Phonological processing dynamics in bilingual word naming.

Canadian Journal of Experimental Psychology / Revue canadienne de psychologie expérimentale 2014, Vol. 68, No. 3, 179 –193

© 2014 Canadian Psychological Association 1196-1961/14/$12.00 http://dx.doi.org/10.1037/cep0000026

Phonological Processing Dynamics in Bilingual Word Naming Deanna C. Friesen and Debra Jared

Corinne A. Haigh

University of Western Ontario

Bishop’s University

The current study investigated phonological processing dynamics in bilingual word naming. EnglishFrench and French-English bilinguals named interlingual heterophonic homographs (i.e., words that share orthography but not meaning or pronunciation across languages), heterophonic cognates (i.e., words that share both orthography and meaning across languages, but not pronunciations), interlingual homophones (i.e., words that share pronunciation, but not orthography or meaning across languages), and single-language matched control words in both English and French naming tasks. Cross-language phonological activation was strongest in bilinguals’ second language. The results provided evidence for feedforward activation of phonological representations in the nontarget language, as well as feedback activation of these phonological representations from semantic representations. Results are interpreted within the more recent Bilingual Interactive Activation (BIA⫹) framework. Keywords: bilingual, phonology, naming, cross-language effects

cross-language effects that are observed in the naming task reflect the impact of the nontarget language on target language processing. We start with a discussion of a theory of bilingual word recognition. Next, we review other studies that have used the naming task to investigate activation of phonology in the nontarget language, and then we present the details of our own research. To first give a brief overview of our study, the participants were EnglishFrench bilinguals (Experiment 1) and French-English bilinguals (Experiment 2). Three types of English-French cross-language words were used (heterophonic homographs, heterophonic cognates, and homophones), as well as matched single language control words. Participants named words, first in one of their languages and then in the other.

Research with monolingual skilled readers has shown that they rapidly activate phonological representations from printed letter strings even when a spoken response is not required (e.g., Perfetti, Bell, & Delaney, 1988; Rastle & Brysbaert, 2006). These phonological representations feed activation back to corresponding orthographic representations (e.g., Pexman, Lupker, & Jared, 2001; Stone, Vanhoy, & Van Orden, 1997), and feed activation forward to corresponding semantic representations (e.g., Jared, Levy, & Rayner, 1999; Newman, Jared, & Haigh, 2012; Van Orden, 1987). Furthermore, semantic representations can send activation back to phonological representations (e.g., Strain, Patterson, & Seidenberg, 1995). A question of interest for bilingual readers is whether the same processing dynamics occur for the nontarget language when reading words in a target language. In the current study, a naming task was used to investigate how activation in the nontarget language flows through the word recognition system of bilinguals. The naming task is more likely to reveal subtle cross-language phonological effects than other tasks, such as the lexical-decision task, because it requires participants to produce a spoken response. Furthermore, the naming task is necessarily a language-specific task because it requires the participant to produce a word in a particular language, unlike the lexicaldecision task where responses can often be made on the basis of activation of words in either language (Smits, Martensen, Dijkstra, & Sandra, 2006). Therefore, we can be more confident that any

The Bilingual Interactive Activation (BIA) Model We adopt the theoretical framework of the BIA model (Dijkstra & van Heuven, 1998, 2002), because this model is currently the most clearly articulated model of bilingual word recognition (Figure 1). Dijkstra and van Heuven proposed that bilinguals do not have separate lexicons for each of their languages; instead, their lexicons are integrated across languages. Upon presentation of a word, sublexical and lexical orthographic representations become active, and they in turn activate associated phonological and semantic representations and the corresponding language node. Semantic representations can send activation back to phonological and orthographic representations. Activation is not specific to the language of the presented word, but depends on overlap of representational codes. For example, all words in each language that share letters in common with the input receive some activation. However, Dijkstra and van Heuven assumed that activation levels depend in part on subjective frequency, and therefore phonological and semantic codes associated with the bilinguals’ second language may be delayed in activation relative to first language codes (see also Gollan, Montoya, Cera, & Sandoval’s, 2008,

Deanna C. Friesen and Debra Jared, Department of Psychology, University of Western Ontario; Corinne A. Haigh, School of Education, Bishop’s University. This research was funded by a Natural Sciences and Engineering Research Council of Canada Discovery grant awarded to D.J. We thank Steve Hernandez and Amal Quatab for assistance in testing the participants. Correspondence concerning this article should be addressed to Debra Jared, Department of Psychology, University of Western Ontario, London, Ontario, Canada N6A 5C2. E-mail: [email protected] 179

180

FRIESEN, JARED, AND HAIGH

ever, it is still not clear whether activated nontarget phonological representations feed activation forward to semantic representations and whether semantic representations then feed activation back to phonological representations in each language and influence naming.

Cross-Language Masked Phonological Priming

Figure 1. The Bilingual Interactive Activation (BIA⫹) model of Dijkstra and van Heuven (2002).

“weaker links” hypothesis for a similar view regarding the impact of the frequency of use of each language in speech production). Because words in both languages are activated, a mechanism is needed to ensure that responses are in the correct language for the task. Initially, the BIA allowed language nodes to inhibit representations in the nontarget language (Dijkstra & van Heuven, 1998). However, in the more recent BIA⫹ (Dijkstra & van Heuven, 2002), a task schema receives the output of the identification system and determines the response that is to be made depending on the demands of the task. Presumably, a naming response is made primarily on the basis of activation of lexical phonology. Smits et al. (2006) elaborated on how the model selects a naming response from activated alternatives. They proposed that the word identification system checks the language membership of each available response beginning with the lexical representation that is most active. If this candidate meets the language requirements of the task, it may be released. Otherwise, a response is delayed until a language-appropriate response is found or until a time criterion is reached.

Evidence for Cross-Language Activation of Phonology In the review that follows, we show that naming studies with bilinguals have provided support for some of the assumptions of the BIA⫹, including the assumption of a shared sublexical phonological store for both languages, and the assumption that when two languages share an alphabet, printed words in one language activate phonological representations in the other language. How-

One paradigm that has been used to investigate cross-language activation of phonology in bilinguals is the masked priming paradigm, in which words in one language are briefly presented and then masked, and are followed by a target word in their other language. Although most studies have used a lexical-decision task with this paradigm (e.g., Ando, Matsuki, Sheridan, & Jared, in press; Dimitropoulou, Duñabeita, & Carreiras, 2011; Gollan, Forster, & Frost, 1997; Nakayama, Sears, Hino, & Lupker, 2012), masked primes in one language have also been shown to facilitate the naming of phonologically similar target words in the other language, both when target words were in bilinguals’ second language (L2; Kim & Davis, 2003; Zhou, Chen, Yang, & Dunlap, 2010), and in their first language (L1; Jouravlev, Lupker, & Jared, 2014; Zhou et al., 2010). Jouravlev et al. showed that facilitation is obtained even when English primes and Russian targets share only onset phonology (e.g., viper – BOБЛA /voblə/). This result, along with findings in the lexical decision studies that the size of the cross-language phonological priming effect is not influenced by word frequency or L2 proficiency, provide evidence of shared sublexical phonological representations across languages. The phonological representation of the target is activated more quickly in the related condition than in the unrelated condition because some of its sublexical phonology has already been activated by the prime. These studies provide evidence that phonological representations for a bilingual’s two languages are in an integrated store, or if they are in separate stores, the stores are highly interconnected. However, Andrews (1997) has pointed out that the consequences of coactivation in the priming paradigm are not necessarily relevant to whatever coactivation occurs in response to a single target word. That is, the activation of the nontarget language may be exaggerated when it appears in a prime. Therefore, the remainder of the review focuses on single-word naming studies. The languages of the bilinguals in these studies use the same alphabet.

Cross-Language Spelling-Sound Consistency Effects Two languages that share an alphabet often have some letters that are pronounced differently. For example, -AIT is pronounced /eit/ in English words such as BAIT, but /ε/ in French words such as FAIT. If bilinguals activate sublexical phonological representations for both of their languages from printed words, then the naming of words with such letters should be hindered by competition from phonological representations activated by knowledge of letter-sound correspondences in the other language. Jared and Kroll (2001) found that French-English bilinguals made more errors naming low-frequency English words such as BAIT than words such as BUMP (-UMP does not exist in French monosyllabic words). This finding suggests that their knowledge of French spelling-sound correspondences was used to activate French phonological representations when reading in English. No

INTERLINGUAL EFFECTS IN WORD NAMING

difference between the two word types was found for EnglishFrench bilinguals who were doing the naming task in their L1. After naming a filler list of French words, participants named a second block of English experimental words. In the second English block, both groups of bilinguals took longer to name BAIT-type words than BUMP-type words. The stronger effect in the second block of English trials compared with the first suggests that nontarget language phonological representations are typically fairly weakly activated from the spelling of a target language word, particularly when they are in the bilingual’s second language, unless the nontarget language has recently been used. In a similar study with Dutch-English bilinguals, Smits, Sandra, Martensen, and Dijkstra (2009) did not find a cross-language consistency effect in an English naming task unless Dutch words were included on the list. Their study included both high- and low-frequency English words, however. It is possible that crosslanguage consistency affects only the naming of low-frequency words.

Interlingual Heterophonic Homographs Interlingual heterophonic homographs are words that have the same spelling, but not the same meaning or pronunciation across languages. For example, the word PAIN occurs in both English and French, but in French it means “bread” and is pronounced /pε˜/. Other examples from English and French are CHAT (“cat,” /ʃa/), SALE (“dirty,” /sal/), COIN (“corner,” /kwε˜/), and DENT (“tooth,” /d␣˜ /). If bilinguals activate phonological representations for both of their languages from printed words, then the naming of such words in one language should be hindered by competition from the phonological representation activated in the other language. Jared and Szucs (2002) conducted an English naming study, comparing performance on interlingual homographs and matched English control words. Homographs were chosen such that the English member of the pair was a low-frequency word and the French member of the pair was a high-frequency word to maximize the likelihood of detecting the influence of French phonology. French-English bilinguals had significantly longer naming latencies and made more errors on homographs than on English control words, and even English-French bilinguals produced a small but significant effect in the error data. After naming a block of French filler words, participants named a second block of English experimental words. Both groups had significantly longer naming latencies and produced more errors on interlingual homophones than matched English control words in the second English block. Similarly, Smits et al. (2006) found that Dutch-English bilinguals produced more errors on interlingual homographs than on control words in an English naming task. Furthermore, in a study with children in Grade 3, English native speakers who were in French Immersion programs at school produced more errors in naming interlingual homographs than control words, both in a French naming task and in an English naming task (Jared, Cormier, Levy, & Wade-Woolley, 2012). These studies provide stronger evidence that activation of phonological representations is not selective for language than did the Jared and Kroll (2001) study. The difference in findings may have been because the French phonological representations were more highly activated when the target stimulus corresponded to an entire

181

word in French (e.g., PAIN) than when it shared only some letters with French words (e.g., BAIT, which is not a French word but shares letters with FAIT, etc.). This explanation attributes the locus of the effects in both studies to feedforward orthographicphonological processing. Jared and Szucs considered another possible explanation for the difference between the studies, which was that the different semantic representations associated with the interlingual homographs may have competed and slowed naming responses. They thought this explanation was less likely because Hino and Lupker (1996) had shown that monolinguals named words with more than one meaning (e.g., LEAN) faster than words with a single meaning (e.g., LUNG), which suggests that, at least within a language, multiply activated meanings do not necessarily slow naming. However, this reasoning may not be quite accurate because the meanings of an ambiguous word such as LEAN are associated with a single pronunciation. In contrast, the interlingual homographs in Jared and Szuc’s (2002) study had more than one pronunciation as well as more than one meaning. A more appropriate comparison to monolingual findings is a study that used heterophonic homographs such as TEAR and WIND. In a monolingual English experiment, Seidenberg, Waters, Barnes, and Tanenhaus (1984) demonstrated that such heterophonic homographs are named more slowly than other words with inconsistent spelling-sound correspondences, indicating that multiple meanings may slow naming when they are associated with multiple pronunciations. Therefore, the interlingual homograph effect may arise as a consequence of both feedforward activation of phonology by orthography and feedback activation of phonology by semantics.

Cognates One way to obtain evidence for a contribution of feedback from semantics to phonology is to compare interlingual homograph effects with effects for other cross-language words that also share orthography and not phonology, but which have the same meaning in the two languages. Such words are called heterophonic cognates. An example from English and French is the word TRAIN, which has the same meaning in both languages, and is pronounced /trein/ in English and /trε˜/ French. Although there are a large number of lexical decision studies showing facilitation for cognates (e.g., Dijkstra, Grainger, & van Heuven, 1999; Dijkstra, Miwa, Brummelhuis, Sappelli, & Baayen, 2010; Font, 2001; Lemhöfer and Dijkstra, 2004; van Hell & Dijkstra, 2002), there are only two studies that we are aware of that used a naming task with adult bilinguals. In one study, De Groot, Borgwaldt, Bos, and van den Eijnden (2002) found no correlation of cognate status and the word naming latencies of Dutch-English bilinguals, either for the participants who read words in L1 or the participants who read words in L2. In the other study, Schwartz, Kroll, and Diaz (2007) had English-Spanish bilinguals name cognates in either their L1 or L2. Cognates differed orthogonally on both phonological and orthographic similarity across languages; thus producing four types of cognates. When these cognate types were collapsed across categories and compared with single language control words, Schwartz et al. found that bilinguals performing the task in their L1 named cognates more slowly than English control words. In contrast, no differences between cognates and matched Spanish control words

182


Facilitatory interlingual homophone effects have been observed in English lexical decision tasks with French-English bilinguals (Haigh & Jared, 2007) and Dutch-English bilinguals (Lemhöfer & Dijkstra, 2004). Similarly, Carrasco-Ortiz, Midgley, and Frenck-Mestre (2012) found that French-English bilinguals produced a reduced N400 for interlingual homophones compared with control words in an English semantic judgment task. However, there are also conflicting findings. Dijkstra et al. (1999) observed that Dutch-English bilinguals had significantly slower English decision latencies and made more errors on the same interlingual homophones and controls used in Lemhöfer and Dijkstra (2004). Furthermore, in naming studies with English monolinguals, lowfrequency homophones with high-frequency mates have been observed to have slower naming latencies than matched control words (Biedermann, Coltheart, Nickels, & Saunders, 2009; Edwards, Pexman, & Hudson, 2004). Thus, more evidence is needed to determine whether interlingual homophone effects in bilingual naming are inhibitory or facilitatory, which in turn will help us understand the extent of activation of nontarget language representations.

were observed when bilinguals read aloud in their L2. However, because more than half of the cognates differed in spelling across languages, it is not entirely clear that the slower naming latencies in the English version of the task were because of competing phonological representations and not a consequence of the different orthographic representations. Furthermore, it is surprising that the effect was larger in the participants’ L1 than their L2, because usually the reverse is found. Schwartz et al. did make a further comparison of performance on orthographically similar cognates that shared a pronunciation and those that did not. They found that cognates with different pronunciations were named more slowly and less accurately than cognates with similar pronunciations, both in English and in Spanish. It is unclear from this comparison whether cognates with different pronunciations are named more slowly than single language control words (i.e., different pronunciations in each language compete), or whether cognates with similar pronunciations are named more quickly than control words (i.e., similar pronunciations in each language facilitate one another), or both. Nonetheless, these results provide further support for the view that nontarget language phonological representations are activated from the orthography of the target language, and provide rare evidence that this can happen even when bilinguals are reading in their L1. Jared et al.’s (2012) study of Grade 3 English native speakers who were in French Immersion programs produced yet a different pattern of results. They found that fewer errors were produced in naming cognates than control words, but only in the L2 naming task. There was no cognate effect in the L1 naming task. Both of these results differ from the children’s performance on interlingual homographs, which incurred more errors than control words, providing some evidence of an influence of feedback from semantic representations to phonology. In the current study, we investigated whether adult bilinguals also show a different pattern of performance for cognates and interlingual homographs. A comparison of the interlingual homograph and cognate effects provides evidence concerning the extent to which nontarget representations are activated. If activation in the nontarget language flows just from orthography to phonology, then interlingual homograph effects and cognate effects should both be inhibitory and of a similar magnitude because both had different pronunciations in English and French.

In summary, the present study investigated interlingual homograph, cognate, and interlingual homophone effects in L1 and L2 word naming with the goal of understanding how activation in the nontarget language flows through the word recognition system of bilinguals. Stimulus sets were developed in both English and French. In Experiment 1, participants were English-French bilinguals and in Experiment 2 participants were French-English bilinguals. The two groups of bilinguals were tested to investigate the generalizability of findings. Participants named words in one language and then in the other. Half of the participants started with each language. An advantage of such a design is that L1 and L2 are within-participant variables in each experiment. A disadvantage, however, is that there may be carry-over effects from naming words in one language to naming words in the other language. Analyses were therefore conducted for each language (regardless of whether it came first or second), and also for each language when it appeared in the first block of trials only, at which point the experiments involved a single language.

Interlingual Homophones

Carry-Over Effects

Another way of investigating cross-language phonological effects is to compare naming latencies for interlingual homophones and matched control words. Interlingual homophones are words that have the same or very similar pronunciation across languages, but a different spelling and meaning in each language. An example of a homophone pair from English and French are the words FOE and FAUX (“false”), both pronounced /fo/. At present, we are unaware of an interlingual homophone naming study with adult bilinguals. Jared et al. (2012) found that Grade 3 English native speakers who were in French Immersion programs named interlingual homophones faster and with fewer errors than control words, but only in the L2 naming task. There was no homophone effect in the L1 naming task. There is some evidence from other tasks of a facilitatory effect for interlingual homophones in adult bilinguals reading in their L2.

We also looked at carry-over effects from the first block to the second block of trials to see whether they could inform the debate about how the appropriate response is selected from activated candidates. Researchers in bilingual speech production have suggested two broad types of views on language selection (see Misra, Guo, Bobb, & Kroll, 2012). One view is to assume that bilinguals can inhibit representations in the language that is not needed in the task, so that representations in the task relevant language are likely to become most highly activated and selected. The original BIA allowed such top-down inhibition from language nodes. Another view is to assume that bilinguals are able to ignore activated candidates from the nontarget language and to consider only target language candidates. In the revised version of the BIA model (BIA⫹), top-down language-to-word inhibition was eliminated and a task schema system was added which makes use of the

The Present Study


output of the word recognition system. When the task requires responses from a single language, the task schema at the decision stage excludes responses from the other language. Therefore, evidence that participants inhibited one language when reading in another would be inconsistent with the BIA⫹. Misra et al. (2012) obtained evidence for inhibition of L1 when participants had previously named a block of pictures in L2 but not the reverse. If bilinguals inhibit their nontarget language when naming words in the target language in the first block of trials, then when there is a language switch and the nontarget language becomes the target language in the second block, naming times for all word types should be slowed because of the need to overcome the inhibition. Carry-over effects could also arise, however, if reading aloud in a target language in Block 1 increases the activation of that language and the heightened activation persists into the subsequent block of trials in the other language. If so, then interlingual words should be more affected than single language control words. In Jared and Kroll (2001) and Jared and Szucs (2002), English word naming latencies for English-French bilinguals were slower after having read a block of French filler trials, however crosslanguage inconsistent (e.g., BAIT) and interlingual homographs were more slowed than single language control words. These findings suggest that naming in L2 increases the activation level of L2 representations for some time, but provide little support for prolonged inhibition of L1 representations. In the current study, we examined whether naming response times differed depending on whether or not participants had just completed a block of naming trials in their other language to provide further evidence regarding whether bilinguals inhibit representations in a nontarget language.

Experiment 1 Method Participants. Thirty English-French young adult bilinguals (22 women, 8 men) participated in the study. Their mean age was 20.2 years (range ⫽ 18 –25). They all attended university where English was predominantly spoken (in Southwestern Ontario). Participants received course credit or were paid for their involvement in the study. Responses on a language experience questionnaire indicated that all had learned English first and French second. Table 1 presents other data regarding their ability in French. Self-report data include participants’ ratings of their current ability to understand, speak, read and write in French on a 10-point scale, where 0 indicates no ability and 10 indicates very fluent. Also included is the mean age at which participants reported beginning to acquire French. As an objective measure of French fluency, participants completed an online picture-naming task in which they named 48 line drawings taken from the Snodgrass and Vanderwart (1980) picture set. This task is the same as used in previous bilingual experiments in our lab (Friesen & Jared, 2012; Haigh & Jared, 2007; Jared & Kroll, 2001; Jared & Szucs, 2002; see Haigh & Jared, 2007, or Jared & Kroll, 2001, for a list of the picture names). Participants were included in the study only if they correctly named at least 24 of the pictures in French (this is the same criterion as used in our previous studies). Stimuli. Critical items were English-French interlingual words (30 heterophonic homographs, 30 heterophonic cognates

183

Table 1 Second Language Ability of English-French and French-English Bilinguals Experiment 1 Experiment 2 English-French French-English Measure Second language picture naming Percentage correct Reaction time (ms) Self-rated ability in L2 (/10) Understanding Speaking Reading Writing Age began second language (in years)

M

SD

M

SD

67.4 1690

12.0 428

81.9 1127

12.3 262

8.1 7.5 8.0 7.4 4.9

1.8 1.5 1.6 1.7 2.3

8.9 8.7 9.2 8.6 8.0

1.5 1.2 1.1 1.6 3.8

and 30 homophones), 45 English control words, and 45 French control words (see Appendix). Half of the cognates, homographs, and homophones had a lower frequency in English than in French. These items and their control words were presented in the English list. The other half of the interlingual stimuli had a lower frequency in French than in English and was presented in the French list. Each item only appeared in one list. Each interlingual word was matched with a single language control word on written frequency, initial phoneme, word length, and number of syllables. English written word frequency was obtained from CELEX norms (Baayen, Piepenbrock, & van Rijn, 1993) and French word frequency was obtained from Lexique norms (New, Pallier, Ferrand, & Matos, 2001). Word characteristics are reported in Table 2. Frequencies are reported per million words. For the English set of stimuli, interlingual words and their control words were also matched on monolingual naming times produced from the English Lexicon Project (Balota et al., 2007). T tests indicated that interlingual homograph, cognate, and interlingual homophone effects were not significant for these English monolinguals (all ps ⬎ .40). Ferrand et al. (2011) published a smaller set of naming times for French words. Over 80% of the words in our interlingual homograph and control groups and in our interlingual homophone and control groups were included in their study. T tests on their latencies for our words indicated that neither of these effects were significant (all ps ⬎ .25). Another 180 words were selected in both English and French to serve as filler words. The large number of fillers was used to reduce the proportion of interlingual words on the list. Fifteen English-only words and 15 French-only words were selected as practice items. Procedure. Participants began by completing a questionnaire about their language experience. They were then seated approximately 60 cm in front of a computer screen. Participants were asked to read words from a single language as quickly and accurately as possible into a microphone. Stimuli were presented one at a time at the center of the screen in 18-point lower case bold Courier New font using E-prime software (Schneider, Eschman, & Zuccollotto, 2002). Each stimulus remained on the screen until the participant began to speak into the microphone. Reaction times (RTs) were recorded from when the word first appeared on the screen to when the participant’s response triggered the voice key. The intertrial interval was 2 s. An experimenter recorded the accuracy of their responses. Errors included dysfluencies (e.g.,


184

Table 2 Word Characteristics for the English and French Naming Tasks as a Function of Word Type and Word Frequency English target words Interlingual

French target words Control

Interlingual

Frequency

Control

Frequency

Word type

English

French

Length

Frequency in English

Length

English

French

Length

Frequency in French

Length

Cognates Homographs Homophones

24.3 18.8 13.5

124.5 150.8 319.1

5.7 4.7 3.4

20.4 17.2 11.6

5.7 4.7 3.4

33.5 43.5 42.8

149.3 197.1 163.7

5.5 4.4 4.5

32.0 40.5 44.7

5.6 4.5 4.8

stopping before completing a pronunciation), naming a cognate or homograph in the wrong language, and misreading the word. Words pronounced with a slight accent were not considered errors. Words within a block were randomized for each participant. Participants read a block of words in one language followed by a block in the other language. Language order was counterbalanced across participants. Participants then completed the online second language picturenaming task. Pictures were presented one at a time on the computer screen, and participants were asked to name the objects aloud in French as quickly and accurately as they could. Pictures remained on the screen until the participant’s response. Naming errors were coded by hand. The session took approximately 45 min.

Results Data from 2.4% of the critical word trials were discarded because the microphone failed to detect a response. A further 2.1% of RTs were considered to be outliers because they were 2.5 SDs above or below the participant’s mean RT in the word’s condition. These data points were removed from all analyses. Mean latencies and error rates for all words are shown in Table 3. Analyses of variance were conducted on response latencies and square-root transformed errors using both participant (F1) and item (F2) means as units of analysis. The data from each language were analyzed separately. The primary analyses addressed whether interlingual words differed from single language control words, and the sec-

Table 3 Mean Naming Latencies and Error Rates on Interlingual and Control Words for English-French Bilinguals in Experiment 1 Naming latency (in ms) Critical

Control

Error rates (%) Effect

Critical

Control

Effect

L1 naming (English) Overall Homographs Cognates Homophones Block 1 Homographs Cognates Homophones Block 2 Homographs Cognates Homophones

564 561 543

540 554 534

24 7 9

7.6 2.0 1.6

0.0 4.0 0.4

7.6 ⫺2.0 1.2

531 546 532

524 537 523

7 9 9

6.2 3.1 0.9

0.0 4.0 0.4

6.2 ⫺0.9 0.5

597 575 554

556 570 546

41 5 8

8.9 0.9 2.2

0.0 4.0 0.4

8.9 ⫺3.1 1.8

L2 naming (French) Overall Homographs Cognates Homophones Block 1 Homographs Cognates Homophones Block 2 Homographs Cognates Homophones

600 572 565

564 598 580

36 ⫺26 ⫺15

6.0 2.0 1.1

1.3 1.3 1.3

4.7 0.7 ⫺0.2

610 580 551

558 604 580

52 ⫺24 ⫺29

7.6 2.2 0.9

1.8 0.9 1.3

5.8 1.3 ⫺0.4

590 564 579

570 592 580

20 ⫺28 ⫺1

4.4 1.8 1.3

0.9 1.8 1.3

3.5 0.0 0.0


ondary analyses examined whether bilinguals inhibit a nontarget language when naming words in their other language. Interlingual versus control words. The interlingual homographs and cognates and their controls were analyzed in one set of analyses, and the homophones in another set because, as indicated in the beginning of this article, the comparison of interlingual homographs and cognates addresses a different issue with respect to crosslanguage phonological effects than interlingual homophones. Interlingual homographs and cognates. The analyses of variance (ANOVAs) had the variables Word Type (Interlingual vs. Control) and Interlingual Word Group (Homographs vs. Cognates). They were treated as within-participants and between-items variables. Of particular interest is whether there was an interaction between word type and interlingual word group. The analyses also included Block (Block 1 vs. Block 2) as a between-participants and within-item variable, but this variable was not the focus of these analyses. In addition, analyses were conducted on the data from Block 1 to determine whether cross-language effects can be seen when the other language has not just been used. When participants named words in L1 (English), the main effect of word type was significant by participants in the latency data, F1(1, 28) ⫽ 10.24, p ⬍ .01, ŋp2 ⫽ .27; F2(1, 56) ⫽ 2.49, nonsignificant (ns), and in the error data, F1(1, 28) ⫽ 10.76, p ⬍ .01, ŋp2 ⫽ .28; F2(1, 56) ⫽ 2.90, p ⫽ .09. Critically, there was an interaction of word type and interlingual word group in the latency data by participants, F1(1, 28) ⫽ 4.52, p ⬍ .05, ŋp2 ⫽ .14; F2 (1, 56) ⫽ 1.06, ns, and in both analyses in the error data, F1(1, 28) ⫽ 38.23, p ⬍ .001, ŋp2 ⫽ .58; F2(1, 56) ⫽ 7.04, p ⬍ .01, ŋp2 ⫽ .11. English homographs were named 24 ms more slowly than matched control words, F1(1, 28) ⫽ 14.33, p ⬍ .01, ŋp2 ⫽ .34; F2(1, 56) ⫽ 3.39, p ⫽ .07, ŋp2 ⫽ .06, and 7.6% more errors were made on homographs than on control words, F1(1, 28) ⫽ 46.98, p ⬍ .001, ŋp2 ⫽ .63; F2(1, 56) ⫽ 9.48, p ⬍ .01, ŋp2 ⫽ .15. In contrast, the cognate effect (9 ms slower on cognates) was not significant in the latency data, F1(1, 28) ⫽ 1.42, ns; F2 ⬍ 1, and in the error data it only approached significance by participants (2.0% fewer errors on cognates), F1(1, 28) ⫽ 3.97, p ⫽ .07, ŋp2 ⫽ .11; F2 ⬍ 1. When just the participants who named English words in Block 1 were included in the analyses, the main effect of word type was not significant in the latency data, F1(1, 14) ⫽ 1.55, ns; F2 ⬍ 1; but it was significant in the analysis by participants in the error data, F1(1, 14) ⫽ 7.30, p ⬍ .05, ŋp2 ⫽ .34; F2(1, 56) ⫽ 2.75, ns. Similarly, the interaction of word type and interlingual word group was not significant in the latency data, Fs ⬍ 1, but it was significant in the participants analysis of the error data, F1(1, 14) ⫽ 15.06, p ⬍ .01, ŋp2 ⫽ .52; F2(1, 56) ⫽ 2.26, ns. The homograph effect was significant in the error data only, F1(1, 14) ⫽ 25.53, p ⬍ .001, ŋp2 ⫽ .64; F2(1, 56) ⫽ 5.00, p ⬍ .05, ŋp2 ⫽ .08, and there was no cognate effect. When participants named words in L2 (French), the main effect of word type was not significant in the latency analyses, Fs ⬍ 1, but it was significant in the error analyses, F1(1, 28) ⫽ 16.75, p ⬍ .001, ŋp2 ⫽ .37; F2(1, 56) ⫽ 9.50, p ⬍ .01, ŋp2 ⫽ .15. Critically, there was a significant interaction of word type and interlingual word group both in the latency data, F1(1, 28) ⫽ 18.47, p ⬍ .001, ŋp2 ⫽ .40; F2(1, 56) ⫽ 8.87, p ⬍ .01, ŋp2 ⫽ .14, and in the error data, F1(1, 28) ⫽ 4.55, p ⬍ .05, ŋp2 ⫽ .14; F2(1, 56) ⫽ 5.52, p ⬍ .05, ŋp2 ⫽ .09. French homographs were responded to 36 ms more slowly than their control words, F1(1, 28) ⫽ 12.70, p ⬍ .01, ŋp2 ⫽

185

.31; F2(1, 56) ⫽ 4.85, p ⬍ .05, ŋp2 ⫽ .08, and produced 4.7% more errors, F1 (1, 28) ⫽ 13.90, p ⬍ .01, ŋp2 ⫽ .38; F2 (1, 56) ⫽ 14.75, p ⬍ .001, ŋp2 ⫽ .21. In contrast, French cognates were responded to 26 ms faster than their control words, F1(1, 28) ⫽ 5.16, p ⬍ .05, ŋp2 ⫽ .16, ŋp2 ⫽ .16; F2(1, 56) ⫽ 4.04, p ⬍ .05, ŋp2 ⫽ .07, although there was no cognate effect (0.7% difference) in the error data, Fs ⬍ 1. For participants who named French words in Block 1, the main effect of word type was not significant in the latency data, F1(1, 14) ⫽ 1.15, ns; F2 ⬍ 1; but it was significant in the error data, F1(1, 14) ⫽ 10.42, p ⬍ .01, ŋp2 ⫽ .43; F2(1, 56) ⫽ 10.14, p ⬍ .01, ŋp2 ⫽ .15. The interaction of word type and interlingual word group was significant in the latency data, F1(1, 28) ⫽ 8.77, p ⬍ .01, ŋp2 ⫽ .39; F2(1, 56) ⫽ 8.65, p ⬍ .01, ŋp2 ⫽ .13, but not in the error data, F1(1, 28) ⫽ 2.14, ns; F2(1, 56) ⫽ 2.90, p ⬍ .10, ŋp2 ⫽ .05. There was a significant inhibitory interlingual homograph effect in both the latency data, F1(1, 14) ⫽ 8.32, p ⬍ .05, ŋp2 ⫽ .37; F2(1, 56) ⫽ 6.28, p ⬍ .05, ŋp2 ⫽ .10, and in the error data, F1(1, 28) ⫽ 7.75, p ⬍ .05, ŋp2 ⫽ .36; F2(1, 56) ⫽ 11.95, p ⬍ .001, ŋp2 ⫽ .18. Although cognates were named 24 ms faster than controls, this difference was not significant, F1(1, 14) ⫽ 1.54, ns; F2(1, 56) ⫽ 2.74, ns, nor was there a cognate effect in the error data, F1(1, 28) ⫽ 1.91, ns; F2(1, 56) ⫽ 1.10, ns. Interlingual homophones. The critical variable in the ANOVAs was homophony (homophone vs. control), which was treated as a within-participant and between-item variable. The ANOVAs also included Block (Block 1 vs. Block 2), which was treated as a between-participants and within-item variable, but again it was not the focus of these analyses. The data from Block 1 were also analyzed using t tests. When participants named words in L1 (English), there was no effect of homophony when all participants were included, in either the latency data, F1(1, 28) ⫽ 2.49, ns; F2 ⬍ 1, or in the error data, F1(1, 28) ⫽ 2.97, ns; F2 ⬍ 1, nor was there a significant effect of homophony just for participants who named words in English in Block 1, in the latency t1(14) ⫽ 1.48, ns; t2(28) ⫽ 0.72, ns, or error data t1(14) ⫽ 0.56, ns, t2(28) ⫽ 0.23, ns. When participants named words in L2 (French), homophones were responded to more quickly than control words in the analysis by participants, F1(1, 28) ⫽ 4.78, p ⬍ .04, ŋp2 ⫽ .15; F2(1, 28) ⫽ 2.11, ns, but there was no homophone effect in the error data, Fs ⬍ 1. For participants who named words in French in Block 1, a facilitatory homophone effect was observed in the latency data, t1(14) ⫽ 2.34, p ⬍ .05; t2(28) ⫽ 1.89, p ⬍ .07, and there was no homophone effect in the error data. Carry-over. The final set of analyses specifically examined whether reading times and accuracy for words differed when they occurred in Block 2 than when they appeared in Block 1. If the nontarget language is inhibited when naming in the target language in Block 1, then when that language becomes the target language in Block 2, naming should be slower. ANOVAs had variables Block (Block 1, Block 2), Word Type (Interlingual vs. Control) and Interlingual Word Group (Homographs, Cognates, Homophones). English words were named 34 ms more slowly on average in the second block compared with the first block. This difference was not significant by participants, F1(1, 28) ⫽ 1.24, ns, but was highly significant by items, F2(1, 84) ⫽ 41.85, p ⬍ .001, ŋp2 ⫽ .33. There was no effect of Block in the error data Fs ⬍ 1. Block did not interact significantly with word type either in the latency data, F1(1, 28) ⫽ 2.25, ns; F2(1, 84) ⫽ 1.34, ns, or in the error data,

186


Fs ⬍ 1. Interlingual words were named 39 ms more slowly in Block 2 than in Block 1, and control words were named 29 ms more slowly in the second block than the first. The three-way interaction of Block by word type by interlingual word group approached significance in the latency data, F1(2, 56) ⫽ 2.70, p ⬍ .08, ŋp2 ⫽ .09; F2(2, 84) ⫽ 2.45, p ⬍ .10, ŋp2 ⫽ .06, and by items in the error data, F1(2, 56) ⫽ 1.21, ns; F2(2, 84) ⫽ 2.60, p ⫽ .08, ŋp2 ⫽ .06. Interlingual homograph effects in particular were larger in Block 2 than in Block 1. Homographs were named 66 ms more slowly in Block 2 than Block 1 whereas their control words were named 32 ms more slowly in Block 2. For French words, there were no effects of Block in either the latency or error data, nor any interactions of Block and word type, all Fs ⬍ 1. French words on average were named only 2 ms more slowly in the second block than in the first, and interlingual words in particular were named 3 ms more slowly in the second block than in the first (mean naming latencies for control words were exactly the same in the two blocks). The three-way interaction of Block by word type by interlingual word group approached significance in the latency data, F1(2, 56) ⫽ 2.79, p ⫽ .07, ŋp2 ⫽ .09; F2(2, 84) ⫽ 2.88, p ⫽ .06, ŋp2 ⫽ .06, and was not significant in the error data, Fs ⬍ 1. Interlingual homograph and homophone effects were actually smaller in Block 2 than in Block 1.

Discussion English-French bilingual participants had more difficulty naming interlingual homographs than matched single language control words. When naming in L1 (English), there was a homograph effect in both the latency and error data when all participants were included in the analyses, although there was a homograph effect only in the error data when just participants who named in English in the first block of trials were included. When reading in L2 (French), there was a homograph effect in the latency and error data, both with all participants and when just participants who named in French in the first block of trials were included. These interlingual homograph effects could have occurred because the pronunciation in each language was activated (even though each block required naming in just a single language), and these activated pronunciations competed. However, it is also possible that the meanings associated with the homographs in each language were activated and these in turn activated their corresponding pronunciations in each language. Performance on the cognates, which also had different pronunciations in each language, but a common meaning, is informative regarding whether there was additional feedback activation from semantic representations. Despite also having a different pronunciation in each language, cognates were not more difficult to name than matched control words. There was no difference between naming times or errors for cognates and matched control words when naming in L1 (English), either when all participants were included or when just those who named in English in Block 1 were included. Furthermore, in L2 (French), cognates were named more quickly than matched controls, although this cognate effect was not significant when just those who named in French in Block 1 were included. It is important that interactions between word type and interlingual word group were observed in both L1 and L2. These were most evident when all participants were included in the analyses, but were also observed in some analyses that included only Block

1, particularly in the L2 latency analyses. Since interlingual homographs and cognates differed on whether they have two meanings or a single meaning across languages, this finding provides evidence that semantic representations associated with the nontarget language reading are activated and in turn activate their corresponding phonological representations. A facilitatory interlingual homophone effect was observed when naming in L2, providing evidence that common phonology with L1 is helpful when reading in L2. No interlingual homophone effect was observed in L1 naming, probably because L1 phonological representations are activated strongly enough from print in L1 that there is little chance of an influence of L2. There was some evidence that participants inhibited L1 representations when reading in L2 in Block 1. When L1 words had to be named in the subsequent block, naming latencies were longer than they were for participants who named in L1 in Block 1. However, the difference was not significant by participants (although was highly significant by items), perhaps either because only some participants produced long-lasting inhibition of L1, or because participants recovered from the inhibition part way through the list of English words. In contrast, no evidence was found indicating that representations in L2 were inhibited when reading in L1 in Block 1. There was no difference in L2 latencies for participants who named L2 words in Block 2 compared with those who named L2 words in Block 1. Of course these data do not rule out the possibility that L2 representations were inhibited when reading in L1 in Block 1 but that participants recovered quickly as they began naming words in L2. There were marginal triple interactions between Block, Word Type, and Interlingual Word Group. The English naming data provided some evidence that naming in French in Block 1 produced long-lasting activation of French that carried over into Block 2. L1 Block 1 versus Block 2 differences were 34 ms larger for interlingual homographs than their matched control words. A more thorough interpretation of these findings and the implication of these results for the BIA⫹ can be found in the General Discussion.

Experiment 2 To investigate the generalizability of our findings from Experiment 1, we tested a second group of bilinguals on the same words, this time with French as their L1 and English as their L2. Not only did these bilinguals differ in which of the two languages was their L1, they also differed in the use of their L2. Both groups of bilinguals were living in a primarily English environment at the time of the study, and therefore the French-English bilinguals in Experiment 2 were using their L2 in their daily lives to a greater extent than the English-French bilinguals in Experiment 1. As the objective picture naming data in Table 2 show, they were more fluent in their L2 than the participants of Experiment 1. Furthermore, English and French differ in the transparency of their spelling-sound correspondences. In Experiment 1, the participants’ L1 was the less transparent language and their L2 was the more transparent language. In Experiment 2, this was reversed.

Method Participants. Thirty French-English young adult bilinguals (21 women, 9 men) participated in the study. They all attended


university where English was predominantly spoken (in Southwestern Ontario). Their mean age was 21.8 years (range ⫽ 18 –29). Participants received course credit or were paid for their involvement in the study. Responses on the language experience questionnaire indicated that all had learned French first and English second. See Table 1 for self-ratings of their abilities in English, the mean age they began to acquire English, and results from the picture-naming task from Experiment 1, which they performed in English. Materials and procedure. The materials and procedure were the same as in Experiment 1.

Results Data from 3.0% of the critical word trials were discarded because the microphone failed to detect a response. A further 1.8% of RTs were outliers and also removed from the analyses. The same set of analyses as those in Experiment 1 were conducted. Mean latencies and error rates for all words are shown in Table 4. Interlingual versus control words. As in Experiment 1, interlingual homographs and cognates were included in one set of analyses, and interlingual homophones in another set. Interlingual homographs and cognates. When participants named words in L1 (French), there was no main effect of word type either in the latency analyses, F1(1, 28) ⫽ 3.23, p ⫽ .08, ŋp2 ⫽ .10; F2 ⬍ 1, or in the error analyses, Fs ⬍ 1. Word type did not interact with interlingual word group in the latency data, F1(1, 28) ⫽ 1.54, ns; F2 ⬍ 1, but did in the error data, F1(1, 28) ⫽ 7.36,

187

p ⬍ .02, ŋp2 ⫽ .21; F2(1, 56) ⫽ 3.29, p ⬍ .08, ŋp2 ⫽ .06. French homographs were named 6 ms faster than matched control words, Fs ⬍ 1, but produced 2.2% more errors than controls, F1(1, 28) ⫽ 5.22, p ⬍ .02, ŋp2 ⫽ .16, F2 (1, 56) ⫽ 1.77, ns. Cognates were named 14 ms faster than control words, F1(1, 28) ⫽ 4.84, p ⬍ .05, ŋp2 ⫽ .15; F2 ⬍ 1, and produced 1.4% fewer errors than controls, F1(1, 28) ⫽ 3.84, p ⫽ .06, ŋp2 ⫽ .12; F2 (1, 56) ⫽ 1.53, ns. When just the participants who named French words in Block 1 were included in the analyses, the main effect of word type was again not significant in the latency analyses, F1(1, 15) ⫽ 1.75, ns; F2 ⬍ 1, or in the error analyses, Fs ⬍ 1. The interaction of word type and interlingual word group was not significant in the latency data, F1(1, 15) ⫽ 1.82, ns; F2 ⬍ 1, but was significant in the participants analysis of the error data, F1(1, 15) ⫽ 4.58, p ⬍ .05, ŋp2 ⫽ .23; F2 ⬍ 1. However, neither the homograph nor cognate effect reached significance in any of the analyses. When participants named words in L2 (English), there was a main effect of word type in both the latency data, F1(1, 28) ⫽ 6.48, p ⬍ .02, ŋp2 ⫽ .19, F2 (1, 56) ⫽ 5.08, p ⬍ .05, ŋp2 ⫽ .08, and in the error data, F1(1, 28) ⫽ 35.08, p ⬍ .001, ŋp2 ⫽ .56; F2(1, 56) ⫽ 2.71, p ⫽ .07, ŋp2 ⫽ .06. Word type did not interact with interlingual word group in the latency data, F1(1, 28) ⫽ 1.07, ns; F2 ⬍ 1, but did in the error data by participants, F1(1, 28) ⫽ 14.51, p ⬍ .01, ŋp2 ⫽ .34; F2(1, 56) ⫽ 2.71, ns. However, when each effect was tested separately, the interlingual homograph effect was significant but the cognate effect was not. Specifically, homographs were named 41 ms more slowly than matched controls, F1(1,

Table 4 Mean Naming Latencies and Error Rates on Interlingual and Control Words for French-English Bilinguals in Experiment 2 Naming latency (in ms) Critical

Control

Error rates (%) Effect

Critical

Control

Effect

L1 naming (French) Overall Homographs Cognates Homophones Block 1 Homographs Cognates Homophones Block 2 Homographs Cognates Homophones

539 531 546

545 545 547

⫺6 ⫺14 ⫺1

4.0 0.4 1.6

1.8 1.8 1.3

2.2 ⫺1.4 0.3

549 529 539

550 544 545

⫺1 ⫺15 ⫺6

2.5 0.4 0.4

0.8 1.7 1.3

1.7 ⫺1.3 ⫺0.9

529 532 554

539 546 550

⫺10 ⫺14 4

5.7 0.5 2.9

2.9 1.9 1.4

2.8 ⫺1.4 1.5

L2 Naming (English) Overall Homographs Cognates Homophones Block 1 Homographs Cognates Homophones Block 2 Homographs Cognates Homophones

609 614 578

568 595 559

41 19 19

14.4 10.9 4.7

2.7 8.9 2.2

11.7 2.0 2.5

571 598 573

568 599 552

3 ⫺1 21

14.3 11.4 4.8

3.8 9.1 2.4

11.1 2.3 2.4

641 627 583

568 591 565

73 36 18

14.6 10.4 4.6

1.7 8.8 2.1

12.9 1.6 2.5

188


28) ⫽ 4.29, p ⬍ .05, ŋp2 ⫽ .13; F2(1, 56) ⫽ 4.02, p ⫽ .05, ŋp2 ⫽ .07, and produced 11.7% more errors, F1(1, 28) ⫽ 44.52, p ⬍ .001, ŋp2 ⫽ .61; F2(1, 56) ⫽ 6.09, p ⬍ .02, ŋp2 ⫽ .10, and cognates were named 19 ms more slowly than matched control words, F1(1, 28) ⫽ 2.99, p ⫽ .09, ŋp2 ⫽ .13, F2(1, 56) ⫽ 1.40, ns, and produced 2.0% more errors, F1(1, 28) ⫽ 2.00, ns; F2 ⬍ 1. When only participants who named English words in Block 1 were included in the analyses, in the latency analyses there was no effect of word type and no interaction of word type and interlingual word group, Fs ⬍ 1. In the error data, there was a main effect of word type that was significant by participants, F1(1, 13) ⫽ 15.32, p ⬍ .01, ŋp2 ⫽ .54; F2(1, 56) ⫽ 2.26, ns, and no interaction of word type and interlingual word group, F1(1, 13) ⫽ 3.12, ns; F2(1, 56) ⫽ 1.53, ns. There was only a homograph effect in the error data, F1(1, 13) ⫽ 13.93, p ⬍ .01, ŋp2 ⫽ .52; F2(1, 56) ⫽ 3.75, p ⬍ .06, ŋp2 ⫽ .06, and there was no cognate effect. Interlingual homophones. When participants named words in L1 (French), there was no effect of homophony when all participants were included, in either the latency data, F1 ⬍ 1; F2(1, 28) ⫽ 1.22, ns, or in the error data, Fs ⬍ 1, nor was there a significant effect of homophony just for participants who named words in French in Block 1, in the latency t1(15) ⫽ 1.04, ns; t2(28) ⫽ 0.51, ns, or error data t1(15) ⫽ 1.00, ns, t2(28) ⫽ 0.59, ns. When participants named words in L2 (English), homophones were responded to 19 ms more slowly than control words, F1 (1, 28) ⫽ 14.07, p ⬍ .01, ŋp2 ⫽ .33; F2 (1, 28) ⫽ 2.17, ns, and produced 2.5% more errors, F1 (1, 28) ⫽ 5.77, p ⬍ .03, ŋp2 ⫽ .17; F2 ⬍ 1. For participants who named words in English in Block 1, an inhibitory homophone effect was observed in the latency data in the analysis by participants, t1(13) ⫽ 3.06, p ⬍ .01; t2(28) ⫽ 1.52, ns, and there was no homophone effect in the error data, t1(13) ⫽ 1.47, ns; t2(28) ⫽ 1.19, ns. Carry-over. For French words, there were no effects of Block in either the latency, F1 ⬍ 1; F2(1, 84) ⫽ 2.20, ns, or error data, F1 ⬍ 1; F2(1, 84) ⫽ 3.38, p ⬍ .07, ŋp2 ⫽ .04. French words on average were named only 1 ms more slowly in the second block than in the first. There were also no interactions of Block and word type, or triple interaction of Block, word type, and interlingual group, Fs ⬍ 2. Interlingual words were named 9 ms more slowly in the second block than in the first and control words were named 7 ms faster in the second block. English words were named 19 ms more slowly on average in the second block compared with the first block. This difference was not significant by participants, F1 ⬍ 1, but was significant by items, F2(1, 84) ⫽ 5.22, p ⬍ .05, ŋp2 ⫽ .06. However, the interaction of Block and word type was significant, F1(1, 28) ⫽ 5.59, p ⬍ .05, ŋp2 ⫽ .17; F2(1, 84) ⫽ 3.70, p ⬍ .06, ŋp2 ⫽ .04. Interlingual words were named 36 ms more slowly when English was named in the second block of trials compared with the first, whereas their matched controls were named only 2 ms more slowly. The three-way interaction of Block by word type by interlingual word group was not significant. There was no effect of Block in the error data, and no significant interactions of Block with the other variables, Fs ⬍ 1.

Discussion French-English bilingual participants in Experiment 2 provided only weak evidence that they activated representations from both languages when naming words in L1 (French). When all partici-

pants were included, there were small interlingual homograph and cognate effects in some analyses by participants. The facilitatory cognate effect seen in the participants analysis of the latency data differs from Experiment 1 where no cognate effect was found in L1 naming. However, these observations are consistent with a lexical decision study by van Hell and Djkstra (2002) in which they found larger cognate facilitation effects in L1 with greater proficiency in L2. As seen in Table 2, participants in Experiment 2 were more proficient in their L2 than those in Experiment 1. There was an interaction between word type and interlingual word group in the error data, because of small effects in errors that were in different directions for homographs and cognates. Because these error effects were in the opposite direction, semantic representations associated with the nontarget (English) reading were perhaps weakly activated and in turn activated their corresponding phonological representations. When just participants who read in French in Block 1 were included, effects were even weaker. More evidence was found for the activation of French representations when naming in L2 (English). When all participants were included, interlingual homographs had significantly longer naming times and produced more errors than matched control words, which matches the results in L2 naming in Experiment 1. Unlike Experiment 1, however, there was no facilitatory cognate effect, in fact there was a nonsignificant trend to an inhibitory effect. This finding is consistent with the results of an eyetracking study by Pivneva, Mercier, and Titone (2014), who also found that increased L2 proficiency reduced cognate facilitation effects in L2. Although the inhibitory interlingual homograph effect was significant and the weaker inhibitory trend for cognates was not, the interaction between word type and interlingual word group was significant only in the participants’ analysis of the error data, providing little evidence that phonological representations were activated by feedback from semantic representations. If the impact of French on English naming is because of especially quick activation of French phonological representations because of well-practiced and transparent spelling-sound correspondences, then a robust difference between interlingual words and their controls would also be expected in the analysis of the data just from participants who named in English first. However, crosslanguage effects were much less evident when only data from these participants were included. There was an inhibitory interlingual homograph effect just in the error data. This finding indicates that much of the evidence for activation of French phonological representations came from participants who had named French words in the first block. There was no interlingual homophone effect when naming in L1, as in Experiment 1, but in contrast to Experiment 1, when naming in L2 interlingual homophones had somewhat longer naming latencies than matched control words. That is, shared phonology with L1 French words hindered the naming of L2 English words. ANOVAs on the Block 1 naming latency data with Experiment (1 vs. 2) and Word Type (homophone vs. control) as variables revealed a significant interaction, F1(1, 27) ⫽ 12.03, p ⬍ .01, ŋp2 ⫽ .31; F2(1, 56) ⫽ 5.91, p ⬍ .05, ŋp2 ⫽ .10. We consider possible explanations for these findings in the General Discussion. There was no evidence here that these bilinguals inhibited L1 French representations when naming L2 English words, as French words were not named more slowly overall in the second block (after reading in English) compared with the first block. In con-


trast, English words were named more slowly when in the second block compared with the first in the analysis by items, but the evidence does not suggest that L2 English representations were generally inhibited when reading in L1 because it was the just the interlingual words and not the English control words that were named more slowly. This finding may instead be because of prolonged activation of French representations from the first block to the subsequent English block.

General Discussion The current study investigated phonological processing dynamics in bilingual word naming using English-French interlingual homographs, cognates, and interlingual homophones and matched control words. We not only sought evidence as to whether bilinguals simultaneously activate phonological representations from a nontarget language when naming target language words, but also to determine whether any such phonological activation flows to other representations. We use the theoretical framework of the BIA⫹ (Dijkstra & van Heuven, 2002) to explain our findings.

Interlingual Homographs and Cognates Differences were found in naming performance on interlingual words and matched controls, with these effects generally being strongest when participants named words in L2, and weakest when participants named words in L1 in the first block of trials. Broadly, these findings indicate that phonological representations of a nontarget language are activated when naming words in a target language, and they provide support for the assumption in the BIA⫹ model that representations are integrated across languages. More specifically, interlingual homographs, which have a different pronunciation and meaning in English and French, were more difficult to name than matched control words. The inhibitory interlingual homograph effect was particularly robust when participants named words in L2, both when L2 was French (Experiment 1) and when L2 was English (Experiment 2). Experiment 2 replicates the findings of Jared and Szucs (2002, Experiment 2), and Experiment 1 extends their findings to naming in French as an L2. When naming in L1, the homograph effect was much weaker, particularly in the first block of trials, and was mostly a small inhibitory effect in error data. This too is consistent with Jared and Szucs (2002, Experiment 1) results, and expected by the BIA⫹ because activation levels depend in part on subjective frequency, and therefore phonological and semantic codes associated with bilinguals’ second language may be delayed in activation relative to first language codes. A comparison of the interlingual homograph and cognate effects provides evidence concerning the extent to which nontarget representations are activated. If activation in the nontarget language flows just from orthography to phonology, then interlingual homograph effects and cognate effects should both be inhibitory and of a similar magnitude because both had different pronunciations in English and French. However, participants did not generally find cognates more difficult to name than their matched controls. The only cognate effect that was significant by both participants and items was in Experiment 1 where there was a facilitatory cognate effect in the latency data when participants were naming in L2 (French), although in Experiment 2, participants did show a nonsignificant inhibitory trend when

189

naming cognates in L2 (English). Performance on interlingual homographs and cognates was different enough to produce a significant interaction between word type and interlingual word group in both L1 and L2 in Experiments 1 and 2 in at least one of the four analyses (RT and errors, each by participants and items) with all participants included. These findings provide evidence that activation of the nontarget language goes beyond feedforward activation from orthography to phonology and extends to semantic representations and back to phonological representations. Because the overall analyses included some participants who had just named a block of words in the nontarget language, the extent of nontarget language activation suggested by these analyses may be greater than what occurs in a monolingual reading situation. Therefore, analyses were also conducted on just the first block of trials, at which point the experiments involved a single language. There was less evidence that activation of the nontarget language extended to semantic representations on those trials, as fewer of the interactions between word type and interlingual group were significant. However, the weaker statistical effects may be partly attributable to the fact that the Block 1 analyses included only half of the participants. Nonetheless, there was particularly clear evidence in Experiment 1 of semantic involvement when Block 1 words were in L2, as a significant interaction between word type and interlingual group was observed in the latency data. A parallel effect was not seen in the latency data in Experiment 2 where participants were more proficient in their L2, although in that experiment the homograph effect was significant in the error data in Block 1 and the cognate effect was not. The differences in findings across the two experiments are probably because of L2 proficiency differences. Pivneva et al. (2014) showed in an eyetracking study that increased L2 proficiency reduced the cognate facilitation effect in L2 but did not influence the inhibitory homograph effect. In the BIA⫹, feedback from semantic representations to lexical phonological representations is likely to increase the difficulty of naming interlingual homographs and may reduce the difficulty of naming cognates. Semantic representations receive activation from lexical orthographic representations as well as lexical phonological representations. Because the homographs and cognates used in our study have the same spelling in English and French, lexical orthographic representations for the word in each language would activate their corresponding semantic representation, as would lexical phonological representations. In the case of homographs, lexical representations in each language would activate different semantic representations (e.g., LIT: ignited, bed), whereas cognates (e.g., TRAIN) would activate the same semantic representation. The two semantic representations associated with the homograph would each send activation to the relevant phonological representations in each language, with the consequence that two lexical phonological representations in each language would receive activation (e.g., English and French phonological representations of LIT /lit/ and /li/, as well as the English phonological representation /bεd/ associated with the “bed” meaning and the French phonological representation /alyme/ associated with the “ignited” meaning). Thus, feedback from semantic representations would not only increase the activation of the phonological representations of the homograph each language, but would add competition from another word in the nontarget language and a word in the target language. It is easy to see how the added competition could slow the naming of homographs relative to single language control words. In contrast, a single semantic representation is activated in the case of cognates (e.g., TRAIN), and this representation feeds back activa-

190


tion to its lexical phonological representation in each language (e.g., /trein/ and /trε˜ /). When naming in a weaker L2, quick and strong activation of the semantic representation from L1 representations would be especially helpful, because the semantic representation would provide stronger feedback to the L2 phonological representation than would occur for a single language control word. However, feedback from the semantic representation would also provide additional activation of the phonological representation in the nontarget language, which would increase competition, particularly if the nontarget language were the participants’ L1. If the word with the highest activation was always selected and produced, then responses in the weaker L2 might never get selected. In the BIA⫹, the lexical phonological representation that is activated most highly is selected first, and if it does not have the correct language membership, then the next most highly activated lexical phonological representation is selected. When a phonological representation with the correct language membership is found, it is produced. In naming a cognate in L2, the L1 lexical phonological representation would probably be selected first because it is likely to be most active, but because it does not have the correct language membership it would not be produced as a response. The next most active phonological representation would then be selected, which is likely to be the L2 pronunciation of the cognate, and if so, it would be produced. To get a cognate facilitation effect, as was found in L2 naming in Experiment 1, requires the assumption that the boost in semantic activation that comes from L1 lexical representations, and consequently in the activation from the semantic representation to the L2 pronunciation of the cognate, allows the L2 pronunciation to be selected faster than the pronunciation of an L2-only word, despite some additional competition from the L1 pronunciation of the cognate. A cognate facilitation effect, therefore, would probably be most likely to occur in an L2 naming task when L2 is weaker than L1. If the nontarget language provides only a small boost to the activation of the semantic representation of a cognate, perhaps because of high proficiency in the target language, then the target phonological representation will in turn get only a small amount of additional activation compared with a target language-only word. Competition from the nontarget lexical phonological representation in the case of a cognate may cancel out this additional activation, and may even produce an inhibitory cognate effect. A computer simulation of the BIA⫹ is needed to determine whether this speculation is correct and to better understand how proficiency in the nontarget language influences the cognate effect in the model. In summary, results from the current study on heterophonic homographs and cognates indicate that interlingual naming effects cannot be explained solely as a consequence of competition from feedforward activation of phonological representations from orthographic representations. Feedback activation of phonological representations from semantic representations also needs to be incorporated into a comprehensive explanation.

Interlingual Homophones An intriguing finding was that the interlingual homophone effect in L2 naming in Block 1 was facilitatory in Experiment 1, but inhibitory in Experiment 2. The facilitatory effect in naming French L2 words in Experiment 1 is consistent with Jared et al.’s (2012) finding with Grade 3 English-French bilingual students,

and the inhibitory effect in naming L2 English words in Experiment 2 is consistent with previous English monolingual studies in which low-frequency homophones with high-frequency mates had slower naming latencies than matched control words (Biedermann et al., 2009; Edwards et al., 2004). The challenge is to explain how these two different patterns of results arise. In the BIA⫹, a printed homophone (e.g., FOE, which has French homophone mate FAUX) would activate two lexical phonological representations, one for each language) from activated sublexical phonological units, and these would compete and slow naming, especially in L2. Furthermore, if the two lexical phonological units each activated their corresponding semantic representation (“enemy,” “false”) and these fed back activation to lexical phonology in each language, two additional lexical phonological representations would be activated, which would increase the inhibitory interlingual homophone effect. Competition between lexical phonological representations can, therefore, account for the inhibitory interlingual homophone effect in Experiment 2, but not the facilitatory homophone effect in Experiment 1. Haigh and Jared (2007) observed a facilitatory interlingual homophone effect in a lexical-decision task for French-English bilinguals reading in English (which is opposite to the naming results for similar bilinguals in Experiment 2 here). They suggested that participants may have responded on some trials on the basis of lexical phonological activation without checking language membership. Similarly, a facilitatory effect might have been produced in Experiment 1 if participants responded on the basis of the L1 lexical phonological representation, which would be expected to be activated more quickly than the L2 representation in unbalanced bilinguals. However in a naming task, a strategy of ignoring the language node would have increased the error rate for interlingual homographs and cognates, which are pronounced differently in English and French. In the first block of Experiment 1, cognates had a slightly higher error rate than control words (2.2% vs. 0.9%), so this explanation is possible, but perhaps not that likely. Of potential relevance is that the lexical phonological representations of the two members of each homophone pair may have been more similar for the English-French participants than for the French-English participants because the English-French participants had less exposure to native speakers of their L2 than did the French-English bilinguals. For example, the pair REAR-RIRE may have been represented as /rir/ in both languages for at least some of the English-French bilinguals, but /rir/ and /ʁiʁ/ for the French-English bilinguals. Another example is SANK-CINQ which may both have been represented as /sæŋk/ by EnglishFrench bilinguals but as /sæŋk/ and /sε˜k/ by the French-English bilinguals. A detailed phonetical analysis of the participants’ responses is beyond the scope of this study, and in any case, participants did not produce both members of any homophone pair. If homophone mates were more similar for English-French participants, when the L2 member was presented, the lexical phonological representation for the homophone in L1 would have been more highly active than for French-English participants. On the surface, this would seem to imply that there should have been a greater inhibitory effect in Experiment 1 than in Experiment 2. Facilitation might result if the lexical phonological representation for the L1 homophone fed strong activation back to the sublexical phonological representations, which in turn fed activation forward to the L2 lexical phonological representation. The single language control


word would not have this extra boost to the L2 lexical phonological representation. For facilitation to occur, this boost to the L2 lexical phonological representation in the case of an interlingual homophone would have to be strong enough to overcome inhibition from the L1 lexical phonological representation. Such a boost may have been particularly helpful for low-frequency L2 words for the English-French bilinguals who were not as fluent in their second language as the French-English bilinguals. Again, a working simulation is needed to test the plausibility of this account.

191

single language. More specifically, interlingual effects cannot be attributed solely to the feed forward activation of phonological representations in both languages. Our findings suggest that semantic representations associated with mates in the nontarget language are also activated and in turn feed activation back to lexical phonological representations. Results from the current study provide an important contribution to our understanding of the processing dynamics in bilingual word identification.

Résumé

Language Inhibition We explored whether our data provided evidence that bilinguals inhibit one language when naming words in the other language. The BIA⫹ does not include such an inhibitory mechanism, although its predecessor the BIA did. Misra et al. (2012) found that participants inhibited L1 when naming pictures in L2, but not the reverse. If the nontarget language was inhibited in Block 1 and it took more effort to reactivate it in Block 2 when it became the target language, then naming times for all word types in Block 2 should have been slowed. Alternatively, if block differences occurred because activation of the target language in Block 1 persisted into Block 2, then interlingual words should be more affected than single language control words in Block 2. In Experiment 1, there was evidence that Block 2 naming in English was influenced by both of these factors. Interlingual words and their controls were named more slowly in Block 2 than in Block 1, although homographs were particularly slowed. The effect of Block was significant just by items, suggesting that only some participants inhibited English when they named words in French in Block 1, or that the inhibition lasted only part of the way through the English list. There was no other evidence of a carryover of inhibition of the nontarget language in Block 1 to naming that language in Block 2. In Experiment 2, English words were also named more slowly in Block 2 compared with Block 1, but it was just the interlingual words that were named more slowly, not the control words. This finding suggests that block differences were because of activation of French representations from Block 1 persisting into the English block. Furthermore, there was no evidence that either group of bilinguals inhibited activation of French when they named words in English in Block 1, as naming times for French words in Block 2 were not slower than in Block 1. The English- French bilinguals in Experiment 1 may have shown more evidence of inhibiting L1 when reading in L2 than the French-English bilinguals in Experiment 2 because their L2 was relatively weaker compared with their L1. In summary, only very limited support was found here that participants inhibited representations from one language when naming words in the other language, consistent with Jared and Kroll (2001) and Jared and Szucs (2002), although our data do not rule out the possibility that there was some inhibition of nontarget representations in Block 1 but that participants recovered quickly as they began naming words in Block 2. Given that the current version of the BIA⫹ does not include an inhibitory mechanism, further research on this issue is needed. Future research investigating this issue could divide lists into carefully matched segments so that the duration of any inhibition after switching languages could be examined.

Conclusion Results from the current study clearly indicate that lexical codes from both languages are activated when bilinguals read aloud in a

Cette étude a examiné la dynamique du traitement phonologique de la dénomination bilingue. Des sujets bilingues anglais-français et français-anglais ont désigné des homographes hétérophones interlinguaux (mots possédant la même orthographe, mais n’ayant pas la même signification ou prononciation dans les deux langues), des mots apparentés hétérophones (mots ayant la même orthographe et la même signification dans les deux langues, mais une prononciation différente), des homophones interlinguaux (mots ayant la même prononciation dans les deux langues, mais pas la même orthographe ni la même signification), et ont assorti des mots contrôles dans le cadre de tâches de dénomination unilingue, en français et en anglais. L’activation phonologique interlinguale était plus forte dans la langue seconde des sujets bilingues. Les résultats mettent en évidence l’activation en aval des représentations phonologiques dans la langue non cible, ainsi qu’une rétroactivation de ces représentations phonologiques a` partir des représentations sémantiques. Les résultats sont interprétés selon le plus récent modèle d’activation interactive bilingue (BIA, Bilingual Interactive Activation). Mots-clés : bilingue, phonologie, dénomination, effets interlinguaux.

References Ando, E., Matsuki, K., Sheridan, H., & Jared, D. (in press). The locus of Katakana-English masked phonological priming effects. Bilingualism: Language and Cognition. Andrews, S. (1997). The effect of orthographic similarity on lexical retrieval: Resolving neighborhood conflicts. Psychonomic Bulletin & Review, 4, 439 – 461. http://dx.doi.org/10.3758/BF03214334B Baayen, R., Piepenbrock, R., & van Rijn, H. (1993). The CELEX lexical database [CD-ROM]. Philadelphia, PA: University of Pennsylvania Linguistic Data Consortium. Balota, D. A., Yap, M. J., Cortese, M. J., Hutchison, K. A., Kessler, B., Loftis, B., . . . Treiman, R. (2007). The English lexicon project. Behavior Research Methods, 39, 445– 459. http://dx.doi.org/10.3758/ BF03193014 Biedermann, B., Coltheart, M., Nickels, L., & Saunders, S. (2009). Effects of homophony on reading aloud: Implications for models of speech production. Language and Cognitive Processes, 24, 804 – 842. http://dx .doi.org/10.1080/01690960802597243 Carrasco-Ortiz, H., Midgley, K. J., & Frenck-Mestre, C. (2012). Are phonological representations in bilinguals language specific? An ERP study on interlingual homophones. Psychophysiology, 49, 531–543. http://dx.doi.org/10.1111/j.1469-8986.2011.01333.x De Groot, A. M. B., Borgwaldt, S., Bos, M., & van den Eijnden, E. (2002). Lexical decision and word naming in bilinguals: Language effects and task effects. Journal of Memory and Language, 47, 91–124. http://dx .doi.org/10.1006/jmla.2001.2840 Dijkstra, T., Grainger, J., & van Heuven, W. J. B. (1999). Recognition of cognates and interlingual homographs: The neglected role of phonology.

192


Journal of Memory and Language, 41, 496 –518. http://dx.doi.org/ 10.1006/jmla.1999.2654 Dijkstra, T., Miwa, K., Brummelhuis, B., Sappelli, M., & Baayen, H. (2010). How cross-language similarity and task demands affect cognate recognition. Journal of Memory and Language, 62, 284 –301. http://dx .doi.org/10.1016/j.jml.2009.12.003 Dijkstra, T., & van Heuven, W. (1998). The BIA model and bilingual word recognition. In J. Grainger & A. M. Jacobs (Eds.), Localist connectionist approaches to human cognition (pp. 189 –225). Mahwah, NJ: Erlbaum. Dijkstra, T., & van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition, 5, 175–197. http://dx.doi.org/ 10.1017/S1366728902003012 Dimitropoulou, M., Duñabeitia, J. A., & Carreiras, M. (2011). Phonology by itself: Masked phonological priming effects with and without orthographic overlap. Journal of Cognitive Psychology, 23, 185–203. http:// dx.doi.org/10.1080/20445911.2011.477811 Edwards, J. D., Pexman, P. M., & Hudson, C. E. (2004). Exploring the dynamics of the visual word recognition system: Homophone effects in LDT and naming. Language and Cognitive Processes, 19, 503–532. http://dx.doi.org/10.1080/01690960344000215 Ferrand, L., Brysbaert, M., Keuleers, E., New, B., Bonin, P., Méot, A., . . . Pallier, C. (2011). Comparing word processing times in naming, lexical decision, and progressive demasking: Evidence from chronolex. Frontiers in Psychology, 2, 306. http://dx.doi.org/10.3389/fpsyg.2011.00306 Font, N. (2001). Rôle de la langue dans l’accèss au lexique chez les bilingues: Influence de la proximité orthographique et sémantique interlangue sur la reconnaissance visuelle des mots [Role of language in lexical access in bilinguals: Influence of interlingual orthographic and semantic proximity on visual word recognition]. Unpublished doctoral dissertation, Université Paul Valery, Montpellier, France. Friesen, D. C., & Jared, D. (2012). Cross-language phonological activation of meaning: Evidence from category verification. Bilingualism: Language and Cognition, 15, 145–156. Gollan, T. H., Forster, K. I., & Frost, R. (1997). Translation priming with different scripts: Masked priming with cognates and noncognates in Hebrew-English bilinguals. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23, 1122–1139. http://dx.doi.org/ 10.1037/0278-7393.23.5.1122 Gollan, T. H., Montoya, R. I., Cera, C., & Sandoval, T. C. (2008). More use almost always a means a smaller frequency effect: Aging, bilingualism, and the weaker links hypothesis. Journal of Memory and Language, 58, 787– 814. http://dx.doi.org/10.1016/j.jml.2007.07.001 Haigh, C. A., & Jared, D. (2007). Phonological activation in bilingual silent reading. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33, 623– 644. http://dx.doi.org/10.1037/0278-7393.33.4.623 Hino, Y., & Lupker, S. J. (1996). Effects of polysemy in lexical decision and naming: An alternative to lexical access accounts. Journal of Experimental Psychology: Human Perception and Performance, 22, 1331– 1356. http://dx.doi.org/10.1037/0096-1523.22.6.1331 Jared, D., Cormier, P., Levy, B. A., & Wade-Woolley, L. (2012). Crosslanguage activation of phonology in young bilingual readers. Reading and Writing, 25, 1327–1343. http://dx.doi.org/10.1007/s11145-0119320-0 Jared, D., & Kroll, J. F. (2001). Do bilinguals activate phonological representations in one or both of their languages when naming words? Journal of Memory and Language, 44, 2–31. http://dx.doi.org/10.1006/ jmla.2000.2747 Jared, D., Levy, B. A., & Rayner, K. (1999). The role of phonology in the activation of word meanings during reading: Evidence from proofreading and eye movements. Journal of Experimental Psychology: General, 128, 219 –264.

Jared, D., & Szucs, C. (2002). Phonological activation in bilinguals: Evidence from interlingual homograph naming. Bilingualism: Language and Cognition, 5, 225–239. http://dx.doi.org/10.1017/S1366728902003024 Jouravlev, O., Lupker, S. J., & Jared, D. (2014). Cross-language phonological activation: Evidence from masked onset priming and ERPs. Brain and Language, 134, 11–22. http://dx.doi.org/10.1016/j.bandl.2014 .04.003 Kim, J., & Davis, C. (2003). Task effects in masked cross-script translation and phonological priming. Journal of Memory and Language, 49, 484 – 499. http://dx.doi.org/10.1016/S0749-596X(03)00093-7 Lemhöfer, K., & Dijkstra, T. (2004). Recognizing cognates and interlingual homographs: Effects of code similarity in language-specific and generalized lexical decision. Memory & Cognition, 32, 533–550. http:// dx.doi.org/10.3758/BF03195845 Misra, M., Guo, T., Bobb, S. C., & Kroll, J. F. (2012). When bilinguals choose a single word to speak: Electrophysiological evidence for inhibition of the native language. Journal of Memory and Language, 67, 224 –237. http://dx.doi.org/10.1016/j.jml.2012.05.001 Nakayama, M., Sears, C. R., Hino, Y., & Lupker, S. J. (2012). Cross-script phonological priming for Japanese-English bilinguals: Evidence for integrated phonological representations. Language and Cognitive Processes, 27, 1563–1583. http://dx.doi.org/10.1080/01690965.2011 .606669 New, B., Pallier, C., Ferrand, L., & Matos, R. (2001). Une base de données lexicales du français contemporain sur internet: LEXIQUE [A lexical database of contemporary French on the Internet: LEXIQUE]. L’Année Psychologique, 101, 447– 462. http://www.lexique.org. http://dx.doi.org/ 10.3406/psy.2001.1341 Newman, R. L., Jared, D., & Haigh, C. A. (2012). Does phonology play a role when skilled readers read high-frequency words? Evidence from ERPs. Language and Cognitive Processes, 27, 1361–1384. http://dx.doi .org/10.1080/01690965.2011.603932 Perfetti, C. A., Bell, L. C., & Delaney, S. M. (1988). Automatic (prelexical) phonetic activation in silent word reading: Evidence from backward masking. Journal of Memory and Language, 27, 59 –70. http://dx.doi .org/10.1016/0749-596X(88)90048-4 Pexman, P. M., Lupker, S. J., & Jared, D. (2001). Homophone effects in lexical decision. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27, 139 –156. http://dx.doi.org/10.1037/0278-7393 .27.1.139 Pivneva, I., Mercier, J., & Titone, D. (2014). Executive control modulates cross-language lexical activation during L2 reading: Evidence from eye movements. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40, 787–796. http://dx.doi.org/10.1037/a0035583 Rastle, K., & Brysbaert, M. (2006). Masked phonological priming effects in English: Are they real? Do they matter? Cognitive Psychology, 53, 97–145. http://dx.doi.org/10.1016/j.cogpsych.2006.01.002 Schneider, W., Eschman, A., & Zuccollotto, A. (2002). E-prime user’s guide. Pittsburgh, PA: Psychology Software Tools. Schwartz, A. I., Kroll, J. F., & Diaz, M. (2007). Reading words in Spanish and English: Mapping orthography to phonology in two languages. Language and Cognitive Processes, 22, 106 –129. http://dx.doi.org/ 10.1080/01690960500463920 Seidenberg, M. S., Waters, G. S., Barnes, M. A., & Tanenhaus, M. K. (1984). When does irregular spelling or pronunciation influence word recognition? Journal of Verbal Learning & Verbal Behavior, 23, 383– 404. http://dx.doi.org/10.1016/S0022-5371(84)90270-6 Smits, E., Martensen, H., Dijkstra, T., & Sandra, D. (2006). Naming interlingual homographs: Variable competition and the role of the decision system. Bilingualism: Language and Cognition, 5, 225–239. Smits, E., Sandra, D., Martensen, H., & Dijkstra, T. (2009). Phonological inconsistency in word naming: Determinants of the interference effect between languages. Bilingualism: Language and Cognition, 12, 23–39. http://dx.doi.org/10.1017/S1366728908003465

INTERLINGUAL EFFECTS IN WORD NAMING Snodgrass, J. G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174 –215. http://dx.doi.org/10.1037/0278-7393.6.2.174 Stone, G. O., Vanhoy, M., & van Orden, G. (1997). Perception is a two-way street: Feedforward and feedback phonology in visual word recognition. Journal of Memory and Language, 36, 337–359. http://dx .doi.org/10.1006/jmla.1996.2487 Strain, E., Patterson, K., & Seidenberg, M. S. (1995). Semantic effects in single-word naming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1140 –1154. http://dx.doi.org/10.1037/ 0278-7393.21.5.1140

193

van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can influence native language performance in exclusively native contexts. Psychonomic Bulletin & Review, 9, 780 –789. http://dx.doi.org/10.3758/ BF03196335 van Orden, G. C. (1987). A ROWS is a ROSE: Spelling, sound, and reading. Memory & Cognition, 15, 181–198. http://dx.doi.org/10.3758/ BF03197716 Zhou, H., Chen, B., Yang, M., & Dunlap, S. (2010). Language nonselective access to phonological representations: Evidence from Chinese-English bilinguals. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 63, 2051–2066. http://dx.doi.org/10.1080/ 17470211003718705

Appendix Experimental Words Used in Experiments 1 and 2 Cognates

Controls

accord arrive bizarre fragile grand image menace pardon pose prince saint terrain train trait voyage

acorn arrest blizzard frenzy guard incline merchant parcel poke print sail toward trust trail vowel

base central change direct fruit large million moral normal parent police second site total vote

bec ceinture chiffre dirige frein larmes maigre mardi nager punir pareil secours songe tiroir veau

Homographs

Controls

Homophones

Controls

English Experimental Stimuli attend awake chat chart coin coil comment collect mince merge manger mangle parole pickle pain park parole pickle pays pegs sale soul sang song seize sneak tire tile vent vest

bet bun cask doe fee foe leer mare oat pen rear sank sue tie tray

bid bud clam dye fan fox loaf maze oak pet reap sink sag tip trap

French Experimental Stimuli bond bain cent soi chair croix course crainte court coude four fier habit homard laid lave once onde ours oeuf ride ruche sort sec stage selon suit soif tape tasse

bague cache chaud chère coule dors fiche flèche glace lac nid nord paix rhume trou

bouge casse chaise chèvre cuir défi feux fleuve glisse loi noix neuf pont reine trente

Received June 13, 2014 Accepted August 8, 2014 䡲

Copyright of Canadian Journal of Experimental Psychology is the property of Canadian Psychological Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Asymmetries in phonological development: the case of word-final cluster acquisition in Welsh-English bilingual children.

Relation between Phonological Processing, Auditory Processing and Speech Perception among Bilingual Poor Readers.

Phonological assessment and analysis of bilingual preschoolers' Spanish and English word productions.

Bilingual Object Naming: A Connectionist Model.

Asymmetric Switch Costs in Numeral Naming and Number Word Reading: Implications for Models of Bilingual Language Production.

Flexibility in Bilingual Infants' Word Learning.

Implicit object naming in visual search: Evidence from phonological competition.

The word-superiority effect and phonological recoding.

Phonological code retrieval during picture naming: Influence of consonant class.

Lexical effects in word naming in spanish children.

Predicting reading ability over the long term: The changing roles of letter naming, phonological awareness and orthographic processing.

Orthographic and phonological neighborhood effects in handwritten word perception.

Hemodynamic response in a geographical word naming verbal fluency test.

Phonological processing in reading: data from alexia.

Selective inhibition and naming performance in semantic blocking, picture-word interference, and color-word Stroop tasks.

Dialect Variation Influences the Phonological and Lexical-Semantic Word Processing in Sentences. Electrophysiological Evidence from a Cross-Dialectal Comprehension Study.

Second language phonology influences first language word naming.

Phonological harmony as a processing problem.

Language specificity of lexical-phonological therapy in bilingual aphasia: A clinical and electrophysiological study.

Rethinking assessment measures of phonological development and their application in bilingual acquisition.

Processing advantage for emotional words in bilingual speakers.

Phonological and semantic factors in the object-naming errors of skilled and less-skilled readers.

Parallel deterioration to language processing in a bilingual speaker.

Brain bases of morphological processing in Chinese-English bilingual children.