Declaration of authenticity
MA Applied Linguistics - 2018/2019 Student name: Lison Hasse Student number: 3761894 Course name: Master’s Thesis in Applied Linguistics Assignment: Master’s Thesis PLAGIARISM is the presentation by a student of an assignment or piece of work which has in fact been copied in whole, in part, or in paraphrase from another student's work, or from any other source (e.g. published books or periodicals or material from Internet sites), without due acknowledgement in the text. TEAMWORK: Students are encouraged to work with each other to develop their generic skills and increase their knowledge and understanding of the curriculum. Such teamwork includes general discussion and sharing of ideas on the curriculum. All written work must however (without specific authorization to the contrary) be done by individual students. Students are neither permitted to copy any part of another student’s work nor permitted to allow their own work to be copied by other students. DECLARATION • I declare that all work submitted for assessment in this course is my own work and does not involve plagiarism or teamwork other than that authorised in the general terms above. • I understand that this declaration covers all work submitted for assessment for Semester 1, 2018/2019, in this course. Signed Lison Hasse Date 20/06/2019P
ROFICIENCY AS A
P
REDICTOR OF THE
V
ARIATION
IN
M
ULTILINGUALS
’
S
ELF
-P
ERCEPTION
ABSTRACT
Research consistently shows that multilinguals feel different when speaking another language. Of the many potential factors for this phenomenon, none have been studied in isolation. This study looks at the specific effect of proficiency on personality change by comparing the English proficiency of Dutch and French natives with their results on two versions of a personality test, in their L1 and L2. A series of analyses revealed a change in personality for the French group only and no effect of proficiency. The study therefore discusses other potential predictors and suggests an inversion of the cause-effect relationship between personality and proficiency.
Key words: identity, individual differences, multilingualism, personality, proficiency, self LISON HASSE S3721894 MA IN APPLIED LINGUISTICS FACULTY OF ARTS SUPERVISOR: DR. HILDE HACQUEBORD SECOND READER: PROF. DR. MEREL KEIJZER JUNE 2019 13 880 WORDS
T
ABLE OF CONTENTS1.
I
NTRODUCTION4
2.
B
ACKGROUND6
2.1. A brief history of “linguistic schizophrenia”
6
2.2. Language and the speaker
7
2.2.1. Linguistics and cultural identity
7
2.2.2. Sapir-Whorf Hypothesis
9
2.2.3. Self or selves?
10
2.3. A dynamic system
11
2.3.1. Individual differences
11
2.3.2. Personality
12
2.3.3. Bilingualism, personality and the self
13
2.4. The present study
15
3.
M
ETHOD16
3.1. Measuring issues
16
3.1.1. Choice of approach
16
3.1.2. Measuring personality
17
3.1.3. Measuring proficiency
18
3.2. Present design
19
3.2.1. Participants
19
3.2.2. Material
19
3.2.3. Procedure
22
3.2.4. Analysis
23
4.
R
ESULTS24
4.1. Descriptive statistics
24
4.2. Variation in personality
27
4.3. Effect of proficiency on self-perception
28
5.
D
ISCUSSION29
5.1. The study
29
5.2. Self-assessed personality
30
5.3. Proficiency as a factor
31
5.4. Language and society
32
5.5. Future studies
34
6.
C
ONCLUSIONS35
1.
I
NTRODUCTIONMost bilinguals know how difficult it may be to express certain emotions in a foreign language, or on the contrary with what ease some things can be uttered in a language other than the mother tongue. A considerable amount of studies on the use of swear words, on linguistic emotionality, on code-switching and on self-perception have looked at the relationship of multilinguals with the various languages they possess.
The concept of multilingualism has raised a lot of debate among researchers. Sometimes considered too easily manipulated for research purposes (Bourdieu, 1991), it has long been defined as the co-existence, within a single individual, of two or more monolinguals. This understanding led to the stigmatisation of multilinguals, supposedly suffering from what is commonly called “linguistic schizophrenia”, namely “the view of bilingualism as a problem of two incompatible identities” (Pavlenko, 2006a). Throughout the instability and political turmoil of the XXth century, bilinguals and multilinguals have been
thought in turn unstable, confused or downright disloyal. By the beginning of the XXIst century, multilingualism had become a general topic of research in
linguistics, psychology and cognitive science. Multilinguals were finally seen as unique individuals, with one mental setting modified by several languages rather than as the sum of several monolinguals in one body (Grosjean, 1982; Pavlenko, 2006a). In the present study, multilingualism will be understood as the capacity to communicate in more than one language, regardless of the age at which and the situation in which they were acquired, and will be used as a synonym for bilingualism.
The question thus became: what is the difference between a monolingual and a multilingual? In order to answer this question, it is necessary to define identity and its related concepts. Psychologists and psycholinguists have extensively studied identity and self, both in monolingual and multilingual contexts (Baumeister, 2006; Ervin, 1954; Khilstrom & Klein, 2002; Pascual y Cabo & Rothman, 2012; Radden, 2006; Rogers, 1961; Ushioda & Dörnyei, 2009). Sometimes seen as the stable core defining the individual, sometimes thought as a fleeting momentary characteristic, identity seems to remain out of the researchers’ grasp. Even more so when compared to the self, an equally debated construct in this field. The self appears to be a more conscious and less stable characteristic of an individual (Baumeister, 2006; Khilstrom & Klein, 2002). However, a definition of the self as mainly declarative knowledge about oneself tends to assimilate self and self-concept (or self-perception). In the present paper, identity will be seen as the inherited and stable sum of all definable characteristics of an individual – age, origin, native language, time and place of birth, and other factors defining a person as unique. The self, on the other hand, will be seen as its fleeting and moving counterpart, whereas self-perception will
be defined as the conscious knowledge or understanding – whether biased or not – of an individual’s self by that same individual. All definitions can be found in the appendix.
Considering the self as an evolving construct, one may imagine not only that a person’s self may change over their lifetime, but also that several selves may co-exist at any moment in time. In order to investigate the factors for such a phenomenon, one needs to look into the field of individual differences (ID). Robinson (2002) described two categories of individual characteristics – affective and cognitive – to which Chan, Chin, Bhatt and Walker (2012) added a third one: variables that have an impact on cognition and affect. Nevertheless, one variable that is largely left aside, possibly because it is harder to define and measure, is personality. As it happens, personality is a core construct of the self and arguably one of the central IDs to consider in language teaching. Consequently, researchers have developed several tests to measure personality, the most important of which are the Big Five Model (McCrae & Costa, 1995) and the Myers-Briggs Type Indicator (Briggs Myers, McCaulley, Quenk, & Hammer, 1998). These tests, along with others and various methods, have often been used to look at the effect of speaking a foreign language on the individual’s personality.
Deciding on a method to define a participant’s self is challenging. Researchers have experimented with various verbal techniques (Ervin, 1954), with interviews and induced recall (Marian & Kaushanskaya, 2004), with autonomic responses (Harris, Ayçiçegi & Gleason, 2003), and with questionnaires (Dewaele & Pavlenko, 2001-2003; Nodoushan & Laborda, 2014). Introspective approaches seem to remain the most accessible and most telling measure of self-perception, which is why the present study uses a self-rated online personality test. Measuring proficiency is no less demanding. Self-rated questionnaires have been shown unreliable, whereas the usually preferred method, a combination of interviews by an independent rater and standardized tests (Pavlenko, 2006b), is logistically taxing.
For the present study, a cloze test was designed and tested in a pilot study. Thirty-six participants (17 Dutch and 19 French) were instructed to take this proficiency test and to fill in two personality questionnaires in their native language (L1) and in their second language (L2, English) with a two-week interval between the two versions of the test. With this design, the study will try to reproduce the findings of previous studies in term of personality shift from one language to the other, and to investigate the effect of proficiency on this personality shift.
The present thesis will attempt to contribute to research on the factors at play in multilinguals’ self-concept changes. Findings in this field could have implications for research into motivation and attitude – by understanding what role personality and variation in personality may play in these concepts and vice
versa – but also into attrition – by studying the psychological phenomenon underlying language dominance. Moreover, since personality type have often been linked to learning types (Dörnyei, 2005; Onwuegbuzie, Bailey & Daley, 2000), understanding the way and the reasons why it evolves throughout the life of a language learner may yield extremely useful information for syllabus designers.
This paper will first review the history of stigmatisation against multilinguals. Then, we will look at some definitions of identity and self, investigate how language may be intertwined with these concepts – linguistic relativity hypothesis – and therefore discuss whether the self should be understood as a single structure. Because language and personality are highly dynamic systems, we will then have a look at the field of individual differences, in particular how personality varies from one individual to the next and review previous research on the interaction between self-perception or self-concept and languages. In the method section, we will look at the various approaches that have been implemented for psycholinguistic research and review personality and proficiency measures. The design of the present paper will then be discussed, in particular the development of the cloze test. Descriptive statistics and analyses results will be provided and duly discussed in regard to the literature. We will reflect on the observed personality changes, on proficiency as a factor for this change and on the possibly major role of society in this matter. Finally, suggestions for future studies will be offered.
2.
B
ACKGROUND2.1.
A
BRIEF HISTORY OF“
LINGUISTIC SCHIZOPHRENIA”
Contrary to what is commonly believed by most western Europeans, an overwhelming majority of the world’s population is multilingual. Although numbers are hard to obtain, it is often supposed that over 60% of people around the globe speak more than one language. Language is an essential component of a nation (Pascual y Cabo & Rothman, 2012) and as a result, multilingualism keeps progressing with globalisation. Globalisation reduces not only physical but also linguistic boundaries, while giving a strong advantage to prestigious languages over minority ones (O’ Riagain, 2012). Learning a new language therefore means evolving in new surroundings, within new communities, which triggers behavioural changes (Pavlenko, 2006a). Auer (1998) noted that although, in bilingual communities, a change of language is expected to be accompanied with a change in verbal or nonverbal behaviour, in monolingual communities, bilinguals are likely to be seen as having “conflicting characters”, in other words to suffer from “linguistic schizophrenia”.
This vision of multilinguals as being the sum of several monolinguals led to a global belief that multilingualism gave rise to conflicting linguistic as well as political and moral allegiances (Pavlenko, 2006a). During and after WWI, immigrants in the US were encouraged to abandon their native language in order to prove their loyalty to their new nation (Pavlenko, 2002; Pavlenko, 2006a). Similarly, Nazi scholars believed bilinguals to experience an “inner split and suffer intellectual and moral deterioration in their struggle to become one” (Pavlenko, 2006a; Henss, 1931), a view that persisted into the second half of the century, with intellectuals affirming that bilingualism could even lead to pathological schizophrenia (Adler, 1977). Even though migration, peace and progressive education have changed this perspective, its negative connotations are still present in the mind of multilinguals themselves who report feelings of guilt, disloyalty, confusion and fear of schizophrenia or illegitimacy (Bilingualism and Emotions Questionnaire, Dewaele & Pavlenko, 2001-2003; Pavlenko, 2006a).
However disturbing this understanding of the multilingual mind may be to us today, it raised general awareness about the issues faced by multilinguals. The concept of multilingualism itself has been hotly discussed, being sometimes deemed too elastic and too easily manipulated to hold any argumentative value, since there is no agreement on the type or amount of knowledge, or the number of languages necessary to qualify as multilingual (Bourdieu, 1991). However, there now is a consensus on the fact that multilinguals are neither the sum of several monolinguals nor similar to a single monolingual (Grosjean, 1982) but that multilingualism is indeed a “unique linguistic and psychological phenomenon” (Pavlenko, 2006a). As a result, research into bilingualism and emotions, bilinguals’ self-perception, and linguistic identity has been further extended by psychologists, neurolinguists, psycholinguists and cognitive scientists. These studies may in fact yield results that could be used in many fields of social and human sciences, and shed more light on the cognitive dissociation experienced by more than half of the world’s population.
2.2.
L
ANGUAGE AND THE SPEAKER2.2.1. Linguistic and cultural identity Identity is a complex concept that researchers in psychology, cognitive sciences and linguistics are still debating today. The main object of disagreement concerns the stability of identity – or lack thereof – over the course of one’s lifetime. In the Encyclopedia of Applied Linguistics, Pascual y Cabo and Rothman (2012) summarised the dilemma by saying that identity had been treated as a “permanent label” whereas it is now seen as an “ever-changing concept”. Similarly, Baumann (2002) defined identity as a sort of process, rather
than a state: “instead of talking about identities, inherited or acquired, it would be more in keeping with the realities of the globalising world to speak of identification as a never-ending, always incomplete, unfinished and open-ended activity”. However, there is no consensus about this definition. The main issue in defining identity is to differentiate it from the concept of “self” (see 2.2.3 of the present paper). Pavlenko (2006a) very explicitly claimed that identity is different from the self, arguing the latter to be a more personal, subjective and moving concept. Nevertheless, in another paper (2006b), she writes about the native language being the “language of the self” rather than the identity language, which would have been more in line with her own definition of the self. In quite the same way, Wierzbicka (2007) questions the existence of an “inherited identity” or “stable core of the self”, there again taking one as a synonym for the other. Even Radden (2006) in the Encyclopedia of Cognitive Science, after defining personal identity as “the sense in which an individual can be understood as a uniquely definable entity at any time and as the same entity through time”, later on assumes that the “self” remains constant throughout life. Nodoushan and Laborda (2014) drew on Pavlenko’s work to undermine the conception of an individual with one unique self and assuming different identities and instead concluded that it is the self that is transformed by the situation, and not identity. In this thesis, I will define identity as the sum of all objective and inherited characteristics ensuring the definition of an individual as a stable entity through time.
Studies further looked at how identity is constructed. Two main factors seem to emerge from research: memories and social role. As early as the XVIIth century,
the philosopher John Locke (1690) linked identity to memories, making identity the sum of everything the individual can remember of the past. Marian and Neisser (2000) used this definition to base their language-dependent memory hypothesis. They successfully showed that memories were more easily retrieved through the language in which they were first encoded, involuntarily giving more evidence in favour of the non-permeability and stability of identity. Finally, it appears the role played in the community is another important factor in shaping identity. Baumeister (2006) calls “social identity” the identity that is formed by recognising oneself part of a group or community. It follows that the language spoken within that group or community must play a part in identity development (Pascual y Cabo & Rothman, 2012). According to Pavlenko (2006a), language and culture are part of a single ensemble. By growing up in a certain circle, the individual needs to conform to certain social and linguistic norms that shape their “cultural personality”. Consequently, language appears one of the dominant factors in identity.
2.2.2. Sapir-Whorf Hypothesis
The apparent connection between language and identity led to the highly discussed concept of linguistic relativity, also know as the Sapir-Whorf hypothesis (SWH). Beginning with Wilhelm von Humboldt’s nineteenth-century Weltanschauungstheorie, cultural linguistics have looked at the relationship between language and culture (Sharifian, 2012). In his posthumous Essai sur les langues du Nouveau Continent (1904), Humboldt declared that “the world in which we live is exactly that into which the language we speak transplants us” (Koerner, 1992). This theory was investigated by a series of researchers among which Edward Sapir who, in The Status of Linguistics as a Science (1929) wrote: “Human beings do not live in the objective world alone, nor alone in the world of social activity as ordinarily understood, but are very much at the mercy of the particular language which has become the medium of expression for their society”. The theory, by then referred to as Sapir-Whorf hypothesis, became somewhat more radical under Benjamin Whorf’s pen who saw reality as a “kaleidoscopic flux of impressions which has to be organized by the linguistic system in our minds” (1956). Whorf not only suggested that the grammatical categories of one’s language supposed a certain perception of the world, but also that they limited such perception (Hussein, 2012).
Interest for the SWH has been quite unstable. A lot of studies have supported the hypothesis (Marian & Kaushanskaya, 2004; Hussein, 2012; Gipper, 1972; Kay & Kempton, 1984). However, some doubt remains. Regier and Xu (2017) noted two main counterarguments. First, they criticised the inconsistent record of replication for findings in support of the hypothesis. Second, they saw the SWH as contradicting the accepted theory of a universal cognitive foundation for human thought. Hussein (2012) noted that Sapir’s hypothesis did not question the existence of an objective world but stated that one’s language influences only the way it is perceived, whereas Whorf’s version did not plan for an objective, universal world. Consequently, rather than understanding linguistic relativity as a set of discrete categories in which individuals of different backgrounds are allowed to develop, it should be seen as a continuum, a scale on which an individual evolves according to the language(s) they speak (Regier & Xu, 2017). Another common criticism raised by the SWH is that of the place of the bilingual within this continuum. For Macnamara (1970), if different languages created different worlds, then bilinguals would be “doomed”. As in the case of linguistic schizophrenia, multilinguals would be unfit for any of their linguistic communities. However, as Athanasopoulos (2012) puts it, it seems that learning a second language does transform the individual’s thinking, hence their cognitive mode and their perception of the world. This could equally be seen as supporting the SWH. All in all, the ultimate validity of the Linguistic Relativity Hypothesis is irrelevant to understand multilingual experience, it is “enough that some
bilinguals relate to their languages in ways that reproduce some version of the hypothesis” (Pavlenko, 2006a). 2.2.3. Self or selves? We previously defined identity as the objective cultural and social characteristics inherited by an individual (see 2.2.1). We opted for an understanding of identity as a unique and stable trait and opposed it to the self. In the Encyclopedia of Cognitive Science, Khilstrom and Klein (2002) defined the self as the “knowledge structure representing one’s declarative knowledge of oneself”, thus making the self a subjective, fluid and moving concept. The self would consequently be a form of metaknowledge of one’s cognitive and behavioural characteristics (Khilstrom and Klein, 2002), an almost voluntary, conscious knowledge about oneself that can only be accessed through introspection. A closely related idea is that of self-concept, which Baumeister (2002) described as “the mental representation of a person by that same person”. Thus, concept (or self-perception) shall here be used as synonyms of self.
Considering the fleeting nature of the self, researchers have questioned the existence a single self throughout the individual’s life and at any given moment in time. According to Khilstrom and Klein (2002), the self is “constantly being actualised, tested and reconstructed by the individual not only over the course of their life but even during social interaction”.
In the context of languages, Pavlenko (2006a) drew on the Sapir-Whorf hypothesis to state that, if languages create extremely different worlds for their speakers, it is normal that the individual should feel their self change as they change languages. A major criticism usually made to this theory is that of the plurality of self the monolingual experiences when changing tone and register. However, researchers in linguistics agree that learning a new language cannot be compared to acquiring a new register, since it involves acquiring a completely new phonetic, lexical and morphosyntactic system (Nodoushan and Laborda, 2014) and evolving in unfamiliar surroundings (Pavlenko, 2006a). Pavlenko (2006b) and Besemeres (2002) studied the testimony of translingual writers, potentially masters of introspection, to illustrate this change of personality and voice between various languages. Since then, other linguists have tried to measure this dissociation in quantitative terms. Nodoushan and Laborda (2014) had 183 Iranian-American individuals fill two versions of the Self Concept Scale (Rogers, 1961) in English and in Persian. They found a statistically significant discrepancy between the real self-concept of their participants in Persian and in English. Marian and Kaushanskaya (2004) led interviews with 47 Russian-English bilinguals during which they measured the number of personal or group pronouns used as a way to indicate a preference for individualism (more commonly associated with English) and collectivism (associated with Russian) according to the language spoken. They found that
only the language of retrieval (as opposed to the language of encoding) of a memory had an effect on the choice between individualism and collectivism. Both studies concluded in favour of the coexistence of multiple language-specific selves within the multilingual individual.
2.3.
A
D
YNAMICS
YSTEM2.3.1. Individual differences
Since the middle of the XXth century, researchers have looked into the reasons
why individuals learn and process languages differently from one another. They uncovered a range of individual differences (ID) that have been described as the notion that “each individual person comprises a unique combination of aspects that might determine learning outcomes” (Murphey & Falout, 2013) or the characteristics that “mark a person as a distinct and unique human being” (Dörnyei, 2005).
Robinson (2002) separated those factors into two categories: affective and cognitive abilities. The former category includes anxiety, motivation, attitude and emotions. Intrinsically related to the other three, motivation is often divided into integrative motivation – based on an interest for the foreign language and its culture – and instrumental motivation – derived from a practical need to communicate in the L2. Attitude is a slightly different construct – namely the learner’s state of mind regarding the target language and its native community – that does not always go in the same direction as motivation. Emotions play an important role in anxiety, motivation and attitude and are sometimes argued to deserve more attention from researchers, as they could potentially explain cases “where action is prevented by emotional arousal” (MacIntyre, 2002). The second category of factors comprises intelligence, aptitude and working memory, all of which are consistently challenging to measure. Sternberg (1990) defined intelligence as the sum of analytic, creative and practical abilities while aptitude is commonly divided into phonetic coding, grammatical sensitivity, inductive learning and memory capacity (Carroll & Sapon, 1959).
Chan et al. (2012) added a third category to Robinson’s model: variables that have an impact on cognition and affect. The sociocultural and socioeconomic backgrounds of an individual fall into this category. So does age, the study of which led to the validation of some form of the critical period hypothesis. In short, this category assembles IDs that emerge from amalgamations of emotional-cognitive states and situational states of individuals (Murphey & Falout, 2013).
Chan et al. (2012) pointed out that researchers focus primarily on factors that are relatively stable, apply to everybody and have an impact on the learning processes and outcomes, possibly because they are more easily measured and
defined as the sum of discreet subparts. However, when thinking about variables that have an impact on cognitive and affect, one factor appears largely left aside: personality (and subsequently, self-perception). Personality may in fact influence how an individual thinks and behaves, and how they reflect on and monitor their own thinking and behaviour. Thus, it appears of utmost importance in understanding both the way a learner can take an active part in their own learning – through academic metaknowledge for example – and the way a multilingual might construct their past, present and future selves within the newly accessible communities (Murphey & Falout, 2013).
2.3.2. Personality
For some researchers, personality is a central, if not the most important, individual characteristic. Determined both by environmental and biological or hereditary factors (Dörnyei, 2005), personality is defined by Pervin and John (2001) as the characteristics of an individual “that account for consistent patterns of feeling, thinking and behaving”. They further described the field of personality psychology as that which “most considers people in their entirety as individuals and as complex beings”. Paradoxically, however, Dörnyei (2005) pointed out that personality is one of the least studied IDs, possibly due to the fact that it is considered to hold less relevance in language teaching as motivation or aptitude for example.
Several models have been proposed to classify personalities. Eysenck’s model (Eysenck & Eysenck, 1985) described three dichotomies: extraversion versus introversion, neuroticism version emotional stability and tough-mindedness versus tender-mindedness. In the 1930s and 1940s, the Big Five Model (B5) gained momentum. Theorised by Goldberg, McCrae and Costa, it kept the first two oppositions of the Eysenck’s model but replaced the third by three more: conscientiousness, agreeableness and openness to experience. Based on Jung’s theories, the Myers-Briggs Type Indicator (MBTI) (Briggs Myers et al., 1998) later gained momentum and became widely used in many domains (see point 3.1.2).
A common question regarding personality is its degree of consistency across time and place. Pervin and John’s (2001) original description does suppose “consistent patterns” of personality. However, taking into account the situational factors may lead to variation of behaviour or even thinking patterns (Dörnyei, 2005). Indeed, if one sees personality as the product of environmental factors – typically in the child’s home –, one may imagine that changes in these factors would lead to changes in personality.
Yet, we have seen that different languages create rather different worlds for their speakers – or at least, worlds that they experience as different. Thus, it seems likely that personality changes according to the language being spoken, a notion familiar to many multilinguals (Dewaele, 2012; Nodoushan and Laborda, 2014;
Pavlenko, 2006a; Pavlenko, 2006b; Wierzbicka, 2004). In the introduction of Emotions and Multilingualism, Pavlenko (2006b) notes that a change in languages implies a change in affective styles and that code-switching is often reduced to an “us versus them” dichotomy. In line with these remarks, many of her subjects – though not all of them – reported experiencing their personality in the L1 as more real and more natural than their personality in the L2, which feels fake and artificial (Pavlenko, 2006a). The psychoanalyst Karl Jung called this aspect of personality the persona, namely “a mask that feigns individuality”, a form of “conscious personality as a more or less arbitrary segment of collective psyche” (Jung, 1967). According to Pavlenko (2006a), the persona is the image we project in public, the social act we put on, opposed to the private self. However, this does not necessarily mean that it is less true of a personality. Finally, Wierzbicka (2004) also notes that the set of concepts available to describe feelings – the very seat of personality (Dörnyei, 2005), along with thoughts and behaviour – is specific to a particular language, there again suggesting that language influences personality.
2.3.3. Bilingualism, Personality and the Self
If personality is perceived as a concept absolutely stable over time and across situations (Pervin, 1980), then the language used should only be seen as the expression of an underlying trait rather than a factor of change (Chen & Bond, 2010). It would therefore play no part in the individual’s personality. Nevertheless, generations of researchers have looked at the potential effect of language on personality. Ervin-Tripp (1964) used a psychological approach and administered the thematic apperception test to French-English bilinguals, a test in which subjects are required to construct a narrative based on ambiguous pictures supposed to reveal their underlying values, concerns and personality. From the themes that she found in the participants’ narratives in French and in English, Ervin-Tripp concluded that multilinguals possess two personalities, at least in their verbal productions.
The emotional weight of a language, or language emotionality, often comes under researchers’ scrutiny. Indeed, the way an individual perceives their various languages is part of how they will then think, feel and behave in that language. Hence, it is part of their personality. Some researchers have operationalized language emotionality as the degree of autonomic response from their participants. Harris, Ayciçegi and Gleason (2003) used measures of skin conductance to study the reaction of Turkish-English individuals to various words in both Turkish and English. They found a significant difference in emotional reaction between the two languages. Marian and Kaushanskaya (2004) looked at the production side of language emotionality by examining the effect of the recall language on the type of memories elicited. They also found a significant difference in emotional expression between the two
languages. One of the most ambitious attempts to describe the relationship between bilingualism and emotions is Dewaele and Pavlenko’s Bilingualism and Emotion questionnaire (Dewaele & Pavlenko, 2001-2003). In this wide-scope survey, they formulated 34 questions to establish precise sociobiographical information about their 1,039 participants and ended with the following open-ended question: “Do you feel like a different person sometimes when you use your different languages?”. They found a vast majority of affirmative answers, 26% of negative answers and 6% of ambiguous answers. They drew a whole range of conclusions, some of the most important being the following (Dewaele, 2012): 1) the “language of the heart”, namely the language of stronger emotionality and truthfulness, can shift depending on the dominant language, the nature of the L2 network, the degree of socialization in the L2, etc.; 2) the frequency of use of the L2 is the variable with the strongest effect on self-perception; 3) even participants who declared a maximal level of proficiency in L1 and L2 and a constant use of both revealed a linguistic preference for inner speech, swearing and emotionality in the L1.
Questionnaires and self-reports remain the preferred tool to investigate personality. Hull (1996) submitted Chinese, Korean and Mexican subjects to the California Psychological Inventory and found differences on all scales between the participants’ native language version and the English version. Since he (Hull, 1990) had obtained similar results with another measure of personality, he eventually concluded that each language known by an individual corresponded to a slightly different version of that person’s personality. Ramirez-Esparza, Gosling, Benet-Martinez, Potter and Pennebaker (2006) had Spanish natives living in the US take two version of the Big Five Model test. They found that the participants answered more “like Americans” to the English version of the test, thus hinting at a potential cultural accommodation effect. One remaining problem is the unreliability of self-reports (McCrae, Yik, Trapnell, Bond & Paulhus, 1998). In order to solve this issue, Nodoushan and Laborda (2014) made use of the self-concept scale (SCS) to measure the discrepancy between the actual and the ideal self-concept of their participants both in English and in Persian. They found a statistically significant difference between the participants’ actual self-concepts in both languages and noted that the subjects had a more realistic understanding of themselves in English than in Persian. This form of double standard measurement insures a higher degree of reliability since the participants become their own point of reference. Chen and Bond (2010) used a similar technique: they had their participants fill in the Big Five questionnaire for themselves, for the typical Chinese- and the typical English-speaker, thus creating a frame of reference for each individual. They equally found an effect of language on all factors.
All studies on the matter seem to indicate a personality shift between the various languages of the multilingual individual. The effect size varies largely from one
study to the next depending on the confounding variables accounted for in the design, but also on the type of cue provided. Oyserman and Lee (2008) performed a meta-analysis on 10 studies using language as a priming method. They found an overall weak effect of language priming, which could suggest that cultural priming should be preferred. In the present study, a simple language cue has been chosen for feasibility reasons. However, proficiency is taken to reflect also the closeness of a participant to the target language community.
2.4.
T
HE PRESENT STUDYIn view of previous research, there seems to be agreement on the fact that multilinguals experience a modification of their self as they change languages. The factors for these changes and their relative degrees of importance are still unclear. Dewaele (2010) emphasized the dynamic aspect of language choice to express emotion. Pavlenko (2004) and Dewaele (2012) suggested a range of variables that could play a role in multilingual self-perception, among which past experience, current use of the various languages and several sociobiographical and psychological factors, including proficiency.
Through their Bilingualism and Emotions questionnaire (Dewaele & Pavlenko, 2001-2003), the two scientists showed that even language dominance is a key factor affecting language choices, and multilingual experience and self-perception (Dewaele, 2010; Dewaele, 2012; Pavlenko, 2004; Pavlenko, 2006a). Anooshian and Hertel (1994) hypothesized that the age of acquisition of a language was more important than the level of proficiency for the way a language was internalized and experienced. This supposition, however, appears to overgeneralize the process of multilingual self-perception by assuming that all individuals are equally affected by those variables. The present exploratory paper is an attempt to investigate the effect of proficiency on multilinguals’ self-perceived personality.
In order to do so, Dutch and French learners of English have been required to perform a cloze test to assess their English proficiency – through vocabulary knowledge and fluency – and to fill in two introspective personality questionnaires in their first language – Dutch or French, respectively –, and in English, with a two-week delay between the two surveys. The variation in the degree to which they exhibited certain personality traits in their own L1 and in English was later compared to their English proficiency results. The research questions are as follows:
• Is there a variation between multilinguals’ self-assessed personalities in the L1 and in the L2?
• Does the L2 proficiency affect self-perception of Dutch and French learners of English in their L1 and in their L2?
• Is there a difference in the extent of the personality shift from the L1 to the L2 between the Dutch and the French participants? And if so, does proficiency in English affect the Dutch and French participants’ self-perception in the same way?
Because a higher level of proficiency in the L2 supposedly brings the learner closer to the target language community, it is hypothesized that the more proficient the learners become in their L2, the further their self-perceived L2 self will be from their L1 self. Considering the overall attitude of the French population toward English and the relationship of the French to their own language, it is supposed that the variation in self-perceived personality between the L1 and the L2 will be greater for the Dutch learners. However, proficiency should affect both groups in a similar way. In other words, if proficiency indeed affects self-perception, it is expected that a higher proficiency will lead to a bigger variation in personality between the L1 and the L2 for French learners as well as for Dutch learners, though possibly not to the same extent.
By attempting to study and quantify the effect of proficiency on the bilingual selves, the present thesis may have implications for research into motivation, attitude, anxiety and possibly attrition.
3.
M
ETHOD3.1.
M
EASURING ISSUES3.1.1. Choice of approach
Introspective approaches seem to dominate the field of self-perception and personality. Dewaele and Pavlenko’s Bilingualism and Emotions Questionnaire (Dewaele & Pavlenko, 2001-2003) belongs to this qualitative research approach. Introspection appears to be one of the fastest and surest ways to estimate the state of mind of the individual on a particular question. The self, as “declarative knowledge of oneself” (Khilstrom & Klein, 2002) follows the same logic and so does the field of research on attitude (Ó Riagáin, 2012), for which other qualitative and declarative procedures are also used to establish the degree of prestige of given languages and the characteristics attributed to their speakers. As a means of comparison for the results obtained with the Bilingualism and Emotions Questionnaire, Pavlenko (2006a) used the introspective testimony of translingual writers about their relationship with their various languages (see also Wierzbicka, 2004). Being accustomed to thinking and writing in several languages, such individuals are more likely to have a clearer and more stable understanding of their experience as multilinguals.
However, these methods do not yield quantifiable results and are liable to more interference with many factors ranging from the researcher’s bias in interpreting the results to the participant’s mood when taking the test. Therefore, Pavlenko (2006a) also compared her qualitative results with empirical data obtained by other researchers, such as verbal behaviour measures (Ervin, 1954; Ervin-Tripp, 1964).
To bridge the gap and solve most flaws of these various approaches, researchers have implemented mixed methods by providing a comparative scale per individual (Nodoushan and Laborda, 2014), by combining verbal measures with emotionality assessment (Marian and Kaushanskaya, 2004) or by combining introspective data with autonomic responses (Harris, Ayçiçegi and Gleason, 2003), for example.
All in all, there cannot be a perfect method for such a dynamic and wide system as the multilingual self. However, Pavlenko (2006b) warns about the importance to combine “experimental tasks with ethnographic approaches, interviews and questionnaires”.
3.1.2. Measuring personality
“Researchers generally believe that personality-trait structure and the manifestations of its factors are universal” (McCrae & Costa, 1997). Psychologists have proposed several models to classify and assess personality traits.
The B5 stems from Hippocrates’s four temperament types: sanguine, phlegmatic, choleric and melancholic. Based on the lexical hypothesis – the idea that traits important to an individual will eventually become part of their language –, it emerged in the 1930s and 1940s and was finally implemented by McCrae and Costa (1995). It consists of five dimensions: openness to experience, conscientiousness, extraversion, agreeableness and neuroticism. The test result is a 5-letter code corresponding to the participant’s type, which supposedly describes the way they make decisions and behave in their professional as much as in their personal life. The B5 was said to be too simplistic and not representative of the entirety of human personality (Dörnyei, 2005). Researchers have argued that it only accounts for 56% of the normal personality traits sphere, which led McAdams (1995) to describe it as a “psychology of the stranger” in reference to the assumption that the model only revealed traits that could be easily observed in a stranger.
Similarly to the B5, the MBTI (Briggs Myers et al., 1998) is a type-based introspective self-report. Based on Carl Jung’s theory of four psychological functions – sensation, intuition, thinking and feeling – modified by two attitude types – extraverted and introverted –, the MBTI created four dichotomies that describe 16 potential personality types. Widely used in recruiting and team management, the MBTI has often been reported as “learning styles” (Onwuegbuzie, Bailey & Daley, 2000). However, Dörnyei (2005) advises to keep
in mind the fact that the MBTI provides “personality dimensions with cognitive style correlates”. The MBTI has equally faced a lot of criticism, being accused of poor validity, poor reliability, lack of independence between the categories and lack of comprehensiveness – principally due to the absence of the neuroticism aspect (Boyle, 1995).
Both the B5 and the MBTI are type-based, as opposed to trait-based, which means they indicate an individual’s preference for a certain trait without examining the extent to which this trait is demonstrated. Moreover, the MBTI can only be performed by a trained assessor and the B5 is not easily accessible online by potential participants.
The more recent test called Neris Type Explorer (Neris Type Explorer, 2011-2019) draws on both the type- and trait-approach to personality assessment in that it uses a string of letters to define a personality type – similarly to the MBTI – but measures the various aspects on a scale in order to indicate a preference for a certain type and the extent to which this type is preferred. The five scales – Mind (introverted vs. extraverted), Energy (observant vs. intuitive), Nature (thinking vs. feeling), Tactics (judging vs. prospecting) and Identity (assertive vs. turbulent) – are based on and adapted from the Big Five model (16personalities, 2017b). A summary of the different traits can be found in the appendix.
3.1.3. Measuring proficiency
Choosing a proficiency measure is equally challenging. In the context of multilingual self-perception, proficiency seems to be a major factor. Indeed, Pavlenko (2006a) declared the level of proficiency to be one of the main sources of varying self-perception from one language to the next – along with linguistic and cultural differences, learning contexts and levels of language emotionality. Proficiency has been perceived in three different ways: 1) as equivalent to the overall language ability; 2) as the ability to use a language in a particular situation and for specific purposes; 3) as the performance indicated by a given task (Shin, 2012). In the present paper, we shall understand proficiency as the “overall communicative language ability” (Shin, 2012), in line with Davies et al. (1999) who noted “considerable overlap between the notion of language proficiency and the term communicative competence”.
Language proficiency can be understood as the sum of linguistic knowledge an individual can muster or as real-life competence. Most existing frameworks – among which the Common European Framework of Reference (CEFR) – favour the second option, which is also the most relevant in the context of self-perception. Recognizing the fact that language is always used in an interactive and social context, Chalhoub-Deville (2003) stated that, performance being co-constructed, language proficiency cannot be seen as a stable trait of the individual. However, such a view would render proficiency practically
impossible to measure and does not explain the generally consistent performance of individuals across tasks (Shin, 2012).
A very large number of ways have been thought out to evaluate proficiency. A common method – supposedly due to its accessibility and important feasibility – is self-assessment. Pavlenko (2006b), however, noted that self-evaluation did not constitute a reliable approach, no more than language preference questionnaires as an index of language proficiency. Instead, she recommended a combination of independent raters’ judgments and standardized tests and measures. In fact, Marian and Kaushanskaya (2004) did find that the self-reported language of preference did not correlate with proficiency as established by independent raters. However, for practicality reasons, it remains a commonly used technique (Harris, Ayçiçegi and Gleason, 2003).
For equally practical reasons and to avoid the bias of a proficiency questionnaire, the present paper shall make use of an English cloze test to assess the participants’ proficiency level (see 3.2.2.1).
3.2.
P
RESENT DESIGN3.2.1. Participants
Thirty-six younger adults (n=36) participated in this study. Age ranged from 20 to 37 years old (M=24.89, sd=3.86). The cohort was constituted of two groups: 17 Dutch natives (ranging from 21 to 37 years old; M=25.65, sd=4.09) and 19 French natives (ranging from 20 to 34 years old; M=24.21, sd=3.61). All participants had started learning English at least ten years prior to the study. Six participants in each group had spent some time in an English speaking country (the average duration of the stay was 3.82 months for the Dutch group and 3.79 months for the French group). In order to account for potential attrition, only participants living in a country where their native language was spoken were selected.
3.2.2. Material
3.2.2.1. Cloze testThe cloze test uses a discrete-point testing technique, opposed to more commonly used integrative language tests, the practicality of which may render it an immensely convenient tool for language assessors (Stubbs & Tucker, 1974). The principle of the cloze test is to select a text specifically for the target group and delete words according to a regular pattern. According to Kobayashi (2002), “the deletion of words at regular intervals ostensibly produces a representative sample of the linguistic features of [a] text” thus representing a potentially “valid measure of the underlying language ability”.
Cloze tests have however faced a range of criticism (Kobayashi, 2002). First, they have been thought to measure lower-order language skills rather than higher-order. Second, it has been shown that even native speakers do not always achieve a perfect grade. Third, the method is believed to require more productive than understanding skills. And fourth, some have argued that other factors than language come into play. While the first objection may very well be an inherent and unavoidable drawback of any discrete-point method, the second does not appear to constitute a real obstacle. There are a variety of reasons why a native speaker may not score perfectly on a given proficiency test that do not question the method’s validity – ranging from educational background to attrition, through multicompetence, to name a few. The third criticism is irrelevant in the present context since productive skills should be the focus of a study touching upon sociology. In terms of the fourth objection, it may be argued that deduction, logical or analytical skills are somehow related to linguistic skills and should therefore be included in the test. In the present cloze test, for example, the fact that some participants sometimes gave grammatically correct but illogical answers may be a sign of a lack of understanding of the general or local context. Overall, cloze tests have given very volatile reliability and validity results (Brown, 2002). However, Stubbs and Tucker (1974) showed cloze tests results to correlate with other English proficiency tests and older studies have proved the consistency of several cloze versions with each other (Bachman, 1985; Jonz, 1990).
When developing a cloze test, two main questions arise: at what interval to delete words and which scoring method to adopt. The first issue is easily solved with a pilot study (see below). The scoring method has raised more debate. The appropriate word scoring method has often been deemed more valid and reliable (Brown, 1980; Kobayashi, 2002). However, it implies the creation of possible answer glossaries by native speakers for each item, and is therefore much more time consuming. Nevertheless, Stubbs and Tucker (1974) found that the results of an exact word scoring method correlated with that of an appropriate word scoring method, thus making the test accessible to non-native teachers and assessors.
The pilot study. In order to design and validate the final test, a pilot study was first carried out. Seven French natives, ranging from 21 to 26 years old (M=23.86, sd=1.57), participated in this pilot phase. Two texts dealing with approximately the same topic – namely international cities’ happiness rankings – were selected. In the first text, every seventh word was deleted while every sixth was taken out of the second text. When the word to delete was a proper noun or a number, the next word was chosen instead (except for locations that had already been mentioned and could be deducted from the context). In both texts, an introduction of about 50 words (47 for the first and 54 for the second) was left untouched. Thus, the first text was 440-word long and comprised 50 blanks
while the second was 455-word long and comprised 61 blanks. Each participant was asked to fill in both text, in a randomised order, and their answers where scored according to both an appropriate (AW) and an exact word (EW) scoring method. A 26-year-old English native assessed the appropriateness of the participant’s answers. The results can be seen in Table 1.
Table 1. Results of the pilot cloze tests according to the exact word (EW) and appropriate word (AW) scoring methods for texts 1 and 2 (out of 50 for text 1 and 61 for text 2).
Participant EW1 AW1 EW2 AW2
1 36 45 36 47 2 32 43 36 49 3 27 42 42 53 4 30 42 42 54 5 23 39 27 42 6 29 45 36 53 7 33 43 31 47 The results for both methods and for both tests were analysed for normality of distribution using a Shapiro-Wilk test and by looking at the skewness and kurtosis of each dataset. All datasets proved to be normally distributed: no skewness or kurtosis and Shapiro-Wilk results over W=.90, p=.34. Thus, all techniques and tests would allow the researcher to properly differentiate between the higher and lower proficiency participants.
Finally, a Pearson r correlation analysis was carried out on both texts to test the existence of a relationship between the two scoring methods. For both texts, the exact word and appropriate word scoring method were strongly and significantly correlated (r(5)=.78, p=.038, for text 1; r(5)=.89, p=.006, for text 2). However, since the exact word scoring method was preferred for time and practicality reasons, and since the correlation appeared stronger and more significant for the second text, it was finally selected for the present study. The final cloze test submitted to the participants as well as the exact answers can be found in the appendix. 3.2.2.2. Neris personality explorer The Neris Type Explorer is freely available online and therefore makes use of an immense database to test and perfect its functioning. The researchers behind this tool have mostly worked on three aspects: internal consistency, test-retest reliability and discriminant validity. For the first, they found that the Cronbach alpha for each scale was comprised between .7 and .9, thus certifying the test’s internal consistency. For the second, the researchers had a cohort of almost 3,000 participants perform the test twice with an interval of 5 to 7 months. The
correlation coefficients for each scale were comprised between .7 and .9 (p<.001). There appears to be little variation in the results of a given individual over time. For the third, the researchers looked at the amount of crossover between the different scales. They found that all correlation coefficients were inferior to .37, the maximum value being between the Energy and Tactics scales, an effect that they argued to have been shown previously by other instruments for similar concepts. Therefore, they concluded that the test showed discriminant validity (16personalities, 2017a).
The strength of the Neris Type Explorer is that, being free and accessible, it can be regularly assessed and improved. Moreover, due to its growing popularity, the test is available in a variety of languages – among which Dutch and French – which are equally reviewed and tested. In regards to these practical aspects and the apparent reliability and validity of the test, the present study will base its personality assessment on the Neris Type Explorer.
3.2.3. Procedure
Each participant was contacted via email and informed that they would take part in a study on language and personality. They were told the study would consist in three parts but did not know the exact purpose of the tests they were about to perform, neither did they know that the last part of the study would be a repetition of the personality test in another language. The participants were then sent the cloze test and a link to the first personality test. The language of this first personality test – Dutch or English for the Dutch participants; French or English for the French participants – was randomly assigned in such a way as to have approximately the same number of participants starting with their L1 as participants starting with their L2. Participants were not given any time limit for either test.The results of the cloze test were manually compiled according to an exact-word scoring method. Consequently, answers were only counted as correct when they corresponded exactly to the original word. As the objective was to focus on communicative skills, spelling errors and typos were counted as correct – i.e. relativily instead of relatively – and so were regional variations – i.e. colors instead of colours. Ambiguous mistakes, which could potentially reveal an inaccurate answer – i.e. wich instead of with – were noted as wrong.
The date at which each participant performed the personality test was noted and a link to the second personality test, in the language they had not yet worked with, was sent to them two weeks later. This two-week interval between the two versions of the test was meant to avoid a training effect. For the personality tests, both the type – the 5-letter code – and the traits results – the percentage of preference for one end of the spectrum – were compiled for analysis.
3.2.4. Analysis
The data were analysed in R (R Development Core Team, 2012). In order to investigate the relationship between the language of the test and the personality results, a series of paired-sample t-tests were performed, first for each individual personality traits and then for the total personality result. The latter was calculated as the sum of the deviation from baseline – that is, from 50 –, since such a score indicates no particular preference for either side of the personality spectrum. Hence, the formula to calculate the total personality result was as follows:
Total = Σ(x − 50)
where x represents the five personality traits studied by the Neris Type Explorer. This was preferred to a simple sum of all results. Indeed, since each trait is independent from the others, such a calculation would not have any logical value. Multivariate analyses were also performed to investigate the possible interaction of the participants’ L1 (Dutch or French) with the results.
In order to look at the effect of proficiency on the shift in personality, the absolute difference between the results in L1 and the results in L2 were calculated for each trait. As a measure of the total shift in personality, the sum of all differences was calculated for each individual, thus creating the following formula:
Total difference = Σ (xL1 − xL2)
where xL1 corresponds to all trait results in the first language and xL2 in the second language.
A series of Pearson r were performed to look for a potential interaction between proficiency and each trait difference, and between proficiency and the total difference. Multiple linear regressions were carried out to assess the potential interaction of participants’ L1 with the previous results.
4.
R
ESULTS4.1.
D
ESCRIPTIVE STATISTICSThe descriptive statistics for age and time spent in an English-speaking country are provided in the following tables. Table 2. Descriptive statistics for age across groups (in years).
Groups Mean Median SD Min. Max. Range
All participants 24.89 24.5 3.86 20 37 17
Dutch 25.65 25 4.09 21 37 16
French 24.21 24 3.61 20 34 14
Table 3. Descriptive statistics for time spent in an English-speaking country across groups (in months). Groups Mean Median SD Min. Max. Range
All participants 3.81 0 7.59 0 36 36
Dutch 3.82 0 8.84 0 36 36
French 3.79 0 6.53 0 24 24
From these two tables, the two groups appear equivalent both in terms of age and of experience abroad.
Table 4 presents the descriptive statistics for proficiency results for all participants and per group.
Table 4. Descriptive statistics for cloze test results (out of 61).
Groups Mean Median SD Min. Max. Range
All participants 38.42 38 5.05 27 49 22
Dutch 39.06 38 5.18 31 49 18
French 37.84 38 5 27 48 21
The cloze tests results followed a normal distribution for both groups and for the whole population, according to skewness and kurtosis measures and following the results of a series of Shapiro-Wilk tests. The normality of distribution was confirmed by a histogram.
The following tables present the personality types measured by the Neris Type Explorer, obtained by Dutch and French participants in the L1 and the L2.
Table 5. Personality types of Dutch participants in L1 and in L2 (I=Introvert, E=Extravert; N=Intuitive, S=Observant; T=Thinking, F=Feeling; J=Judging, P=Prospecting; A=Assertive, T=Turbulent*). Personality traits N=in Dutch (tot=17) N’=in English (tot=17) INTJ-A/-T N=0 N’=0 INTP-A/-T N=0 N’=0 ENTJ-A/-T N=1 N’=0 ENTP-A/-T N=1 N’=2 INFJ-A/-T N=2 N’=5 INFP-A/-T N=2 N’=1 ENFJ-A/-T N=1 N’=0 ENFP-A/-T N=2 N’=5 ISTJ-A/-T N=0 N’=0 ISFJ-A/-T N=3 N’=2 ESTJ-A/-T N=0 N’=0 ESFJ-A/-T N=2 N’=0 ISTP-A/-T N=0 N’=0 ISFP-A/-T N=2 N’=2 ESTP-A/-T N=0 N’=0 ESFP-A/-T N=1 N’=0 * The Identity trait (assertive vs. turbulent) is considered as additional information rather than part of the personality type. Table 6. Personality types of French participants in the L1 and in the L2 (I=Introvert, E=Extravert; N=Intuitive, S=Observant; T=Thinking, F=Feeling; J=Judging, P=Prospecting; A=Assertive, T=Turbulent*). Personality traits N=in French (tot=19) N’=in English (tot=19) INTJ-A/-T N=1 N’=1 INTP-A/-T N=1 N’=2 ENTJ-A/-T N=0 N’=0 ENTP-A/-T N=0 N’=0 INFJ-A/-T N=4 N’=4 INFP-A/-T N=3 N’=4 ENFJ-A/-T N=2 N’=0 ENFP-A/-T N=1 N’=2 ISTJ-A/-T N=1 N’=1 ISFJ-A/-T N=4 N’=3 ESTJ-A/-T N=0 N’=0 ESFJ-A/-T N=1 N’=1 ISTP-A/-T N=1 N’=0 ISFP-A/-T N=0 N’=1 ESTP-A/-T N=0 N’=0 ESFP-A/-T N=0 N’=0 * The Identity trait (assertive vs. turbulent) is considered as additional information rather than part of the personality type.
The average results for each trait was calculated per group in the L1 and in English. The results can be seen in Table 7 and 8.
Table 7. Average trait measures in the L1 (Dutch or French), per group (compared to baseline). Group Mind Energy Nature Tactics Identity
Dutch 1.94% more introverted 3.76% more intuitive 12.82% more feeling 4.29% more judging 2% more turbulent French 14.74% more introverted 6.26% more intuitive 15.95% more feeling 8.74% more judging 13.95% more turbulent The average personality of the Dutch and the French participants in their native language were on the same side of each personality trait spectrum. In other words, the Dutch and the French showed a preference for the same side of each dimension. However, differences of up to about 13% were detected between the two groups, implying an important cultural difference.
Table 8. Average trait measures in the L2 (English), per group (compared to baseline).
Group Mind Energy Nature Tactics Identity
Dutch 0.29% more introverted 6.59% more intuitive 11.88% more feeling 2.29% more judging 5.47% more turbulent French 18.74% more introverted 6.63% more intuitive 16.53% more feeling 0.68% more judging 19.16% more turbulent In English, both group preferred the same end of the spectrum for each trait. The extent of that preference, however, was modified by the language of the test. In order to see this modification more clearly, the average percentage of change between English and the L1 was calculated for each group. The results are shown in Table 9.
Table 9. Personality in English compared to the personality in the L1 (in English, participants were on average …. than in their native language).
Group Mind Energy Nature Tactics Identity
Dutch 1.65% more extraverted 2.82% more intuitive 0.94% more thinking 2% more prospecting 3.47% more turbulent French 4% more introverted 0.37% more intuitive 0.58% more feeling 8.05% more prospecting 5.21% more turbulent
The previous analyses were meant as a way to investigate the direction of a potential personality shift. In order to more closely study the extent of such a shift, the absolute difference between the results obtained in the L1 and the results obtained in the L2 was calculated for each trait. The descriptive statistics of the results are presented below.
