The Foreign Language Effect and Disembodied Cognition: The Complexity of Emotional Boundaries and Linguistic Factors
IN THIS ARTICLE
Cognitive psychology research informs on the complexities of human functioning and behavior and thereby, simultaneously, extends our agency to harness its potential malleability. Our various cognitive processes (e.g., decision-making, emotion, language) furthermore point to complex, interrelated relationships that coalesce into human intricacies. Indeed, the subfield of bilingual cognition points to a special premise of “disembodied” cognition such as in Keysar, Hayakawa, and An’s (2012) proposed foreign language effect (FLe) eliciting emotional and cognitive distance. The purpose of this review is to explore the scope of this premise via proposed mechanisms (e.g., System 1/System 2) and domains (e.g., emotional, morality, logical) of the FLe and overall linguistic factors (e.g., age and context effects). Amid a complex picture, the foreign language effect appears largely constricted to emotion-driven contexts as related to the interface between altered System 1 and System 2 processes. Moreover, a nuanced focus on age and context effects holds promise in capturing the variable patterning of disembodied cognition within groups with diverse language trajectories such as U.S. Spanish-English bilinguals.
The literature on bilingual cognition historically points to potential cognitive differences that arise from bilingual or multilingual language use, coming to momentum with Keysar et al. (2012) proposition of the foreign language effect. The authors performed three experiments and a control version of a framing risk paradigm on a college population: 121 English-Japanese Bilinguals, 144 Korean-English bilinguals, 103 English-French bilinguals, and 84 English-Spanish Bilinguals, respectively. All participants acquired their foreign language in a classroom setting and underwent several assessments (e.g., writing, comprehension) to determine qualifying proficiency. In each experiment, participants completed vetted translations of a modified “Asian disease” problem entirely in their foreign (FL) or native language (NL). A comprehension task preceded both conditions (NL and FL) but finished with an NL translation of materials for the latter. The Asian disease problem reveals asymmetrical human tendencies in the framing of risk despite similar expected values. Individuals tend to be risk-averse in the domain of gains (i.e., lives saved) and risk-seeking in the domain of losses (i.e., lives lost) and thus point to biases that deter rational analyses. The results of the first portion of experiments demonstrate that FL use unravels framing biases, leading to more rational, description independent choices. Furthermore, Korean-English bilinguals and English-Spanish bilinguals completed experiments gauging loss aversion; the former participated in a betting task with an equal proportion of low and high stakes while the latter dealt with money in a bet/investment task. Coupled together, the experiments illustrate FL use may reduce loss aversion even when involving tangible resources.The above findings prompted Keysar et al. (2012) to introduce an increased-systematicity account whereby foreign language use encourages System 2 processes through emotional and cognitive distance. A presumably reduced emotionality and taxing use of the FL may result in a reliance on mental resources deployed in an analytical and rule-governed manner. Alternatively, the cognitive load that the FL arises could promote a default to the largely automatic and emotionally ridden System 1 processes, but unsupported in Keysar et al. (2012). Given that “an emotional reaction sometimes induces a less systematic decision,” the authors propose the FLe weakens “decision biases [rooted] in an emotional reaction” (Keysar et al., 2012, p. 666-667). The contribution of the FLe to the literature inspires various subsequent studies such as those assessing its replicability and mechanisms in moral quandaries.
Assessing Mechanisms and Replication of the FLe
The FLe in Morality Domains
The FLe replicates in morality paradigms such as the trolley and footbridge dilemmas involving emotional appeal and/or components (Cipolletti, Macfarlane, & Weissglass, 2015; Geipel, Hadjichristidis, & Lurian, 2015; Hayakawa, Tannenbaum, Costa, Corey, & Keysar, 2017). These paradigms involve moral quandaries tied to emotionally aversive options of sacrificing a human life for a greater good although varying in involvement. Their prime scenario involves a runaway train menacing human lives and a respective, morally challenging option: pressing a button with the collateral damage of a life (trolley) versus directly pushing a person to save the rest (footbridge). The former typically bears less emotional weight or moral conflict as compared to the latter. Responses are further framed in terms of unwavering adherence to moral principles (i.e., deontological -no) or maximizing the potential benefits for the greater good (i.e., utilitarian/consequentialist- yes). Deontological responses are typically associated with emotion and thus characteristic of System 1 whereas utilitarian responses recruit rational, deliberate thinking consistent with System 2. As Cipolletti et al. (2016) warn, however, the exclusive affiliation of responses and prominent characteristics (i.e., emotional vs rational) with either processing type may oversimplify reality and thus begets caution in interpretation.
While the study of these morality paradigms supports the assertion of an FL’s influence, differing methodology leads to variable framing of the FLe, specifically regarding its mechanism. All utilize undergraduate populations subjected to exclusionary criteria (i.e., measures of proficiency) and attest to moderate generalizability to variant groups when combined; Cipolletti et al. (2015) focuses on 160 English-Spanish bilinguals although over representative of Native English speakers and Hayakawa et al. (2017) on a whopping total of 1,277 of English-German bilinguals and Spanish-English bilinguals with overrepresentation for the former. Lastly, Geipel et al. (2015) appraise a total of 338 participants composed of Italian-English, Italian-French bilinguals, German-English bilinguals, and Chinese-English bilinguals. Hayakawa et al. (2017) employ an elaborate experimental design with a process-disassociation moral task whereby half of the dilemmas were traditionally incongruent (utilitarian vs deontological) and the other half were congruent or in agreement (e.g., only a mildly injured person in the paradigms). The wording varied with experiments such as in personal commitment to completing an action or finding it generally acceptable, coupled with individual measures (e.g., Interpersonal reactivity index) for three of the experiments. Similarly, Geipel et al. (2015) introduce distress ratings and contrasting moral dilemmas in a third experiment: returning a lost wallet (impersonal-low emotion) and sacrificing the life of your crying baby to avoid detection by soldiers (personal-high emotion).
The results of the three studies reveal that FL use leads to an increased willingness to accept utilitarian action in moral paradigms considered to be more morally controversial (i.e., footbridge dilemma). Cipolletti et al. (2015) frame the FLe mechanism as a potential change in the cognitive process given that a mere language change cannot account for an FLe effect. Thus, they posit a dual-process approach (System 1 and System 2 processes) adequately “explain[s] [the representation of] details more or less vividly, as it explicitly appeals to distinct processing types.” (Cipolletti et al., 2015, p. 34). The findings allude to the recruitment of systematic thinking, but the authors remain wary of a conclusive role of emotion, citing the need to directly study moral emotions absent in their design. Taking this call to action, Geipel et al. (2015) distress ratings illustrate FL attenuation in both dilemmas and of which did not appear to mediate associations between language and moral judgment. Furthermore, the presence of the FLe in both impersonal dilemmas (i.e., lost wallet) and personal dilemmas (i.e., footbridge) fails to support contingency on emotional appeal. Thus, the authors conclude obstructed accessibility to clear social and moral rules (e.g., do not steal, do not kill) in certain paradigms may underlie the MFLe mechanism rather than pure emotional attenuation.
Yet, Hayakawa et al. (2017) disentanglement of deontological (D) and utilitarian (U) responses via their process-disassociation task and U and D parameters for each participant unveil the blunting of emotional reactions. That is, the FL decreases deontological responses thereby stunting emotional processing associated with deontological rules. Conversely, FL use does not increase utilitarian responses theorized to arise from deliberation tied to an increased cognitive load. Instead, the pattern of decreasing FL utilitarian responses illustrates the opposite effect of cognitive load. The resulting premise of the FLe as not because individuals “think more but because they feel less” informs the likely possibility it only “attenuates decision biases associated with System 1” (Hayakawa et al., 2017, p. 1396). Thus, the literature paints a complex picture of the Moral Foreign Language effect with a possible interplay of System 1 and System 2 as it modulates the former and relates to the access of normative knowledge.
Setting Boundaries: FLe as Emotion Driven or Logical Buffer?
With the supposed account of FL potentially prompting deliberate and thus mostly rational thinking, further research investigates its generalizability to a then seemingly compatible domain – logical paradigms or biases correlated with cognitive ability. Parallel to the findings of the MFLe, however, the research leads to an index of clear boundaries as the FLe fails to emerge in logical paradigms devoid of direct emotional components (Costa, Foucart, Arnon, Aparici, & Apesteguia, 2014; Vives, Aparici, & Costa, 2018). Costa et al. (2014) utilize the Cognitive-Reflection test “designed to assess…the ability to suppress an incorrect intuitive answer triggered by System 1 [to a] logical answer [by] System 2” (Costa et al., 2014, p. 249). Six hundred and thirty Spanish- English and English-Spanish bilinguals (M age = 20.1; M age = 20.6, respectively ) answered three questions such as: If 5 machines take 5 minutes to make 5 keyboards, how long would it take 100 machines to make 100 keyboards? The results show the number of correct answers and intuitive/incorrect answers did not significantly differ for NL and FL language conditions in both groups; Spanish-English Bilinguals (NL = 64%; FL = 58%) and English-Spanish Bilinguals (NL = 53; FL = 58) incorrectly answered the above question.
Logic driven bias
Vives et al. (2018) center on three core biases: the outcome bias, and the conjunction and base-rate neglect fallacies within 6 experiments and more than 2000 participants. Participants were recruited from Spaniard universities and were Spanish-English bilinguals who acquired their FL, English, in a classroom setting. The outcome bias involves judging decisions as appropriate if they result in a desirable outcome although knowledge of the outcome is unavailable during decision-making. The conjunction and base-rate neglect fallacies relate to the representative heuristic in which probability judgments rely on similarity to a target population (e.g., stereotypes) thereby neglecting other critical information. Furthermore, the conjunction fallacy considers the presence of two events to be more probable than one of its constituents while the base-rate fallacy fails to consider a populations’ base rate.
Examples of these paradigms included: the assessment of personal, hypothetical investment decisions after a monetary gain or loss; the probability of someone suffering a heart attack or someone suffering a heart attack and being older than 55 years old; ascribing a person’s profession (e.g., lawyer) based upon the presentation of congruent (i.e., stereotypical) or incongruent information and a given population base-rate (e.g., 90% lawyers and 10% engineers). The results indicate participants were all susceptible to these biases regardless of language context. Interestingly, the authors’ sixth study involving an emotional (e.g., cancer) base-rate fallacy versus a neutral version had no language effect nor interaction with emotion and congruency conditions. This suggests the FLe likely hinges on an emotionally based decision-making context and is thus inaccessible in a mere emotional addition to a mostly logical context.
These findings bolster the notion that FLe boundaries entail a limitation to “contexts in which emotionality is a key factor driving such biases” (Costa et al., 2014, p. 252). Vives et al. (2018) remain partially wary of this due to absent support in an emotion-addition paradigm and certain regret ratings, but we must note emotion was not at the forefront. Conversely, Costa et al. (2014) replication of the FLe in various emotionally rooted contexts such as differential framing paradigms and ambiguity aversion attest to its generalizability within its boundaries. In terms of its mechanism, it seems the context must begin with preferential processing by System 1 that an FL then attenuates, leaving a weakened System 1 to interface with a rational System 2.
Ivaz, Costa, and Dunabeitia (2015) attest to the FLe specifically on System 1 processes through the well-established self-bias effect. The self-bias effect is the tendency to perceive self-related stimuli as more salient, prompting a shift in one’s perceptual focus to “prioritize the processing of self-related stimuli over others” (Ivaz et al., 2015, p. 5). The nature of this self-prioritization mirrors the natural affiliation we have for personally related phenomena and thus embodies automatic, emotional processing with little linguistic demands. In consideration of such, the self-paradigm utilized involves the simple pairing of geometric shapes (i.e., circle, triangle, and a square) with written labels varying in self-relatedness for each NL or FL condition (e.g., English – “you,” “friend,” and “other”); Two groups of thirty-nine native Spanish speakers with high proficiency in English (FL) were split roughly in half to complete the task in their NL or FL. An additional group of 48 Spanish-English bilinguals (M age = 24.08; 35 females) completed a single multilingual version combining the stimuli with their respective associations in both languages. Both experiments contained an even division of randomized mismatching and matching trials concerning learned associations between shapes and linguistic tags. In turn, response times were measured as a function of participants’ accuracy.
The results illustrate a self-bias effect twice as large in the native language (114ms vs 54ms faster) and thus point to its significant reduction in latency and accuracy in the FL. The multilingual version corroborates this finding albeit to a lesser extent and only digresses to establish a familiar effect, the faster response to familiar than other stimuli for both NL and FL. As Ivaz et al. (2015) state “[this] reveal[s]…highly automatic emotional reactions are lessened” and thus suggesting a reduced emotional reactivity to the FL stimuli (p. 17).
Affective Valence Account
There is substantial support that the FLe corresponds to affective processes and its modulation via the blunting of emotion – the latter being insinuated or empirically supported. However, Hadjichristidis, Geipel, and Savadori (2015) pose that rather than an elusive blunting of emotion, the FLe influences an affective valence containing both positive and negative emotions. In one of their experiments, 123 participants (M age = 25.33) rated 26 specific hazards (e.g., airplane travel) with regards to risk and benefit to Italian society in the FL (English) or NL (Italian) context. Furthermore, participants provided ratings of their positive and negative feelings toward the presented hazards on a 5- point scale. The main findings yield lessened perceptions of risk in the FL (M = 3.96) compared to the NL (M = 4.22) and similar ratings to benefit judgments. Complementarily, there was a significant increase in negative feelings experienced and an amplification of positive feelings in the FL.
Fitting with the aforesaid FLe boundaries and its pinning to affective processes, Hadjichristidis et al. (2015) refute the notion of an FL prompting analytical thinking. Indeed, the exceptionally high and stable correlations between risk-benefit judgments and positive-negative feelings support the predominant involvement of System 1 than System 2 processes. The study’s cornerstone, however, lies in that the resulting “net increase in positive affect mediated the [FLe]…a general indication that [FL] sways the balance of feelings [positive and negative] toward the positive side” (Hadjichristidis et al., 2015, p. 124). Contrary to the past focus on an FL attenuation of negative emotions, a more emotionally complex picture arises in which an FL may trigger a positivity bias and/or reduce a negativity bias. That is, FL may preferentially activate positive over negative associations tied to a positively tinged FL acquisition climate (e.g., academic). It may also reduce the ratio and/or relative weight among positive and negative associations when inclined to weigh negative events more than positive ones (negativity bias).
Cognitive Control Account
Although FLe support gains traction in its replicability and role in emotional processing, there is competing empirical evidence that denounces a “pure” FLe phenomenon. Oganian, Korn, and Heekeren (2016) fail to replicate the FLe in the classical framing paradigm of the Asian disease problem given to 744 native German speakers with English or French as their foreign language. Experiment two entails the second group of 420 German bilinguals who received the instructions in German (NL) or English (FL) followed by the Asian disease problem in a language congruent (same language, no switch) or incongruent (switch) condition. The significant interaction between switch and frame reveals language switching minimized framing effects irrespective of the actual language utilized. In the English language baseline, the choice of a sure option increased in a loss frame (33% vs 45 %) and decreased in a gain frame (68% vs 57%) among corresponding no switch and switch conditions. The reduction of framing effects – risk aversion in a gain frame and risk-taking in a loss frame - after a language switch although to a lesser extent into L1 suggests: “[This] results from [a] transient change in the level of cognitive control…not from general [NL and FL] processing differences” (Oganian et al., 2016, p. 146). Given the methodological power inherent in the use of various tasks with a comparable bilingual population, the study exalts the need to survey other contextual factors related to FL use and types of decision-making.
Affective (Emotional) Processing in Bilinguals: What Factor(s) Matter?
A discussion on contextual factors affecting the dynamics of FL use and its application is at the heart of the larger body of research that the FLe relates to. That is, the continuous study on the affective processing of bilinguals and multilinguals pointing to a premise of “disembodied” cognition via reduced emotionality in an FL. However, this phenomenon is hardly straightforward and is subject to debate with numerous inconsistencies and other contingencies. There is also the yet to be fully discussed question of how and why an FL may become less emotionally entrenched in comparison to an NL or vice versa. Caldwell-Harris (2015) and Pavlenko (2012) pattern a complex picture of bilingual and multilingual emotional processing, alluding to a myriad of contributing factors, related hypotheses, and implications.
Bilingual Heterogeneity and Language Embodiment
To begin, definitions pertinent to bilingual and multilingual groups must be established and are mostly derived from Pavlenko (2012). Conceptually, Bilingual and multilingual speakers are those who use two or more languages simultaneously or sequentially (e.g., immigrants). Bilinguals and multilingual populations can be further broken down by order, age, and context of language acquisition, language dominance, and levels of language proficiency. Order of language acquisition follows a chronological approach which may be sequential (i.e., first being L1, second as L2, etc.) or simultaneous (i.e., both in childhood). The distinction between an FL and general second language (L2 or LX) speaker is that the former learns the L2 in a classroom devoid of a naturalistic or mixed context. Age of acquisition (AoA) refers to the age at which L2 or LX learning begins and fall into categories of simultaneous, childhood/early, or late bilinguals. The context of acquisition (CoA) involves FL language or instructed context, L2 or naturalistic contexts, and mixed contexts. Language proficiency, as evident in the FLe studies, is commonly examined via standardized proficiency tests and self-reports. As endpoints along a continuum, balanced bilinguals are those with relatively similar skills in acquired languages whereas dominant bilinguals have greater ease and access to one such language. While the dominant language may also be the most proficient, both may be variable across areas of language use. Relatedly, a “shift in use and dominance in L2, accompanied by declining use and inhibition of the L1” may lead to L1 attrition whereby disfluency, access, and other linguistic components suffer (Pavlenko, 2012, p. 408). Lastly, Pavlenko (2012) forewarns treating an L1 as the NL due to L2 influence and thus overgeneralizing despite differential language trajectories.
In the purview of inconsistent findings on reduced emotionality, the literature points to enhanced reliability when “L1 is the dominant language and L2 (or LX) is a later-learned and less proficient language.” (Caldwell-Harris, 2015, p. 215). This seems to at least partially stem from differences in the CoA whereby naturalistic contexts pose heightened opportunities for various types and intensity of exposure as well as authentic interactions. The idea then of emotional grounding through a naturalistic context is further enforced with coinciding crucial and memorable developmental points in an early AoA. Relatedly, the lack of emotional grounding in a later learned L2 or LX may render its negative words as decontextualized and emotionally distant. Positive words do not fare similarly given social interactions in adulthood and/or an FL instructive context are positively tinged (Caldwell-Harris, 2015). Pavlenko (2012) and Caldwell-Harris (2015) ermbody these phenomena with compatible theories of language embodiment and aceprecrued emotional resonance, respectively. The idea is languages “acquire both affective and auto-biographical dimensions” when it concurrently develops among these systems (Pavlenko, 2012, p. 421). While commonly linked to this scenario but certainly not limited to it, emotional resonances accrue upon language use and learning within emotional contexts. Thus, two bilinguals utilizing the same language with similar proficiency may experience different levels of emotionality such as when one uses an L2 primarily at work/school and the other in familial settings (Caldwell-Harris, 2015).
L1/ L2 Advantage and Malleability
The four factors mediating the perception of language emotionality and language choice for emotional expression are the order of acquisition, language dominance, AoA, and CoA. The primary acquisition of an L1 leads to higher ratings of emotionality, provide naturalistic or mixed contexts, and is more likely to be the chosen mode of expression for positive and negative affect. These factors coalesce into the L1 advantage whereby automaticity of affective processing increases and the L2 advantage in which this automaticity lessens. However, language embodiment is malleable to the inherent diversity and combination of these factors such as a shift in dominance, frequent L2 use within a naturalistic context, language attitudes, and CoA of an L2. In Pavlenko’s (2012) interpretive language processing, language attitudes -- such as an internalized low social status of a mother tongue or associated negative experiences to an L1 -- may influence emotionality (e.g., positively or negatively) and in choice for expression. The differences in affective properties of languages then rely on age effects (i.e., order of acquisition + AoA) and context effects (i.e., CoA, frequency of use, and language dominance). With this renewed focus and the need for material high in ecological validity, Pavlenko’s (2012) takeaway of “it depends” on emotionality differences calls for nuanced exploration. This is especially true for non-prototypical cases far from an L1/L2 dichotomy such as U.S. Spanish-English bilinguals with a myriad of language trajectories.
Patterning Trends of “Disembodied” Cognition Amongst U.S. Spanish-English Bilinguals
A comprehensive look into the language trajectories of early childhood Spanish-Bilinguals in the U.S. brings to life the age and context effects that may culminate into emotional differences. An analysis of factors such as the societal climate, embedded institutional influences (i.e., school), and its profound effects in language use and acquisition provides a baseline to better-attuned experiments especially in terms of design paradigms. An enhanced methodology bears both theoretical and applied interests, with the latter perhaps being most urgent. The undue infringement on the Spanish language, a core cultural element of a disenfranchised population, poses significant challenges within the community and to general minority issues that touch on emotion. Elucidating on heterogenous language trajectories and corresponding emotional allocation may abet social justice efforts and collectivism within the Latinx population.
Context and Age Effects
The context of acquisition for early childhood Spanish-English bilinguals centers upon interaction with the prominent institution of the U.S. educational system. As an extension of the sociopolitical climate, schools function according to notions of the Spanish language and its corresponding populations’ inferiority. Thus, these childhood bilinguals contend and ultimately oblige -- albeit to varying degrees -- to the pressing, normative influence towards English dominance and subsequent Spanish attrition. This phenomenon permeates generations of Latinx and embodies problematic divides and alienation within communal and familial dynamics, especially for children of first-generation immigrants (Worthy & Rodriguez-Galindo, 2006; Zentella 1997). As Zentella (1997) states, it begins with a “subtractive instead of [an] additive approach… [viewing] the standard English dialect…as a substitute for all [the children’s] varieties of Spanish and other nonstandard dialects of English” p. (123).
English Hegemony/Spanish Attrition
Even in ‘well-intended’ bilingual programs, the above ideology permeates and is paired with standard Spanish that is incongruent with the dialects of the diverse Latinx community. Zentella’s (1997) observation of generational language shift toward English dominance and monolingualism in East Harlem Latinx exalts the erosion of community cohesiveness and mutual trust. The diminished competence in Spanish leads to discordance whereby children miss out on multifaceted cultural transmission as they are at odds to communicate and bond with key transmitters in participatory activities. Overall, the hampered acquisition of Spanish due to rampant, institutional English hegemony “severs its vital connections to Latin[x] children’s cognitive, academic, and emotional growth” (Zentella, 1997, p. 126). This occurs despite the wishes of the Latinx parents and community at large to rightfully maintain the beneficence of bilingualism but unfortunately lack the resources to consummate.
Indeed, Worthy & Rodriguez-Galindo’s (2006) survey of Latinx immigrant parents on their children’s bilingualism and complementary observations of the students illustrate the conflict of bilingualism itself. The perspectives of these parents on bilingualism represent an ideal balance between acculturation and cultural/familial preservation as a double force to social and economic advancement. Yet, the study finds that in less than two years, six out of the thirteen now rising 7th graders showed significant Spanish erosion (i.e., vocabulary, fluency) and resistance. Despite strategies employed by the parents -- monitoring both languages’ proficiency, implementing an English-free home, and adopting roles as Spanish home-instructors -- most were unable to actively intervene given their all-consuming economic situations. Further research on the trajectory of these bilingual’s language acquisition and maintenance illustrates the potential outcomes of a language foundation skewed toward Spanish attrition.
Acquisition and Maintenance
Bohman, Bedore, Pena, Mendez-Perez, and Gilliam’s (2010) survey of seven hundred and fifty-seven Hispanic prekindergartners and kindergarteners reveal critical factors in initial language learning and subsequent linguistic knowledge. Regarding language traction, the authors find language output (i.e., production) was significant for both English and Spanish performance in semantic and morphosyntax domains. However, input (i.e., reception) only significantly predicted Spanish semantic and morphosyntax. This may reflect the delayed English input for certain sequential Spanish-English bilinguals up until school exposure and of which influences English performance. Complementarily, the association found between low SES, a proxy for ‘acculturation’, and satisfactory Spanish performance attest to early maintenance primarily among Latinx immigrants and their children. In terms of adding to language knowledge, output in each language led to higher scores on performance but input bore a complex picture as it was only correlated with English morphosyntax. Bohman et al. (2010) posit input here may not relate to the development of Spanish morphosyntax due to general early exposure and use thus leaving practice and exposure to propel English morphosyntax. Yet, the associations found between age, SES, and parent education with higher performance on English semantics and morphosyntax suggest certain simultaneous bilinguals’ generational English dominance via acculturation. This coupled with language output driving language maintenance lends to Montrul and Potowski (2007) on Spanish-English bilinguals in a dual-immersion school.
Montrul and Potowski (2007) compare the command of gender agreement in Spanish across monolingual children in Mexico (N = 29) and 6 to 11-year-old Spanish-English bilinguals in the U.S. The latter group was heritage speakers from Spanish-speaking homes (N = 28) and L2 learners of Spanish from English-speaking homes (N =28) who attended a two-way Spanish-English immersion school. Gender categorization and agreement are part of native language proficiency and are semantically and formally (e.g., phonology) based in Spanish with some nouns categorized as feminine or masculine (Montrul & Potowski, 2007). All groups completed an oral narrative task on gender determiners (i.e., little red riding hood retelling) and a puzzle task on gender agreement with adjectives. The results illustrate both groups of Spanish-English bilinguals perform less accurately than Spanish monolingual children (error rates of 5% on oral task and more than 30% on the puzzle task). Although Spanish L2 learners performed the least well, abundant error rates among heritage speakers reveal gender agreement is unreliable. Differences among heritage speakers arose in that sequential bilinguals (i.e., Spanish precedes English exposure) performed better on the gender agreement puzzle task than simultaneous bilinguals (i.e., concurrent Spanish and English exposure) who patterned with Spanish L2 learners. Although resembling findings on incomplete Spanish acquisition of adult, simultaneous Spanish-English bilinguals, the authors advance there are no robust signs of Spanish loss.
A dual immersion context promotes a pattern of sustained acquisition (i.e., older children performed better) and one of language maintenance in which simultaneous heritage speakers and L2 learners improve in accuracy with age while sequential bilinguals slightly decrease. Thus, gender marking and agreement are “affected in a language contact situation, but [the] degree of affectedness [relates] to [the] age of onset of bilingualism and input factors” (Montrul, 2007, p. 322). As such, Montrul and Potowski (2007) suggest simultaneous Spanish-English bilinguals may never fully trigger a gender agreement rule due to reduced Spanish input engendering incomplete Spanish grammars across the lifespan, thus alluding to the need for early support.
Curricula for bilinguals often involves English-only instruction or initial Spanish instruction as the precursor to English, with the latter rising due to increases in dense, Latinx populations. Duursma, Romero-Contreras, Szuber, Proctor, Snow, August, and Calderon (2007) illustrate how these options interact with language support within the home and subsequent vocabulary development in English and Spanish. Their investigation on predictors of Spanish and English among 96 fifth-grade Latinx English language learners show proficiency in Spanish and English necessitate differential supports (Duursma et al., 2007). In line with research on generational language maintenance, the authors find language preference at home relates to their choice in preferred language instruction and proficiency in both languages.
Paternal preference for English predicted both Spanish and English vocabulary, with it specifically influencing children receiving initial Spanish instruction to have higher scores on English vocabulary than those without such preference. In contrast, prediction for Spanish vocabulary among students with English instruction reveals both paternal and maternal language preference was significant. More specifically, children earned higher scores on Spanish vocabulary when both parents held a preference for Spanish despite English-only instruction. Personal English literacy supports significantly predicted English vocabulary for those receiving initial English instruction and might reflect parent’s positive attitudes toward education of the dominant language (Duursma et al., 2007). The finding that language preference for sibling interaction had a larger effect than parent preference on English proficiency while the reverse was true for Spanish underscores its need for intensified support. And thus, patterns with the overall fragility of Spanish support systems within the scope of family, school, and community (Duursma et al., 2007). Lastly, Latinx girls outperforming boys in vocabulary may reflect the motivations of Spanish maintenance for familial purposes while simultaneously valuing English attainment for educational success (Duursma et al., 2007). Jaramillo (1996) reflects this skewed gender responsibility in which Latinx women lead the shift towards more open behavior (i.e., tu use with extended kin) but simultaneously uphold older cultural traditions (i.e., usted). Thus, gender socialization and roles may moderate the need for robust instructional and social support.
Support needs in Spanish acquisition and maintenance of U.S. Spanish-English bilinguals are markedly evident yet the disproportionate risk of those with a simultaneous classification garners more attention. As such, Montrul (2006) assess potential Spanish language loss and English dominance via grammar proxies in Spanish-English simultaneous bilinguals (M age = 21.54; N = 18) as compared to matched Spanish and English monolinguals. Four experiments on the knowledge and processing of lexical-semantics and syntactic competence in English and Spanish followed Spanish and English proficiency tests. Although the perception of bilinguals’ linguistic skills matched proficiency tests (i.e., English dominance), the results illustrate syntactic knowledge of unaccusativity was strong in both languages. Montrul (2006) contends the early development of this knowledge (i.e., Before age four), unaffected by input fluctuation or literacy development during the school years, may underlie its resilience. While displaying differences with specific language constructions, simultaneous bilinguals exhibited similar linguistics representation as monolinguals in both languages and experienced longer processing times that contradict an English advantage. Furthermore, their performance with the English core and peripheral verbs patterned with English monolinguals but displayed the Spanish processing pattern. Overall, resilient knowledge within certain grammatical domains leads to Montrul’s (2006) conclusion: The universal aspects of language sustain language loss whereas other specific aspects of language maintain heightened vulnerability to late acquisition or errors.
Reclaiming Agency and Validity in Language Trajectories
Proficiency tests may obscure certain language maintenance (i.e., universal aspects) as well as perpetuate the superiority ascribed to “standard” language varieties. Standardized principles are incompatible with the language varieties Latinx subgroups possess and thus reflect unrecorded inquiry on their ‘disembodied’ cognition or incorrectly presented as unfavorable findings. As Showstack (2012) explains, monoglossic language ideologies center on “the assumption that any [variation] from the standard in linguistic repertoires of bilinguals represents impunity and have a profound impact on language education” (p. 3). Notions to correct “improper” Spanish of Latinx or overall denigration may engender their classification of the stigmatized varieties as a deficit or value and affect emotional grounding.
Ascribed illegitimacy curtails the symbolic power that affords cultural capital while also affecting categorization within or outside essentialized notions of Latinx identity tied to language (Showstack, 2012). A culmination of these factors underlies language maintenance and use and is evident among Showstack’s (2012) study’s population: Texan Spanish-English bilinguals enrolled in Spanish university courses for heritage speakers with motivations of maintaining cultural ties and/or resolving language/identity ambiguity. The findings illustrate how “in some contexts, [Spanish-English bilinguals] associated value with hybrid cultural experiences, but also presented the legitimacy of [such] linguistic repertoires [e.g., ‘Spanglish,” Chicanx varieties] as questionable” (Showstack, 2012, p. 12). Thus, the prevalence of a monoglossic language ideology led to dominant deficit perspectives for non-standard varieties and subsequent categorization outside of essentialized notions of Latinx identity. While some contend these ideologies by empowering differing contextual uses of Spanish and language contact (CoA), its prevailing reign attests to the need for increased consciousness. In a similar vein, research must incorporate the non-standard language varieties that merit respect and, in turn, boost validity.
At the center of prominent non-standard language varieties, U.S. English bilinguals employ forms of ‘Spanglish’ rooted in a code-switching phenomenon. Rather than deviant language mixing, this language variety embodies a sensible adaptation arising from tumultuous language contexts and its consequences. Heredia and Altarriba (2001) asses the nature and explanations for code-switching in which a word or a phrase in one language substitutes its counterpart in a second language (e.g., I want café). A switch from Spanish to English may reflect a “problem of retrieval affected by a combination of closely related factors such as language use and word frequency (Heredia & Altarriba, 2001, p. 165). Patterns of language dominance specifically relevant to this population may lead to asymmetrical language intrusion. In early bilingual stages, code-switching may involve intrusions of the first language during second language use but may reverse with a dominance shift. Nonetheless, a marked surge in the positive reclaiming of code-switching is evidenced by Montes-Alcala (2000).
Rather than reflective on language loss or subpar language proficiency, the college-educated Spanish-English bilinguals in this study ascertain it promotes a sense of representation and is a significant part of their identity. Furthermore, the study’s inquiry into intrasentential (i.e., language switch within a sentence) and intersentential (i.e., language switch after a sentence in one language) code-switching attest to its grammatic governance in both languages and complexity. That is, the more complex and elaborate intrasentential type had a higher prevalence in both written and oral modes even among those holding negative views of code-switching. Thus, the handicap imposed on Latinx code-switching communities obscures the unique and robust non-standard language varieties which carry an emotional resonance we must explore.
Conclusion and Future Directions
Amid a complex picture, the FLe appears largely constricted to emotion-driven contexts as related to the interface between an altered System 1 and System 2 processes. Yet, the FLe relies on a bilingual’s disembodied cognition and of which patterning varies according to age (i.e., order of acquisition, AoA) and context effects (i.e., CoA, frequency of use, language dominance). A nuanced focus considering such may propel appropriate inquiry of groups that carry convoluted language trajectories such as early U.S. Spanish-English bilinguals to understand the how and why of Pavlenko’s (2012) “it depends.” The literature on childhood U.S. Spanish-English bilinguals illuminates various interrelated factors to account for; notably, simultaneous vs sequential categorization, subtractive education, the minority status of Spanish and its speakers, acculturation pressures, and non-standard language varieties. In the creation of research materials gauging emotional grounding, the use of universal language aspects and high-frequency grammar structures may boost ecological validity. In the same vein, research paradigms should comprehensively integrate non-standard language varieties to correspondingly create better-attuned measures, pointed research questions, and identify subgroups. With such modifications, future studies may then employ the language(s) true to the core of bilinguals while remaining privy to the heterogeneity of language histories and trajectories within groups.
Bohman, T. M., Bedore, L. M., Pena, E. D., Mendez-Perez, A., & Gilliam, R.B. (2010). What you hear and what you say: Language performance in Spanish-English bilinguals. International Journal of Bilingual Education and Bilingualism, 13(3), 325-344.
Caldwell- Harris, C. L. (2015). Emotionality differences between a native and foreign language. Current Directions in Psychological Science, 24(3), 214-219.
Cipolletti, H., Mcfarlane, S., & Weissglasss, C. (2015). The moral Foreign-Language Effect. Philosophical Psychology, 29(1), 23-40.
Costa, A., Foucart, A., Arnon, I., Aparici, M., & Apesteguia, J. (2014). “Piensa” twice: On the foreign language effect in decision making. Cognition, 130(2), 236-254.
Duursma, E., Romero-Contreras, S., Szuber, A., Proctor, P., Snow, C., August, D., & Calderon, M. (2007). The role of home literacy and language environment on bilinguals English and Spanish vocabulary development. Applied Psycholinguistics, 28(1), 171-190.
Geipel, J., Hadjichristidis, C., & Surian, L. (2015). The Foreign Language Effect on Moral Judgement: The role of emotions and norms. Plos One, 10(7), 1-17.
Hadjichristidis, C., Geipel, J., & Savadori, L. (2015). The effect of foreign language in judgements of risk and benefit: The role of affect. Journal of Experimental Psychology: Applied, 21(2), 117-129.
Hayakawa, S., Tannenbaum, D., Costa, A., Corey, J. D., & Keysar B. (2017). Thinking more or feeling less? Explaining the Foreign-Language Effect on moral judgement. Psychological Science, 28(10), 1387-1397.
Heredia, R. R., & Altarriba, J. (2001). Bilingual language mixing: Why do bilinguals code-switch? Current Directions in Psychological Science, 10(5), 164-168.
Ivaz, L., Costa, A., & Dunabeitia, J. A. (2016). The emotional impact of being myself: Emotions and foreign-language processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 42(3), 489-496.
Jaramillo, J. A. (1996). Tu and Usted: Address etiquette in the Mexican American family. Hispanic Journal of Behavioral Sciences, 18(4), 522-532.
Keysar, B., Hayakawa, S. L., & An, S. G. (2012). The Foreign-Language Effect: Thinking in a foreign tongue reduces decision bias. Psychological Science, 23(6), 661-668.
Montes-Alcala, C. (2000). Attitudes toward oral and written code-switching in Spanish-English bilingual youths. In A. Roca (Ed.), Research on Spanish in the United States: Linguistic Issues and Challenges (pp. 218-227). Sommerville, MA: Cascadilla Press.
Montrul, S. (2006). On the bilingual competence of Spanish heritage speakers: Syntax, lexical-semantics, and processing. International Journal of Bilingualism, 10(1), 37-69.
Montrul, S., & Potowski, K. (2007). Command of gender agreement in school-age Spanish-English Bilingual children. International Journal of Bilingualism, 11(3), 301-328.
Oganian, Y., Korn, C. W., & Heekeren, H. R. (2016). Language switching – But not foreign language use per se – Reduces the framing effect. Journal of Experimental Psychology: Learning, Memory, and Cognition, 42(1), 140-148.
Pavlenko, A. (2012). Affective processing in bilingual speakers: Disembodied cognition? International Journal of Psychology, 47(6), 405-428.
Showstack, R. E. (2012). Symbolic power in the heritage language classroom: How Spanish heritage speakers sustain and resist hegemonic discourses on language and cultural diversity. Spanish in Context, 9(1), 1-26.
Vives, M.-L., Aparici, M., & Costa, A. (2018). The limits of the foreign language effect on decision-making: The case of the outcome bias and the representativeness heuristic. Plos One, 13(9). 1-14.
Worthy, J., & Rodriguez-Galindo, A. (2006). “Mi hija vale dos personas”: Latino immigrant parents perspectives about their childrens bilingualism. Bilingual Research Journal, 30(2), 579-601.
Zentella, A. C. (1997). Latino youth at home, in their Communities, and in School. Education and Urban Society, 30(1), 122-130.