., carte in<br />
French could mean map, menu or card in English). Words with accents were excluded, to<br />
prevent the eventual identification of French items in the prime position and (important in<br />
Experiment 2 where all targets are in caps and in French) because the use of accents in<br />
upper case French words is non-standardized (état could be written ETAT or ÉTAT, or<br />
PASSE could be passe or passé). Finally all stimuli were morphemically simple items.<br />
The non-critical stimulus pairs (used in probe trials) were formed by combining four to<br />
eight letter English animal names with unrelated non-animal words.<br />
P | 112
L'amorçage masqué par équivalent de traduction<br />
For both the critical and probe trials, the first stimulus was referred to as the prime<br />
and the second as the target. Primes were presented in lower case letters and targets in<br />
upper case (this was done in order to minimize the physical similarity between repeated<br />
items). Stimulus lists consisted a pseudorandom mixture of trials where the target was a<br />
full repetition of the prime in L2 (e.g. beach - BEACH), trials where the prime was an L1<br />
translation of the target (e.g. plage - BEACH), trials where the prime and target were<br />
unrelated L2 words (e.g. sleep - BEACH) and trials where the prime was an L1 word<br />
unrelated to the L2 target (e.g. miel - BEACH). Lists were formed so as participants saw a<br />
subset of 40 of the 400 critical items in each condition. Across lists (and participants),<br />
each target word appeared in each of the four conditions (REPETITION,<br />
TRANSLATION, UNRELATED WITHIN LANGUAGE, UNRELATED ACROSS<br />
LANGUAGE), but within lists each target stimulus was presented only once. An<br />
important feature of this design is that the prime and target ERPs in the repeated,<br />
translation and unrelated conditions are formed from exactly the same physical stimuli<br />
(across participants) which should reduce the possibility of ERP effects across conditions<br />
due to differences in physical features or lexical properties.<br />
Each list also contained 36% non-critical trials, half of which had an English animal<br />
name in the prime position and an English filler word in the target position and the other<br />
half of which had an unrelated filler word in the prime position and an English animal<br />
name in the target position. The animal names served as probe items in a go/no-go<br />
semantic categorization task in which participants were instructed to rapidly press a<br />
single button whenever they detected an animal name. Participants were told to read all<br />
other words passively without responding (i.e., critical stimuli did not require an overt<br />
response). Probe items were placed in the prime position to serve as a measure of prime<br />
detectability and provide an objective measure of the effectiveness of the masking<br />
P | 113
L'amorçage masqué par équivalent de traduction<br />
procedure. A practice session was administered before the main experiment to familiarize<br />
the participant with the task.<br />
Procedure Visual stimuli were presented on a 15” monitor set to a refresh rate of 60 Hz<br />
(which allows 16.67 ms resolution of stimulus control) and located 143 cm directly in<br />
front of the participant. Stimuli were displayed at high contrast as white letters on a black<br />
background in the Verdana font (letter matrix 20 pixels wide x 40 pixels tall). Each trial<br />
began with a forward mask of 12 hash marks (############) presented for a duration of<br />
200ms. The forward mask was replaced at the same location on the screen by a lower<br />
case prime item for 50ms. The prime was then immediately replaced by a 10 character<br />
uppercase random consonant string backward mask (ZJGRRFMXHG). The backward<br />
mask remained on the screen for 17ms (one frame) and was immediately replaced by the<br />
target in upper case letters for a duration of 300 ms. All target words were followed by a<br />
1000ms blank screen which was replaced by a blink stimulus (see <strong>Fig</strong>ure 1). The<br />
participants were instructed to blink only during the 1500ms that this stimulus was on the<br />
screen. The blink stimulus was followed by 500ms of blank screen after which the next<br />
trail began.<br />
<strong>Fig</strong>ure 1. A typical trial. Here an L1-L2 translation prime.<br />
P | 114
L'amorçage masqué par équivalent de traduction<br />
EEG recording procedure Participants were seated in a comfortable chair in a sound<br />
attenuated darkened room. The electroencephalogram (EEG) was recorded from 29 active<br />
tin electrodes held in place on the scalp by an elastic cap (Electrode-Cap International –<br />
see <strong>Fig</strong>ure 2). In addition to the 29 scalp sites, additional electrodes were attached to<br />
below the left eye (to monitor for vertical eye movement/blinks), to the right of the right<br />
eye (to monitor for horizontal eye movements), over the left mastoid bone (reference) and<br />
over the right mastoid bone (recorded actively to monitor for differential mastoid<br />
activity). All EEG electrode impedances were maintained below 5 kΩ (impedance for eye<br />
electrodes was less than 10 kΩ). The EEG was amplified by an SA Bioamplifier with a<br />
bandpass of 0.01 and 40 Hz and the EEG was continuously sampled at a rate of 200 Hz<br />
throughout the experiment.<br />
<strong>Fig</strong>ure 2. Electrode montage and four analysis columns used for ANOVAs.<br />
Data analysis Averaged ERPs time-locked to target onset were formed off-line<br />
from trials free of ocular and muscular artifact (less than 10% of trials) and were<br />
bandpass filtered at .5 and 15 Hz. Four types of targets were formed from two levels of<br />
PRIMING-TYPE (within language v. between language) and two levels of REPETITION<br />
P | 115
L'amorçage masqué par équivalent de traduction<br />
(repeated vs. unrelated, note that between language repetition is L1-L2 translation<br />
priming, e.g., plage-BEACH). The main analysis approach involved measuring mean<br />
amplitudes in three temporal epochs surrounding the two primary repetition ERP effects<br />
reported by Holcomb and Grainger: the N250 and N400. To best capture activity in the<br />
N250 epoch we selected a typical window surrounding the component from 200-300 ms.<br />
To best capture N400 activity, which varied in its time-course as a function of language<br />
and priming (see <strong>Fig</strong>ures 3a and 3b), we selected two windows, the first of which was<br />
within the range typical for this component in many previous L1 studies (350ms-500ms)<br />
and a second which was later (500-650 ms) and more in line with the later time-course of<br />
the N400 for L2 translation priming reported by Alvarez et al (2003). Separate repeated<br />
measures analyses of variance (ANOVAs) were used to analyze the data in each of these<br />
three epochs. The Geisser-Greenhouse (1959) correction was applied to all repeated<br />
measures with more than one degree of freedom in the numerator. Separate follow-up<br />
analyses for the within- and between-language conditions were performed in cases of<br />
significant REPETITION by PRIME-TYPE interactions.<br />
In order to thoroughly analyze the full montage of 29 scalp sites we employed an<br />
approach to data analysis that we have successfully applied in a number of previous<br />
studies (e.g., Holcomb et al., 2005). In this scheme the 29 channel electrode montage is<br />
divided up into seven separate parasagittal columns along the antero-posterior axis of the<br />
head (see <strong>Fig</strong>ure 2). The electrodes in each of three pairs of lateral columns and one<br />
midline column are analyzed in four separate ANOVAs. Three of these analyses (referred<br />
to as Column 1, Column 2 or Column 3) involved an anterior/posterior ELECTRODE-<br />
SITE factor with either three, four or five levels, as well as a HEMISPHERE factor (Left<br />
vs. Right). The forth “midline” analysis included a single anterior/posterior<br />
ELECTRODE-SITE factor with five levels.<br />
P | 116
L'amorçage masqué par équivalent de traduction<br />
Results<br />
Visual Inspection of ERPs<br />
The ERPs time locked to targets from 29 electrode sites for the repetition and<br />
translation priming conditions are plotted in <strong>Fig</strong>ure 3a&b. <strong>Fig</strong>ure 4 presents enlargements<br />
of the CP1 and CP2 sites from <strong>Fig</strong>ure 3. <strong>Fig</strong>ures 5 and 6 are the voltage maps for the<br />
repetition and translation priming conditions for time periods surrounding the N250 and<br />
N400. As can be seen in <strong>Fig</strong>ure 3, ERPs in the target epoch varied substantially as a<br />
function of both the REPETITION and the PRIMING-TYPE factors. In looking over<br />
these plots one must keep in mind that at the short SOA used in this experiment, the ERPs<br />
time locked to the target are a composite of neural activity generated by both the target<br />
and the immediately preceding prime and masking stimuli. The influence of the pre-target<br />
stimuli can be seen in the sequence of positive and negative peaks which start prior to the<br />
vertical calibration bar and run through the first 150 ms of the target epoch. However,<br />
starting at about 150-200 ms the morphology of the waveforms more closely conform to<br />
typical target ERP components which include the anteriorally distributed P2 peaking at<br />
approximately 180 ms post-target onset and more posterior negativities probably<br />
reflecting N1 target activity also peaking just before 200 ms. Starting at approximately<br />
180 ms the various independent variables appear to start exerting their influence on the<br />
ERPs. In the following sections we detail these effects in each of four measurement<br />
windows.<br />
Analyses of ERP Data<br />
N250: 200 - 300 ms epoch. In this epoch an omnibus ANOVA on the mean amplitudes<br />
between 200 and 300 ms produced a main effect of PRIMING-TYPE at all columns<br />
P | 117
L'amorçage masqué par équivalent de traduction<br />
(midline: F(1,19) = 14.694, p = 0.001; C1: F(1,19) = 10.034, p = 0.003; C2: F(1,19) =<br />
11.953, p = 0.001; C3: F(1,19) = 15.535, p < 0.000) indicating that target ERPs were<br />
more negative going for trials with a language switch between the prime and target (i.e.,<br />
L1-L2) than for trials where both the prime and target were in the same language (i.e.,<br />
L2-L2). The main effect of REPETITION was also significant at all columns (midline:<br />
F(1,19) = 5.071, p = 0.031; C1: F(1,19) = 5.162, p = 0.029; C2: F(1,19) = 5.562, p =<br />
0.024; C3: F(1,19) = 7.746, p = 0.009) indicating that targets following unrelated primes<br />
were more negative-going than targets that were repetitions or translations of the primes.<br />
Although there was not a significant interaction between REPETITION and<br />
PRIMING-TYPE, examination of <strong>Fig</strong>ures 3 to 6 suggests that the negativity in this<br />
measurement window has a somewhat different time-course for the within-language and<br />
between-language conditions. To assess this visual impression we employed an analysis<br />
strategy used by Phillips et al (2006) for analyzing the time-course of bilingual priming.<br />
We divided the theN250 epoch into two sub-windows, an earlier 200-250 ms epoch and a<br />
later 250-300 ms epoch. TIME EPOCH was then entered as an additional factor into the<br />
ANOVA. These analyses resulted in significant TIME EPOCH (early vs. late) by<br />
REPETITION by PRIMING-TYPE by HEMISPHERE interactions in the C2 and C3<br />
columns (C2: F(1,35) = 5.08, p = .031; C3: F(1,35) = 6.29, p = .017; marginal C1:<br />
F(1,35) = 2.90, p = .098). Following these significant interactions we then used separate<br />
follow-up analyses of the two priming types to better characterize the interactions in each<br />
sub-epoch for the three lateral analysis columns (C1, C2 and C3). In the earlier window<br />
(200-250 ms) only the within-language comparison resulted in significant REPETITION<br />
effects (C1: F(1,19) = 3.635, p = 0.065; C2: F(1,19) = 5.606, p = 0.024; C3: F(1,19) =<br />
5.477, p = 0.025). However, in the later epoch (250-300 ms) there were no significant<br />
REPETITION effects for the within-language condition (all Fs < 3; p > .098), but in the<br />
P | 118
L'amorçage masqué par équivalent de traduction<br />
between language condition there were significant interactions between REPETITION<br />
and HEMISPHERE at all three lateral columns (C1: F(1,19) ) = 4.617, p = 0.039; C2:<br />
F(1,19) = 4.237, p = 0.047; C3: F(1,19) = 6.793, p = 0.013). As can be seen in <strong>Fig</strong>ures 4<br />
and 6 these later effects were due to the between-language REPETITION effect being<br />
larger over the right hemisphere in this later epoch.<br />
N400 early: 350-500 ms epoch. There was a REPETITION by ELECTRODE SITE<br />
interaction at the midline sites (F(4, 140) = 4.34, p = .026; marginal at C3:F(4,140) =<br />
3.51, p = .054) indicating that the small REPETITION effect in this window tended to be<br />
larger at central and posterior sites. There were no interactions between REPETITION<br />
and PRIME_TYPE in this epoch (all Fs < 1.0).<br />
N400 late: 500-650 ms epoch. There were robust effects of REPETITION across the<br />
four analysis columns (midline: F(1,35) = 5.18, p = .029; C1: F(1,35) = 5.28, p = .028;<br />
C2: F(1,35) = 4.81, p = .035; C3: F(1,35) = 4.90, p = .033) with unrelated targets<br />
producing more negative-going responses. At the midline and more lateral columns this<br />
effect tended to be larger at posterior sites as indicated by significant REPETITION by<br />
ELECTRODE SITE interactions (midline: F(4,140) = 6.39, p = .005; C2: F(3, 105) =<br />
5.92, p = .011; C3: F(4,140) = 6.59, p = .007). There were no significant REPETITION<br />
by PRIME-TYPE interactions in this epoch (all Fs < 2.2).<br />
Behavioral Data<br />
Participants detected 81.5% (SD = 6.6) of animal probes in the target position and<br />
had on average 1.9 false alarms (SD = 2.4). No participants detected, or pressed to any<br />
animal probes in the prime position and no participants reported seeing any primes,<br />
during debriefing at the end of the experiment.<br />
P | 119
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 3a-b. ERPs time locked to L2 target onset in the repeated (solid) and unrelated (dashed)<br />
conditions for the: (a) within-language (L2-L2) and (b) between-language (L1-L2) comparisons. In<br />
this and all subsequent ERP figures negative voltages are plotted upward and target onset is<br />
indicated by the vertical calibration bar marked by the arrow with a “T” below it in the lower left<br />
time scale legend (prime onset, which was 67 ms earlier, is marked by the arrow with a “P” below it).<br />
Refer to <strong>Fig</strong>ure 2 for electrode locations.<br />
P | 120
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 5. Voltage maps calculated from difference waves (Unrelated-Repeated) for the withinlanguage<br />
condition at each of nine time points encompassing the N250 and N400 epochs (L2 – L2<br />
priming).<br />
<strong>Fig</strong>ure 4. Enlargement of CP1 and CP2 sites from <strong>Fig</strong>ure 3a & b. ERPs are time-locked to targets in<br />
L2.<br />
P | 121
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 6. Voltage maps calculated from difference waves (Unrelated-Translation) for the betweenlanguage<br />
condition at each of nine time points encompassing the N250 and N400 epochs (L1 – L2<br />
priming).<br />
Discussion<br />
As in several recent ERP studies (e.g., Holcomb & Grainger, 2006; Chauncey et al.,<br />
2008), two negativities; the first peaking at around 250 ms (N250) and the second after<br />
400 ms (N400) were modulated by repetition. The N250 has been argued to reflect the<br />
mapping of sublexical orthography onto lexical representations and the N400, the<br />
mapping of lexical form onto meaning (Grainger and Holcomb, in press). The effects on<br />
the N400 are consistent with two previous studies of unmasked repetition priming in L2<br />
and translation priming from L1 to L2 (Alvarez et al., 2003; Phillips et al., 2006).<br />
Likewise the extended time-course of the N400 translation priming effect found here is<br />
also consistent with these previous studies. 3<br />
New here is the observation that when<br />
3 The continuation of the priming effects into the 550-650 ms range might appear too late to be an N400<br />
effect. Therefore it would be reasonable to entertain the alternative interpretation that the effects may reflect<br />
a late positive component. However Alvarez et al. (2003) and Phillips et al. (2006) both showed prolonged<br />
N400 effects in the case of L1-L2 translation priming and to a lesser extent for L2-L2 priming. Moreover at<br />
P | 122
L'amorçage masqué par équivalent de traduction<br />
primes are masked repetition effects occur within a second language (L2-L2) on both the<br />
N400 and the earlier N250. This is important proof that our L2 prime words were being<br />
effectively processed in the extreme masking conditions used in the present study.<br />
Also new here is the finding of a clear modulation of both the N250 and N400<br />
components when the prime words were presented in L1 and the targets in L2. This is a<br />
critical finding given that there was minimal form overlap in the non-cognate translation<br />
equivalents used in the present study. This therefore suggests a semantic influence on<br />
processing that is reflected in the N250 component. The architecture of the BIA model<br />
(Grainger & Dijkstra, 1992; van Heuven et al., 1998) enables such semantic influences on<br />
lexical form representations. For the RHM (Kroll & Stewart, 1994), on the other hand,<br />
such influences could arise via direct connections between lexical representations of<br />
translation equivalents. However, given the asymmetrical nature of these connections, the<br />
RHM predicts relatively weak effects from L1 to L2, which should be amplified when<br />
primes are in L2 and targets in L1. Experiment 2 tests this prediction by presenting all<br />
targets in L1.<br />
Experiment 2<br />
In Experiment 2, ERP masked repetition priming effects of both within and crosslanguage<br />
primes (L1 – L1 repetition priming and L2 to L1 translation priming) on L1<br />
targets were measured in a 67ms SOA prime-target paradigm that was procedurally<br />
identical to Experiment 1 except that the targets were in participants’ L1 while masked<br />
primes were in both L1 and L2.<br />
least one previous study has concluded that LPC repetition effects are non-existent or greatly attenuated in<br />
masked priming (Misra, 2003), So it would seem that the most parsimonious explanation is that the late<br />
negative difference is due to a prolonged N<br />
400 for words processed in L2.<br />
P | 123
L'amorçage masqué par équivalent de traduction<br />
Methods<br />
Participants The same 36 participants who served in Experiment 1 also participated in<br />
this experiment.<br />
Stimuli and procedure. The critical and non-critical stimuli were the same as in<br />
Experiment 1 but the lists were constructed and counter-balanced with all of the targets in<br />
L1 (French). The lists were constructed with subsets of 400 critical items from each<br />
language so that there would be no repetition of primes or targets for any given<br />
participant between Experiment 1 and 2. The order of the experiments was<br />
counterbalanced across participants. The timing of the masked priming paradigm, the<br />
procedure and the laboratory were the same as in Experiment 1, as well as the EEG<br />
recording, electrode montage and the data analysis. The time between the experiments<br />
was varied according to the availability of the participants, from same day to four weeks.<br />
The order of the experiments was varied across participants.<br />
Data analysis The same approach to data analysis as Experiment 1 was used for<br />
Experiment 2.<br />
Results<br />
Electrophysiological Data<br />
Visual Inspection of ERPs<br />
The ERPs time locked to targets from 29 electrode sites for the repetition and<br />
translation priming conditions are plotted in <strong>Fig</strong>ure 7a&b. <strong>Fig</strong>ure 8 presents enlargements<br />
of the CP1 and CP2 sites from <strong>Fig</strong>ure 7. <strong>Fig</strong>ures 9 and 10 are the voltage maps for the<br />
P | 124
L'amorçage masqué par équivalent de traduction<br />
REPETITION effect in the within-language priming condition and the between-language<br />
priming condition for time periods surrounding the N250 and N400. As can be seen in the<br />
<strong>Fig</strong>ure 7 plots, ERPs in the target epoch varied substantially as a function of both the<br />
REPETITION and the PRIMING-TYPE factors.<br />
Analyses of ERP Data<br />
N250: 200 - 300 ms epoch. The omnibus ANOVA on the mean amplitudes between 200<br />
and 300 ms produced an interaction between REPETITION and PRIMING-TYPE at all<br />
columns (midline: F(1,19) = 5.046, p = 0.031; C1: F(1,19) = 4.873, p = 0.034; C2:<br />
F(1,19) = 5.482, p = 0.025; C3: F(1,19) = 7.110, p = 0.012). To better understand this<br />
interaction separate follow-up analyses of the two priming types were run. In these<br />
analyses there were significant effects of REPETITION at all columns for the withinlanguage<br />
condition (midline: F(1,19) = 8.203, p = 0.007; C1: F(1,19) = 8.339, p = 0.007;<br />
C2: F(1,19) = 8.502, p = 0.006; C3: F(1,19) = 12.371, p = 0.001). No significant effects<br />
were found for the between-language condition in this epoch (all Fs < 1 – see <strong>Fig</strong>ures 7, 8<br />
and 10).<br />
N400 early: 350-500 ms epoch: In this window there were again significant<br />
REPETITION by PRIME-TYPE interactions (midline: F(1,35) = 4.69, p = .037; C1:<br />
F(1,35) = 6.14, p = .018; C2: F(1,35) = 6.59, p = .015; C3: F(1,35) = 5.78, p = .022). To<br />
better understand these interactions separate follow-up analyses of the two priming types<br />
were run. For the within-language condition there were significant REPETITION effects<br />
(midline: F(1,35 = 7.77, p = .009; C1: F(1,35) = 10.28, p = .003, p = ; C2: F(1,35) =<br />
10.75, p = .002; C3: F(1,35) = 7.63, p = .009) and significant REPETITION by<br />
ELECTRODE SITE effects (midline: F(4,140) = 6.69, p = .003; C2: F(3,105) = 4.50, p =<br />
.023; C3:F(4,140) = 4.99, p = .015). Examination of <strong>Fig</strong>ures 7a, 8 and 9 reveals that the<br />
P | 125
L'amorçage masqué par équivalent de traduction<br />
unrelated targets were more negative-going across the scalp than were the repeated<br />
targets. Unlike the previous epoch there were now also significant REPETITION by<br />
ELECTRODE SITE effects for the between-language condition (midline: F(4,140) =<br />
3.56, p = .034; C3: F(4,140) = 4.10, p = .032). However, while the within-language<br />
REPETITION effect was due to consistently more negative going potentials for unrelated<br />
compared for repeated targets across the scalp, for the between-language REPETITION<br />
there was a negative-going effect at the back of the head (unrelated more negative than<br />
translation targets), but a reversed positive-going pattern at more frontal sites (see <strong>Fig</strong>ures<br />
8b and 10). This latter effect took the form of a more positive potential for unrelated<br />
compared to translation targets.<br />
N400 late: 500-650 ms epoch: In this temporal window there were again significant<br />
interactions between REPETITION, PRIME-TYPE and ELECTRODE SITE at the<br />
midline and the more lateral sites (midline: F(4,140) = 4.17, p = .023; C2: F(3,105) =<br />
3.30, p = .061; C3: F(4,140) = 3.98, p = .03). Separate follow-up analyses for the two<br />
prime types did not reveal any significant REPETITION effects for the within-language<br />
contrasts (all Fs < 1). However, like the previous epoch the between-language contrast<br />
revealed a posterior negativity for unrelated compared to translation targets, and an<br />
anterior positivity for this same contrast (REPETITION by ELECTRODE SITE effect,<br />
midline: F(4,140) = 4.54, p = .02; C3: F(4,140) = 3.88, p = .039).<br />
P | 126
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 7a-b. ERPs time locked to L1 target onset in the repeated (solid) and unrelated (dashed)<br />
conditions for the: (a) within-language (L1-L1) and (b) between-language (L2-L1) comparisons. All<br />
else is as in <strong>Fig</strong>ure 3.<br />
P | 127
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 8a. Enlargement of sites CP1 and CP2 from <strong>Fig</strong>ure 7a & b. Targets in L1.<br />
<strong>Fig</strong>ure 8b. Enlargement of sites Fpz and Oz from <strong>Fig</strong>ure 7a & b. Targets in L1.<br />
P | 128
L'amorçage masqué par équivalent de traduction<br />
<strong>Fig</strong>ure 9. Voltage maps calculated from difference waves (Unrelated-Repeated) for the withinlanguage<br />
condition at each of nine time points encompassing the N250 and N400 epochs (L1 – L1<br />
priming).<br />
<strong>Fig</strong>ure 10. Voltage maps calculated from difference waves (Unrelated-Repeated) for the betweenlanguage<br />
condition at each of nine time points encompassing the N250 and N400 epochs (L2 – L1<br />
priming).<br />
P | 129
L'amorçage masqué par équivalent de traduction<br />
Behavioral Data<br />
Participants detected 95.7% (SD = 5.5) of animal probes in the target position and<br />
had on average 0.75 false alarms (SD = 0.8). No participants reported seeing any primes,<br />
even when prompted. Two participants detected, or pressed to one animal probe in the<br />
prime position each. These were presumably false alarms to the targets associated to these<br />
animal probe primes.<br />
Discussion<br />
As in Experiment 1 we again found robust ERP repetition priming effects when the<br />
prime and target were within the same language. These effects replicate those of<br />
Holcomb and Grainger (2006) finding similar N250 effects and N400 effects for<br />
repetition priming. Moreover, the N400 effect for L1-L1 priming did not extend past the<br />
traditional N400 window as it did in Experiment 1 for L2-L2 priming. Also, unlike<br />
Experiment 1, but compatible with the bulk of the behavioral literature examining L2-L1<br />
priming (e.g., Jiang, 1999), we found no evidence of a translation priming in the N250<br />
epoch, although there was evidence of such priming in both the early and late N400<br />
measurement epoch. The nature of these N400 effects was not like that found in<br />
Experiment 1 for translation priming. In Experiment 1 both within and between-language<br />
priming in the N400 windows appeared to modulate a broadly distributed negativity. In<br />
Experiment 2, within-language priming followed this same pattern (although it terminated<br />
earlier). However, the between-language contrast revealed a very different pattern with<br />
what appeared to be a posterior N400 effect but a reversed anterior positive-going effect.<br />
Finally, it is important to contrast the total absence of between-language priming<br />
(L2-L1) in the N250 component in Experiment 2 with the presence of within-language<br />
P | 130
L'amorçage masqué par équivalent de traduction<br />
(L2-L2) priming found in Experiment 1. The absence of translation priming effects in this<br />
ERP component when primes are in L2, in association with the robust effects found in<br />
Experiment 1 when primes are in L1, counters the predictions of the RHM and provides<br />
support for the BIA model. The full implications of these findings are discussed below.<br />
General Discussion<br />
The present study tested within-language repetition priming and between-language<br />
non-cognate translation priming in second language learners using the masked priming<br />
paradigm and ERP recordings. It was argued in the Introduction that this paradigm offers<br />
a strong test of the time-course of interactivity at the lexical and semantic level of<br />
representation between a bilingual’s two language systems. It was also pointed out that<br />
non-cognate translation priming is an ideal testing ground for semantic priming effects in<br />
the absence of form overlap across primes and targets. On the basis of prior work<br />
combining masked priming and ERPs (e.g., Holcomb & Grainger, 2006) we expected to<br />
observe within-language repetition effects in an early component (N250), thought to<br />
reflect sublexical processing, and both within-language repetition priming and between<br />
language translation priming in the later N400 component (thought to at least partly<br />
reflect semantic-level processing).<br />
The N400 was found to be sensitive to non-cognate translation priming, in both<br />
directions, although the distribution of the effect and the presence of a temporally<br />
coincident anterior positive effect in Experiment 2 suggest that there are likely<br />
differences in the mechanisms supporting priming in the two directions. The findings on<br />
the N400 then are in line with the general consensus that this component reflects<br />
processing at a semantic / conceptual level. The fact that the pattern of the effect with L2<br />
P | 131
L'amorçage masqué par équivalent de traduction<br />
primes and L1 targets is somewhat different, could be due to the amount of processing<br />
necessary to generate such semantic-level effects (i.e., primes would be less efficiently<br />
and less rapidly processed in L2 than L1 in our participants who are learners and not<br />
balanced bilinguals).<br />
The N250 was found to be sensitive to within-language repetition priming in both<br />
L1 and L2, and most important, non-cognate translation primes were also found to<br />
modulate N250 amplitude when primes were in L1 and targets in L2, although the timecourse<br />
of this effect was somewhat later. No such translation priming effect on N250<br />
amplitude was found in Experiment 2 when primes were in L2 and targets in L1. The fact<br />
that the N250 ERP component was found to be sensitive to non-cognate translation<br />
priming when primes were in L1 and targets in L2 (Experiment 1) is perhaps the key<br />
finding of the present study. Holcomb and Grainger (2006) suggested that the N250<br />
reflects purely form-level processing – the mapping of sublexical form (letters and letter<br />
clusters) onto whole-word orthographic representations. The present results provide a<br />
clear demonstration that semantic overlap across primes and targets, in the absence of<br />
form overlap, can influence N250 amplitude. In the present study, this semantic influence<br />
occurred in the later phase of the N250, peaking at around 300 ms post-target onset (see<br />
<strong>Fig</strong>ure 6). This therefore provides us with an upper limit for the onset of semantic<br />
influences during visual word recognition. Based on the delayed time-course of the N250<br />
in L1-L2 translation condition this suggests that there is a rather short lag (around 50 ms)<br />
between sublexical form processing and availability of semantic information. This is in<br />
line with cascaded activation accounts of lexical processing (e.g., McClelland &<br />
Rumelhart, 1981), according to which higher level codes are activated with a minimal lag<br />
relative to lower-level processing. In a similar vein pointing to the possibility of cascaded<br />
processing, Grainger, Kiyonaga, and Holcomb (2006) have also shown a lag of<br />
P | 132
L'amorçage masqué par équivalent de traduction<br />
approximately 50 ms between the earliest modulation of the N250 by an orthographic<br />
manipulation (transposed-letter priming), and a later effect produced by a phonological<br />
manipulation (pseudohomophone priming).<br />
The observed effect of translation primes on the N250 fits nicely with a recent<br />
finding reported by Morris, Franck, Holcomb, and Grainger (2007) that the N250 is<br />
sensitive to the semantic transparency of morphologically related primes and targets. In<br />
this study, a distinct pattern of N250 priming effects was found for transparent primetarget<br />
pairs, where the prime has a clear semantic relation with the target (e.g., baker -<br />
bake) compared with opaque pairs, where there is no obvious semantic relation between<br />
primes and targets (e.g., corner - corn). Morris et al. argued that these semantic influences<br />
on the N250 likely reflect interactive processing, whereby higher-level semantic<br />
information feeds back to influence on-going form-level processing. In such an account,<br />
semantic information is rapidly accessed during visual word recognition, after some<br />
minimal form-level processing. This information is then fed back to lower levels of<br />
processing in order to generate a resonant activation state allowing a single form-meaning<br />
association to emerge as the best interpretation of the stimulus.<br />
In terms of models of bilingual word recognition, the present results fit well with<br />
the bilingual extension of McClelland and Rumelhart’s (1981) interactive-activation<br />
model – the BIA model (Grainger & Dijkstra, 1992; van Heuven et al., 1998). L1 primes<br />
would rapidly activate the corresponding semantic representation that would in turn<br />
feedback information to appropriate form-level representations in L1 and L2, hence<br />
modifying processing of L2 targets that are translates of the L1 prime. One could argue,<br />
however, that this translation priming effect reflects direct connectivity between the form<br />
representations of translation equivalents, as postulated in the RHM (Kroll & Stewart,<br />
1994). In this case, translation priming would not reflect semantic-level processing, as<br />
P | 133
L'amorçage masqué par équivalent de traduction<br />
argued above. The fact that the N250 translation priming effect was asymmetrical (i.e.<br />
observed only when translation primes were in L1 in Experiment 1 and not observed<br />
when translation primes were in L2 in Experiment 2) as has been found in behavioral<br />
studies of translation priming, would appear to be strong evidence against a lexically<br />
mediated (form-based) account of these effects as per the RHM. According to Kroll and<br />
Stewart’s model, direct connectivity across translation equivalents is stronger from L2 to<br />
L1 than vice versa, particularly in second language learners such as the participants of the<br />
present study. Therefore, according to a direct connectivity account of translation<br />
priming, we ought to have observed stronger effects with L2 primes and L1 targets.<br />
It is nevertheless possible that the absence of translation priming from L2 to L1 in<br />
the present study is simply a reflection of the inability of our participants to rapidly<br />
process briefly presented primes in their L2. This handicap in processing briefly<br />
presented primes in L2 would be further exaggerated when all visible words (i.e., targets)<br />
are in L1. This could arise via a global inhibition operating between languages, such that<br />
all L2 representations would be partially suppressed when processing in a seemingly<br />
monolingual L1 context. This would explain why we obtained robust effects on N250<br />
amplitude when both primes and targets were in L2 (repetition priming), since L2 prime<br />
processing would be enhanced in this context. If this interpretation is correct, then we<br />
ought to be able to increase L2-L1 translation priming effects by having a large number<br />
of L2 targets intermixed with the L1 targets. This would disable any kind of monolingual<br />
processing mode and therefore prevent the global inhibition of L2 representations.<br />
One problem with the above characterization is that although there were no N250<br />
effects for translation priming in Experiment 2, there were significant priming effects<br />
later during the N400 window. If L2 primes in an L1 processing mode are not being<br />
processed, then they should not have produced these later effects. Therefore it seems<br />
P | 134
L'amorçage masqué par équivalent de traduction<br />
more likely that the L2 primes in Experiment 2 were processed but that because<br />
processing of L2 items is less efficient and slower (as indicated by the later N400 effects<br />
to L2 targets) there was not sufficient time in this short SOA paradigm for feedback from<br />
the semantic level to the form level to exert its influence on the L1 N250. If this is correct<br />
than lengthening the SOA so as to allow the prime processing to progress further might<br />
boost feedback enough to support N250 type effects with L2 primes and L1 translation<br />
targets.<br />
P | 135
L'amorçage masqué par équivalent de traduction<br />
REFERENCES<br />
Alvarez, R., Grainger, J. & Holcomb, P.J. (2003). Accessing word meaning in primary<br />
and secondary languages: A comparison using event-related brain potentials.<br />
Brain & Language, 87(2), 290-304.<br />
Chauncey, K., Holcomb, P.J. & Grainger, J. (2008). Effects of stimulus font and size on<br />
masked repetition priming: An ERP investigation. Language and Cognitive<br />
Processes, 23(1), 183-200.<br />
De Groot, A. M. B. (1992). Determinants of word translation. Journal of Experimental<br />
Psychology: Learning, Memory, and Cognition, 18, 1001-1018.<br />
De Groot, A. M. B. & Nas, G. L. (1991). Lexical representation of cognates and noncognates<br />
in compound bilinguals. Journal of Memory & Language, 30(1), 90-123.<br />
Deacon, D., Hewitt, S., Yang, C.M. & Nagata, M. (2000). Event-related potential indices<br />
of semantic priming using masked and unmasked words: evidence that the N400<br />
does not reflect a post-lexical process. Cognitive Brain Research. 9, 137–146.<br />
Finkbeiner, M., Forster, K., Nicol, J. & Nakamura, K. (2004). The role of polysemy in<br />
masked semantic and translation priming. Journal of Memory & Language, 51(1),<br />
1-22.<br />
Forster, K. I. (1976). Accessing the mental lexicon. In E.C.J. Walker & R.J. Wales (Eds.),<br />
New approaches to language mechanisms (257–287). Amsterdam:North-Holland.<br />
Forster, K. I., Mohan, K. & Hector, J. (2003). The mechanics of masked priming. In: S.<br />
Kinoshita & S.J. Lupker (Eds), Masked priming: State of the art. Macquarie<br />
monographs in cognitive science. New York, NY: Psychology Press.<br />
Gollan, T. H., Forster, K. I. & Frost, R. (1997). Translation priming with different scripts:<br />
Masked priming with cognates and non-cognates in Hebrew-English bilinguals.<br />
Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(5),<br />
1122-1139.<br />
Grainger, J. & Dijkstra, A. (1992). On the representation and use of language information<br />
in bilinguals. In: Harris, R.J. (Ed.), Cognitive Processing in Bilinguals.<br />
Amsterdam: Elsevier Science Publications, 207-220.<br />
Grainger, J. & Frenck-Mestre, C. (1998). Masked priming by translation equivalents in<br />
proficient bilinguals. Language and Cognitive Processes, 13(6), 601-623.<br />
Grainger, J. & Holcomb, P.J. (in press). Neural constraints on a functional architecture for<br />
word recognition. In P. Cornelissen, P. Hansen, M. Kringelbach & K. Pugh (Eds.),<br />
The neural basis of reading. Oxford University Press: Oxford<br />
Grainger, J., Kiyonaga, K. & Holcomb, P.J. (2006). The time-course of orthographic and<br />
phonological code activation. Psychological Science, 17, 1021-1026.<br />
Greenhouse, S.W. & Geisser, S. (1959). On methods in the analysis of profile data.<br />
Psychometrica, 24, 95-112.<br />
Grosjean, F. & Miller, J. (1994). Going in and out of languages: an example of bilingual<br />
flexibility. Psychological Sciences 5(4), 201-206.<br />
P | 136
L'amorçage masqué par équivalent de traduction<br />
Grossi, G. (2006). Relatedness proportion effects on masked associative priming: An<br />
ERP study Psychophysiology, 43(1), 21-30.<br />
Holcomb, P.J. & Grainger, J. (2006). On the time course of visual word recognition: An<br />
event-related potential investigation using masked repetition priming. Journal of<br />
Cognitive Neuroscience, 18(10), 1631-1643.<br />
Holcomb P.J. & Grainger, J. (in press). ERP Effects of Short Interval Masked Associative<br />
and Repetition Priming. Journal of Neurolinguistics.<br />
Holcomb, P.J., Reder, L., Misra, M. & Grainger, J. (2005). Masked priming: An eventrelated<br />
brain potential study of repetition and semantic effects. Cognitive Brain<br />
Research, 24(1), 155-172.<br />
Jiang, N. (1999). Testing processing explanations for the asymmetry in masked crosslanguage<br />
priming. Bilingualism: Language and Cognition, 2(1), 59-75.<br />
Kiefer, M. (2002). The N400 is modulated by unconsciously perceived masked words:<br />
Further evidence for an automatic spreading activation account of N400 priming<br />
effects. Cognitive Brain Research, 13(1), 27-39.<br />
Kiyonaga, K., Grainger, J., Midgley, K.J. & Holcomb, P.J. (2007). Masked cross-modal<br />
repetition priming: An ERP investigation. Language and Cognitive Processes,22,<br />
337-376<br />
Kounios, J. & Holcomb, P. (1992). Structure and process in semantic memory: Evidence<br />
from event-related brain potentials and reaction times. Journal of Experimental<br />
Psychology: General, 121(4), 459-479.<br />
Kounios, J. & Holcomb, P.J. (1994). Concreteness effects in semantic processing: Eventrelated<br />
brain potential evidence supporting dual-coding theory. Journal of<br />
Experimental Psychology: Learning, Memory and Cognition, 20(4), 804-823.<br />
Kroll, J. F. & Stewart, E. (1994). Category interference in translation and picture naming:<br />
evidence for asymmetric connections between bilingual memory representations.<br />
Journal of Memory and Language, 33, 149-174.<br />
Kutas, M. & Hillyard, S.A. (1980). Reading senseless sentences: Brain potentials reflect<br />
semantic incongruity, Science, 207, 203–205.<br />
McClelland, J.L. (1979). On the time relations of mental processes: an examination of<br />
systems of processes in cascade, Psychological Review, 86, 287–330.<br />
McClelland, J.L. & Rumelhart, D. E. (1981). An interactive activation model of context<br />
effects in letter perception: Part 1. An account of basic findings. Psychological<br />
Review, 88, 375–407.<br />
Meyer, D.E. & Schvaneveldt, R.W. (1971). Facilitation in recognizing pairs of words:<br />
Evidence of a dependence between retrieval operations. Journal of Experimental<br />
Psychology, 90, 227-234.<br />
Misra, M., Holcomb, P J. (2003). Event-related potential indices of masked repetition<br />
priming. Psychophysiology, 40(1), 115-130.<br />
Morris, J., Franck, T., Grainger, J. & Holcomb, P.J. (2007). Semantic transparency and<br />
masked morphological priming: AN ERP investigation. Psychophysiology, 44(4),<br />
506-521.<br />
P | 137
L'amorçage masqué par équivalent de traduction<br />
New, B., Pallier, C., Ferrand L., Matos, R. (2001). Une base de données lexicales du<br />
français contemporain sur internet: Lexique. L'Année Psychologique, 101, 447-<br />
462.<br />
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh<br />
inventory. Neuropsychologia, 9(1), 97-113.<br />
Phillips, N. A., Klein, D., Mercier, J. & de Boysson, C. (2006). ERP measures of auditory<br />
word repetition and translation priming in bilinguals. Brain Research, 1125(1),<br />
116-131.<br />
Rastle, K., Davis, M.H., Marslen-Wilson, W.D. & Tyler, L.K. (2000). Morphological &<br />
semantic effects in visual word recognition: A time-course study. Language and<br />
Cognitive Processes, 15(4/5), 507–537.<br />
Sanchez-Casas, R. M., Davis, C. W. & Garcia-Albea, J. E. (1992). Bilingual lexical<br />
processing: Exploring the cognate/non-cognate distinction. European Journal of<br />
Cognitive Psychology, 4, 293—310.<br />
van Heuven, W.J.B., Dijkstra, A. & Grainger, J. (1998). Orthographic neighborhood<br />
effects in bilingual word recognition. Journal of Memory and Language, 39(3),<br />
458-483.<br />
Voga, M. & Grainger, J. (2007). Cognate status and cross-script translation priming.<br />
Memory & Cognition, 35, 938-952.<br />
P | 138
Discussion générale<br />
Chapitre 6<br />
Discussion générale<br />
P | 139
Discussion générale<br />
In the four articles presented in chapters two through five of this dissertation we<br />
presented studies using electrophysiological measures with the goal of adding to the<br />
existing body of research on the nature of the bilingual lexicon. These chapters addressed<br />
important questions about similarities and differences of processing in L1 and L2,<br />
language selectivity and the level of integration and interactivity of the bilingual lexicon.<br />
Chapter two presented studies that sought similarities and differences in processing<br />
in L1 and L2 and this in both late learners and in late bilinguals. We observed that the<br />
processing of L1 and L2 words in late L2 learners and late bilinguals diverged in two<br />
distinct ways as reflected in the ERP waveforms generated by these words during silent<br />
reading for meaning. For learners, an anterior part of the N400 component showed a<br />
distinct latency shift with L2 amplitudes peaking later than L1 amplitudes while the<br />
posterior part of the N400 revealed larger amplitudes to L1 compared with L2 words.<br />
Even the late bilinguals showed an anterior latency shift with L2 amplitudes peaking later<br />
than L1 amplitudes, although this latency shift was smaller. However the bilinguals<br />
showed no significant difference in amplitude between L1 and L2 in posterior regions.<br />
We attempted explanations for these effects in learners with factors known to affect<br />
the N400 component in monolingual processing but to no avail. We could find no<br />
possible subjective frequency account that fit our data nor was there any straightforward<br />
AoA explanation, nor was there a viable orthographic neighborhood explanation.<br />
We were led to conclude, given the pattern of observed effects that word<br />
recognition in L2 involves distinct mechanisms compared with the first language, at least<br />
P | 141
Discussion générale<br />
in the relatively early phases of L2 acquisition in late learners of an L2. This conclusion<br />
steered us away from the BIA model (Grainger & Dijkstra, 1992; van Heuven et al.,<br />
1998) that posits an integrated lexicon with processing much like a monolingual IA<br />
model, and towards a model that predicts processing differences, due to different<br />
mechanisms, between L1 and L2, like the RHM (Kroll & Stewart, 1994).<br />
Another important observation from chapter two had to do with the N400 as a<br />
measure of proficiency. The posterior N400 language effect observed in learners was not<br />
present in proficient bilinguals. However, the anterior N400 latency shift observed for<br />
learners was also observed for proficient bilinguals, but the latency difference was much<br />
smaller. This would suggest that the posterior N400 language effect reflects competence<br />
in L2, but the anterior latency shift of the N400 effect continues to reflect differences in<br />
L1 and L2 processing even in relatively competent bilinguals; i.e., even competent late<br />
learners of an L2 conserve some form of asymmetry. Thus there remains an empirical<br />
question: would early bilinguals show any pattern of anterior latency shift of the N400?<br />
And if so would it be modulated by language dominance?<br />
In the studies in chapter three the participants saw lists of words in one language<br />
only and were therefore in appropriate conditions for using language-specific selection<br />
processes. The presented words were of differing neighborhood density in the nonpresented<br />
language. We observed effects of cross-language neighborhood density for both<br />
L1 and L2. This cross-language neighborhood effect appeared earlier (in the 175–275 ms<br />
epoch) and was more widely distributed across the scalp when the target words were in<br />
L2 and the neighbors in L1. Effects of cross-language neighborhood on L1 words only<br />
appeared in the 300–500 ms epoch and were limited to the most posterior electrode sites.<br />
These results contradict the notion of early language-specific selection in bilingual<br />
language comprehension while providing support for the non-selective access hypothesis<br />
P | 142
Discussion générale<br />
embodied in the BIA+ model (Dijkstra & van Heuven, 2002). In fact, simulations run on<br />
the BIA+ model with the stimuli form the experiments showed that cross-language<br />
neighborhood effects are stronger when targets are in L2 compared with targets in L1.<br />
These results provide further evidence in favor of non-selective access to a highly<br />
integrated and interactive lexicon in bilinguals.<br />
In addition to any observations about language non-selectivity these results show<br />
that L2 neighbors have a later and less widely distributed effect on L1 target processing<br />
than L1 neighbors have on L2 target processing. This is in line with one major principle<br />
implemented in all connectionist models of language processing – that frequency of<br />
exposure determines connection strength. In late bilinguals L2 words have much lower<br />
exposure overall than L1 words. This exposure difference is thus reflected in word<br />
frequency differences between L1 and L2. This could account for why the effects of L2<br />
neighbors are weaker, less widely distributed, and since more time is required for<br />
propagation, appear later than the effects of L1 neighbors. With this observation of timecourse<br />
differences it is important to point out that these participants were proficient<br />
bilinguals and late learners of L2. If the difference of subjective frequency is what is<br />
driving the results it follows that this will continue to have effects in late learners of an L2<br />
even if they are competent bilinguals.<br />
The two first articles of this dissertation therefore appear to come down on opposite<br />
sides of a supposed opposition between the RHM and the BIA model. One way of<br />
integrating the findings is to remark that the BIA model better reflects processing in<br />
relatively proficient bilinguals, while the RHM is a better model of lexical processing in<br />
beginning bilinguals. However our findings with late bilinguals lead us to believe that the<br />
BIA model must account for an on-going asymmetry in bilinguals.<br />
P | 143
Discussion générale<br />
The results of these studies lead us to the observation that although the BIA model<br />
may be a preferable model in that it is an extension of monolingual models, it needs some<br />
mechanism for explaining the developmental story of late learners of a second language –<br />
especially since our results seem to say that the asymmetry of late learners is never<br />
completely vanquished. In the RHM there exists the idea that during L2 acquisition, the<br />
L2-L1 translation route would be gradually replaced by L2 lexical representations that<br />
become part of an integrated network along with L1 representations establishing strong<br />
L2 lexical-semantic links. In a bilingual IA model this could be achieved by gradually<br />
increasing the L2 resting levels, and gradually strengthening the weights of the<br />
connections between L2 word-forms and semantics. Thus as proficiency develops in L2,<br />
lexical processing in L2 becomes more and more akin to lexical processing in L1.<br />
In order to further pursue questions of language integration and interactivity in the<br />
article presented in chapter four we recorded and compared ERPs to cognate and noncognate<br />
words while participants were processing blocked lists of words in their L1 and<br />
L2. ERP negativities in the region of the N400 component were found to be sensitive to<br />
cognate status in both language blocks with non-cognates being more negative than<br />
cognates; i.e., larger N400s reflecting more difficult processing for non-cognates. This<br />
fits with the general hypothesis that the mapping of form to meaning is facilitated in<br />
cognate words. An important finding of this article was the influence of cognate status on<br />
word recognition in L1 because prior behavioral research had provided mixed findings on<br />
this particular issue. Using electrophysiological measures, we were able to observe<br />
cognate effects in relatively low proficiency language learners processing words in their<br />
L1.<br />
Furthermore we hypothesized that the cognate advantage would reflect an<br />
accumulation of the benefits of exposure to a given form-meaning association across two<br />
P | 144
Discussion générale<br />
languages. This would presumably benefit L2 more than L1, particularly in the case of<br />
learners. Thus we predicted to see stronger cognate effects in L2 compared to L1 and we<br />
did not.<br />
However, it is notable that the time-course of the N400 effects differed between the<br />
two language blocks. The cognate effects began to emerge about 200 ms earlier in L1. In<br />
addition to predicting cognate effects in both L1 and L2, we predicted earlier effects in<br />
L1. This is because the locus of the cognate advantage, the mapping between form and<br />
meaning, should arise earlier in the processing of an L1 word than an L2 word.<br />
Our results can, at first blush, be taken as evidence for a factor, known to affect<br />
monolingual word processing; frequency. If we are arguing that our effects are the result<br />
of the accumulation of exposure to a given form-meaning association we are putting forth<br />
an argument of a frequency account of our effects. But the difference in timing of these<br />
effects could point to something else; different mechanisms underlying the observed<br />
differences in time-course of cognate effects between L1 and L2.<br />
Within the framework of the RHM, one could argue that cognate words are<br />
particularly easy to process in the early phases of L2 acquisition given that their<br />
translations equivalents have the same or very similar orthographic forms. Non-cognate<br />
words would be harder to process because of the increased difficulty in mapping the L2<br />
form onto its translation equivalent in L1. This would therefore explain why the cognate<br />
effect arises quite late in processing L2 words, but it cannot explain why there is an effect<br />
in L1.<br />
Differences in the time-course of the N400 component were observed in the studies<br />
in chapter two, three and four. An anterior delay in the N400 for L2 processing was<br />
observed in the language effects article. In chapter three, the article regarding the effects<br />
of cross-language neighborhood found an earlier influence of L1 neighbors on L2 words.<br />
P | 145
Discussion générale<br />
The cognate manipulation in chapter four also produced time-course differences. The<br />
cognate effect was earlier in L1 than in L2. In order to more clearly observe and specify<br />
any differences in the timing of processing between an L1 and an L2 we chose to use a<br />
masked priming paradigm that manipulated two crucial factors in word recognition; form<br />
and meaning. Chapter five presents these studies in which we tested within-language<br />
repetition priming and between-language non-cognate translation priming in second<br />
language learners using the masked priming paradigm.<br />
In the case of within-language repetition priming for both L1 and L2 both the N250<br />
and the N400 components were found to be sensitive, thus much resembling and<br />
reproducing Holcomb and Grainger (2006) results for both L1 and L2. (See Table 1 for a<br />
summary of results.) At this point it is helpful to remember that Holcomb and Grainger<br />
propose that the N250 component reflects form based processing and the N400<br />
component reflects semantic processes.<br />
In the case of translation priming the N400 effect for the L1-L2 direction was robust<br />
and widespread and in the predicted direction, while in the L2- L1 priming direction an<br />
anterior effect in the opposite direction was found. N250 amplitude was found to be<br />
modulated by translation primes only when primes were in L1 and targets in L2.<br />
Table 1. summary of results of priming in N250 and N400 components.<br />
within language repetition translation priming<br />
L1-L1 L2-L2 L2-L1 L1-L2<br />
N250 x x x<br />
N400 x x ~ x<br />
x = clear significant result expected as per Holcomb & Grainger<br />
~ = results in opposite direction<br />
In terms of models of bilingual word recognition, the present results fit well with a<br />
bilingual IA model such as the BIA model. L1 primes would rapidly activate the<br />
corresponding semantic representation that would in turn feedback information to<br />
P | 146
Discussion générale<br />
appropriate form-level representations in L1 and L2, hence modifying processing of L2<br />
targets that are translations of the L1 prime. This would account for N400 effects in L1-<br />
L1 repetition priming and in L1-L2 translation priming.<br />
This pattern of results for the N250 component is most interesting. N250 effects<br />
were predicted and found in L1-L1 and L2-L2 priming replicating Holcomb and Grainger<br />
(2006). However, the fact that the N250 component was found to be sensitive to<br />
translation priming when primes were in L1 and targets in L2 is perhaps the key finding<br />
of this study. Holcomb and Grainger suggested that the N250 reflects purely form-level<br />
processing. The present results provide a clear demonstration that semantic overlap across<br />
primes and targets, in the absence of form overlap, can influence N250 amplitude. These<br />
semantic influences on the N250 likely reflect interactive processing, whereby higherlevel<br />
semantic information feeds back to influence on-going form-level processing. In<br />
such an account, semantic information is rapidly accessed during visual word recognition,<br />
after some minimal form-level processing. This finding is interesting for word<br />
recognition in general.<br />
It is notable that we obtained robust effects on both the N250 component and the<br />
N400 component in the case of L2-L2 priming and an absence of these effects in L2-L1<br />
priming. The participants were not unable to process L2 primes masked at 50 ms, they<br />
were only unable to process these L2 primes when overtly processing a list of L1 words.<br />
It is possible that at 50 ms there was not sufficient time in this short SOA paradigm for<br />
feedback from the semantic level to the form level to exert its influence on the L1 item<br />
given the slow processing of L2 and the rapid processing of L1. If this is correct then<br />
lengthening the SOA so as to allow the prime processing to progress further might boost<br />
feedback enough to support effects with L2 primes and L1 translation targets. Also our<br />
P | 147
Discussion générale<br />
results were found in a population of L2 learners. Perhaps proficient bilinguals would<br />
show translation priming at this SOA.<br />
A note on the RHM - according to Kroll and Stewart’s model, direct connectivity<br />
across translation equivalents is stronger from L2 to L1 than vice versa, particularly in<br />
second language learners such as the participants of this study. Therefore, according to a<br />
direct connectivity account of translation priming, we ought to have observed stronger<br />
effects with L2 primes and L1 targets and we observed effects only with L1 primes and<br />
L2 targets.<br />
In Conclusion<br />
The studies in this dissertation and the body of existing research lead us to draw<br />
some tentative conclusions about the nature of the bilingual lexicon. First of all, the<br />
research in this dissertation suggests that even though L1 and L2 are processed in much<br />
the same way there are nevertheless clear differences in L1 and L2 processing, notably in<br />
learners engaged in L2 acquisition. Moreover, it seems that subtle differences persist<br />
between L1 and L2 processing even in proficient bilinguals that are late learners of their<br />
second language. A second conclusion that can be drawn is that the bilingual lexicon can<br />
be regarded as a highly integrated, interactive system. And access to this integrated<br />
lexicon appears to be non-selective. A third conclusion is that the acquisition of a second<br />
language restructures certain aspects of the existing L1 lexicon. If, as we have argued, the<br />
structure of this integrated lexicon of the late bilingual is similar to that of a monolingual<br />
then if follows that many of the things that we know about ways that monolinguals<br />
achieve visual word recognition apply. However, it is important to keep in mind that the<br />
persisting asymmetry associated with late learning adds to the complexity of the bilingual<br />
lexicon.<br />
P | 148
Discussion générale<br />
In the introduction we stated that many of the existing accounts of lexical<br />
processing describe monolingual processing with rare contingencies for describing<br />
bilingual processing. We also presented two models that attempt descriptions of the<br />
bilingual lexicon and that are very different in nature. It is our idea that both of these<br />
models capture some essence of the bilingual lexicon. We favor a model of bilingual<br />
word processing that is based solidly on known word processing phenomena with<br />
contingencies for handling the case of second language acquisition and the asymmetries<br />
that persist as well as the case of simultaneous early bilingualism. A bilingual IA model<br />
such as the BIA model appears to us to be a good candidate for a model of the bilingual<br />
lexical processing providing that the model includes contingencies for describing both the<br />
evolution of the lexicon in the case of acquisition and the case of simultaneous<br />
bilingualism.<br />
A true universal account of language processing must parsimoniously describe both<br />
monolingual processing and bilingual processing and the representations and processes<br />
involved. To achieve this grand ideal it is clear that more empirical work must be done to<br />
completely inform any description of the bilingual lexicon and it is our position that<br />
electrophysiological methods can provide a wealth of information to this end.<br />
P | 149
Bibliographie<br />
Bibliographie<br />
P | 151
Bibliographie<br />
Altarriba, J. (1992) The representation of translation equivalents in bilingual memory. In<br />
R.J. Harris (Ed.), Cognitive processing in bilinguals (157-174). Amsterdam:<br />
Elsevier.<br />
Alvarez, R., Grainger, J. & Holcomb, P.J. (2003). Accessing word meaning in primary<br />
and secondary languages: A comparison using event-related brain potentials.<br />
Brain & Language, 87(2), 290-304.<br />
Andrews, S. (1989). Frequency and neighborhood effects on lexical access: Activation or<br />
search? Journal of Experimental Psychology: Learning, Memory, and Cognition,<br />
15(5), 802-814.<br />
Andrews, S. (1997). The effect of orthographic similarity on lexical retrieval: Resolving<br />
neighborhood conflicts. Psychonomic Bulletin Review, 4(4), 439-461.<br />
Baayen, R.H., Piepenbrock, R. & Gulikers, L. (1995). The CELEX lexical database.<br />
Philadelphia, PA: Linguistic Data Consortium.<br />
Beauvillain, C. & Grainger, J. (1987). Accessing interlexical homographs: Some<br />
limitations of a language-selective access. Journal of Memory and Language,<br />
26(6), 658-672.<br />
Bentin, S., McCarthy, G. & Wood, C.C. (1985). Event-related potentials, lexical decision,<br />
and semantic priming. Electroencephalography and Clinical Neurophysiology,<br />
60(4), 343-355.<br />
Bentin, S., Mouchetant-Rostaing, Y., Giard, M.H., Echallier, J.F. & Pernier, J. (1999).<br />
ERP manifestations of processing printed words at different psycholinguistic<br />
levels: time course and scalp distribution. Journal of Cognitive Neuroscience,<br />
11(3), 35–60.<br />
Bowers, J. S., Mimouni, Z. & Arguin, M. (2000). Orthography plays a critical role in<br />
cognate priming: Evidence from French/English and Arabic/French cognates.<br />
Memory & Cognition, 28, 1289-1296.<br />
Brown, C.M. & Hagoort, P. (1993). The processing nature of the N400: Evidence from<br />
masked priming. Journal of Cognitive Neuroscience, 5, 34-44.<br />
Brysbaert, M., Van Dyck, G. & van de Poel, M. (1999). Visual word recognition in<br />
bilinguals: Evidence from masked phonological priming. Journal of Experimental<br />
Psychology: Human Perception and Performance, 25(1), 137-148.<br />
Caramazza, A. & Brones, I. (1979). Lexical access in bilinguals. Bulletin of the<br />
Psychonomic Society, 13, 212-214.<br />
Carreiras, M., Perea, M.,] & Grainger, J. (1997). Effects of the orthographic<br />
neighborhood in visual word recognition: Cross-task comparisons. Journal of<br />
Experimental Psychology: Learning, Memory, and Cognition, 23, 857-871.<br />
CELEX English database (Release E25). (1993). Available: www.mpi.nl/world/celex<br />
Chauncey, K., Grainger, J., & Holcomb, P.J. (2008). Code-switching effects in bilingual<br />
word recognition: A masked priming study with event-related potentials. Brain &<br />
Language, 105, 161-174<br />
P | 153
Bibliographie<br />
Chauncey, K., Holcomb, P.J. & Grainger, J. (2008). Effects of stimulus font and size on<br />
masked repetition priming: An ERP investigation. Language and Cognitive<br />
Processes, 23(1), 183-200.<br />
Chen, H-C. & Leung, Y-S. (1989). Patterns of lexical processing in a non-native<br />
language. Journal of Experimental Psychology: Learning, Memory and Cognition,<br />
15(2), 316-325<br />
Coltheart, M., Davelaar, E., Jonasson, J.T. & Besner, D. (1977). Access to the internal<br />
lexicon. In S. Dornic (Ed.), Attention and Performance VI (535-555). London:<br />
Academic Press.<br />
Cristoffanini, P., Kirsner, K. & Milech, D. (1986). Bilingual lexical representation: The<br />
status of Spanish–English cognates. Quarterly Journal of Experimental<br />
Psychology, 38A, 367-393.<br />
de Bruijn, E.R.A., Dijkstra, T., Chwilla, D.J. & Schriefers, H.J. (2001). Language context<br />
effects on interlingual homograph recognition: evidence from event-related<br />
potentials and response times in semantic priming. Bilingualism: Language and<br />
Cognition, 4, 155-168.<br />
de Groot, A. M. B. (1992). Determinants of word translation. Journal of Experimental<br />
Psychology: Learning, Memory, and Cognition, 18, 1001-1018.<br />
de Groot, A. M. B., Delmaar, P. & Lupker, S. J. (2000). The processing of interlexical<br />
homographs in translation recognition and lexical decision: Support for<br />
nonselective access to bilingual memory. Quarterly Journal of Experimental<br />
Psychology, 53A, 397-428.<br />
de Groot, A.B.M. & Kroll, J.F. (Eds.). (1997). Tutorials in Bilingualism, Psycholinguistic<br />
Perspectives. Mahwah, NJ: Lawrence Erlbaum Associates,<br />
de Groot, A.M. & Nas, G.L. (1991). Lexical representation of cognates and noncognates<br />
in compound bilinguals. Journal of Memory and Language. 30, 90-123.<br />
Deacon, D., Hewitt, S., Yang, C.M. & Nagata, M. (2000). Event-related potential indices<br />
of semantic priming using masked and unmasked words: evidence that the N400<br />
does not reflect a post-lexical process. Cognitive Brain Research. 9, 137–146.<br />
Dijkstra, A., van Jaarsveld, H. & Ten Brinke, S. (1998). Interlingual homograph<br />
recognition: Effects of task demands and language intermixing. Bilingualism:<br />
Language & Cognition, 1(1), 51-66.<br />
Dijkstra, T., Grainger, J. & van Heuven, W.J.B. (1999). Recognition of cognates and<br />
interlingual homographs: The neglected role of phonology. Journal of Memory<br />
and Language, 41, 496-518.<br />
Dijkstra, T., Timmermans, M. & Schriefers, H. (2000). On being blinded by your other<br />
language: Effects of task demands on interlingual homograph recognition. Journal<br />
of Memory and Language, 42(4), 445-464.<br />
Dijkstra, T. & van Heuven, W.J.B. (1998). The BIA model and bilingual word<br />
recognition. In J. Grainger A. M. Jacobs (Eds.), Localist connectionist approaches<br />
to human cognition (189-225). Mahwah, NJ: Lawrence Erlbaum Associates.<br />
P | 154
Bibliographie<br />
Dijkstra, T. & van Heuven, W.J.B. (2002). The architecture of the bilingual word<br />
recognition system: From identification to decision. Bilingualism: Language and<br />
Cognition, 5, 175-197.<br />
Dijkstra, T., van Heuven, W.J.B. & Grainger, J., (1998). Simulating cross-language<br />
competition with the bilingual interactive activation model. Psychologica Belgica,<br />
38(3-4), 177-196.<br />
Dyer, F.N. (1973). The stroop phenomenon and its use in the study of perceptual,<br />
cognitive, and response processes. Memory & Cognition, 1(2), 106-120.<br />
Finkbeiner, M., Forster, K., Nicol, J. & Nakamura, K. (2004). The role of polysemy in<br />
masked semantic and translation priming. Journal of Memory & Language, 51, 1-<br />
22.<br />
Forster, K. I. (1976). Accessing the mental lexicon. In E.C.J. Walker & R.J. Wales (Eds.),<br />
New approaches to language mechanisms (257–287). Amsterdam:North-Holland.<br />
Forster, K. I., Mohan, K. & Hector, J. (2003). The mechanics of masked priming. In: S.<br />
Kinoshita & S.J. Lupker (Eds), Masked priming: State of the art. Macquarie<br />
monographs in cognitive science, New York, NY: Psychology Press.<br />
Fox, E. (1996). Cross language priming from ignored words: Evidence for a common<br />
representational system in bilinguals. Journal of Memory and Language, 35, 353-<br />
370.<br />
Gerard, L.D. & Scarborough, D.L. (1989). Language-specific lexical access of<br />
homographs by bilinguals. Journal of Experimental Psychology, 15(2), 305-315.<br />
Gollan, T. H., Forster, K. I. & Frost, R. (1997). Translation priming with different scripts:<br />
Masked priming with cognates and non-cognates in Hebrew-English bilinguals.<br />
Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(5),<br />
1122-1139.<br />
Grainger, J. (1990). Word frequency and neighborhood frequency effects in lexical<br />
decision and naming. Journal of Memory and Language, 29(2), 228-244.<br />
Grainger, J. & Beauvillain, C. (1987). Language blocking and lexical access in bilinguals.<br />
Quarterly Journal of Experimental Psychology, 39A, 295-319.<br />
Grainger, J. & Dijkstra, T. (1992). On the representation and use of language information<br />
in bilinguals. In R.J. Harris (Ed.), Cognitive processing in bilinguals (207-220).<br />
Amsterdam: North-Holland.<br />
Grainger, J. & Frenck-Mestre, C. (1998). Masked priming by translation equivalents in<br />
proficient bilinguals. Language and Cognitive Processes, 13, 601-623.<br />
Grainger, J. & Holcomb, P.J. (in press). Neural constraints on a functional architecture for<br />
word recognition. In P. Cornelissen, P. Hansen, M. Kringelbach & K. Pugh (Eds.),<br />
The neural basis of reading. Oxford: Oxford University Press.<br />
Grainger, J. & Jacobs, A. M. (1996). Orthographic processing in visual word recognition:<br />
A multiple read-out model. Psychological Review, 103(3), 518-565.<br />
Grainger, J. Kiyonaga, K. & Holcomb, P.J. (2006). The time-course of orthographic and<br />
phonological code activation. Psychological Science, 17, 1021-1026.<br />
P | 155
Bibliographie<br />
Grainger, J., Midgley, K.J. & Holcomb P.J. (in preparation). Identical words are read<br />
differently in a bilingual’s two languages.<br />
Grainger, J., Muneaux, M., Farioli, F., & Ziegler, J. (2005). Effects of phonological and<br />
orthographic neighborhood density interact in visual word recognition. Quarterly<br />
Journal of Experimental Psychology, 58A, 981-998.<br />
Grainger, J., O'Regan, J., Jacobs, A. M. & Segui, J. (1989). On the role of competing<br />
word units in visual word recognition: The neighborhood frequency effect.<br />
Perception Psychophysics, 45(3), 189-195.<br />
Greenhouse, S.W. & Geisser, S. (1959). On methods in the analysis of profile data.<br />
Psychometrica, 24, 95-112.<br />
Grosjean, F. (1982). Life with two languages. Cambridge: Harvard University Press.<br />
Grosjean, F. (1988). Exploring the recognition of guest words in bilingual speech.<br />
Language and Cognitive Processes, 3(3), 233-274.<br />
Grosjean, F. (1989). Neurolinguists beware! The bilingual is not two monolinguals in one<br />
person. Brain & Language, 36, 3-15.<br />
Grosjean, F. & Miller, J. (1994). Going in and out of languages: an example of bilingual<br />
flexibility. Psychological Sciences 5(4), 201-206.<br />
Grossi, G. (2006). Relatedness proportion effects on masked associative priming: An<br />
ERP study Psychophysiology, 43(1), 21-30.<br />
Guttentag, R.E., Haith, M.M., Goodman, G.S. & Hauch, J. (1984). Semantic processing<br />
of unattended words by bilinguals: A test of the input switch mechanism. Journal<br />
of Verbal Learning & Verbal Behavior, 23(2), 178-188.<br />
Hahne, A. & Friederici, A.D. (2001). Processing a second language: late learners'<br />
comprehension mechanisms as revealed by event-related brain potentials.<br />
Bilingualism: Language and Cognition, 4, 123-141.<br />
Holcomb, P.J. (1993). Semantic priming and stimulus degradation: Implications for the<br />
role of the N400 in language processing. Psychophysiology, 30(1), 47-61.<br />
Holcomb, P.J. & Grainger, J. (2006). On the time course of visual word recognition: An<br />
event-related potential investigation using masked repetition priming. Journal of<br />
Cognitive Neuroscience, 18(10), 1631-1643.<br />
Holcomb, P.J. & Grainger, J. (2007). Exploring the temporal dynamics of visual word<br />
recognition in the masked repetition priming paradigm using event-related<br />
potentials. Brain Research, 1180, 39–58.<br />
Holcomb P.J. & Grainger, J. (in press). ERP Effects of Short Interval Masked Associative<br />
and Repetition Priming. Journal of Neurolinguistics.<br />
Holcomb, P.J., Grainger, J. & O'Rourke, T. (2002). An electrophysiological study of the<br />
effects of orthographic neighborhood size on printed word perception. Journal of<br />
Cognitive Neuroscience, 14(6), 938–950.<br />
Holcomb, P.J. & Neville, H.J., (1990) Auditory and visual semantic priming in lexical<br />
decision: A comparison using event-related brain potentials. Language and<br />
Cognitive Processes, 5(4), 281-312.<br />
P | 156
Bibliographie<br />
Holcomb, P.J., Reder, L., Misra, M. & Grainger, J. (2005). Masked priming: An eventrelated<br />
brain potential study of repetition and semantic effects. Cognitive Brain<br />
Research, 24(1), 155-172.<br />
Jared, D. & Szucs, C. (2002). Phonological activation in bilinguals: Evidence from<br />
interlingual homograph naming. Bilingualism: Language and Cognition, 5, 225-<br />
239.<br />
Jiang, N. (1999). Testing processing explanations for the asymmetry in masked crosslanguage<br />
priming. Bilingualism: Language and Cognition, 2, 59-75.<br />
Keatley, C. & de Gelder, B. (1992). The bilingual primed lexical decision task: Crosslanguage<br />
priming disappears with speeded responses. European Journal of<br />
Cognitive Psychology, 4, 273-292.<br />
Keatley, C., Spinks, J. & de Gelder, B (1994). Asymmetrical semantic facilitation<br />
between languages. Memory & Cognition, 22, 70-84.<br />
Kerkhofs, R., Dijkstra, T., Chwilla, D.J., & de Bruijn, E.R.A. (2006). Testing a model for<br />
bilingual semantic priming with interlingual homographs: RT and N400 effects.<br />
Brain Research, 1068, 170-183.<br />
Kiefer, M. (2002). The N400 is modulated by unconsciously perceived masked words:<br />
Further evidence for an automatic spreading activation account of N400 priming<br />
effects. Cognitive Brain Research, 13(1), 27-39.<br />
Kiyonaga, K., Grainger, J., Midgley, K.J. & Holcomb, P.J. (2007). Masked cross-modal<br />
repetition priming: An ERP investigation. Language and Cognitive Processes,22,<br />
337-376<br />
Kotz, S.A., (2001). Neurolinguistic evidence for bilingual language representation: a<br />
comparison of reaction times and event-related brain potentials. Bilingualism:<br />
Language and Cognition, 4, 143–154.<br />
Kotz, S.A. & Elston-Guttler, E. (2004). The role of proficiency on processing categorical<br />
and associative information in the L2 as revealed by reaction times and eventrelated<br />
brain potentials. Journal of Neurolinguistics, 17, 215-235.<br />
Kounios, J. & Holcomb, P. (1992). Structure and process in semantic memory: Evidence<br />
from event-related brain potentials and reaction times. Journal of Experimental<br />
Psychology: General, 121(4), 459-479.<br />
Kounios, J. & Holcomb, P.J. (1994). Concreteness effects in semantic processing: Eventrelated<br />
brain potential evidence supporting dual-coding theory. Journal of<br />
Experimental Psychology: Learning, Memory and Cognition, 20(4), 804-823.<br />
Kroll, J.F. & Curley, J. (1988). Lexical memory in novice bilinguals: The role of<br />
concepts in retrieving second language words. In M.M. Gruneberg, P.E. Morris &<br />
R.N. Sykes (Eds). Practical Aspects of Memory Vol 2 (389-395). London: John<br />
Wiley.<br />
Kroll, J.F. & Sholl, A. (1992). Lexical and conceptual memory in fluent and nonfluent<br />
bilinguals. In R.J. Harris (Ed.), Cognitive processing in bilinguals (191-204).<br />
Amsterdam: Elsevier.<br />
P | 157
Bibliographie<br />
Kroll, J.F. & Stewart, E. (1994). Category interference in translation and picture naming:<br />
evidence for asymmetric connections between bilingual memory representations.<br />
Journal of Memory and Language, 33, 149-174.<br />
Kutas, M. & Hillyard, S.A. (1980). Reading senseless sentences: Brain potentials reflect<br />
semantic incongruity, Science, 207, 203–205.<br />
Kutas, M. & Hillyard, S.A. (1984). Brain potentials during reading reflect word<br />
expectancy and semantic association. Nature, 307, 161-163.<br />
Lalor, E. & Kirsner, K. (2001). The representation of “false cognates” in the bilingual<br />
lexicon. Psychonomic Bulletin & Review, 8, 552-559.<br />
Lau, E.F., Phillips, C. & Poeppel, D. (2008). A cortical network for semantics:<br />
(de)constructing the N400. Nature Neuroscience. 9, 920-933.<br />
Lemhöfer, K. & Dijkstra, T. (2004). Recognizing cognates and interlingual homographs:<br />
Effects of code similarity in language-specific and generalized lexical decision.<br />
Memory & Cognition, 32(4), 533-550.<br />
Lemhöfer, K., Dijkstra, T. & Michel, M.C. (2004). Three languages, one ECHO: Cognate<br />
effects in trilingual word recognition. Language & Cognitive Processes, 19, 585-<br />
611.<br />
Lemhöfer, K., Dijkstra, T., Schriefers, H., Baayen, H., Grainger, J., & Zwitserlood, P.<br />
(2008). Native language influences on word recognition in a second language: A<br />
mega-study. Journal of Experimental Psychology: Learning, Memory, and<br />
Cognition, 43(1), 12-31.<br />
Luck, S. J. (2005). An introduction to the event-related potential technique. MIT Press,<br />
Cambridge, MA.<br />
Macnamara, J. (1967). The linguistic independence of bilinguals. Journal of Verbal<br />
Learning & Verbal Behavior 6, 729-736.<br />
Macnamara, J., & Kushnir, S. L. (1971). Linguistic independence of bilinguals: The input<br />
switch. Journal of Verbal Learning & Verbal Behavior, 10, 480-487.<br />
Marian, V. & Spivey, M. (2003). Competing activation in bilingual language processing:<br />
Within- and between-language competition. Bilingualism: Language and<br />
Cognition, 6, 97-115.<br />
McClelland, J.L. (1979). On the time relations of mental processes: an examination of<br />
systems of processes in cascade, Psychological Review, 86, 287–330.<br />
McClelland & J.L., Rumelhart, D.E. (1981). An interactive model of context effects in<br />
letter perception: Part i. An account of basic findings. Psychological Review, 88,<br />
375-407.<br />
McLaughlin, J., Osterhout, L. & Kim, A. (2004). Neural correlates of second-language<br />
word learning: Minimal instruction produces rapid change. Nature Neuroscience,<br />
7, 703-704.<br />
Meyer, D.E., & Schvaneveldt, R.W. (1971). Facilitation in recognizing pairs of words:<br />
Evidence of a dependence between retrieval operations. Journal of Experimental<br />
Psychology, 90, 227-234.<br />
P | 158
Bibliographie<br />
Midgley, K.J., Holcomb P.J. & Grainger, J. (in press-a). Language Effects in Second<br />
Language Learners and Proficient Bilinguals Investigated with Event-Related<br />
Potentials. Journal of Neurolinguistics.<br />
Midgley, K.J., Holcomb P.J. & Grainger, J. (in press-b). Masked Repetition and<br />
Translation Priming in Second Language Learners: A Window on the Time-<br />
Course of Form and Meaning Activation using ERPs. Psychophysiology.<br />
Midgley, K.J., Holcomb P.J., van Heuven, W.J.B. & Grainger, J. (2008). An<br />
Electrophysiological investigation of cross-language effects of orthographic<br />
neighborhood. Brain Research,1246, 123-135<br />
Misra, M. & Holcomb, PJ. (2003). Event-related potential indices of masked repetition<br />
priming. Psychophysiology, 40(1), 115-130.<br />
Moreno, E. & Kutas, M. (2005). Processing semantic anomalies in two languages: an<br />
electrophysiological exploration in both languages of Spanish–English bilinguals.<br />
Cognitive Brain Research, 22, 205-220.<br />
Moreno, E.M., Rodríguez-Fornells, A. & Laine, M. (2008). Event-related potentials<br />
(ERPs) in the study of bilingual language processing. Journal of Neurolinguistics,<br />
21, 477-508.<br />
Morris, J., Franck, T., Grainger, J., & Holcomb, P.J. (2007). Semantic transparency and<br />
masked morphological priming: AN ERP investigation. Psychophysiology, 44,<br />
506-521.<br />
Morrison, C. M., & Ellis, A. W. (2000). Real age of acquisition effects in word naming<br />
and lexical decision. British Journal of Psychology, 91, 167-180.<br />
Morrison, C. M. & Ellis, A.W. (1995). Roles of word frequency and age of acquisition in<br />
word naming and lexical decision. Journal of Experimental Psychology:<br />
Learning, Memory, & Cognition, 21, 116-133.<br />
Münte, T., Wieringa, B.M., Weyerts, H., Szentkuti, A., Matzke, M. & Johannes, S.<br />
(2001). Differences in brain potentials to open and closed class words: class and<br />
frequency effects, Neuropsychologia, 39(1), 91-102.<br />
Nas, G. (1983). Visual word recognition in bilinguals: Evidence for a cooperation<br />
between visual and sound based codes during access to a common lexical store.<br />
Journal of Verbal Learning & Verbal Behavior, 22(5), 526-534.<br />
Neville, H. J, Coffey, S. A, Lawson, D. S, Fischer, A. (1997) Neural systems mediating<br />
American Sign Language: Effects of sensory experience and age of acquisition.<br />
Brain and Language, 57(3), 285-308.<br />
Neville, H. J, Mills, D. L, & Lawson, D. S. (1992) Fractionating language: Different<br />
neural subsystems with different sensitive periods. Cerebral Cortex, 2(3), 244-<br />
258.<br />
New, B., Pallier, C., Ferrand L. & Matos, R. (2001). Une base de données lexicales du<br />
français contemporain sur internet: Lexique. L'Année Psychologique, 101, 447-<br />
462.<br />
New, B., Pallier, C., Brysbaert, M. & Ferrand, L. (2004). Lexique 2: A new French<br />
lexical database. Behavior Research Methods, Instruments & Computers, 36(3),<br />
516-524.<br />
P | 159
Bibliographie<br />
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh<br />
inventory. Neuropsychologia, 9(1), 97-113.<br />
Osterhout, L. & Holcomb, P.J. (1995). The electrophysiology of language<br />
comprehension. In: M.D. Rugg, & M.G.H. Coles, (Eds.), Electrophysiology of<br />
Mind (171-215). Oxford: Oxford University Press.<br />
Phillips, N.A., Klein, D., Mercier, J. & de Boysson, C. (2006). ERP measures of<br />
auditory word repetition and translation priming in bilinguals. Brain Research,<br />
1125, 116-131.<br />
Phillips, N.A., Segalowitz, N., O'Brien, I., & Yamasaki, N., (2004). Semantic priming in<br />
a first and second language: evidence from reaction time variability and eventrelated<br />
brain potentials. Journal of Neurolinguistics. 17, 237–262.<br />
Rastle, K., Davis, M.H., Marslen-Wilson, W.D., & Tyler, L.K. (2000). Morphological &<br />
semantic effects in visual word recognition: A time-course study. Language and<br />
Cognitive Processes, 15, 507–537.<br />
Rodríguez-Fornells, A., Rotte, M., Heinze, H.J., Nosselt, T. & Munte, T.F. (2002). Brain<br />
potential and functional-MRI evidence for how to handle two languages with one<br />
brain. Nature, 415(6875), 1026-1029.<br />
Rugg, M.D. & Coles, M.G.H. (Eds.). (1995). Electrophysiology of mind: Event-related<br />
brain potentials and cognition. Oxford: Oxford University Press.<br />
Ruschemeyer, S., Nojack, A. & Limbach, M. (2008). A mouse with a roof? Effects of<br />
phonological neighbors on processing of words in sentences in a non-native<br />
language. Brain and Language, 104(2), 113-121.<br />
Sanchez-Casas, R.M., Davis, C.W. & Garcia-Albea, J.E. (1992). Bilingual lexical<br />
processing: Exploring the cognate/non-cognate distinction. European Journal of<br />
Cognitive Psychology, 4, 293–310.<br />
Soares, C. & Grosjean, F. (1984). Bilinguals in a monolingual and a bilingual speech<br />
mode: The effect on lexical access. Memory Cognition, 12(4), 380-386.<br />
Tanenhaus, M.K., Magnuson, J.S., Dahan, D. & Chambers, C. (2000). Eye movements<br />
and lexical access in spoken-language comprehension: Evaluating a linking<br />
hypothesis between fixations and linguistic processing. Journal of<br />
Psycholinguistic Research, 29(6), 557-580.<br />
Thierry, G. & Wu, Y.J. (2007). Brain potentials reveal unconscious translation during<br />
foreign-language comprehension. Proceedings of the National Academy of<br />
Sciences. 104, 12530-12535.<br />
Thomas, M. S. & Allport, A. (2000). Language switching costs in bilingual visual word<br />
recognition. Journal of Memory and Language, 43(1), 44-66.<br />
van Hell, J.G. & de Groot, A.M.B. (1998). Conceptual representation in bilingual<br />
memory: Effects of concreteness and cognate status in word association.<br />
Bilingualism: Language & Cognition, 1(3), 193-211.<br />
van Hell, J. G. & Dijkstra, T. (2002). Foreign language knowledge can influence native<br />
language performance in exclusively native contexts. Psychonomic Bulletin &<br />
Review, 9(4), 780-789.<br />
P | 160
Bibliographie<br />
van Heuven, W.J., Dijkstra, T. & Grainger, J. (1998). Orthographic neighborhood effects<br />
in bilingual word recognition. Journal of Memory and Language, 39(3), 458-483.<br />
van Heuven, W.J.B., Schriefers, H., Dijkstra, T. & Hagoort, P. (2008). Language conflict<br />
in the bilingual brain. Cerebral Cortex, 39(11), 2706-2716.<br />
Van Petten, C. & Kutas, M. (1990). Interactions between sentence context and word<br />
frequency in event-related brain potentials. Memory & Cognition, 18, 380-393.<br />
Voga, M. & Grainger, J. (2007). Cognate status and cross-script translation priming.<br />
Memory & Cognition, 35, 938-952.<br />
Weber-Fox, C.M. & Neville, H.J., (1996). Maturational constraints on functional<br />
specializations for language processing: ERP and behavioral evidence in bilingual<br />
speakers. Journal of Cognitive Neuroscience, 8, 231–256<br />
Zevin, J.D. & Seidenberg, M.S. (2004) Age-of-acquisition effects in reading aloud: Test<br />
of cumulative frequency and frequency trajectory. Memory & Cognition, 32, 31-<br />
38.<br />
P | 161