Multilingual Clients’ Experience of Psychotherapy

The present study focuses on the experiences of 182 multilingual clients who had been exposed to various therapeutic approaches in various countries. An on-line questionnaire was used to collect quantitative and qualitative data. The analysis of feedback from clients with multilingual therapists showed that clients use or initiate significantly more code-switching (CS) than their therapists, and that it typically occurs when the emotional tone is raised. Gender was unrelated to CS frequency. CS is used strategically when discussing episodes of trauma and shame, creating proximity or distance. CS allows clients to express themselves more fully to the therapist, adding depth and nuance to the therapy. The therapist’s multilingualism promotes empathy and clients’ own multilingualism constitutes an important aspect of their sense of self. Multilingual clients benefit from a therapeutic environment where multilingualism is appreciated, and where they can use CS.


Introduction
Over the past decade various attempts have been made to address the design and relevance of social and health services for an increasingly diverse population. Researchers have focused on the nature and challenges of offering psychological support and therapy across cultures (Eleftheriadou, 2010;Fernando, 2010;Fernando & Keating, 2009;Lago, 2006;Lowe, 2013;Martinovic & Altarriba, 2013). Gradually more attention has been paid to the role of multilingualism in therapy and the exploration of the relationship between language and therapy (Amati-Mehler, Argentieri & Canestri, 1993;Kokaliari, Catanzarite & Berzoff, 2013;Perez Foster, 1998;Santiago-Rivera & Altarriba, 2002;Szekacs-Weisz, 2004). Marcos (1976) is one of the pioneering studies in this domain. He considered interactions of non-native English-speaking (LX) clients with their English native speaker therapist. Being forced to use an LX has both advantages and disadvantages for clients. They risk encountering encoding difficulties 'in integrating emotions and experiences, the displacement and blocking of affects, and the reinforcement of obsessive resistances [which] may give rise to misinterpretations and distortions of their problems'. (p. 552) However, the emotional distancing linked to the use of the LX, which he calls the detachment effect, 'may facilitate the verbalization of highly charged material by clients who feel "protected" by the linguistic detachment'. (p. 552) More recently there has been an increase in research into the experiences of multilingual therapists providing therapy in a language in which they have not been trained (Verdinelli & Biever, 2009). However, very little research to date seems to have incorporated the voice and adopted the perspective of the multilingual clients (see however Espin, 2013). Such an emic approach allows researchers to explore participants' perspective, their interpretation of behaviour, events and situation using their own language (Pike, 1964).
The present study aims to fill this gap. It was prompted by an initial investigation into the experiences of monolingual and multilingual therapists with their multilingual clients (Costa & Dewaele, 2012. The research described in this paper aims to open up the discussion about multilingualism and therapy; to listen to and to convey the voice of the multilingual client; and to compare and contrast the views of multilingual clients with the views of multilingual therapists. The literature reviewed below characteristically presents the opinion that multilingualism can be a useful resource which psychotherapy can draw upon. Where there is a possibility of code-switching (defined as 'changes from one language to another in the course of conversation' (Li Wei, 2007, p. 14)), this choice can be 'strategically used as a unique therapeutic technique.' (Pitta, Marcos & Alpert, 1978, p. 255). As already mentioned, there is little or no published research from the multilingual client's point of view. The review therefore draws together a number of themes which emerge from the literature regarding multilingual therapists' experiences as well as the experiences of clients. Antinucci (2004) proposes that common language and cultural identity can become part of a collusive process between the client and the therapist. She believes that it is the therapists' task to address their cultural identity and fantasies that surround the desire to collude. In this way, the professional therapeutic frame can be maintained. Similarly, Akhtar (2006) uses the term 'nostalgic collusion' to denote a clinical situation where the client and the therapist collude in avoidance. Mourning and idealisation of a lost culture is not addressed, possibly due to the therapist's unacknowledged grief over certain aspects of his or her lost cultural history and language. The 101 multilingual therapists interviewed in Costa andDewaele (2012, 2013) acknowledged the potential for collusion. One of the therapists suggested that there was a joint unspoken union: 'We know nobody else understands us, it's only us ' (p. 7). This could lead clients to feel that there was a special and intimate relationship with the therapists, which transcended the professional role. The therapists highlighted the importance of attending to boundaries in a way that shaped clients' expectations about the limits of their role. The therapists also believed that, although there were concerns about possible collusion, the benefits were that they were able to help clients to feel more connected and less isolated. Similar views emerged from Verdinelli and Biever's (2009) research with 13 Spanish-English bilingual psychotherapists. The psychotherapists reported that they felt more connected to clients with whom they shared their native language (L1) and that this had a positive effect on the therapy. Amati-Mehler (2004) suggests that it is not only languagematching between client and therapist which enhances the capacity for attunement of the multilingual therapist. She suggests that as a speaker of multiple languages, her ability to hold multiple meanings and associations may make her more available for the complexity and disordered language of some of her patients. A multilingual person may experience less anxiety and greater ability to suspend immediate comprehension and to tolerate ambiguous meanings. In a similar vein, Jimenez (2004) suggests that his multilingualism allowed him to treat patients in Germany despite his imperfect knowledge of German. He suggests that his ability to move between 'the confusion of tongues' and the 'gift of tongues' contributed to the patient's movement from confusion to clarification.

Frustration because of lack of training
The majority of the therapists who participated in the Verdinelli and Biever (2009) study had taught themselves or had learned to provide bilingual therapy services by trial and error. They all acknowledged the lack of attention in current training and supervision to multilingual issues in therapy. The therapists in Costa andDewaele (2012, 2013) also mentioned lack of training and the difficulties they experienced working in their L1 in therapy where they did not have access to the professional vocabulary or experience in relating professionally in their L1: 'Well, when I was thinking about coming to do this interview, I wondered whether I wasn't really a fraud, because although I do speak various languages, I've always been trained in only one..., I find it incredibly difficult to explain, because I've never picked up a French textbook about CBT.' (2013, p. 8).

Language gap and switching
Although no therapist in Costa andDewaele (2012, 2013) had tried out inviting other languages into the therapy, they were interested and saw the potential of trying this. This is consistent with the experiences of therapists in Verdinelli and Biever (2009). These therapists did engage in code-switching (CS), but only 2 of the 13 participants had learned formally about CS. The others had discovered the advantage of its use by experience and by trial and error. Although Pitta et al. (1978) regard CS in therapy to be a useful tool, they caution that the client can choose languages which support their resistance to the therapy. This was a view shared by some therapists in Verdinelli and Biever (2009). Pitta et al. (1978) suggest that the dominant language has richer emotional structures which can capture greater richness of experience, while others feel that this language may not access the client's intellectual resources for making sense of experience and that the emotional potency of the L1 can impede cognitive processes. Szekacs-Weisz (2004) describes how some of her Hungarian clients who had followed her to London from Budapest used CS occasionally in therapy: 'I learned to pay attention to English words suddenly popping up in the verbal environment of the mother tongue. He said the first word in English while recalling a dream about a cockroach. Translating the word in Hungarian: svabbogar (Svab being the name of ethnic Germans in Hungary, bogar meaning bug), it became obvious that the pun in the dream is hiding painful associations to Germans. Following on this line, he became able to talk for the first time about silenced secrets and painful childhood memories of anti-Semitism'. (p. 26) Altman, Schrauf and Walters (2013) focused on CS in the autobiographical memories of mature immigrants. They point out that their personal memories include monolingual as well as bilingual events, and that they can be told in either language of their repertoire. Some narratives are monolingual while others are peppered with CS (p. 212). In their study of 12 English-Hebrew bilinguals aged 64-79, the researchers found that 40% of recalled memories where "crossovers", i.e. a different language from the language of the experimental session. These crossover memories had more frequent CS (p. 228). There were more than three times as many crossovers from L2 to L1 than from L1 to L2 (p. 230). The multilingual therapist's position with regard to the language gap can represent the transitional process between the old situation, which has been left, and the new situation in which the patient is living (Kitron, 1992). This type of therapeutic encounter may be a useful tool in helping to detect 'significant aspects of the patient's unconscious motivations' and ways of 'working through the relevant conflicts and resistances.' (p. 10). The language gap was also identified as a creative tension by multilingual therapists. Being able to tolerate uncertainty and ambiguity is a key skill for therapists. The gap produced by not-knowing can be a source of therapeutic spontaneity and creativity. Winnicott (1971) referred to this as the "potential space." Dewaele's (2013) study on language preferences for emotion among 1569 multilinguals revealed that participants reported CS significantly more frequently when topics being discussed were personal or emotional compared to neutral topics. Dewaele argues that when powerful emotions need to be verbalized quickly, and the speaker realizes that it would take too much time to express them in the weaker output language, possibly with unwanted pragmatic effects, CS might seem like an acceptable option (p. 215). One participant, Vally (Greek L1, English L2, French L3, Turkish L4) reported: 'I think when I talk about emotional topics I tend to code-switch to English a lot. I remember when I was seeing a psychologist in Greece for a while I kept code-switching from Greek to English. We never really talked about this (...) To my mind it may have been some distancing strategy because at the time I was trying to figure out what to do with my life ' (p. 206). Participants also reported that CS was often linked to a change in emotional tone (p. 205). This could include a change in intonation and prosody, increased volume, and faster speaking rate, conveying "emotional information above and beyond semantic linguistic content" (Nygaard & Queen, 2008, p. 10).
Dewaele & Li Wei (in press, a) found that their 2070 multilinguals had a generally positive attitude towards CS. In a further study on the same database Dewaele & Li Wei (in press, b) argued that CS can be a creative discourse strategy used by multilinguals in real-life interactions in order to achieve effective communication. It allows the insertion of threads in different colours in the exchange and hence insert an element of novelty, uniqueness and surprise. They found a considerable amount of interindividual variation in their study of self-reported CS. The degree of multilingualism in the participants' work Language and Psychoanalysis, 2013, 2 (2), 31-50 http://dx.doi.org/10.7565/landp.2013.0005 35 environment was linked to self-reported frequency of CS in a variety of contexts. CS was also linked to higher levels of multilingualism and early bi-and multilingualism. Female participants, extraverts and participants with high levels of Cognitive Empathy also reported to engage more frequently in CS.
One therapist interviewed in Costa andDewaele (2012, 2013) referred to her ability to play with understanding and lack of comprehension, asking her clients to define and explain what they meant by an idiomatic expression like "fish-wife" (p. 8). She felt that by not sharing a common L1 with the clients she was able to allow herself the freedom to be flexible and curious. There may also be a freedom for the client in the therapy where the client can move, through switching languages, between feelings of closeness and distance, power and powerlessness (Kitron, 1992). Szekacs-Weisz (2004) points out that CS can allow clients to find a way out: 'Patients can feel that therapy in their native language binds them in a position they want to move away from' (p. 27).  proposes that one of the ways in which multilinguals cope is by creating new selves for each of the languages spoken. Imberti migrated from Argentina to New York as a young woman and refers to the new self she had to create: 'When we change languages, both our worldview and our identities get transformed. We need to become new selves to speak a language that does not come from our core self, a language that does not reflect our inner-connectedness with the culture it represents.' (p. 71).

Identity
The bilingual therapists interviewed in Verdinelli and Biever (2009) refer to their sense of living in two worlds and that each language establishes and maintains separate cultures. Pavlenko (2005), referring to the data on language preferences for emotion among 1569 multilinguals, noticed that for those who were still in the process of second language socialisation, expressing affect in that language felt a bit fake, like the 'emotions of a different person' (p. 134). However, she does reject the essentialist view of the bilingual as having two languages completely insulated from each other, pointing out that languages and identity are dynamic (p. 189). De Zulueta (2006) pointed out that language is intrinsically linked with our sense of identity. Just as the mother tongue gives us a particular sense of self, a second language learned after puberty can forge a protective identity which can defend us from experiences which are too painful and overwhelming. The therapists in Costa andDewaele (2012, 2013) mentioned the shared identity with multilingual clients of living between two cultures '(…) a monolingual won't have that experience, of going home or thinking that home is elsewhere, or being bored as I was as a child, being dragged back home and thinking 'Oh but I really would like to go like everyone else (on holiday) to Portugal.' (p. 8).

Trauma and Shame
Research conducted by Tehrani and Vaughan (2009) focused on the way in which language can play a part in psychic repair where a traumatic event has occurred. They demonstrated how bilingual differences and CS in therapy can increase emotional mastery and how exploring past problems in a new light can be aided by a new language: '(...) where an individual is equally fluent in two languages the most significant factor in increasing the quality and emotional content of the recall is the language and context in which the incident was encoded.' (p. 11) Language and Psychoanalysis, 2013, 2 (2), 31-50 http://dx.doi.org/10.7565/landp.2013.0005 36 Various commentators on the life of Samuel Beckett have observed that it was by writing in French that Beckett was able to find his creative voice (Casement, 1992). Clare (2004) observes that for some people a foreign tongue 'can give voice to the words that could not be spoken in their own language. ' (p. 184.) The emotional potency of the L1 can impede cognitive processes: 'Sometimes the mother tongue is too close to home and is not conducive to thought' (p. 184).
The therapy case reviewed in Pitta et al. (1978) concurs with the therapists in Verdinelli and Biever (2009) who say that clients may use their second language to avoid painful and shameful memories. One therapist commented that she pays attention to this strategy and uses language to help clients to refocus and to stay with a difficult issue. Dewaele (2013) in his study of emotions in multilinguals, relates that several Arab and Asian participants reported that they switch to English to escape the social taboo in their native languages and cultures. We can conjecture that this may occur because the additional language can circumvent the superego (as embedded and encoded in the L1) and so taboo words or emotions can be allowed to be expressed in a way that would not be allowed in the L1.

Early memories and emotions
A review by Altman, Schrauf, and Walters (2013) on research into immigrant autobiographical memory showed that autobiographic memory associations and retrievals for events from childhood and youth (in the country of origin) are more numerous, more detailed and more emotionally marked when remembering is done in the L1 rather than in a subsequent language. Pavlenko (2012) reviewed clinical, introspective, cognitive, psychophysiological, and neuroimaging studies of affective processing in bilingual speakers in order to find out more about firstly, increased automaticity of affective processing in the L1 and heightened electrodermal reactivity to L1 emotion-laden words; and secondly the decreased automaticity of affective processing in the LX, which reduces interference effects and lowers electrodermal reactivity to negative emotional stimuli. Pavlenko concludes that L1 and LX affective processing 'in some bilingual speakers, in particular late bilinguals and foreign language users, respective languages may be differentially embodied, with the later learned language processed semantically but not affectively ' (2012, p. 405). However, the strong emotional associations of the L1 are not systematically positive. Pavlenko (2005) talks about Russian, the language of her early years: 'It is also a language that attempted to constrain me and obliterate me as a Jew, to tie me down as a woman, to render me voiceless, a mute slave to a hated regime. To abandon Russian means to embrace freedom. I can talk and write without hearing echoes of things I should not be saying. I can be me. English is a language that offered me that freedom (…)' (p. 22).

Research questions
1) Does CS in interactions with multilingual therapists originate in clients or therapists? 2) Is there a gender difference in self-reported frequency of CS by clients or therapists? 3) Does CS in therapy involve a change of emotional tone? 4) To what extent do the clients agree with 28 statements on linguistic practices with mono-or multilingual therapists, perceptions and attitudes towards mono-and multilingual interactions?

Participants
Former or current multilingual clients of psychotherapists were invited to participate in an online survey on their experiences. The only requirement was that they had to be bior multilingual. The authors specifically avoided approaching people known to themselves or to colleagues as clients for ethical reasons. The data was thus collected through non-probability, snowball sampling, i.e. recipients of the call were asked to forward it to friends and colleagues.
Over 200 participants agreed to fill out a short sociobiographical questionnaire and the main instrument. After eliminating those who had not completed all vital parts of the questionnaire, we retained 182 respondents. The sociobiographical questionnaire contained questions about sex, age, nationality, education level, language history and present language use, and the theoretical orientation of their therapist. A majority of participants are women (N = 141). The mean age is 42 yrs (SD = 12), ranging from 21 to 71. Participants are generally highly educated: 8 reported having a Diploma, 24 a Bachelor's degree, 72 a Master's degree, and 77 a PhD 2 . This majority of highly educated, mostly female participants is typical for this kind of data collection (Wilson & Dewaele, 2010).
The median score on a 5-point Likert scale for ethnic diversity during the participants' childhood was 2. However, the median score for ethnic and linguistic diversity in the participants' workplace was higher (4 and 3 respectively). Most participants had received therapy in the Psychodynamic approach (N = 71), followed by the CBT approach (N = 47), the Humanistic Integrative approach (N = 21), the Systemic approach (N = 10), Gestalt (N = 1), the remaining 32 participants were unsure about the approach or did not answer the question. Participants reported having had between 1 and 8 therapists, the largest group reported having had one therapist (N = 52), followed by two therapists (N = 49), three therapists (N = 43) with the remaining 38 clients reporting having had 4 or more therapists. To the question whether any of their therapists were multilingual 84 clients answered "yes", the remaining participants answered either "no" or "unsure". Among the clients who had multilingual therapists, 23 remembered having established this before the start of the therapy, 16 at the beginning and 8 in the middle of the therapy. The statistical analysis will only include the data of the 84 clients (64 women, 19 men) who had had a multilingual therapist.

Instrument
The main questionnaire was exploratory in nature. It contained 28 items in the form of statements with 5-point Likert scales (Strongly disagree, Disagree, Neutral, Agree, Strongly agree). The items covered linguistic practices with mono-or multilingual therapists, perceptions and attitudes towards mono-and multilingual interactions. Some statements were phrased as personal statements ('I', 'me') while others were statements about 'therapists' in general, the agreement with which, we assumed, would be coloured by clients' personal experience. The questionnaire also contained four open questions inviting participants to recall an instance in therapy where a language switch was significant, to reflect on a therapeutic benefit of a language switch, to remember the feeling of hearing (or not hearing) that the therapist was multilingual. This resulted in a database of around 11000 words. The questionnaire was pilot-tested with 10 clients. This led to the deletion of some items and the reformulation of others. The final version of the questionnaire was put on-line on Survey monkey and an open call was also addressed to multilinguals, including those who had participated in previous studies, asking them to forward the call to friends, colleagues or students. The questionnaire was anonymous. Because the Likert scale data are ordinal rather continuous, we have used non-parametric statistical techniques including Wilcoxon Signed Ranks test, Mann Whitney test and Friedman ANOVA tests.
The research design and questionnaire obtained approval from the Ethics Committee of the School of Social Sciences, History and Philosophy at Birkbeck College.

Results
A Wilcoxon Signed Ranks Test revealed that CS originated more from the 84 multilingual clients than from their multilingual therapists. The median score of therapists is 1.0 compared to 1.5 for the clients, in other words, the median difference between both groups is 50%, which is a highly significant difference (Z = -3.9, p < .0001). A Mann Whitney test showed no gender differences in self-reported frequency of CS initiated by the client and the therapist (Mann Whitney U = 595, Z = -.15, p = ns and Mann Whitney U = 588, Z = -.22, p = ns respectively). About a quarter of participants (N = 23) did not answer the question about CS with their therapists and changes in emotional tone. Twenty-four participants (39%) answered "no" to the question. The remaining 37 participants (61%) agreed that their CS was linked to a raised emotional tone. Friedman's ANOVA test for related samples revealed significant differences between the level of agreement with the 28 statements (N = 84, Chi 2 = 612, df = 27, p < .0001) (see Table 1). Table 1 Items (as participants saw them) ranked according to degree of agreement Items Mean Rank I consider that my language plays a role in how I behave in therapy.
21.9 I think the ability of a therapist is improved by working with people who speak a different first language (L1) from their own. 20.4 I think it is an advantage for the therapist to be familiar with a client's culture.
19.7 I think the therapist being from the same culture as the client is an advantage.
19.5 It is easier to form a therapeutic relationship with someone who shares a L1.
19.2 It is easier to form a therapeutic relationship with someone who does not share a L1.
18.5 I think that therapists with bilingual skills are able to understand clients in a different way from therapists who are monolingual.
17.6 I use more non-verbal forms of communication with people who do not share my L1.
17.1 I consider that the languages used by the therapists in therapy play a role in how they behave as therapists.
16.8 I think that the L1 of the client is not relevant in therapy.
16.1 I think that therapists' ability to speak more than one language attunes them more to cultural differences.
15.5 I avoid certain topics when talking to a therapist with whom I do not share a L1.
15.4 I feel that being able to work in a non-native language would give me more freedom to express myself.
15.4 I think therapists who speak more than one language are able to communicate more effectively with clients from different linguistic backgrounds.
15.1 I think that therapists who speak more than one language can understand clients whose L1 is not that of the therapist .
14.8 Therapists who speak more than one language can accommodate different languages more easily in therapeutic work with a client.
14.6 I think there are advantages to using a non-native language for the client in therapy.
14.0 Therapists with whom I share a first language relate differently from therapists with whom I do not share a L1.
13.0 I think a therapist feels less able to challenge clients if they share the same culture or language.
13.0 I think clients can use a non-native language as a distancing device in therapy (if the therapist understands that language). 12.7 I think that the L1 of the therapist is not relevant in therapy.
12.6 I think my proficiency in my L1 affects the way a therapist views me.
10.8 I think therapists can use a non-native language as a distancing device in therapy (if the client understands that language). 10.4 I think how the client relates to the therapist and the transferential projections are likely to be affected by the client's choice of languages used in therapy. 3

9.8
Working with the transference and the therapeutic relationship is easier when the therapist and client share a L1. 4 8.7 I avoid certain topics when talking to a therapist with whom I share a L1.
8.1 From my experience I believe that levels of empathy between clients and therapists are affected by the language in which the therapy takes place.
8.1 It is easier to express strong feelings and emotions in a non-native language. 7.8 The 17 items from the quantitative study with a mean rank of over 14.0 were distributed into three overarching themes: 1) The multilingualism of the therapist promotes greater empathic understanding; 2) Clients view their multilingualism as an important aspect of their sense of self and of their therapy; 3) Language switches in therapy are more frequent when the emotional tone is raised. With these three themes in mind, we investigated the 182 clients' responses to the open questions.

The multilingualism of the therapist promotes greater empathic understanding
Comments which referred to attunement, connection and the relationship were considered particularly relevant. C38 5 commented that the discovery that his therapist was multilingual increased his sense of connection: 'When I found out that my therapist was multilingual, I felt that I had much more in common with him regardless if we continued our interview in English. I felt that I could relate and reveal certain problems to him rather than being analyzed by a uni-lingual doctor.' (C38, French, Dutch, Italian, Spanish, English, German) It is interesting that the previous client does not state that he shared the same language as his therapist. This appreciation of multilingualism in the therapist per se is reiterated by another client: 'Being multilingual myself, I would feel they would understand me better (not even depended on the languages they speak), as I think growing up and living in the society as a multilingual person is very different from monolingual, so I would feel a bit closer to my therapist.' (C75, Polish, English, French, Spanish, Italian, German) However, a small minority of clients disagreed. This example refers to the level of language ability necessary for effective therapy to take place. The client describes an encounter with a non-native English-speaking therapist whose English skills he judged to be 'lower than my previous therapist'. This was the consequence: 'I didn't feel I could talk freely about my feelings and experiences in whatever words felt best, but I had to choose my words more carefully so that she would understand. I found this very distracting and off-putting…' (C115, English, French) The majority of clients' statements did concur with the finding from the statistical analysis that the multilingualism of the therapist led to greater empathy and understanding. One client referred to a sense of enhanced appreciation of cultural nuances by multilinguals: '…I believe that multilingualism enriches the mind and makes it easier to understand viewpoints, particularly some that may be very culture dependent.' (C63, Bulgarian, French, English, Russian, Dutch) Another client referred to her sense of connectedness through sharing a multilingual identity with the therapist: 'It would make me feel a connection with them because being multilingual changes you. It is something in common and can help communication because you have more ways to explain things even if you only share one language because there is an internal system change when one becomes multilingual that is shared even if the languages aren't.' (C14, English, Spanish, Portuguese, French) For this client, although the shared experience of multilingualism was valued, this was not related to the function of therapy. If therapy was not effective, multilingualism could not add much value: 'I gave up on the therapy after 4 weeks when I realised my One client valued the increased empathy and understanding, which she associated with the therapist's multilingualism: 'It means they have an interest in an "otherness" enough to learn the language, or might be multicultural themselves. It also makes aware of the nuances that might be missed and that the patient might have other contextual realities.' (C150, German, English, French, Spanish). However she added that this really should be the case for monolingual therapists too: 'But this should be realised within a single language relationship as well.'

Clients view their multilingualism as an important aspect of their sense of self and of their therapy
This broad theme was divided into 2 sub-themes: 1) Identity 2) Early memories and Emotions.

Identity
One client describes how she experiences herself differently in different languages, and that this had significant implications for therapy: 'I feel like a huge part of me just doesn't go to therapy with me. I have different personas with each language I speak so only speaking in English in therapy isn't helpful... If I have to translate into English… it just isn't the same for me.' (C14, English, Spanish, Portuguese, French) Identity and language are wrapped up with culture. de Zulueta's (1995, p. 179

'Early memories' and 'Emotions'
Many clients commented on the way in which their early memories were encoded in their L1. They felt it was important to be able to access these memories in therapy: 'My therapist did not understand my [L1]. However she asked me to talk about my childhood, which seemed irrelevant in the therapy in English. However when I mixed in some words from my [L1], it started to make more sense talking about my childhood. As if English language did not let my memories come back efficiently enough, and I just needed some key words in [my first language] to bring memories back.' (C76, Russian, Lithuanian, English, Spanish, French, Italian) Others commented: 'It [speaking in her L1] was very beneficial for my getting in touch with how I feel and felt in the past about my mother with whom I only spoke my mother tongue, and so I needed to "speak" with her (in my mind) in that language. I could not begin to really feel what I would "say" to her unless I imagined the words in my native tongue.' (C166, Slovenian, English, German, Croatian, Spanish) 'Once, I was describing my stepfather's house and the words for a lot of elements in the building just came out in Spanish. It was a bit like unblocking of visual and emotional memory…the colours of the elements returned to my mind and gave me the emotional resonance of the place.' (C6, Spanish, English)

Code-switches in therapy were common when the emotional tone was raised
Clients gave numerous examples of CS to create greater emotional proximity or distance. Most welcomed this as a resource to deploy in their therapy. Some were less convinced. Most notably those who had had psychoanalytic therapy, found CS served only to further Language and Psychoanalysis, 2013, 2 (2), 31-50 http://dx.doi.org/10.7565/landp.2013.0005 43 complicate their struggles to make themselves understood: 'I think there would have been more "misunderstandings" and word (power) games. Actually I felt misunderstood enough in a monolingual setting as it was… I endured it for a year, five times a week.' (C66,Spanish,English,French,Italian,German) Another client saw no point in CS if the therapist did not share the same languages: 'I suspect that undergoing therapy in a language not your own is like pisser dans un violon' [urinate in a violin] (C182, French, English, German, Spanish) The majority, however, enjoyed CS, providing the therapist was open to the idea: 'The therapist was not necessarily fluent in the language but was simply present as I expressed myself. It felt liberating and allowing. I have applied this sometimes with clients myself. It doesn't matter whether the therapist understands the actual language spoken: there comes a point where I as a client am invited to hear and listen to myself. This is very helpful, in the presence of another benevolent being.' (C168, French, Italian, English, German, Spanish) One client, however, cautions that as well as being open to the idea, the therapist needs to be skilful in working with the CS. This client explains how she was invited to switch languages in order to increase the emotional resonance of her description. However: 'I didn't find much comfort in confiding this to someone who probably didn't understand its various connotations -I felt more alone than when I explained in English how I felt about the incident.' (C178, English, French, German, Italian, Japanese) C114's experience was more positive. He describes how the CS changed the emotional tone both for him and for the therapist: 'It was easier to "let myself go" in Spanish and easier for the therapist to notice that I was NOT a stiff upper lip…as long as we were speaking in English both of us were less ready to express emotions. We used more formulaic expressions for conventional small talk phrases, like "I am not at my best" instead of Spanish "me siento como un perro mordido" (I feel like a bitten dog)…Spanish allowed for code switching.' (C113, Polish, German, English, Spanish) C63 feels that CS is a natural way of expressing herself: '… describing a situation or a sentiment idiomatically in one language provides better approximation to the "real" thing and expresses more subtle nuances. In my experience this happens automatically. If in one of the languages I speak there is an expression like that, it does come to my lips whether I want it, or not. Then it's up to me to let the lips share it, which I usually do.' (C63, Bulgarian, French, English, Russian, Dutch) Interestingly, although, some therapists expressed concerns about the safety of code switching in Costa & Dewaele (2013), many clients had different views. Some said they found the ability to switch languages helped them feel safer when talking about very difficult topics: 'For me using English when describing something very delicate or important to me is like having a safety net or a parachute. If I cannot say it properly in German, I know I shall be able to say it properly in English. ' (C28,French,English,Spanish,German) A number of clients made reference to the emotional charge in different languages with reference to the transference: 'The mother tongue i.e. the language your mother spoke to you in is highly significant in the transference.' (C126,Guajarati,English,French,Spanish) Others referred to the languages in which they conducted different relationships: 'It helped to be able to switch languages when talking about things that transpired between me and my ex-boyfriend who was Mexican. It helped to be able to describe some of our interactions using the language (Spanglish) that we interacted in.' (C73,English,Spanish) Some made reference to the power differentials when different languages are spoken with varying levels of fluency by the therapist and clients in couples/family therapy. One client described a couples' therapy intervention where the therapy was conducted in his wife's first language and his second language. He had tried to switch languages but was not understood and felt he could not express himself: '…so in the end I had to use "nextbest" phrase strategies (…) the lack of switching was what was significant for me.' (C42,English,Turkish,French,Spanish,Hausa) There are two particularly significant areas singled out by clients for code switching in therapy: trauma and shame. In both these areas code switching is employed to achieve an increase or decrease in emotional tone.

Trauma
Being able to access more than one language for the re-telling of a traumatic experience can be a useful resource for therapy. For some it can provide access to a less emotionally charged medium. As C71 said: 'I felt more comfortable speaking about traumatic events in my non-native tongue. I feel that in my particular case I was able to let go of pain easier thus.' (C71, Hungarian, German, English) For others, the increased intensity of a language can help them to cope with a traumatic event. C168 found that she was better able to process the trauma by describing it in the language in which it occurred: 'I remember being given permission/being asked to express a traumatic incident in the language in which it happened. This I found very liberating. ' (C168,French,Italian,English,German,Spanish) Shame Clients frequently mentioned the use of CS to avoid cultural constraint associated with using their own language. C81 describes using a second language for: '… speaking about topics which I was ashamed of. It is a way to put facts in the distance' (C81, French, English, Spanish, German) Another mentions a freedom from taboo in his LX: 'I was not carrying as much cultural baggage when I spoke French in therapy. I felt more at ease talking about "taboo subjects" [sex] in therapy in French than in English. I felt I was more distanced from the "controversial subjective" and probably culture-based aspects of sex. No sex please we're British vs. a freer attitude. ' (C50,English,French,Spanish,Swedish) Sometimes, not sharing a L1 with the therapist can allow the client some distance from the experience of shame while allowing the maximum toleration of feeling. C184 mentions: '…having said a heavy swear word in my mother tongue and felt not as ashamed in thinking my therapist would not understand exactly the heaviness nevertheless understanding its connected sense of feeling and favouring a cathartic episode. ' (C184,German,Sardinian,Italian,English,Spanish) Alternatively, using a second language may temper the strength of the expression of an emotion: 'I've never switched the languages but, saying this, I can say that I struggled to express my anger on a few occasions. I know I could easily do this by using, for example, swear words in my native language. Swear words in a second language do not have as much strength. ' (C92,Polish,English)

Discussion
In response to our first research question, statistical analysis revealed that participants who had had multilingual therapists reported switching language in interactions with their therapists significantly more frequently than therapists. In other words, clients who knew they had that possibility were more likely to switch. Recommendations are made in the literature (Pitta et al., 1978;Verdinelli; for the therapist to initiate CS in therapy and thus avoid the use of language-switching as a form of resistance. However, clients in this study welcomed the ability to initiate CS themselves as a way to connect them with the intensity or to allow them the distance they needed in a given moment. That is to say that clients valued the potential to manage the emotional flow themselves. This is consistent with the argument put forward by Dewaele (2013) in his study on language preference for expression of emotion. He suggests that people often choose to verbalise strong emotions in a language which allows them speed of expression so that the potency is maintained, illustrated in the current study by the phrase: 'let the lips share it' (C63). Clients also referred to the way in which switching to another language could help them to 'break free from cultural boxes' (C153). Similarly, the therapists in Costa andDewaele (2012, 2013) regarded the language gap as a source of creativity.
The answer to the second research question is negative as no difference emerged between male and female clients in their self-reported frequency of CS initiated by themselves or by their therapist.
The answer to the third research question is less clear-cut, although it went in the expected direction: 39% of the clients who had had multilingual therapists and answered the question felt that CS was not linked an increased emotional tone, the remaining 61% of clients reported that CS in therapy involved an increased emotional tone.
The answer to the fourth research question was based on an analysis of the Likert scale values for 28 statements on linguistic practices with mono-or multilingual therapists, perceptions and attitudes towards mono-and multilingual interactions. The statistical analysis showed significant differences in levels of agreement with the statements. Three broad themes emerged from the 17 items with the most positive ratings: 1) the therapist's multilingualism promoting empathy; 2) the clients' multilingualism as an important aspect of sense of self and of the therapy; and 3) the increased CS when the emotional tone was raised, especially when dealing with trauma and shame. These themes also appeared spontaneously in the feedback of clients. Unsurprisingly, many of these themes have been observed in earlier studies. Costa andDewaele (2012, 2013) found that the majority of the multilingual therapists believed that their multilingualism increased their ability for attunement and empathic understanding with their client. Amati-Mehler (2004) believes her ability to attune to complexity is enhanced by her ability to hold Language and Psychoanalysis, 2013, 2 (2), 31-50 http://dx.doi.org/10.7565/landp.2013.0005 46 multiple meanings as a multilingual. The majority of clients felt that language-matching with a therapist was not necessary but they felt more connected to a therapist if they knew that they were multilingual, an opinion shared by the therapists interviewed in Costa andDewaele (2012, 2013). Clients also stated that a multilingual therapist would be more aware of 'nuances that might be missed and that the patient might have other contextual realities.' (C140) Many clients also reflected on the importance they attributed to their multilingualism in terms of their sense of self and their identity and a sense of having multiple personas (C14). This is a popular theme in multilingualism research, especially the richness and the freedom that cultural hybridity provides (Dewaele, 2013).
Clients also referred to the way in which their early memories were encoded in their early languages and how it was important to be able to express their feelings in the corresponding language. (C76, C166). This is consistent with the findings from studies on autobiographic memory associations (see the review in Altman et al., 2013) and Pavlenko's (2012) argument that bilingual speakers may process the later learned language semantically but not affectively.
The answer to the final research question, namely the importance for the therapist to create an environment where issues of multilingualism can be addressed in the therapy, was largely affirmative. Clients reported positive experiences and welcomed and valued an environment where multilingualism had a place in the therapy. They appreciated it when a monolingual or multilingual therapist invited them to speak in other languages whether or not the therapist understood and they felt they were able to engage more fully in the therapy (C76, C166, C6, C168). Antinucci (2004) and Akhtar (2006), both psychoanalytically trained therapists, recommend that the multilingual therapist should be mindful of the potential to collude with the client's mourning or idealisation of a lost culture or language. Antinucci (2004) suggests that the therapist should address the fantasy around the shared "otherness" with the client. Clients of psychoanalysts who took part in this study had different experiences. For example: 'There was no such disclosure and you are not allowing for other kinds of analysis -indeed, I don't think such disclosure would have been permitted, or at least I cannot recall that there was an offer that it was going to be.' (C66, Spanish, English, French, Italian, German).

Conclusion and implications for practice
The present study aimed to focus on the voice of the multilingual client in therapy. Having collected quantitative and qualitative data via an on-line questionnaire from a relatively large sample of multilingual clients (N =182) from all over the world, who had been exposed to a wide variety of therapeutic approaches, we have a database with strong ecological validity (Wilson & Dewaele, 2010). In other words, "local" effects are unlikely to have overly influenced the patterns that we uncovered. However, because our respondents were generally highly educated and only a minority had traumatic experiences of migration, we cannot claim that our findings apply to all multilinguals in therapy. More research is needed on multilinguals who have experienced traumatic migration and for whom language differences are not seen as benign but may have been part of the traumatic experience (for example the languages in which torture or political strife may have been conducted). Another limitation of our research is the fact that not all respondents had had therapy with a multilingual therapist. Respondents left responses Language and Psychoanalysis, 2013, 2 (2), 31-50 http://dx.doi.org/10.7565/landp.2013.0005 47 blank if they had not had the experience and made it clear in their Open Box answers if they were writing about real or imagined experiences. Only real experiences were analysed and reported on. Some did report the very real frustration of not having had a multilingual therapist.
The patterns that the current research revealed were that clients report more frequent CS than their therapists, and that CS occurs more frequently when the emotional tone is raised. Further analysis revealed that clients use CS strategically when discussing episodes of trauma and shame. CS allows them both to gain proximity or distance according to the need. Clients reported that being able to switch between languages allowed them to express themselves more fully to the therapist, being heard in "stereo" rather than in "mono". The ability to do so added depth and nuance to the therapy, a view defended in previous research (Espin, 2013;Kokaliari et al., 2013;Verdinelli, 2009). Many of the views of the multilingual clients correspond with the views of multilingual therapists. Unlike the therapists however, the clients did not perceive their language switching as a means of avoiding emotional depth, there was general agreement that the therapist's multilingualism promotes empathy and that a client's multilingualism constitutes an important aspect of the client's sense of self, which has strong implications for the therapy. Multilingual clients benefit from a therapeutic environment where multilingualism is appreciated, and where CS is possible (Martinovic & Altarriba, 2013).
Although clients may not be aware of the lack of training as cited by multilingual therapists (Verdinelli, 2009;Costa & Dewaele, 2012 some of their comments pointed to this lack. One mentioned the aloneness they felt when invited to speak in their language with a therapist who lacked comprehension (C178). Another alluded to the fact that the qualities she valued in a multilingual therapist: 'should be realised within a single language relationship as well.' (C140) Three key implications for practice which emerge from this study are: the inclusion of multilingualism and implications for therapy into the curriculum of psychotherapy and supervisor trainings and accreditations; attention to the role that multilingualism can have when working psychotherapeutically with trauma; attention by monolingual and multilingual therapists to the way in which they invite a client's multilingual personas into the therapeutic space. The responses of the multilingual clients in this study remind us of the role languages have in regulating affect, expression of emotion and reflexivity. For all therapists, from all language backgrounds, this is a clear message.