DISCOURSE MARKERS AS A STRATEGY OF CODE-MIXED DISCOURSE IN A GALICIAN-SPANISH-ENGLISH COMMUNITY

The present paper will try to study the point a specific bilingual community (Spanish-Galician-English in London) is at in the bilingual continuum, whether it is in transition to code mixing or rather there is an emergence of a mixed code, which we can infer from the use of functional elements such as discourse markers and interactional signs. The data consists of four conversations among Spanish/Galician/English bilinguals. All participants belong to the bilingual community under study and can be considered complete/full bilinguals (i.e. with fluency in all languages used). One of the key issues in the data is that there are significant patterns of discourse that were separated from markers framing it in language: English markers framing Spanish discourse. This is an indication of code mixing – where a structural pattern that has been grammaticised at the textual level can be perceived.


INTRODUCTION
The term "bilingual" includes different definitions and concepts. Generally speaking, the term refers to the ability of communicating in two languages. However, there are many individual differences between ability and use that vary across a continuum. In order for linguists to define the term, the moment of acquisition of the different linguistic varieties has always been key in determining the type of bilingualism, i.e. a person who learns two languages from birth is denominated "simultaneous bilingual", as opposed to a person who learns a second language which is referred to as a "consecutive bilingual". In addition, if the definition is applied to the competence in each of the varieties at use, the term is generally described on a continuum, in which the ideal bilingualism stage would be the "balanced bilinguals", who are those equally fluent in two languages. However, even if someone is highly proficient in two or more languages, his or her so-called "communicative competence" or ability may not be and in fact hardly ever is fully balanced. As Baker (2006) states, defining exactly who is or is not bilingual is essentially elusive and ultimately impossible. For this reason, many theorists are now beginning to view bilingualism as a spectrum or continuum of bilingualism that runs from the relatively monolingual language learner to highly proficient bilingual speakers who function at high levels in both languages (Garland 2007).
One of the most striking features of bilingual speech is the alternation of the different linguistic varieties that speakers have at their disposal. Code alternation typology includes "code-switching", as well as "code-mixing". Ascertaining whether we are dealing with code mixing or code switching in bilingual code alternation is again complex. The starting point for this paper must thus include an explanation of how these concepts are understood, specially as it is not rare to read "code mixing" being used synonymously with "code switching". Research from a decade ago has given new meaning to the term "code mixing" which had not been explained until then; Maschler (1998: 125), among others, defined code mixing or a mixed code as "using two languages such that a third, new code emerges, in which elements from the two languages are incorporated into a structurally definable pattern", whereas the term "code switching" would be used for juxtapositions with pragmatic and functional meaning (Auer 1999). Taking into account the different types of code alternation phenomena, research has been carried out in order to find out which type of alternation occur in specific bilingual communities (Pena 2006). In this direction, discourse markers and interactional signs play a highly significant role, as they could be determinant to reach conclusions about the bilingual stage and continuum at which individuals and, subsequently, their community may be.

THE COMMUNITY UNDER STUDY
The present paper will try to study at which point in the continuum a specific bilingual community is; whether they are in transition to code mixing or rather if there is an emergence of a mixed code that we can infer from the use of these functional elements. For this reason, a bilingual community was chosen to examine its use of discourse markers. The data consists of four conversations among Spanish/Galician/English simultaneous bilinguals in London. Each segment is about one hour long and are taken from casual conversations that took place in London in 2000. All participants belong to the bilingual community under study and can be considered complete/full bilinguals (i.e. fluency in all languages used).
The data used for the research was based on face-to-face interaction. Participant observation and tape recordings were made of conversations occurring in actual social encounters. Although this type of fieldwork procedure normally concentrates on the stylistic dimension of linguistic variation (e.g. same speaker's language choice in a range of situational contexts), for the purpose of this research, various speakers who supposedly have similar linguistic features were studied in similar situational contexts but in conversations ranging from topics such as work to family or friends. The data gathering approach chosen was a mixture of 'overt' 1 and 'covert' 2 , where one of the participants was in charge of the recording and, accordingly, knew they were being tape recorded (in all cases, she managed to hide the tape recorder) but the rest of the participants did not. After the recordings, permission was asked to all participants in order to have their approval to publish the data without their names, as some instances of the conversations contain personal comments.

THE GALICIAN COMMUNITY IN LONDON
Galicia is located in north-western Spain. It has two co-official languages, Spanish and Galician. It is taught bilingually alongside Spanish in both primary and secondary education, and is used as the primary medium of education at universities in Galicia. Most of its population is bilingual.
During the 1960s and 70s a very large number of Spanish immigrants settled in London. The majority of these were from Galicia. Today, it is estimated that there are 35,000 Galicians living in London. Most of these people had no studies and went to work as waiters, cooks or cleaning personnel in hotels and hospitals.
Although by the end of the 80s many went back to their towns in Galicia, a large number settled down in their adoptive country. Some own their own houses, have their own businesses and have adapted to London fairly well. There is a Spanish bilingual school in London (I.E.S. Vicente Cañada Blanch, founded in 1972) with almost 500 students, of which 80% are second and some third generation Galicians in London.
The first generation settled in London, has given way to a large second generation who has mostly studied and proceeded to higher education. The difference between them and their parents is that they are all fluent in English and Spanish, they normally speak English at work and with their English friends and Spanish or Galician at home or with Spanish friends. Their friends are not limited to other Galician or Spanish people, and their lifestyles are normally the same as that of any other British Londoner. This second generation is employed within a broad spectrum of industries.
The second generation members used for this study consisted of 10 speakers (eight females and two males) in the age range of 23 to 28. They all studied at the Spanish bilingual school in London, have lived in this city all their lives, their 1 Approach used for the gathering of data in which the fieldworker asks formally for permission to carry out the fieldwork to the participants and is normally considered an outsider. 2 Data gathering approach in which the fieldworker first assumes some participant role provided by the setting and begins formal research when some kind of informal and mutually beneficial relationship is established with the people in the field. parents are Galician, they speak either Galician or Spanish at home and while on holiday and during various trips to Spain. They use English at work and in their daily activities. They all have university studies and work or study in London.

DISCOURSE MARKERS
As Stenström (1994) points out, when people talk they use a set of extremely frequent (single and multi word) items to start, carry on, and terminate a conversation. Some of these items constitute turns of their own, others link turns together, while still others serve as stallers, frames and empathizers within turns and many of them can do more than one thing in discourse. Discourse markers serve to start a conversation, to introduce and mark the end of a topic, to introduce a digression and mark the resumption of an old topic and to signal the end of a conversation.
According to Maschler's careful categorization (1998: 127): …in order to be considered a discourse marker, the utterance in question is required first of all to have a metalingual interpretation in the context in which it occurs (that is, rather than referring to the extralingual world, it must refer metalingually to the realm of the text or to the interaction between its participants). The second requirement has to do with structure: the utterance in question must appear at intonation-unit initial position, either at a point of speaker change, or in samespeaker talk, immediately following any intonation contour which is not continuing intonation (i.e. not after a comma in the transcription). It may occur after continuing intonation or at non intonation-unit initial position only if it follows another marker in a cluster.
In the data many instances were found of items, which could be considered what Poplack (1981) or Romaine (1995), among others, considered 'tag-switching' and which are also called discourse markers. According to Romaine (1995: 122) "tag-switching involves the insertion of a tag in one language into an utterance which is otherwise entirely in the other language" and Poplack (1981) points out instances of tag-switching, when speakers insert a tag -a non-functional, usually unrestrained utterance, as in the English 'you know', 'I mean', 'like' -from one language into another. They are subject to minimal syntactic restrictions and therefore, can be easily inserted at different points in a monolingual utterance without violating syntactic rules.

DISCOURSE MARKERS IN THE DATA
In the recorded conversations 208 discourse markers were found. 197 were English words inserted into Spanish utterances, 5 were Galician words inserted  (6) were Spanish words inserted into English.
The most frequently used markers were found mostly in long turns where the participant spoke about emotional or private issues and the highest frequency tags (you know, yeah) were used by the same person. The following table (Table 1) lists all the lexical items found in switches, incorporated due to the importance discourse markers are proving to have in determining whether we are dealing with code switching or code mixing, as previously explained.
The most frequently used item is you know, which is a discourse marker which can be described as an emphatizer, used to engage the listener and make her/him feel part of the conversation. It often appears at the beginning or end of a turn, but also to change footing, to change the frame of events: Example: Yo me crié / you know / me conoce desde que era pequeña 3 The frequent use of you know implies the participant's need to achieve a positive response from the interlocutor and to bring her/him into the conversation, onto closer ground where the other participant feels and understands the exact point of what the speaker introducing the discourse marker is trying to convey.
Example: P1: mamá mamá you know / a mí lo que me jodía es que mis amigas me decían ah tu eres hija única P2: yeah In return, the listener uses 'yeah' (the second most frequently used item) as an interactional signal, to let the speaker know that s/he is following the conversation and remains interested.
In the participants' conversations, most discourse markers and interactional signals are switched into a code different from the one being used in the utterance. Not only are they used as in monolingual speech, to serve their discourse function, but also to emphasise and assert this function. In the case of you know inserted into a Spanish utterance, the speaker could very well have used the Spanish sabes which has the same structural and argumentative properties. Discourse markers, be they fillers (I mean, you know...), frames (right, anyway, now...), hedges (actually, obviously, really...) or links (and, but, because, so...), as well as interactional signals (yes, sure, gosh) are thus used to express a change of footing or used as a contextualisation cue. As a result, one might erroneously conclude that they indicate the alternations being used correspond to code switching; however, they are not lexical but structural elements which acquire a different value by being used in a different language to the surrounding one. They serve to structure a different language code than the others at work, i.e. they structure the new mixed code.
Other studies on the topic present similar conclusions. For example, Oesch Serra (1998: 101-122) analyses Italian/French mixing and discusses the emergence of a mixed code by showing that structural elements, such as discourse markers and connectives, acquire different values in bilingual speech from those in the monolingual modes; and Maschler (1998: 125-155) shows that the emerging English/Hebrew mixed code can be recognised by the separation of discourse markers and conjunctions through language, i.e. discourse markers in Hebrew and conjunctions in English.
As Stroud (1992: 124) notes, when analysing Taiap-Tok Pisin bilingual data, the function of discursive elements is "to maintain the floor, and structure information into bounded units... a means of segmenting... talk and signalling the introduction of new information". As stated above, in the particular bilingual community under study, they are mainly used to mark a contrast between the codes being used, especially to set off a change of footing and for the purpose of creating a new structural mixed code.
Following the work of Oesch Serra (1998) and Maschler (1998), as referred to above, we can conclude that discourse markers tend to occur in the same position and for the same function as they would in monolingual use (i.e. they are placed according to the surrounding language's grammatical norms) and this fact can prove that we are dealing with a mixed language mode which is at an initial stage.

BILINGUAL SPEECH
The term 'code alternation' covers all instances of locally functional usage of two languages in an interactional episode. It may occur between two turns or turninternally, it may be restricted to a well-defined unit or change the whole language of interaction and it may occur within a sentence or between sentences. Two types of code alternation phenomena are 'code switching' (code alternation with functional and contextualisation meaning) and 'code mixing' (code alternation where there is no functional or contextualisation meaning but it becomes a code of its own). Furthermore, the mixed code can evolve into grammaticalised forms (Spanglish from the US has many examples, such as "me dieron el saco" "they sacked me") denominated 'fused lects' 4 .

DISCOURSE MARKERS IN BILINGUAL SPEECH
According to Maschler (1998: 137) there are three issues when dealing with the question of when code switching grammaticises into a new mixed code: if the ad hoc switch is a recurrent pattern, if there is some structural pattern which can be distinguished in the new code and, thirdly, if the switched elements in the mixed code maintain the function they had in the language in which they are uttered or are used as in the surrounding language. From his analysis of Hebrew/ English discourse markers in bilingual data, he states that the options for choosing a marker from either language decrease the more specialised that marker is, i.e. the more specialised a marker is, "the more likely it is to be used exclusively in only one of the languages and therefore the more we would expect it to have undergone grammaticisation in the mixed code" (1999: 141).
In the bilingual participants under study's language behaviour these issues are clearly reflected in the data. The pattern of use of discourse markers being juxtaposed is recurrent; elements are almost exclusively in English surrounded by Spanish conversation. There is evidence that juxtapositions are following a pattern which is on its way to becoming structured as such, although it has not fully sedimented, the speech behaviour seems to be at a point at which code switching has definitely been implanted and code mixing occurs but not as much, or not in the same way, as in other more stable bilingual communities with a longer language alternation tradition. Evidence of this is that many of the alternations in the code mixing utterances carried functional meaning, which made it very difficult to ascertain where they belonged; they were not prototypical code mixing instances, but somewhere in between code switching and code mixing.
In the data, a change of activity type normally includes a discourse marker: a move to a new conversational action can be made by the use of a discourse marker, which can introduce a different language of interaction. Schiffrin (1987) and other linguists studying discourse markers have shown that the need to create conversational boundaries motivates the use of discourse markers and when no conversational action boundary takes place (as in the case of the use of conjunctions) there is no language switch.
In order to investigate how code mixing turns into fused lects we can either check where it initially originated (i.e. how the particular community being analysed used language alternation in the past and how this language alternation evolved), which would be difficult to tell unless the analyst has been gathering data from a bilingual community for entire generations, or we can analyse particular cases of code mixing which are moving to fused lects to try to establish what is prompting this transition. This transition usually has a starting point, such as the juxtaposition of relatively unbound elements of grammar, like discourse markers, conjunctions, or certain adverbials which serve to modify the utterance (i.e. with discursive rather than referential function); for example, in Chicano data we may find all Spanish discursive elements replaced by English ones, so that the Spanish system of discourse and text organisation is replaced by the English one.
The following example clearly shows an English discourse marker replacing a Spanish one ('you know' is inserted into the Spanish language of interaction being used at that point): 1 P2: estaría guay que en esa época hubiese yo qué sé un concierto o algo y con you know § 2 P1: § El Último de la Fila // 3 P2: ¿está? This discourse marker, which can be described as an emphatizer (it involves the listener), is the most widely used in all the data; it can be used to engage the listener and make her/him feel part of the conversation, as in the above example. It often appears at the beginning or end of a turn, but also to change footing, to change the frame for events as in the following extract: P1: ... yo me crié you know like // me conoce desde que era pequeña ... Translation: I was brought up you know like / / he knows me since I was small It is also used elsewhere when the speaker appeals for feedback. It sometimes appears in long turns, usually to call attention to the listener: 5 P1: ... mamá mamá you know / a mí lo que me jodía es que mis amigas me decían ah tu eres hija única 6 P2: yeah In return, the listener uses 'yeah' (the second most frequently used discursive item) as an interactional signal, to let the speaker know that she is following the conversation and remains interested.
As stated above, in the participants' conversations, most discourse markers and interactional signals are switched into a code different from the one being used in the utterance. Not only are they used as in monolingual speech to serve their discourse function, but also to emphasise and assert this function. In the case of 'you know' inserted into a Spanish utterance, the speaker could very well have used the Spanish 'sabes' which has the same structural and argumentative properties.
Interjections and discourse markers are normally inserted in the 'other' language being used. Examples of this are frequent throughout the data: P8: ahhh / okay! 8 P7: yeah! entonces cuando la tía se iba a volver / ellos no querían que yo me fuera / me estaban ofreciendo mogollón de chollos en la compañía / pero nada / no había nada fijo / and I just went / come on / you know! 9 P8: yeah! 10 P7: ¡no! ¡a dónde voy a ir / aquí me mandan a Madrid y a Méjico man ! 11 P7: ¡y a mí me da el palo! el viernes / cogí un cabreo / cogí un cabreo con ella // y ella al final me decía venga salimos // era la una de la mañana / y

DISCOURSE MARKERS AS A STRATEGY OF CODE-MIXED DISCOURSE IN A GALICIAN-SPANISH-ENGLISH COMMUNITY
estábamos aquí / y le decía // oh wait! ha pasado la una de la mañana pero tiene sentido! 12 P7: pero me cabreó (4") ya le voy un día le voy a decir es que siempre un sábado vamos a ir y le voy a decir / no man! me vas a dejar you know a las dos de la mañana porque tú tienes que irte a casa P7: y gente que está en Camariñas (( )) yeah! The same also applies to discursive markers, which tend to be used in the other language available to bilinguals (throughout the data, the participants tend to use English discourse markers in Spanish surrounding passages). Below are instances from these conversations, in which discourse markers in alternation are very frequent. It is especially noticeable that almost all of them belong to P10, who uses them constantly: 14 P10:¡yo que sé / hombre // pero no hay chollos ain't it? 15 P9:shittt! /// ¡qué chulo! 16 P10:yeah! /// 17 P9:¿y Yolanda? 18 P10: pero una pasada / fue / cuando ella vino de Salamanca /// nos encon-salimos a tomar algo ella Ana y yo // yeah? // y después / después I went on my year abroad y volví // anddd // yo volví en su cumpleaños ese noviembre y yo volví en diciembre y le traje un regalo de cumpleaños // se lo dejé en casa de Montse // 19 P10: § está hasta las pelotas de ella yeah? / y ahora Yoli le dice ¿¡sííí por qué te acuerdas!? y Montse is just like / yeah yeah! you know / se nota mogollón y ella nada 20 P10:es que todo es una mierda though / cierra a las dos y media o así 21 P10:fue fue ¡ufff! / fue una pasada mira / él / desde semana santa él tenía // ¡una mala tos! / él cada dos segundos él hacía ahm ahm / yeah and that was it / and then mi tío mi tía le comió el coco vete al médico vete al médico / fue al médico y le dijo / ah / es / un catarro mal curado yeah? /// y después fuimos a España / yeah? // porque además el año anterior yo me lo había pasado siempre con él / como estaba en La Coruña In turn 22, below, the participant uses English in line 3 to add extra information (that she was studying) to set the situation, surrounded by Spanish. She then inserts a few English words that she probably normally uses in this language (recruitment agency, manager) or it may be a lack of competence in that particular situation (job hunting, which she has always done in London and therefore might not know the equivalent in Spanish). She also uses discourse markers in English, as well as reported speech. In isolation these alternations do have pragmatic force; however, the turn as a whole loses this force and the contextualisation cues become very subtle. It is quite clear that this is not a code switching exchange, but a mix due to the structural load and loss of meaningful juxtapositions: P7: yo antes fumaba sólo Marlboro // y cogí un catarro / antes de Navidades /// ¡fatal! / y no me curé bien // y el día de Navidad me volvió pero fatal // llegué I was studying at the time yeah? /// y llegué /// y yo trabajaba // dos días a la semana trabajaba en aaa / en el recruitment / en el recruitment agency // yyy / y yo cogiera el trabajo en Decaux yeah? y me dijera el agency que yo llegara el lunes y ya me sentía mal no pero / pero era el manager // Sophie no vengas mañana because you know I'd rather you go to Decaux / then come to us because you know / it's more important for you to keep your job in Decaux and you're not gonna get it again // y yo yo / no / I'll be fine! // llegué el martes / fatal /// pero tomé las pastillas yeah y fui a trabajar / en el recruitment agency / y el miércoles me levanté / la cabeza // me daba / mira de esto que mueves un poco y te hace / bum bum bum bum bum The same occurs in the following turns, in which P7 mixes English for reported speech and discourse markers, yet the changes from Spanish to English (and vice versa) are not predictable at all: P7: ¡claro / es que quedó cao! / you know / y paso de eso pero es una cosa que aprendí y yo sabía que yo podía tomarme otro y tal / y ella no / you had enough // I went / well I'll have it! / he just looked up / super enfadado and I went saying fucking (( )) and she goes // venga vamos come on // let's go! / and I said ¡vamos! and we went to get money y aparece él // oh how did it go? and he goes oh I'm not getting anything like these // and he / apparently last Friday pay day and they went out and apparently ella no se acuerda // apparently que él le dijo (( )) P7: yyy / entonces cuando volví la llamé yeah? and I said come by tal y cual and we made up (( )) and then she goes me marcho pa' la semana or whatever and I go fuck I just got in and she goes / escríbeme P7: ¡claro! and she goes / escríbeme and I said ¡sí joder escríbeme y te escribo! / y me escribió // and I didn't like that // and I went at least two weeks ago again and anteayer ví a los padres /// y yo estaba // ¡hola hola! you know // Antonio y tal /// you know ¿qué tal está Julio? y ellos ¡ahhh está muy bien! // se viene en agosto y yo y ¿cuándo viene? a finales // ¡me va matar! / ¿por qué? / porque escribió dos veces y aún no le escribí // y el padre ¡cómo te pasas! y yo ¡no! ¡he empezado la carta / but I'm just come on! shit!

CONTEXTUALIZATION CUES
Every day, speakers use a repertoire of vocal and non-vocal (gestures, prosodic cues, etc.) means to express themselves. These semiotic resources, which can be

DISCOURSE MARKERS AS A STRATEGY OF CODE-MIXED DISCOURSE IN A GALICIAN-SPANISH-ENGLISH COMMUNITY
denominated 'contextualisation cues', are crucial in order to create the proper context in which the verbal message is to be understood. Contextualization is vital for communicating effectively and it can include such aspects of context as the speech genre, speech act, key (i.e. the style, tone or manner in which a speech act is carried out), topic, roles of speakers, social relationship between participants, modality, etc. Gumperz & Tannen (1979) used the term 'contextualization' in the study of code alternation for the first time, and stated the importance of an effective context in order to communicate. Context, therefore, is not just given as such in an interaction, but is the outcome of participants' joint efforts to make it available. It does not consist of a collection of material or social facts (such as the interaction taking place in such-and-such locality, between speakers with their particular social roles, etc.), but of a number of cognitive schema(ta) or model(s) about what is relevant for the interaction at any given point in time. What is relevant in this sense may exclude or include certain facts of the material and social surroundings of the interaction as they might be stated by an objective on-looker who tries to describe context without looking at what takes place in it, but it may also include information which is not explicit before the interaction begins, or independently of it. These context parameters refer to distinct linguistic activities which are not predictable from the actual utterances or social environment of the interaction at all, and also to knowledge which may be shared by co-participants from the very beginning, but has to be turned from invisible (and interactionally irrelevant) cognitive dispositions of the participants into commonly available grounds on which to carry out the interaction.

CONCLUSIONS
The data thus proves that discourse markers are very often accompanied by the bilingual strategy of code alternation. The frequent juxtaposition of these elements in the different codes at the participants' disposal suggests that they are perceived as a unique category by bilinguals. In the particular data used for this paper they are clearly used in English surrounded by Spanish speech.
One of the key issues in the data is that there are significant patterns of discourse, which were separated from markers framing it in language: English markers framing Spanish discourse. This is an indication of code mixing -where a structural pattern that has been grammaticised at the textual level can be perceived. They are recurrent in pattern, they are mainly used for framing speech shifts, to mark a contrast between the codes being used, and especially to set off a change of footing and for the purpose of creating a new structural mixed code. The fact that they conform to the structural behaviour of the language they are used in and yet highlight the code alternation suggests that there is a grammaticization process taking place. These alternations are frequently repeated and therefore one discourse marking system could end up being replaced by another (Spanish by English).
Discourse markers function as contextualization cues which tie different discourse parts together, highlighting the contrast between the different codes at use. Although contextualization cues would be a key to defining a particular type of language use as code switching, the use of discourse markers as contextualization cues is so frequent and regular that they lose their pragmatic force and functionality, separating their language from the metalingual frame of discourse markers and thus become code mixing. We can conclude, therefore, that the community under study is at a stage of code mixing in which a code of its own, as the use of alternation of discursive elements proves, has evolved and will probably continue to evolve.