Covert bilingual language activation through cognate word processing: An eye-tracking study

Abstract

The present study examined effects of cross-linguistic overlap and language proficiency on bilingual parallel language activation. Recognition of cognates (e.g., English hen, German Henne) was compared to recognition of non-cognates (e.g., English bike, German Fahrrad) in English-German bilinguals, German-English bilinguals, and monolinguals. In an eye-tracking paradigm, participants identified English cognate and non-cognate targets in picture displays that also contained German competitors with shared phonological onsets. The duration of phonological overlap between targets and competitors was manipulated. English-German bilinguals co-activated German competitors of cognate targets, while German-English bilinguals co-activated German competitors of non-cognate targets, as well as German competitors of cognate targets with high phonological overlap. These results suggest that proficiency in target and non-target languages, as well as cross-linguistic overlap, determine the extent of bilinguals' parallel language activation.

Full text

Covert Bilingual Language Activation through Cognate Word Processing:
An Eye-tracking Study
Henrike K. Blumenfeld (k-blumenfeld@northwestern.edu)
Department of Communication Sciences and Disorders, 2240 Campus Drive
Evanston, IL 60208 USA
Viorica Marian (v-marian@northwestern.edu)
Department of Communication Sciences and Disorders, 2240 Campus Drive
Evanston, IL 60208 USA
Abstract
The present study examined effects of cross-linguistic overlap
and language proficiency on bilingual parallel language
activation. Recognition of cognates (e.g., English hen,
German Henne) was compared to recognition of non-cognates
(e.g., English bike, German Fahrrad) in English-German
bilinguals, German-English bilinguals, and monolinguals. In
an eye-tracking paradigm, participants identified English
cognate and non-cognate targets in picture displays that also
contained German competitors with shared phonological
onsets. The duration of phonological overlap between targets
and competitors was manipulated. English-German bilinguals
co-activated German competitors of cognate targets, while
German-English bilinguals co-activated German competitors
of non-cognate targets, as well as German competitors of
cognate targets with high phonological overlap. These results
suggest that proficiency in target and non-target languages, as
well as cross-linguistic overlap, determine the extent of
bilinguals’ parallel language activation.
Keywords: bilingualism; psycholinguistics; eye-tracking;
spoken language processing
Introduction
Bilinguals have been shown to simultaneously activate both
languages during auditory word recognition. For instance,
covert and automatic activation of the non-target language
was found in bilinguals during spoken word recognition
tasks in monolingual settings (e.g., Marian & Spivey,
2003a,b; Spivey & Marian, 1999). Using an eye-tracking
paradigm (Tanenhaus et al., 2000), bilinguals’ eye
movements were monitored while they identified target
items (e.g., a marker) in a display that also contained
competitor items with shared phonological onsets in the
other language (e.g., a stamp, Russian marka). Bilinguals
looked more at between-language competitors than control
items, suggesting parallel activation of the two languages.
The current study extends this paradigm to study
cognates’ contribution to bilingual parallel language
activation. Cognate words are opportune translation
equivalents that sound highly similar across two languages
(e.g., English cactus, German Kaktus) and provide
processing advantages for bilinguals. For example,
children’s immersion in a second-language (L2)
environment can improve native language (L1) vocabulary
via cognates learned in the L2 (Cunningham & Graham,
2000). Bilingual anomic aphasic patients perform better in
naming tasks when targets are cognates than when they are
non-cognates (e.g., Roberts & Deslauriers, 1999). Cognate
targets may co-activate phonological cohorts in the non-
target language to a greater extent than non-cognate targets;
such co-activation can be gleaned from different patterns of
competitor activation for cognate and non-cognate targets.
Proficiency and Cross-Linguistic Overlap
The architecture of the bilingual lexicon is dynamic, in that
it changes throughout the language learning process (Kroll
& Stewart, 1994). During initial learning stages, bilinguals
establish form-level associations between translation
equivalents. Once an L2 word’s L1 translation has been
activated via form-association, semantic access occurs
through form-meaning links in the L1. Highly proficient
bilinguals have also established direct form-meaning links
in their L2. Recent research by Silverberg and Samuel
(2004) using an L1 lexical decision task found that late
bilinguals with high L2 proficiency showed form L1-L2
priming (suggesting L1-L2 connectivity at the form level),
but no semantic L1-L2 priming (suggesting that semantic
representations were not shared). However, late bilinguals
with low L2 proficiency showed no priming (suggesting
weak connections, and lack of between-language
interactivity). These findings suggest that language
proficiency influences parallel language activation, and that
late bilinguals may not share semantic representations for
non-cognate translation equivalents.
Cognates, on the other hand, have been found to share
semantic representations. Greater effects of cross-language
association priming were found for cognates than for non-
cognates (De Groot & Nas, 1991), and responses to
association generation were more consistent across
languages for cognates than for non-cognates (Van Hell &
De Groot, 1998). Moreover, in a monolingual setting,
Dutch-English-French trilinguals, with highly proficient
English but less proficient French, showed shorter
association generation and lexical decision times for Dutch-
English cognates than for non-cognates and for Dutch-
French cognates (Van Hell & Dijkstra, 2002). These

findings suggest that only proficient non-target languages
contribute to cognates’ shared semantic representations
enough to facilitate cognate selection.
In addition to semantic representations, phonological
representations are also more convergent for cognates than
for non-cognates. Studies of word-retrieval during
production, where phonological form is accessed last, have
ascribed facilitation effects for cognates vs. non-cognates to
shared phonological-level cognate representations (Costa,
Caramazza & Sebastian-Galles, 2000). Similarly, receptive
language tasks that rely on phonological activation, such as
lexical decision (Dijkstra, Van Heuven, & Grainger, 1998;
Van Hell & Dijkstra, 2002) and cross-script repetition
priming (Nakayama, 2000), have yielded shorter reaction
times for cognates than for non-cognates.
The Current Study
To examine the role of proficiency in parallel language
activation, we recruited two groups of late bilinguals. One
group was more proficient (native) in the target language,
English, and the other group was more proficient (native) in
the non-target language, German. To examine the role of
cross-linguistic overlap, we used cognate and non-cognate
target stimuli, and manipulated the extent of between-
language phonological overlap between targets and
competitors. Parallel activation of the non-target language
was probed covertly using English-only auditory cues. In an
eye-tracking task, participants identified pictures whose
names they heard in the target language, while pictures of
between-language competitors were also present. German
co-activation was measured as the percentage of looks to
competitors relative to controls. For non-cognate targets, we
predicted that co-activation of a non-target L1 would be
stronger than co-activation of a non-target L2. For cognate
targets, form-level and meaning-level integration across
languages were expected to modulate this pattern. We
specifically predicted that form-level integration of cognate
targets would result in increased sensitivity to phonological
overlap for competitors of cognate targets compared to
competitors of non-cognate targets.
Methods
Participants
= 28.7, SD = 12.9; 7 females) and 13 monolingual controls
(Mean age = 28.5, SD = 9.4; 7 females) participated. An
additional monolingual was excluded due to exposure to
German at work. The three groups did not differ in age [F(2,
39) = 0.4, p = .7]. Participants completed a Language
Experience and Bilingual Status Questionnaire (Marian,
Blumenfeld, & Kaushanskaya, 2003). All bilinguals
considered themselves dominant in their L1. The ages when
they started learning the second language [English-German
bilinguals: M = 11.8, SD = 8.6; German-English bilinguals:
M = 10.9, SD = 2.4; t(27) = 0.4, p = .7], and became fluent
in it [English-German bilinguals: M=17.4, SD =10.0;
German-English bilinguals: M = 19.1, SD = 6.8; t(27) = 0.5,
p = .6] were similar across groups. At the time of study,
English-German bilinguals had more exposure to the target
language, English (M = 86.2%, SD = 7.3) than German-
English bilinguals [M = 74.0%, SD = 16.3; t(27) = 2.6, p =
.02]; German-English bilinguals had more exposure to the
non-target language, German (M = 23.1%, SD = 16.3) than
English-German bilinguals [M = 11.1%, SD = 6.8; t(27) =
2.6, p = .02]. On the English Peabody Picture Vocabulary
Test (PPVT-III, Dunn & Dunn, 1997), English-German
bilinguals performed best (M = 195.4, SD = 3.9), followed
by English monolinguals (M = 190.9, SD = 6.8), and
German-English bilinguals [M = 172.7, SD = 15.2; F(2, 39)
= 30.2, p < .001; LSD post hocs: p < .05). A German
translation of the PPVT (version B, which consists of
different pictures) was administered to all bilinguals, and
German-English bilinguals (M = 193.9, SD = 7.6)
performed better than English-German bilinguals [M =
178.6, SD = 18.2; F(1, 27) = 8.9, p = .006]. The bilingual
groups did not differ in their L1 [t(27) = .7, p = .5] or L2
proficiency levels [t(27) = 1.0, p = .3].
Stimuli and Design
Stimulus displays consisted of panels with four pictures and
a central fixation cross. Concurrent with pictures, auditory
stimuli were “click on the [target picture] and the [filler
picture].” The name of the target picture was presented 400
ms after picture onset. A total of 128 trials were prepared. In
half of all trials, target pictures were cognates, and in the
other half they were non-cognates. For both, cognate and
non-cognate targets, a 2 x 3 x 3 design was employed, with
between-language competitor (competitor, control) and
phonological overlap (low, medium, high) as within-
Fourteen English-German bilinguals (Mean age = 25.6, SD
= 8.9; 5 females), 15 German-English bilinguals (Mean age
Table 2: Sample stimuli across conditions. Phonological onset overlap between targets and competitors is underlined.
Target status
English Target
Phonological Overlap
German Competitor
Control
Cognate
coffee (Kaffee)
Low
Kreis (circle)
Eichel (acorn)
hen (Henne)
Medium
Hemd (shirt)
Pfeil ( arrow)
file (Feile)
High
Pfeil ( arrow)
Korb (basket)
Non-cognate
pumpkin (Kürbis)
Low
Puppe (doll)
Hase (rabbit)
Curtain (Gardine)
Medium
Kirsche (cherry)
Bagger ( digger)
pickle (saure Gurke)
High
Pickel (zit)
Decke ( blanket)

subjects factors, and group as a between-subjects factor.
For an overview and sample stimuli, see Table 2.
Between-Language Competitor Within half of all trials
(64 trials), target pictures were accompanied by pictures of
between-language competitors, whose onset in German was
phonologically similar to the target’s onset in English. In the
other half of all trials, phonologically dissimilar controls
were inserted into the panel positions of the between-
language competitors. Each between-language competitor’s
picture served as a neutral filler in a different trial.
Picture stimuli were selected from the IMSI Master Clips
electronic database and the Alta Vista search engine, or
hand-drawn. Pictures were black line drawings with gray
shadings, balanced for thickness of line and shades within
each display. Positioning of the pictures in quadrants I-IV of
the visual display was controlled across trials, and the order
of presentation of trials was counterbalanced across
participants and conditions.
Phonological Overlap Phonological overlap between target
words and between-language competitors was manipulated
categorically with three levels, low, medium, and high.
Generally, these categories corresponded to one-phoneme,
two-phoneme and three-phoneme overlap, but final
grouping was based on measurements of overlap in
milliseconds, due to the time-sensitive nature of cohort
activation (e.g., Marslen-Wilson, 1987). Acoustic overlap
over time was measured using Sound Studio speech analysis
software, and recordings of English experimental stimuli
were compared to recordings of German competitors.
Between both cognate and non-cognate targets, and their
respective competitors, there were significant differences
between the low (cognates: M = 33.9ms, SE = 6.0ms; non-
cognate: M = 65.0ms, SE = 6.8ms), medium (cognates: M =
116.2ms, SE = 9.4; non-cognates: M = 120.1ms, SE = 4.7),
and high (cognates: M = 253.5ms, SE = 26.2; non-cognates:
M = 247.6ms, SE = 21.6) overlap conditions [cognates: F(2,
30) = 45.5, p < .001; non-cognates: F(2, 28) = 45.2, p <
.001; LSD post-hocs: p < .05].
Recordings of auditory stimuli were made in a sound-
proof booth (44, 100 Hz, 16 bits) by a female speaker of
American English. Stimulus sets were balanced for spoken
word frequency using the CELEX lexical database (Baayen,
Piepenbrock, & Van Rijn, 1995; F(9, 310) = 0.4, p = .9).
Procedures and Apparatus
All participants were welcomed into the lab using English as
the sole language of communication, and were told that the
goal of the experiment was to find out how they processed
English speech. After informed consent was obtained,
participants were fitted with a head-mounted eye-tracker
(ISCAN) and calibrated. They then read instructions on a
computer screen (G4 Macintosh, 11″x 13.5″), after which
the experimenter verbally reinforced important points, and
answered questions. A five-trial practice session on neutral
stimuli that did not re-occur during the experimental session
allowed participants to become familiar with the task. The
experimental session lasted approximately 20 minutes.
Following the session, participants were administered the
English Peabody Picture Vocabulary Test (PPVT).
Bilingual participants were administered a German
translation of the PPVT, version B, after a brief
conversation to establish the language and ensure
participants’ comfort with it. Finally, all participants filled
out the Language Experience and Bilingual Status
Questionnaire (LEABS-Q).
Coding and Analyses
Eye-tracking data, in the form of video output (participants’
field of view, fixation cross-hairs indicating position of
gaze, and auditory instructions) were manually coded at a
temporal resolution of 33.3 ms/frame using Final Cut
software. Eye-movements to pictures were coded as looks if
they entered the picture’s quadrant and remained there for at
least one frame. A second coder coded approximately 15%
of all data; inter-rater reliability computed using Pearson’s
R was 93.5%.
A total of 9.6% of data were excluded from analyses. Of
these, 3.6% were excluded due to participants noticing
German items, and 7.8 % were excluded due to problematic
pictures, which drew more looks to competitors than
controls in all groups, including monolinguals.
Cognate and non-cognate trials were analyzed using
2x3x3 repeated measures ANOVAs. Follow-up
comparisons were conducted for significant interactions.
Results
Cognate Targets
Between-group Comparisons A direct comparison across
the three groups revealed a 3-way interaction of Competitor,
Phonological Overlap, and Group [F(2, 39) = 3.8, p = .03, h
= .2], as well as a 2-way interaction of Competitor and
Group [F(2, 39) = 3.1, p = .048, h = .1]. These interactions
were followed-up with pair-wise comparisons. Three-way
interactions of Competitor, Phonological Overlap, and
Group were found, both when comparing monolinguals to
English-German bilinguals [F(1, 25) = 4.6, p = .04, h = .2],
and when comparing monolinguals to German-English
bilinguals [F(1, 26) = 6.6, p = .02, h = .2]. Further,
comparing monolinguals to English-German bilinguals
yielded a 2-way interaction of Competitor and Group [F(1,
25) = 6.0, p = .02, h = .2, see Figure 1]. Comparing English-
German bilinguals to German-English bilinguals yielded a
main effect of competitor [F(1,27) = 12.2, p = .002, h = .3],
where bilinguals looked more at competitors (M = 59.2%,
SE = 2.4%) than at controls (M = 51.7%, SE = 2.6%).
Moreover, an interaction of competitor and phonological
overlap was found [F(1, 27) = 16.0, p < .001, h = .4]. There
was no 3-way interaction between competitor, phonological
overlap and group [F(1, 27) = 0.2, p = .7, h = .006],
suggesting that the same interaction patterns of competitor

and phonological overlap existed across both bilingual
groups. The significant interactions obtained in between-
group comparisons were followed-up with within-group
ANOVAs and are reported below.
English Monolingual Controls Monolingual participants
looked at between-language competitors 58.5% (SE = 2.8)
of the time, and at controls 60.3% (SE = 4.3) of the time.
Within-group ANOVAs yielded no significant effects of
Competitor [F(1, 12) = 0.3, p = .6, h = .02] or interactions
of Competitor and Phonological Overlap [F(1, 12) = 0.4, p =
.5, h = .03], suggesting no differences across conditions.
English-German Bilinguals English-German bilingual
participants looked at competitors 55.9% (SE = 2.9) of the
time, and at controls 46.8% (SE = 3.5) of the time, yielding
a main effect of Competitor [F(1, 13) = 9.3, p = .009, h =
.4]. In addition, an interaction of Competitor and
Phonological Overlap was found [F(1, 13) = 6.1, p = .03, h
= .3], with an equal percentage of looks to competitors (M =
54.8%, SE = 4.3) and controls (M = 54.0%, SE = 5.3) at low
phonological overlap [t(13) = 0.2, p = .9], a greater
percentage of looks to competitors (M = 55.7%, SE = 4.4)
vs. controls (M = 45.1%, SE = 3.8) at medium phonological
overlap [t(13) = 2.6, p = .02], and a greater percentage of
looks to competitors (M = 57.4%, SE = 4.5) vs. controls (M
= 41.2%, SE = 5.0) at high phonological overlap [t(13) =
2.9, p = .01].
Figure 1: Percentage of looks to competitors and controls in
the low, medium, and high phonological overlap conditions,
for monolinguals (Ms), English-German bilinguals (EGBs),
and German-English bilinguals (GEBs) when the targets
were cognates.
German-English Bilinguals German-English bilingual
participants looked at competitors 62.5% (SE = 3.7) of the
time, and at controls 56.6% (SE = 3.9) of the time; the
difference did not reach significance [F(1, 14) = 3.6, p = .08,
h = .2]. However, as in the English-German bilingual group,
an interaction of Competitor and Phonological Overlap was
observed [F(1, 14) = 10.2, p = .007, h = .4]. There was an
equal percentage of looks to competitors (M = 59.3%, SE =
4.8) and controls (M = 61.5%, SE = 4.5) at low
phonological overlap [t(14) = 0.5, p = .6], an equal
percentage of looks to competitors (M = 62.6%, SE = 3.6)
and controls (M = 59.2%, SE = 3.8) at medium
phonological overlap [t(14) = 0.8, p = .4], and a higher
percentage of looks to competitors (M = 65.6%, SE = 5.0)
vs. controls (M = 49.1%, SE = 5.7) at high phonological
overlap [t(14) = 3.1, p = .007].
Non-cognate Targets
Between-group Comparisons A direct comparison across
the three groups revealed a 2-way interaction of Competitor
and Group [F(2, 39) = 3.2, p = .049, h = .1]. Follow-up pair-
wise comparisons revealed that the competitor-group
interaction was significant only between monolinguals and
German-English bilinguals [F(1, 26) = 4.0, p = .047, h = .1],
but was not significant between monolinguals and English-
German bilinguals [F(1, 25) = 0.005, p = .9, h < .001], or
between English-German bilinguals and German-English
bilinguals [F(1, 27) = 3.6, p = .07, h = .1].
English Monolingual Controls Monolingual controls
looked at competitors 55.1% (SE = 2.9) of the time, and at
controls 54.6% (SE = 3.8) of the time. A follow-up
ANOVA within the monolingual group yielded no main
effect of competitor [F(1, 12) = 0.1, p = .8, h = .006].
English-German Bilinguals English-German bilingual
participants looked at between-language competitors 46.4%
(SE = 3.3) of the time, and at controls 45.7% (SE = 3.6) of
the time. No main effect of competitor was found [F(1, 13)
= 0.08, p = .8, h = .006], suggesting that English-German
bilinguals did not co-activate the non-target language when
the target was a non-cognate. At low phonological overlap,
English-German bilinguals looked at competitors 40.0%
(SE = 4.9) of the time and at controls 30.5% (SE = 4.9) of
the time. At medium phonological overlap, they looked at
competitors 47.2% (SE = 5.7) of the time and at controls
54.8% (SE = 5.9) of the time. At high phonological overlap,
they looked at competitors 52.0% (SE = 5.3) of the time and
at controls 52.0% (SE = 4.2) of the time. No interaction of
competitor, phonological overlap, and group (Monolingual,
English-German bilingual) was present.
German-English Bilinguals German-English bilingual
participants looked at competitors 65.9% (SE = 3.2) of the
time, and at controls 56.2% (SE = 3.6) of the time, yielding
a main effect of competitor [F(1, 14) = 5.6, p = .03, h = .3].
This suggests that German-English bilinguals co-activated
the non-target language when the target was a non-cognate.
At low phonological overlap, German-English bilinguals
looked at competitors 64.7% (SE = 4.7) of the time and at
controls 52.1% (SE = 5.1) of the time. At medium
phonological overlap, they looked at competitors 65.8% (SE
= 5.5) of the time and at controls 62.2% (SE = 5.5) of the
time. At high phonological overlap, they looked at
0
10
20
30
40
50
60
70
80
Lo Med Hi Lo Med Hi Lo Med Hi
Competitor Control
Ms
EGBs
GEBs
*
*
*
Percentage of Looks (%)

competitors 67.0% (SE = 4.8) of the time and at controls
54.3% (SE = 4.9) of the time. No interaction of competitor,
phonological overlap, and group (Monolingual, English-
German bilingual) was present.
Discussion
The present study employed eye-tracking to investigate
covert activation of a non-target language. When the target
language was an L1 (i.e., English for English-German
bilinguals), competitors from the non-target L2 were co-
activated in the presence of cognate targets, but not in the
presence of non-cognate targets. When the target language
was an L2 (i.e., English for German-English bilinguals),
competitors from the non-target L1 were co-activated in the
presence of cognate targets only when high phonological
overlap existed between the competitors and the targets.
Further, competitors were consistently co-activated in the
presence of non-cognate targets. These findings suggest that
parallel activation of two languages depends on relative
proficiency levels in the target and non-target languages, as
well as on cross-language overlap.
Cognate and Non-cognate Status
The pattern of competitor activation with cognate and non-
cognate targets differed when the non-target language was
an L1 compared to when it was an L2. L2 competitors were
co-activated in the presence of cognate targets, but not in the
presence of non-cognate targets, suggesting that cognate
targets facilitated L2 competitor activation. L1 competitors
were co-activated in the presence of cognate targets only
when there was high phonological overlap, while being
consistently activated in the presence of non-cognate
targets.
Cognate Status We propose that during cognate processing
in L2, L1 co-activation contributes strongly to activation of
shared semantic representations, making competition from
the L1 competitor unlikely. However, during cognate
processing in L1, L2 co-activation contributes less to
activation of shared semantic representations, which makes
competition from the L2 competitor more likely. How does
this L1-L2 ‘asymmetry’ evolve? The Revised Hierarchical
Model (RHM, Kroll & Stewart, 1994) suggests that late
bilinguals learn their L2 via form-association, resulting in
strong form-level connections between the two lexicons.
While within-language form-meaning connections are
strong in the L1, and L2 meaning is initially accessed via
L1, the fluent late bilingual has also developed direct form-
meaning links within L2. Other findings suggest that in late
bilinguals semantic representations are generally not shared
(Silverberg & Samuel, 2004), while cognates may be the
exception to this rule (e.g., De Groot & Nas, 1991). Due to
cognates’ high form-level similarity, L2 cognate acquisition
may rely heavily on existent L1 representations. Semantic
access of L2 cognates may proceed through L1
representations for a longer time. When direct form-
meaning links are established in L2, parallel semantic
access through the strong L1 pathway (and onto shared
representations) may prevail (Dijkstra & Van Hell, 2002).
Thus, L1s continuous and consequently strong activation
during cognate processing may account for its influence on
semantic activation (and selection) of target L2 cognates.
Further, strong co-activation of the non-target L1 pathway,
leading onto shared semantic representations, may reduce
the chance that L1 competitors will be co-activated to the
level necessary to compete.
Non-cognate Status For non-cognate targets, consistent co-
activation of L1 competitors, but not L2 competitors was
found. This pattern is likely due to activation levels of the
non-target language. Two determining factors for activation
levels are overall proficiency and exposure. Since German-
English bilinguals were more proficient in the non-target
language, German competitors may have been more readily
activated in this group. The lack of L2 co-activation in
English-German bilinguals is consistent with Weber and
Cutler (2004) who, using a similar eye-tracking paradigm,
did not find L2 activation in Dutch-English bilinguals with a
less proficient L2. Furthermore, at the time of study,
German-English bilinguals had more exposure to the non-
target language than English-German bilinguals. A series of
eye-tracking studies suggest that recent exposure to a
language may influence the extent of parallel language
activation (Marian & Spivey, 2003b; Spivey & Marian,
1999). Asymmetric co-activation was found in a study in
which bilinguals were immersed in their L2 country: while
L2 was co-activated during L1 processing, L1 was not co-
activated during L2 processing (Spivey & Marian, 1999). In
other words, only the non-target language that was subject
to high exposure was co-activated. In a follow-up
experiment, preliminary measures were taken to boost L1
exposure, and results yielded L1 co-activation during L2
processing (Marian & Spivey, 2003b). In the current study,
the higher levels of recent exposure to the non-target
language in German-English bilinguals compared to
English-German bilinguals, provide another reason for L1
co-activation in the former group.
The results in the non-cognate condition confirm previous
findings, and suggest that non-target language resting
activation, as influenced by proficiency and recent exposure,
determines the degree of parallel activation. Findings with
cognate targets suggest that where a non-target language is
not prominent enough to be activated in parallel, such co-
activation does occur with highly integrated lexical
representations across languages.
Phonological Overlap and Lexical Status
The current study suggests that the effect of phonological
overlap between targets and competitors depends on the
lexical status of the target. When the target was a cognate,
both bilingual groups looked more at competitors, relative
to controls, as phonological overlap increased, suggesting
that increased phonological overlap leads to increased
parallel language activation. On the other hand, when the

target was a non-cognate, increased parallel activation with
increased phonological overlap was not found. Cognates
have close form-level connections between languages,
created by both, phonological overlap, as well as early
learning by form-association. This may result in higher
sensitivity to phonological overlap during competitor
activation. Moreover, co-activation of the cognate’s
translation equivalents leads to phonological overlap
between the target and the competitor within the non-target
language. For non-cognate targets, where translation
equivalents are not similar at the form level, and
phonological overlap is strictly between languages, the
amount of phonological overlap does not appear to
influence the extent of between-language competitor
activation as directly. In sum, our findings for phonological
overlap manipulations suggest language co-activation
through integrated form-level representations in the case of
cognates, but less so for non-cognates.
Conclusion
In the present study of bilingual parallel language activation,
findings with cognate targets, non-cognate targets, and
varied phonological overlap suggest that proficiency levels,
as well as overlap between languages, determine the extent
of parallel language activation. We suggest that the pathway
of second language word learning relies on form similarity
to translation equivalents in the native language and shapes
within-language and between-language representations. In
turn, these mappings in the bilingual lexicon influence the
nature of parallel language activation.
Acknowledgments
This work was supported by grant NSF BC-0418495 to the
second author, and by a Northwestern Graduate Research
Grant to the first author. Thanks also go to James Booth,
Cynthia Thompson, Margarita Kaushanskaya, Judith Kroll
and members of the Northwestern Bilingualism and
Psycholinguistics Laboratory for helpful comments.
References
Baayen, H., Piepenbrock, R., & Van Rijn, H. (1995). The
CELEX lexical database (CD-ROM). Philadelphia, PA:
University of Pennsylvania Linguistic Data Consortium.
Costa, A., Caramazza, A., & Sebastian-Galles, N. (2000).
The cognate facilitation effect: Implications for models of
lexical access. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 26(5), 1283-1296.
Cunningham & Graham (2000). Increasing native English
vocabulary recognition through Spanish immersion:
Cognate transfer from foreign to first language. Journal of
Educational Psychology, 92(1), 37-49.
De Groot, A. M. B., & Nas, G. L. J. (1991). Lexical
representation of cognates and noncognates in compound
bilinguals. Journal of Memory and Language, 30, 90-123.
Dijkstra, T., Grainger, J., & Van Heuven, W. J. B. (1999).
Recognition of cognates and interlingual homographs: the
neglected role of phonology. Journal of Memory and
Language, 41, 496-518.
Dijkstra, T., Van Heuven, W. J. B., & Grainger, J. (1998).
Simulating cross-language competition with the bilingual
interactive activation model. Psychological Belgica, 38,
177-196.
Dunn, L. M. & Dunn, L. M. (1997). Peabody Picture
Vocabulary Test (PPVT). American Guidance Service,
Circle Pines, MN.
Kroll, J. F. & Stewart, E. (1994). Category interference in
translation and picture naming: Evidence for asymmetric
connections between bilingual memory representations.
Journal of Memory and Language, 33, 149-174.
Marian, V., Blumenfeld, H. K., & Kaushanskaya, M.
(2003). Developing a tool for assessing language
experience and bilingual status: LEABS-Q. Poster,
Midwestern Psychological Association, Chicago, IL.
Marian, V. & Spivey, M. (2003a). Bilingual and
monolingual processing of competing lexical items.
Applied Psycholinguistics, 24(2), 173-193.
Marian, V. & Spivey, M. (2003b). Competing activation in
bilingual language processing: Within- and between-
language competition. Bilingualism, 6(2), 97-115.
Marslen-Wilson, W. D. (1987). Functional parallelism in
spoken word recognition. Cognition, 25(1-2), 71-102.
Nakayama, M. S. (2002). The cognate status effect in lexical
processing by Chinese-Japanese bilinguals. Psychologia,
45, 184-192.
Roberts, P. & Deslauriers, L. (1999). Picture naming of
cognate and non-cognate nouns in bilingual aphasia.
Journal of Communication Disorders, 32(1), 1-23.
Silverberg, S. & Samuel, A. G. (2004). The effect of age of
second language acquisition on the representation and
processing of second language words. Journal of Memory
and Language, 51, 381-398.
Spivey, M., & Marian, V. (1999). Cross-talk between native
and second languages: Partial activation of an irrelevant
lexicon. Psychological Science, 10, 281-284.
Tanenhaus, M., Magnuson, J., Dahan, D. & Chambers, C.
(2000). Eye movements and lexical access in spoken
language comprehension: Evaluating a linking hypothesis
between fixations and linguistic processing. Journal of
Psycholinguistic Research, 29, 557-580.
Van Hell, & De Groot, A. M. B. (1998). Conceptual
representation in bilingual memory: Effects of
concreteness and cognate status in word association.
Bilingualism: Language and Cognition, 1(3), 193-211.
Van Hell, J. G., & Dijkstra, T. (2002). Foreign language
knowledge can influence native language performance in
exclusively native contexts. Psychonomic Bulletin &
Review, 9(4), 780-789.
Weber, A., & Cutler, A. (2004). Lexical competition in non-
native spoken word recognition. Journal of Memory and
Language, 50, 1-25.