In the past few decades, research has shown that bilingual individuals shift their personality when language is used as a prime; and that they change their personality in accordance to the personality differences associated to the culture of that language (Ross et al., 2002; Ramírez-Esparza et al., 2006; Chen and Bond, 2010; Chen et al., 2014). For example, Ramírez-Esparza et al. (2006) asked Spanish-English bilinguals of Mexican descent to participate in two language sessions to trigger two language modes. In the Spanish session, the participants were interviewed in Spanish and then completed the Big Five Inventory in Spanish (Benet-Martínez and John, 1998). The same procedure was repeated in the English session and the participants responded to the Big Five Inventory in English (John and Srivastava, 1999). Results demonstrated that bilinguals were more extraverted, agreeable and conscientious in English than in Spanish. Furthermore, these differences were consistent with the personality displayed in the culture associated with each language. That is, U.S. Americans also scored higher than Mexicans in Extraversion, Agreeableness, and Conscientiousness.

Ramírez-Esparza et al. study (2006) demonstrated that bilinguals shift their personality depending on the language mode, and that they switch their personality to match the personality of the two cultures they identify with. However, the results also raised other questions. For example, it was puzzling that personality differences between U.S. Americans and Mexicans were inconsistent with well-known stereotypes about these cultures. Mexicans are polite and kind; they show respect toward others, avoid conflict by emphasizing positive behaviors and deemphasizing negative behaviors. Cultural scientists have used the cultural script Simpatía to label this kind of social interaction among Mexicans and Latinos (Triandis et al., 1984; Díaz-Loving and Draguns, 1999; Holloway et al., 2009). However, when Mexicans and bilinguals respond to the questionnaire in Spanish, they saw themselves as less agreeable than U.S. Americans and bilinguals, when responding to the questionnaire in English. What could account for this paradoxical finding? Ramírez-Esparza et al. (2006) hypothesized that bilinguals expressed changes in their cultural values on self-reports, as opposed to changes in their personality. In other words, response to self-reports interacts with cultural values in bilinguals. Since an underlying characteristic of the cultural script Simpatía is modesty, when bilinguals read a question in Spanish, the modesty value is activated. Individuals' need to sound modest leads them to diminish their standing on agreeableness and other traits (see also Ramírez-Esparza et al., 2008).

Besides the discrepancies noted above between self-reports and behavior with Mexicans, self-reports have broader limitations that are worth discussing. First, the usage of these instruments usually follows the “imposed etic” approach. This strategy operates by administering measures that were developed under a particular cultural worldview that is then implicitly superimposed on members of different cultural groups. While this facilitates the comparison of a construct across cultures, it might also leave out other important constructs that are specific to the cultural group being examined (Benet-Martínez and Waller, 1997). Second, another limitation involves the translation of assessment materials. It is often incorrectly assumed that the act of directly translating measures to another cultural group's language makes the construct valid for them. In other words, translation of instruments can be correct from a linguistic point of view, but the constructs themselves might not resonate psychologically with the members of the cultural group being analyzed (van de Vijner and Tanzer, 2004).

In light of these important challenges, it is worth asking what Mexican-American bilinguals have to say, about themselves, using their own words in each language. By using this approach, we can capture language specific domains of self-definition and bypass the methodological cultural biases that result from using self-reports. Ramírez-Esparza et al. (2006, 2008, 2014) and Ramírez-Esparza and García-Sierra (2014) have proposed that to collect and analyze culture- and language-specific domains, it is important to use alternative methods. In our current study, we use one such alternative approach. That is, we allow individuals to spontaneously express their values, thoughts, and feelings associated to their most salient self-schemas when describing their personality.

Ross et al. (2002) conducted a study in which they randomly assigned a group of bilingual Chinese-Canadians to describe their personality in an open-ended way in two different languages: English and Chinese. Their goal was to test the assumption that language prompts shifts in cultural frames; for example, when asked to describe their personality in Chinese, participants would be inclined to define their personality more frequently in terms of their relationships with others, their cultural background, and their group memberships, with much less focus on themselves and their traits, as opposed to English. They asked two independent coders to rate how much these four themes appeared in the self-descriptions. The authors indeed found that in Chinese, participants mentioned their cultural background more, they had increased references to others, more group membership references and less statements about themselves, as opposed to English.

Although, Ross et al.'s study (2002) used an alternative measure to extract self-schemas, it has some limitations. First, bilingual participants were assigned either to two independent conditions (English language description or Chinese language description). This does not answer the question of a possible shift of self-schemas within the same individual. Also, the authors were set to look for 4 pre-determined self-schemas in the self-descriptions (references to others, cultural background, group memberships and to oneself). This leaves out other self-schemas that naturally emerge and that might fit in the definition of independent and interdependent selves (e.g., focus on individual achievement vs. relational goals). Moreover, the richness of open-ended data is lost if the goal of the researcher is simply to look for the occurrence of a particular self-schema that is only known to him or her. A final limitation has to do with its usage of the traditional method to analyzing open-ended text. In this approach, independent coders are trained to look for particular themes in the text; this requires time and energy to develop complicated coding schemes and it is not exempt of biases (Chung and Pennebaker, 2008).

In this study, we address these gaps by analyzing the open-ended self-descriptions of a sample of Mexican-American bilinguals in two different languages at two different moments in time, to tap into intra-individual variability. The data is analyzed using an innovative text analytic strategy called the Meaning Extraction Method (MEM; e.g., Chung and Pennebaker, 2008; Boyd, 2015; Stanton et al., 2015; Pulverman et al., 2016). This text analytic approach uses factor analyses to observe how groups of content words cluster together. Each cluster can be thought as an underlying self-schema that guides word choice when describing one's personality. These coherent word clusters will suggest latent dimensions, which will be analyzed qualitatively to define and name the factors. This has been called the inductive-qualitative approach (Chung and Pennebaker, 2008). Relevant to this study, self-schemas have been successfully defined with monolinguals U.S. American participants (Chung and Pennebaker, 2008) and in both monolingual U.S. American participants and monolingual Mexican participants (Ramírez-Esparza et al., 2012).

More explicitly, Ramírez-Esparza et al. (2012) collected open-ended personality descriptions in the same way we intend to do in this study, from a sample of 560 English-speaking U.S. Americans and 496 Spanish-speaking Mexicans. The results from the MEM showed that 8 factors for U.S. Americans and 6 factors for Mexicans brought together a group of content words that are psychologically meaningful and coherent. The authors used previous findings with U.S. American participants (i.e., Chung and Pennebaker, 2008) and an inductive-qualitative approach to name the factors and to observe which factors were found only in one culture (i.e., culture-specific self-schemas) and which ones were similar across cultures (i.e., cross-cultural self-schemas). The results from these analyses showed that some factors were comparable across cultures (e.g., Values, Sociability, and Emotionality). Although, emotions in the U.S. Americans were reflected in two separate factors (i.e., Positivity and Negativity), they appeared together as a single emotional factor in the Mexicans (i.e., Emotionality). Other factors were similar, but were labeled differently because U.S. Americans talked more about their daily activities (i.e., Daily Activities) and Mexicans talked about their specific interests (i.e., Hobbies). Two culture-specific self-schemas were found for Mexicans (i.e., Relationships and Simpatía), and three in the U.S. Americans (Fun, Existentialism, and College Experience). We use findings from this particular study to shape our expectations on the kind of self-schemas that can emerge in Mexican-American bilinguals.

In this study we explore Mexican-American's self-schemas when describing their personality to an open-ended question in English vs. in Spanish. In order to achieve this goal, first we use the MEM, and then we rely on both previous studies (i.e., Chung and Pennebaker, 2008; Ramírez-Esparza et al., 2012) and a inductive-qualitative approach to formulate expectations, name the factors and to observe if there are factors that are specific of a language (i.e., language-specific) and if there are factors that are found across languages (i.e., cross-language). We expect that the language mode will prime individuals to shift their self-schemas to match the cultural values associated with that language. When bilinguals describe their personality in English, their U.S. American cultural self-schemas will emerge, whereas when they describe their personality in Spanish, their Mexican cultural self-schemas will be most salient (i.e., Ramírez-Esparza et al., 2006).

A secondary goal of this investigation is to observe if the resulted self-schemas in English are comparable to the self-schemas in Spanish. In order to achieve this goal we use a quantitative approach following guidelines found in Ramírez-Esparza et al. (2012). Essentially, this approach allows us to examine the degree to which dimensions in one language relate with dimensions in the other language. If they relate, this means that the dimensions are meaningful in both languages and are not language-specific. If they do not relate, this means that the dimensions are meaningful in only one language. Using this quantitative approach Ramírez-Esparza found that some dimensions were culture-specific (that is, they were only relevant for Mexicans or U.S. Americans) and other dimensions were cross-cultural (that is, they were relevant for both cultures). We expect that for the bilinguals we will also find some dimensions that are language-specific and others that are relevant across languages.

Bilinguals in the English mode are expected to show a more independent self-construal characterized by themes related to being sociable, and achievement-oriented (Markus and Kitayama, 1991). For example, Ramírez-Esparza et al. (2012) found in the factor Fun words such as outgoing, fun, hang, and girls (see also Chung and Pennebaker, 2008). Moreover, in the dimension College Experience, the U.S. Americans used words such as clean, organized, college, intelligent. Furthermore, we expect that in the English mode, bilinguals will describe themselves in more abstract and global ways, in accordance with the independent self-construal (Markus and Kitayama, 1998). For example, Ramírez-Esparza et al. (2012) and Chung and Pennebaker (2008) reported a dimension that was labeled as Existentialism and included words such as traits, attitude, hope, understand, personality.

In contrast, bilinguals in the Spanish mode will reflect themes related to an interdependent self-construal, characterized by an emphasis on relationships and significant others (Markus and Kitayama, 1991; Triandis, 1995). For example, Ramírez-Esparza et al. (2012) found more interdependent themes in the Mexican participants; the Relationships factor included words such as parents, family, friend, boyfriend with everyday activities such as career, university, school, eat.

We expect that bilinguals will show different patterns of emotional expression in their descriptions depending on the language mode. In the English mode, bilinguals will either use negative emotion words or positive emotions words, while in the Spanish language mode, bilinguals will show both positive and negative emotions within a single self-schema. This hypothesis would be consistent with the idea that individuals from interdependent cultures tend to display increased emotional complexity (i.e., co-occurrence of positive and negative emotions), compared to individuals of independent cultures (Spencer-Rodgers et al., 2010; Grossmann et al., 2016). For example, Ramírez-Esparza et al. (2012), reported that for U.S. Americans emotions were separated in either negative self-schemas (e.g., they used words like upset, angry, mad) or positive self-schemas (e.g., they used words like happy, laugh, good). Conversely, in the Mexicans only one self-schema emerged that contained both negative emotion words (e.g., hate, sad) and positive emotion words (e.g., cheerful, happy).

We expect that the Simpatía schema will be salient only in the Spanish mode (Triandis et al., 1984; Díaz-Loving and Draguns, 1999; Ramírez-Esparza et al., 2008; Holloway et al., 2009). This would be consistent to what Ramírez-Esparza et al. (2012) reported: Mexicans used words like kind, noble, affectionate, honest; but U.S. Americans did not use words that reflect Simpatía.

Other self-schemas that were found in Ramírez-Esparza et al. (2012) that were comparable across cultures but slightly different were Daily Activities for the U.S. Americans and included words as play, sports, years, school, day, stay, whereas the Hobbies for the Mexicans included words such as music, read, travel, movies, parties. Therefore, it is possible that bilinguals in the English mode they will talk more about their daily activities and in Spanish they will talk more about their hobbies.

Finally, we expect to find dimensions that will emerge both in the English and Spanish mode. This will be consistent with the idea that bilinguals are affected by norms, environment, and institutions of the context they are immersed (e.g., Rentfrow et al., 2008). For example, Ramírez-Esparza et al. (2012), found that Values was a meaningful self-schema for both U.S. Americans and Mexicans. U.S. Americans used words such as family, life, God, respect; similarly Mexicans used words such as learn, work, money, life, God.

The initial sample consisted of 310 bilingual participants, which self-identified as bilingual and Mexican-American. Out of these initial sample, 236 completed participation in all phases of the study. Out of these participants, 74 were not included in the study because they completed either the essay in English only (n = 44) or in Spanish only (n = 30). Participants were additionally excluded on the basis of not identifying as Mexican-American; failure to show an adequate level of Spanish-English bilingualism; and whose self-descriptive essay did not have a total word count of at least 50 words, which suggests that they had not taken the 15-min self-description writing task seriously. The final sample comprised of 193 bilingual participants (44 men and 149 women) with a mean age of 23.43 (SD = 4.92) years old. Participant's self-identified socioeconomic status was 62.10% working to lower-middle class, 27.40% middle-class, and 9.50% upper-middle to upper class. Per our sampling, 100% of participants self-identified as bicultural (Mexican living in the US or born in the US from Mexican descent).

Seventy-four percent (n = 142) of participants indicated that they have been born in the United States, while the remaining 26.40% (n = 51) indicated Mexico as their country of birth. All participants indicated a level of confidence of at more than 90% regarding hearing, speaking and reading in both English and Spanish.

Participants resided in various parts of the United States and were recruited to participate via online postings in different outlets, such as Latino and/or Chicano studies centers, flyers, and by sending emails to a wide audience that self-identified as Latino within a university located in southwestern U.S. Participants were offered $20 in gift cards from a popular online store in exchange for participation. Prospective participants were sent a link to an online bilingualism assessment and a demographic questionnaire. This was done as a first step, to ensure participant's eligibility in terms of language proficiency. Participants were included in the study if their self-identity matched inclusion criteria and if they indicated at least 75% confidence in reading, hearing and speaking in both languages. As a second step, eligible participants were sent a link to either one of the two parallel online questionnaires that were created in both English or Spanish to establish the language modes. This was done in a counterbalanced way (e.g., participant 1 was sent the survey in English, while participant 2 was sent the survey in Spanish).

Thirdly, participants that completed the first survey waited for approximately a week until they received the second survey. The language of the second survey depended on the language of the first administration.

After reading a description of the study and giving their consent, participants were asked to describe their personality for 15 min. The instructions for the English personality description were as follows:

Both online questions had the same physical appearance, the only difference was the language in which they were presented. A 15-min timer was presented on the writing web page, which included a large blank text field for typing. Each of the self-descriptions was formatted as a single plain text file. A spell-check was performed for each text file. The mean word count was 186.30 (SD = 111.27) for the English text files and 157.67 (SD = 92.00) for the Spanish text files.

Table 1 shows the Top 40 most frequently used content words by the bilinguals in both English and Spanish. Note that many of the Top 10 most frequent words in the English text files were also within the Top 10 most frequent words in the Spanish text files (i.e., time, family, work, life, friend, and learn). Interestingly, in the essays in English, bilinguals used a larger amount of words about being outgoing and sociable (e.g., social, outgoing, meet, friendly, quiet, shy); whereas in Spanish, they used more words related to their roles (e.g., education, university, work, woman), hobbies (e.g., read, travel, study) and bicultural experience (Spanish, Mexico, culture).

A principal components extraction with varimax rotation was performed independently on the words from the English self-descriptions and on the words from the Spanish self-descriptions. Factors were extracted, based on a scree of Eigenvalues for the principal components (Cattell, 1966). Factors at the elbow bend that had Eigenvalues above 1, and provided a significant increment to the cumulative percent variance were selected. As in Chung and Pennebaker (2008) and Ramírez-Esparza et al. (2012) studies, words with factor loadings ≥ |0.20| or higher were retained.

The first 6 factors accounted for 14.5% of the total variance. As can be seen in Table 2, 6 factors bring together a group of content words that are psychologically meaningful and coherent. Four factors were comparable to those found by Ramírez-Esparza et al. (2012) in monolingual U.S. Americans: Factor 1 (College Experience), which included words such as class, university, health, job, and competitive; Factor 2 (Values) which included words such as belief, money, country, pride; Factor 4 (Existentialism) includes words such as important, future, responsible, dream; and Factor 6 (Sociability) which includes words such as friendly, social, group, introvert.

Two factors were unique to the bilinguals and were not comparable to the U.S. American monolinguals from Ramírez-Esparza et al. (2012): Factor 3 (Bicultural Identity), which includes words such as culture, Mexican, American, language; and Factor 5 (Daily difficulties) which includes words such as hard, stress, afraid, and words such as organize, strive, time, perfectionist, stop.

The first 5 factors accounted for 14% of the total variance. As shown in Table 3, the 5 factor solution yielded a set of coherent dimensions. Five factors were comparable to those found by Ramírez-Esparza et al. (2012) in monolingual Mexicans: Factor 1 (Relationships) includes words related to relationships, and everyday activities, such as father, daughter, mother, siblings, education, work, write student, goal; Factor 3 (Values) includes words such as improve, change, open, understand, friends, chat, psychology; Factor 4 (Simpatía), which included words such as honest, affectionate, friendly, sincere, humble; and Factor 5 (Hobbies) which included words such as music, read, dance, study, travel.

One factor was unique to the bilinguals and was not comparable to the Mexican monolinguals from Ramírez-Esparza et al. (2012), but was similar to those found in the English self-descriptions was: Factor 2 (Bicultural Identity) and included words such as American, Mexico, culture, live, identify, countries, roots.

Even though this paper uses an inductive-qualitative approach to extract and name self-schemas within the essays by language, a secondary goal of this investigation was to use a quantitative approach. This approach was used to better understand the degree to which self-schemas are language-specific (i.e., they are only relevant in that language) or interdependent (i.e., they are relevant in both languages). In order to achieve this goal we used the Linguistic Inquiry Word Count software (LIWC2015; Pennebaker et al., 2015). LIWC2015 counts the words that belong to a particular category of words within text files. A user-defined dictionary directs LIWC2015 as to which words or categories of words to search for. In order to quantify if word usage is comparable across languages, we used what we called the quantitative approach (also see Ramírez-Esparza et al., 2012). This approach consisted in using a translation process to be able to relate dimensions from the Spanish language with the English language and vice versa. Specifically we performed the following four steps:

LIWC2015 program was used to quantify the percentage of words that belong to each factor that appear in a given essay. In order to do this, a dictionary was created where each word that belonged to a factor comprised a single LIWC2015 category. For example, the dictionary would include a category called College Experience and each word within the category (e.g., university, finish, job, school) would become part of this dictionary “College Experience.” This step was done for each of the resultant schema themes in both languages.

Table 4 shows the correlations between the English dimensions and the Spanish dimensions. Correlations from English text files are shown in the top half of the table; correlations from the Spanish text files are shown in the bottom half of the table. Only those correlations that have moderate or above effect sizes are bolded (r ≥ 0.30; Cohen, 1988). Positive bolded correlations between the factors and the translated versions of the factors indicate that there is a relation across languages or more interdependence. In contrast, weaker correlations indicate that factors are language-specific or more independent; and they are only relevant in that language.

We speculated that bilinguals in the English mode would show a more independent self-construal characterized by themes related to being sociable and achievement oriented (Markus and Kitayama, 1991); and they would also describe their personality in more abstract ways (Markus and Kitayama, 1998). In accordance with our expectations, in the factor Sociability the bilinguals used words such as introvert, gathering, quiet, social, group, shy. This demonstrates that when bilinguals are in their English mode they think about being sociable, outgoing, proactive and about maintaining interactions with others. This feature is common within an independent self-construal, where interactions with others are driven by a need for individuals to be leaders and assert themselves in front of others (Ramírez-Esparza et al., 2006). This factor is very similar to the ones reported by Ramírez-Esparza et al. (2012) and Chung and Pennebaker (2008) in monolingual U.S. American participants.

Furthermore, the College Experience dimension presents a series of words that are overall oriented toward achievement (e.g., university, finish, attention, work, competitive). This theme falls in line with more individualistic ideals, where a person must challenge him or herself and live up to his or her full potential (Markus et al., 2006). This factor resembles closely the Education factor reported by Chung and Pennebaker (2008) and to the College Experience factor reported by Ramírez-Esparza et al. (2012) in monolingual U.S. American participants.

Finally, the Existentialism factor provides additional insights for an independent self. As Markus and Kitayama (1998) noted, through the lens of an independent self, individuals tend to think about themselves in more abstract and global ways. This is reflected by words such as role, important, strong, public, close, dream, passionate. Chung and Pennebaker (2008) and Ramírez-Esparza et al. (2012) reported a similar factor in monolingual U.S. American participants.

Conversely, we hypothesized that bilinguals in the Spanish mode would reflect a more interdependent self, by valuing social relationships and connection to others (Markus and Kitayama, 1991). Our results provide support to this hypothesis. On the Relationships dimension, bilinguals mentioned their relationships (i.e., they used words such as father, siblings, mother, parents, relationships) in hand in hand fashion with their everyday activities (i.e., they used words such as, house, work, university, write, goal, achieve, career, school, student), just like the Mexicans (see Ramírez-Esparza et al., 2012).

We expected that bilinguals would show differential patterns of emotional expression according to the language mode. Following Ramírez-Esparza et al. (2012), words denoting negative and positive emotions would emerge in separate dimensions in English, while in Spanish, these words should appear within a single self-schema. In this study no clear positive or negative valenced dimensions emerged. One of the possible reasons why a not clear positive-valenced and/or negative-valenced factor emerged as in Ramírez-Esparza et al. (2012), is because the bilinguals gave priority to describing their thoughts and feelings about being biculturals. In this sample, we find that in both the English mode and the Spanish mode, bilinguals described their experience of being biculturals, which we will discuss more in detail below.

We expected that a Simpatía schema would emerge only when bilinguals describe their personality in the Spanish mode. Our MEM results confirmed this expectation. Importantly, the quantitative approach demonstrated that Simpatía was only relevant in the Spanish language, further supporting the idea that this dimension becomes only relevant when bilinguals are in the Spanish language mode.

The Simpatía dimension mirrored a well-established cultural-script of Mexicans and Latinos (Triandis et al., 1984; Holloway et al., 2009) where a person high on Simpatía is likeable, easygoing, polite, fun to be with, affectionate, and likes to share feelings with others. Indeed, the bilinguals in the Spanish mode, just like the Mexicans, used words such as honest, responsible, cheerful, friendly, reliable, sincere, humble, kind. This finding is important, given that previous studies have reported that both Mexicans and bilinguals when answering to self-reports in Spanish, tend to rate themselves as less agreeable than U.S. Americans (Ramírez-Esparza et al., 2006, 2012). However, when Mexicans and bilinguals are asked to describe their personality in Spanish in an open-ended question, the cultural biases that are primed when answering to self-reports are not prevalent. This shows that alternative methods such as factor analyses of meaningful words can be informative about cultural values associated with communality and maintenance of harmony, which underlie Simpatía (Ramírez-Esparza et al., 2008).

We also speculated in line with Ramírez-Esparza et al.'s (2012) findings that bilinguals in the English mode would talk more about their daily activities, whereas bilinguals in the Spanish mode would talk more about their hobbies. Our results provide partial support for this hypothesis. While we did find a separate self-schema related to hobbies in the Spanish mode (i.e., bilinguals used words such as music, travel, read, dance, play, study, exercise) we did not find a self-schema related to daily activities in the English mode. Rather, we found a self-schema that we labeled as Daily Difficulties. This factor includes words about everyday activities (e.g., organize, time, walk, music, stop, give, action) and words that have a negative connotation (e.g., afraid, hate, timid, hard, stress). It is worth noting, that this factor also contained words pertaining to being conscientious and diligent (e.g., organize, perfectionist, power, strive, fight). The Daily Difficulties dimension could be a reflection of a well-established individualistic, cultural ideology present in U.S. Americans: The Protestant Work Ethic. This philosophical system underlies individualism in the United States (Weber, 1958) and proposes that individuals are, above all, capable to work autonomously toward their life goals. Under this lens, hard work, self-discipline and the exercise of personal responsibility are the key to success in life (Feather, 1984). Thus, it is possible that in the English mode, the Protestant Work Ethic is activated and the bilinguals think about the activities they have to do in order to accomplish their goals, including being diligent and conscientious.

Finally, we expected to find dimensions that will emerge both in the English and Spanish mode. This will be consistent with the idea that the bilinguals will be affected by norms, the environment, and institutions of the context they are immersed (Rentfrow et al., 2008). Accordingly, we found two factors to be similar across languages: Values and Bicultural Identity. However, the quantitative approach confirmed that only bicultural identity was relevant in both languages, but Values was not. This latter finding can be better understood if one analyses closely the words used in each of the Values dimensions.

Although, bilinguals in the English and Spanish mode express their values as a salient dimension, there are some apparent differences depending on the language mode they were using. For example, in the English language mode, the factor Values seemed to reflect independence (in words such as support, connect, money, goal, confident, independent, plan, seek, see), and an effort to improve oneself (in words such as confident, relate, challenge, failure, change, volunteer, learn, knowledge) On Table 5 is an example of an essay that scored high on Factor 2 (Values) in the English mode. The independent focus that transpires in this essay can be explained in terms of an analytic thinking style (Nisbett et al., 2001). This style is characterized by a cognitive focus on an object that occupies a given context; the object itself is the main source of interest for the perceiver, regardless of the environment that the object is embedded in. In the sample essay, we get the sense that the writer situates herself as the main focus of the narrative. The text references someone that is agentic and dominates her context.

In contrast, the factor Values in the Spanish mode also seemed to reflect value for self-improvement (in words such as improve, change, learn, understand); rather than showing a value for independence, it shows a value for interdependence or connectedness (in words such as new, places, open, friends, chat, loyal). On Table 6 is an example of an essay (translated to English) that scored high on Values factor in the Spanish mode. The interdependent focus that transpires in this essay can also be explained as a holistic thinking style (Nisbett et al., 2001). Rather than seeing the objects as distinct, separate entities on a given context, holistic thinking is characterized by a cognitive focus on the context where an object is embedded. Furthermore, holistic thinking style focuses on the relationships that objects establish (Nisbett et al., 2001). In this example essay, we get a sense that the writer positions herself in function of her context; she is in search of strengthening relationships with others.

In summary, the Values factor in both languages talks about what is important for that particular person, but it is interesting that the expression of these values is driven by the nuances directed by either the independent or the interdependent self-construal. Furthermore, results can also be explained by both analytic and holistic thinking styles, which have been found to be related to independent and interdependent social orientations (Varnum et al., 2010).

Finally, it is worth noting that both English and Spanish writing samples showed a factor related to one's show bicultural identity. However, biculturalism in English is more referred toward a student identity (e.g., graduate, student, education, reliable, busy) while in Spanish, the words refer more toward a general introspection/appreciation of the bicultural identity (e.g., think, feel, identify, roots, mind, love, respect). What is interesting about this dimension is how the bicultural identity is salient, regardless of language. The fact that themes are equally salient and appear in both English and Spanish can signal a bicultural identity that is fully integrated into the experience of the self, enriching it and making it stronger. The quantitative approach seemed also to support this saliency; given the significant relation this factor had with its translated version in both English and Spanish text files.

Having a bicultural identity that is salient and integrated within one's experience has been operationalized as Bicultural Identity Integration construct (BII) (Benet-Martínez and Haritatos, 2005). This construct refers to the capacity of an individual to integrate two different cultural experiences and see them as complementing parts of the self, rather than parts that are in conflict or opposition (Benet-Martínez and Haritatos, 2005). It has been reported that individuals high in BII show advantages in terms of their subjective well-being. For example, Chen et al. (2008) gathered data on the BII levels and well-being of undergraduates in a Hong Kong university that came from mainland China. The authors reported that those students with higher levels of BII showed higher levels of psychological adjustment (composite measure of indicators of self-esteem, life satisfaction, anxiety, depression and loneliness) compared to those with lower levels of BII. These effects were detected above and beyond the influence of demographic characteristics and personality factors like neuroticism levels. More recently, Schwartz et al. (2015) explored the longitudinal association between BII and levels of psychological adjustment and family functioning in 9th grade Hispanic adolescents that had recently migrated to the United States. Data on BII levels was gathered across five different time points for the adolescents. In another study, data was gathered on various indicators of psychological well-being and family functioning. The authors report that two trends were detected across the 5 initial time points: high and low levels of BII. At the sixth time of measurement, the adolescents with high levels of BII also tended to show higher levels of self-esteem, optimism, and more fulfilling family relationships, compared to those lower in BII (Schwartz et al., 2015). Hence, there is support for the idea that a salient and strong bicultural identity might signal full integration of this identity within the self. In turn, a full integration is associated with psychological advantages in terms of subjective well-being. This is an important and interesting question that should be addressed in future studies. Specifically, it is important to observe how dimensions of bicultural identity that emerge using an open-ended approach relate to self-reports of BII and other well-being variables.

An important line of research that needs further exploration is the relationship of the themes explored here and other variables related to individual success such as psychological well-being, academic achievement and community engagement. As we outlined before, previous studies provide support for the idea that individuals that have incorporated their bicultural identity tend to feel better psychologically (e.g., Chen et al., 2008). Other evidence points out to the potential association between this “double personality” and academic performance. Brannon et al. (2015) reported evidence that when African Americans engaged in cultural activities relevant to their identity, they are more prone to activate their interdependent self-schema, which in turn, appeared related to enhanced academic performance. Naumann et al. (2017) found that Mexican Americans might shape their support for certain social causes depending on their identification with either their Euro-American identity or their Mexican identity. It would be interesting to find evidence of the linkage between the independent and interdependent self of Mexican Americans and these indicators of individual success.

Another line of research worth pursuing is to test if the noted differences in independent vs. interdependent self-construal can cause differences in thinking or cognitive styles. Previous evidence points out that social orientations (independent vs. interdependent) and cognitive styles (analytic vs. holistic) are distinct constructs that have a positive relationship (Grossmann and Na, 2014).

A final line of future research that is worth exploring has to do with complementing these findings with behavioral data. Our study shows that there are certainly differences regarding how bilinguals describe themselves according to language, but what about real life? How do bilinguals switch frames in real life situations? Previous studies have been applied successfully to gather behavioral data on Mexican and U.S. American participants; interesting comparisons have been made on how sociable members of this culture are, as opposed to how they behave in the real world (Ramírez-Esparza et al., 2009). Behavioral data on how bilinguals behave and frame-switch according to cultural cues that occur in a naturalistic environment would add significant richness to the findings that we have exposed here. In addition, future research should also follow a multimethod approach. For example, studies should test if the dimensions reported here converge with reports from outside observers, along with self-reported data. This design has been used successfully in the past to replicate the structure of self-report measures such as the NEO-PI (McCrae et al., 2005).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.