Inter-language tissue and why nothing is ever just borrowed

Inter-language tissue and why nothing is ever just borrowed This contribution is challenging the traditional concept that languages in contact are separate (or separable) entities. An alternative model is proposed, inspired by an approach from science and technology studies: agential realism (Barad 2007). The central concept is that of inter-language mappings, the correspondences between structures of languages which get stabilised in the contact speakers’ community and make whole subsets of the source language lexicon in principle borrowable and potentially already borrowed. Importantly, the inter-language mappings cannot be properly assigned to any of the languages in contact and they enforce a profound rethinking of the way the most important processes of language contact - borrowing and code-switching - are conceptualised. Using examples from a variety of languages, I am showing how borrowing needs to be conceptualised not as adaptation to a new system, but as creation of a relation which cannot be properly situated in any of the varieties in contact. Key words: language contact, agential realism, inter-language mappings, inter-language tissue, contact speaker 0. The in-between: the primacy of relations over relata The primary goal of this contribution is to re-visit the deep-rooted conceptualisation of language contact as a matter of combining elements of clearly separable entities – languages in contact. An alternative model is proposed, which enables viewing language contact in its own materiality and temporality, with its own speakers’ competences, communities, histories and other properties, which are not derivable from the “languages in contact” or any less real than they are. The inspiration for this intervention comes from the field of science and technology studies (STS), more specifically from the approach called agential realism (Barad 2007). Drawing mostly from insights on quantum phenomena, agential realism actively challenges the idea that there are entities which precede their relations in some temporal or ontological 1 sense. In this sense, for agential realism, there is an ontological primacy of relations over relata and it is the isolation of the entities that requires an addition move – an agential cut. Language contact seems an especially challenging area for implementing this reversal of ontological primacy, since such a move goes against all habits of mind which are constitutive to main discourses on language, including the scientific ones. If there is a common property of everything worth the name of a language, then it is its separability from everything else that it occurs together with. Whether a language is real is not as much a question of having an army and navy (as famously quoted by Max Weinreich), although these attributes certainly cannot harm. Rather, what makes a linguistic variety a language is its ability to stand alone and serve the purposes that every full-fledged language can. This self-sufficiency requirement is the flipside of Benveniste’s famous claim that “no type of language can by itself alone foster or hamper the activity of the mind” (1971: 63-64, for an interesting STS discussion, see Kirby 1997). In this discourse, in order to prove not to hamper the activity of the mind, languages are well-advised to be strictly separable from the “outside” world which they are “representing”, but also from all the other languages. I have challenged the separation between language and the “outside world” elsewhere (see MY_NAME 2012). The main focus of this article is challenging the presumed separability of languages in contact, by showing that every specific contact has its own materiality and temporality, which cannot be properly assigned to either of the “separate languages”. This reality of language contact is iteratively materialised and reconfigured both in production and in the linguistic competence of real speakers, who are in this sense crucially contact speakers. This conceptualization puts the logic of languages in contact on its head: rather than saying that languages A and B have contact, we are saying that contact has languages A and B. In Baradian terms, the languages in contact are co-constitutive and therefore entangled within the same phenomenon. Wherever we make a cut between them, we need to justify the apparatus which we used to make the cut. The proposed view of contact is trying to introduce a new cut into the phenomenon language contact/entangled languages. The proposed re-conceptualisation of language contact/entangled languages may seem a pure gedankenexperiment, which proposes an (at best) equally adequate rendition of the reality. But my intervention is not about defeating and replacing the classical model. Rather, it is about showing that in order to provide an analysis of language contact/entangled languages phenomena, some approximating, fixating and excluding has to take place. In 2 Baradian terms, an apparatus is necessary for an agential cut to be made. By showing an alternative apparatus, I am indirectly reworking the classical one, showing under which conditions it works and which cuts it presumes. Situating the classical apparatus, I shall also point out what it crucially leaves out and how this “outside” has fought its way back into the proposed models. I shall show that the inter-language tissue, represented in speakers as inter-language mappings, have been emerging in linguists’ accounts of phenomena, but as epiphenomenal/external to the basic model. Here, is an important aspect of agential realism is put to use: rather than dissolving every discussion into relativism/monism and phobia of boundaries, agential realism recognises boundaries and asks under which apparatus they matter. My intervention is limited in scope, in the sense that I shall try to deviate from the usual linguistic discourse only where I deem necessary. Also, the presentation will be built around a single concept, that of inter-language mappings. This is the concept at which will be motivated by asking a very generativist/mentalist question: how is contact represented in contact speakers’ minds? However, I would like to have my readers keep in the back of their head that language entanglement may very well not be limited to classical cases of language contact. It may rather be the case that the multiplicity of varieties, which is usually analysed under the rubrics of linguistic repertoire, registers, dialects etc., is part of the same phenomenon. In this sense, the proposed reconceptualisation may help us move beyond the insight that language entanglement triggers much of what is usually analysed as diachronic change, and see language not as a frozen an monolithic entity, but as a simultaneous multiplicity, an ever entangled state of mutually constituting languages, which have a relentless potential for on-going differing. Before moving on to the details of inter-language mappings, I would like to pay the necessary tribute to a recently developed line of thought in linguistics, which, as far as I can see, has strived for the same goal: questioning the conceptualisation of clear boundaries between languages as a pre-given: the theory of languaging (first introduced in Jørgensen 2004). According to this approach, humans are said to language (crucially without a direct object of the verb) and they do this it using features, which are often associated with specific sociocultural constructs called languages. I find this approach or, to speak Baradian once more, the apparatus constituted this way, extremely useful. However, languaging seems to be too holistic to enable us to analyse the cases where sets of features are systematically assigned to different languages and paired in a way which makes them “derivable” from one 3 another. I shall show in the following section that this leads to insights which should be read into (and not against) the classical approaches. This is simultaneously setting a boundary to my contribution: in cases where there is no clear boundary between languages, there will be no (clear) inter-language mappings. As I am showing in section 2.1., inter-language mappings are virtually always present in contact data that have been of interest to linguists. In this sense, limiting the attention to cases where features are assigned to languages is useful, but should not be misinterpreted as an attempt of universalising this situation. On the other hand, it is important not to preclude the option that boundaries are constitutive to language, i.e. that an inherent polyglossia, differing and normative adherence to whatever is falls within certain boundaries is part of the very matter language is made of. The rest of this contribution is organised as follows. Section 1 presents the central concept of inter-language mappings and presents two examples of it. Section 2 shows how inter-language mappings are on the one hand omnipresent in contact data analysed by linguists (and even constitutive to them), but also how, on the other hand, they are systematically excluded from the basic model of language contact. Section 3 discusses three further cases of inter-language mapping, showing how another habit of mind, thinking in terms of adaptation, has obscured mappings. Section 4 summarises the findings and signals some possibilities for further research which this contribution hopes to have opened. 1. Prelude: inter-language mappings as the synchronic side of language contact In this section, the central concept, that of inter-language mappings, is introduced. This concept is the answer to a classical questions of post-Saussurean linguistics in (1). (1) Is the process in question synchronic or diachronic? Is the process part of the synchronic grammar? Is the process in question a reflection of the speakers’ knowledge or emergent in the community? Is the process mentally represented? Applied to the issues of language contact/entanglement, these questions turn into those in (2). (2) What is it that borrowers/code-switchers/code-mixers know? 4 Is there a synchronic/I-language side to language contact? Is there a “grammar of contact” shared by the contact speakers? These questions have been given less thought that it may first appear. The linguistic discourses on language contact, both generative (e.g. Kang 2011) and “traditional” (e.g. Matras 2009), implicitly frame language contact as diachronic. In this picture, there is a clear input, the word in the source language, which precedes the output, the word in the recipient language. Moreover, the process is seen as a result of workings of elements which are typically either properties of the recipient language or of all language, rather than contactspecific: Markedness, Universal Grammar, Native Grammar, phonological perception etc. In such a picture, there is no room for the generalisations which the speakers can extract from many pairs of words in which a structure in language A is iteratively mapped onto a corresponding (or, rather becoming-corresponding) structure in language B. In other words, in such a picture there is no room for the synchrony of language contact, which I argue to be expressed the inter-language mappings. Inter-language mappings are the synchronic “grammar of contact”. They are internalized by members of the bilingual community based on exposure to the existing correspondences. In other words, contact speakers typically know what to do: they know about the categories in both languages and how to map the ones to the others. In the remainder of this section, I shall present two examples of inter-language mappings in order to facilitate further discussion. The first case comes from Serbo-Croatian. All the standard languages which have arisen after the split of Serbo-Croatian have a straight-forward mapping which can turn any English verbal stem into a biaspectual SC verb. However, Serbian and Croatian use different integration suffixes here1, as illustrated in (3). (3) 1 Serbian Croatian daunlodovati daunlodati “to download” spemovati spemati “to spam” šerovati šerati “to share” Arguing for or against the unity of Serbo-Croatian is not my goal here. However, this case shows that it once again depends on the used apparatus. On the one hand, the fact that both Serbian and Croatian have converged on a single suffix each and that this suffix is, unlike in native derivations, used for the creation of biaspectual verbs can be seen as an indication that Serbo-Croatian is very much a single system. On the other hand, the different selection of suffixes can be seen as a sign that Serbian and Croatian have “moved on”. 5 (4) Serbian VovatiBIASPECTUAL (+ whole paradigm) VatiBIASPECTUAL (+ whole paradigm) V Croatian As shown in recent neologisms and nonce derivations (MY_NAME 2010, MY_NAME, in preparation), Serbo-Croatian does not have a default verbaliser and a plethora of suffixes (and prefixes) emerges in elicitation experiments. Moreover, the newly derived verbs which are not based on English verbs typically have single aspect. In other words, the level of stability of the used suffixes, as well as their function, is a property of the on-going contact with English, not of Serbo-Croatian (or Serbian or Croatian). However, it is important not to replace one essentialism with another and claim that these mappings are external to SerboCroatian in the sense that they cannot influence native derivations or the inter-language mappings of future contacts. Such developments are possible, but by no means guaranteed. Also, it is important to note that verbs produced by these mappings are slowly but surely finding their way into the standard languages, which are also used by speakers with no command of English. In this sense, the whole language is entangled, albeit through the mediation of a more or less clearly defined community of contact speakers. In order not to create the wrong impression that mappings require highly specific knowledge which only exists in bilingual communities, we are now turning to a case of hyperforeignisms (Janda et al. 1994), where there is typically no knowledge of the source language. A case at hand is the English treatment of Japanese prosody (in words ending in a vowel), where Japanese loanwords get borrowed into English with penultimate stress, regardless of the original stress pattern (5). Janda et al. (1994) conjecture that this pattern is extended from Spanish/Italian words, which often end in a vowel as well. (5) 6 Japanese English Natsuko Tsujimura Natsuko Tsujimura Hiroshima Hiroshima sake sake In this case, the entanglement crosses many language boundaries, so that the same mapping seems to be operating in Dutch, German, Serbo-Croatian, Slovenian and many other languages. An important property of the concept of mappings is that it cannot be properly cut as belonging to either of the languages. Rather, a mapping is a feature of a contact speakers’ community. Moreover, since mappings are generalisations, the existence of a mapping always enforces productivity, it always pushes the pattern to be used on new items. In other words, the existence of a mapping makes subsets in the source lexicon in principle borrowable and potentially already borrowed. Hualde (2000), describing Spanish-Basque facts which will be discussed in section 3, describes the status of the words which contribute to/are derived by mappings as follows. A Basque speaker who uses a word [] will not necessarily know whether s/he is repeating a Basque word which s/he has learned of whether s/he has in fact adapted it spontaneously from his/her Spanish lexicon. The cases we have just seen and the concept they are used to support destabilise quite a few classical generalisations about the nature of linguistic contact facts. In the following section, I am turning to the problem of how inter-language mappings have been excluded from the basic model of language contact and how they kept re-emerging in different disguises. 2. Inter-language mappings: Everywhere in language and nowhere in linguistics In this section, I am putting the newly introduced concept back into the existing accounts of language contact. Subsection 2.1. shows how the existence of inter-language mappings is presumed in borrowing and code-switching data. How borrowing and code-switching, as the most analysed contact processes, conceptually exclude inter-language mappings, and how the mappings can be implemented back into these processes is the subject matter of subsection 2.2. 2.1. So obvious that you don’t see it: No mappings, no linguistic interest 7 The existence of productive mappings is implicated in everything that linguists would consider contact data: presentability of facts as systematic mappings between two languages is required in order to be able to account for them. As a matter of fact, mappings seem to take a while to be established and linguists are generally not interested in the pre-systematic soup, although they know that there is one. The most influential statements in this area were made by Einar Haugen, half a century or longer ago (Haugen 1950, 1956, 1969). Haugen differentiates between three types of bilingual communities with different types of borrowing. During the pre-bilingual period, speakers adapt loanwords to the native patterns without establishing regular patterns. The period of childhood bilingualism introduces more regular integrations and some sounds from the donor language. Finally, in the period of adult bilingualism the loanwords are integrated systematically and with many non-adaptations. It should be noted that Haugens’s generalisations were made based on the developments in immigrant languages in the US, in communities which all eventually switched to English. There seem to be two axes along which Haugen’s scenario proceeds – the duration of contact (which 2 brings regularity ) and higher proficiency in source language (which brings more nonadaptations). These two aspects are inseparable in the communities observed by Haugen, but their conceptual separation is essential if we are to compare most of the contact situations in the modern world, characterised by long-term pre-bilingual communities. In such cases speakers establish what Haugen calls “interlingual identification” (Haugen 1956: 44) – one-to-one mappings between the languages. Haugen described these in the following way. When languages are in contact, there is a strong tendency for speakers to equate items of the one with items of the other. Some items thereby acquire associations different from those they have for monolinguals without such experience. (Haugen 1956: 44/45) From the perspective of the inter-language mappings, much of what has been done in the linguistic analysis of borrowing can be read as an attempt to account for the specific interlanguage mappings using the properties of the source language (or language in general). While predicting outcomes of languages aligning their structures can be useful, the general 2 A terminological caveat is in place here. The irregularity contingent to the initial period of bilingualism should not be confused with the variation which characterises the early life of borrowings. Haugen (e.g. 1956: 55) argues that the latter type of irregularity stems from the speaker’s awareness of the origin of the borrowed item, which wears out with time. 8 conclusion seems to be that there is no destiny that can be read off the available structures: the mappings which will get established are not predictable only from the language combination. What further problematises this endeavour is the fact that once mappings become stable, they do not depend on any synchronic motivation in any way. Hualde (2000) makes this point about phonological adaptations in Spanish-Basque contact. Once a pattern of adaptation becomes conventionalized, phonetic/phonological considerations no longer apply. There have been quite a few formal analyses presenting data which are inviting a mapping analysis. Yip (2002, 2006) shows how in the adaptation of English names Taiwan Mardarin has consistent deletion where Mainland Mandarin has epenthesis. Boersma & Hamann (2009) are clear about what falls outside of their model, well-summarised in the title of their article “Loanword adaptation as first-language phonological perception”: it is the way in which bilinguals filter out candidates containing certain segments because they have already analysed the source language as not having them. For instance, Korean speakers are assumed to have analysed English as not having tense plosives, which excludes all the candidates containing these segments as possible outputs of the perception grammar, which in their model equals what is in charge of borrowing. As with the most formal accounts of borrowing, all the necessary mechanisms are already present before the contact (i.e. as part of the entity recipient language) and the existence of mappings is precluded in the basic model, so that it can only be included in the form of stipulated filters. We shall return to this latter point later. For the moment, it is important to make clear that the concept of inter-language mappings offers a reanalysis of many phenomena, including borrowing and code-switching. In the following section, we are turning to the issue of why inter-language mappings are constitutively excluded from the discussion of borrowing and code-switching. 2.2. Borrowing and code-switching and how they preclude the mappings In this section, borrowing and code-switching, the most prominent (conceptualisations of) contact phenomena, are re-visited in the light of inter-language mappings. As will be shown, inter-language mappings cannot be made compatible with the essentialist definition of 9 either of the processes, but also not with many features which are implied by the use of exactly these metaphors to speak of contact/entanglement. First, both borrowing and code-switching are predicated on the monolithic and stable idea of languages in the sense of words clearly being present in or absent from a certain lexicon at a certain moment: if codes are switched between, the underlying assumption is that one is in a single code at every moment and if words are to be borrowed, they first have to not belong to a code, and then to start belonging to it. This clashes with the defining property of inter-language mappings to make whole sets of words borrowable and potentially already borrowed, so crucially in-between presence and absence. From the perspective of the interlanguage mappings, nothing can be entirely new either to the recipient discourse (which is a precondition for code-switching) or to the recipient lexicon (in the case of borrowing), because contact speakers always already have had experience with the structures in question. Moreover, the perennial issue of the difference between code-switching and borrowing is obviated once we integrate inter-language mappings into the picture. The insight that words are never simply inserted from the source lexicon into either the recipient language discourse (code-switching) or the recipient lexicon (borrowing), but that this insertion is always partially mediated by what is already known, indicates that this dichotomy is yet another case of different apparatuses producing different cuts, which cannot be used both at the same time. As we have already seen in the previous subsection, excluding the fact that the agents of borrowing/code-switching are actually experienced in dealing with the relevant structures is one of the hallmarks of the linguistic analysis of loanwords. Introducing an excellent overview of the current state of the art, Kang (2011) states that borrowing can be considered “a real-life Wug test (Berko 1958) which can enable us to probe into the grammatical knowledge of speakers in ways that native data alone cannot”. This idea relies on the assumption that borrowing can be seen as a “spontaneous reaction” of the recipient language to borrowed items. However, inter-language mappings point at exactly the opposite direction: the bilinguals know what to expect and how to treat it. The introduction of mappings also leads to different empirical scope on language contact/entanglement research. For instance, the whole class of hyperforeignisms shows that there can be borrowing patterns which follow neither from the properties of the borrowed item nor from the borrowing language. Another feature of the formal approaches to loanwords which we have already seen is investing what we are arguing to be the inter-language tissue in the grammar of the recipient language. Kang (2011) states that: “the patterns that emerge in loanword adaptation often 10 reveal aspects of native speakers’ knowledge that are not necessarily obvious in data of the native language and, as a result, loanword data can inform our analysis of the native phonology”. In the same vein, Calabrese and Wetzels (2009:1) maintain that there is a “general consensus” that the nativisation of loanwords provides a window for studying “the true synchronic phonology of L1 by observing its phonological processes in action”. It seems that this approach to borrowing is made more plausible because the commonly used borrowing metaphor makes an interesting implicit assumption about the “agents” of contact: that one borrows from others rather than form oneself and then makes the borrowed item one’s own. Much of the tradition will assume that speakers who introduce loanwords “hear” the form from the source language, a scenario which is in the heart of the “loanword adaptation is perception” hypothesis. Of course, it is possible to construe a situation in which the source form is presented only orally and only to speakers who become minimal bilinguals by virtue of this exposure. However, larger loanword data sets are drawn from situations in which at least some of the borrowers are symmetrical bilinguals and/or “native speakers of the contact”, that is speakers who know how to handle structures from the source language due to lasting contact between the languages. Here, the borrowing metaphor seems to lead into a faulted conceptualisation. There has been quite some critique of the metaphors used to discuss language contact, mostly in the sociolinguistic, but only rarely in the “formal” linguistic literature. An noteworthy exception to the latter generalisation is Janda et al.’s (1994) discussion of hyperforeignisms, These authors deconstruct the borrowing metaphor in a very fruitful way, illustrated by the following quote. [B]oth "borrowing" and "stealing" imply that what is taken into an L1 from an L2 must necessarily cease to exist in L2, whereas "loanwords" ("thefts") clearly need not disappear from their donor languages. Hyperforeignization, though, gives the coup de grâce to such proprietary metaphors, since one obviously cannot either borrow or steal what doesn't exist. Rather, hyperforeignisms demonstrate that biological analogies like "replication" or "cloning" are much more apt for lexical interchanges between languages. Indeed, there exists a terminology along these lines which goes all the way back to the work of Haugen (1950) and Weinreich (1953): in particular, the target of copying from one language into another can be considered as a model, and the result of such copying as a replica. And, just as mutations can potentially occur in the 11 course of biological replication or cloning, so one can misperceive or even, in a sense, hallucinate one's linguistic model, thereby creating either a deformed copy or a copy of nothing — a hyperforeignism. (p. 74 - 75) Clearly, hyperforeignisms are an extreme case of abstraction from experience (real or imagined) that also inter-language mappings are based on. Interestingly, Janda et al. also conclude that hyperforeignisms are an instance of hypercorrection. After a discussion of Labov’s native data, they state the following. We must conclude, then, that hypercorrection is a potentially pervasive phenomena which must always be taken into account in situations where there is contact between speakers of different language-varieties. And so, even in the case of exotic-seeming hyperforeignisms which may involve interactions between genetically unrelated languages, we are never really far from the garden-variety hypercorrection that occurs so commonly back home, in our first, native language. (p. 77) This brings us very close to the programmatic hope expressed at the end of the introductory section above. It could be that inter-language mappings are actually better termed intra-language mappings, because they operate on the inherent polyglossia that language is. Due to the limited scope of this contribution, rather than elaborating the transversal potential of the proposed model, I shall use the remaining space to discuss a number of language contact cases entertaining the new discourse. The most important feature of the proposed framing is seeing languages in contact as entangled, and consequently destabilising the idea that words belong to languages in contact in an exclusive way. 3. Inter-language mappings in action In this section, the concept of mappings is employed to provide an account of three data sets, which would usually be considered to fall under the auspices of the field of loanword adaptation. One of the goals here is divorcing borrowing from the adaptation logic, which means that I shall propose doing away with the telicity bias, which enforces seeing the integration of words from another language as adaptation. 3.1. [æ] in Dutch: same language, different mappings 12 A prominent difference between Netherlandic and Belgian Dutch (Flemish) is the way the English [æ] in words such as badminton, laptop, stand and rap is integrated. (6) Belgian Netherlandic b[ɑ]dminton b[ɛ]dminton “badminton” l[ɑ]ptop l[ɛ]ptop “laptop” st[ɑ]nd st[ɛ]nd “stand” r[ɑ]ppen r[ɛ]ppen “to rap” The native phonologies seem not to point in this direction in any clear way. Also in L2 English, [æ] is most commonly rendered as [ɛ] in both varieties. Furthermore, for Flemish, there are normative interventions which clearly state that [ɛ] would be the pronunciationbased adaptation (e.g. Taalnet 2000). In most adaptation approaches, the Flemish values would be explained away as extralinguistically conditioned spelling pronunciations, or possibly influences from French, whereas the Netherlandic values would be seen as pronunciation-based. While all these considerations on the source of the mapping are interesting for the present model, they are not full-fledged accounts. Rather, they can be taken as pointers at representations which need to be taken into consideration, because borrowers are aware of them. For both varieties, the input is not simply [æ], it is [æ] plus the graphemic value <a>, plus the Dutch English value [ɛ] and this input is linked to the output [ɑ] and [ɛ] in Belgian and Netherlandic Dutch, respectively. In this situation, the Belgian speakers do not simply base their outcome on the spelling, they also take the pronunciation and nativisation into account and lexicalise “against” them. The same is true of Netherlandic Dutch speakers, who are well aware of the spelling. In sum, the differences are not based on the speaker’ knowing different things, it is rather the case that the communities have converged on different forms. It is by now clear that in the mapping approach, there is nothing inherently preferable about mappings based on the pronunciation, the representation of mappings does not suffer from its different grounding. The mapping is then as shown in (7). 13 (7) Belgian Dutch (Flemish) <a> [ɑ] <a>, [æ], [ɛ]NATIVIZATION Netherlandic Dutch <a> [ɛ] While this example requires no reference to adaptation, it is also not entirely incompatible with the adaptation approach in the sense that both [ɑ] and [ɛ] are segments of Dutch, whereas [æ] is not. The following subsection provides a case in which the encountered pattern is not compatible with adaptation. 3.2. Spanish-Basque Inter-language Mappings Unlike, the Dutch mapping which refers to segments (and graphemes) in a context-free way, in this section we are considering an example of a mapping which is morphologically circumscribed. This is not the first case of such kind: the Serbo-Croatian facts from the previous section referred only to verbs. In an article which can be read as a veritable precursor of the present approach, Hualde (2000) presents surprising correspondences which hold between Spanish nouns and their counterparts (borrowed from the former) in Basque. For instance, nouns which end in the sequence -on generally have -oi in their Basque correspondents. (5) Spanish Basque camión kamioi electrón elektroi "truck" "electron" If observed from the perspective of adaptation, these facts will appear surprising, because very common words in Basque end in the "repaired" Spanish sequence e.g. gizon "man" and on "good". As Hualde explains, this correspondence has a historical explanation the Latin sequence -one was borrowed into Basque faithfully as -one, and has consequently had different destinies in Spanish (-one > -on) and Basque (-one > -oni > -oi), yet the mapping between the related forms has remained. 14 This example shows that mappings do not have to be motivated in synchronic terms, and that the outputs are not to be analysed as adaptations. Note that this does not mean that anything can be a mapping. Rather than being derivatives of the synchronic grammar, mappings are conventionalised, sedimented out of the experience with the same structures within the contact speakers’ community. These conventionalized mappings still have to refer to (meta)linguistically accessible units - they just do not have to relate them in a way which would seem (phonetically) natural or predictable from the current state of the languages. An interesting aspect of this specific case is that the sequences –on and -oi are not morphemes in these words. However, equally arbitrary morpheme-to-morpheme correspondences can bleed the general correspondence. So, as we learn from Hualde (2000), the nouns which end in the Spanish -cion/-sion get borrowed with a final -zio/-sio in Basque. (6) edición edizio "edition" conclusión konklusio "conclusion" Here it may be a matter of discussion whether the units -cion/-sion and -zio/-sio are real morphemes in all the words in which they occur in both languages. In the following section, we are turning to a case where another element which is typically considered external to language, the norm of the standard language, plays a constitutive role. 3.3. Syllabic r in English Borrowings into Serbian English rotacised vowels, realised as [ɚ], are typically mapped to the Serbian syllabic r by the speakers. This is a very prominent feature of the Serbian accent in English and of spontaneous nativisations, very obvious from the phoneticised English, common in colloquial Serbian. In the following example, taken from a blog, this accent is being ridiculed. The underlined letters are realisations of ɚ/vocalic r. (7) Srbs hed inaf of filti, pezent, primitivli komplikejted end ekstrimli agli Srbian lengvič. Lajk evidens za ovu tvrdnju, Aj vil mejk šr tu provajd ju sam bjutiful ekzempls from evridej lajf, nešnal kalčr end medijas, vič mejd mi da tinkujem o jedna veri seriozna reforma na jezik. Det kajnd of reforma iz not jednostavna tu establiš, zbog čega gesujem det bi se mastovala sprovoditi step baj step. ...Enivej, fajnal gol vud bi tu supres Srbian lengvič (vokabjuleri, 15 speling, pronansiješn, artikjulejšn end so on...) et ol instances: strating from evridej komjunikejšn, preko skuls end edjukejting programs, tu ekonomi, politiks, kalčr, literačr, rilidžn end mor. These examples are an indication that there is a clear awareness of the mapping from [ɚ] to the syllabic r in Serbian speakers. In standard Serbian, however, there are no examples of syllabic r in English loans and the words containing [ɚ] are adopted with [Vr] sequences, whereby the vowel is usually taken from the graphical form. (8) Standard Serbian Colloqual Serbian Firt/Fert Frt “Firth” Berns/Burns Brns “Burns” Hurtsi/Hartsi Hrtsi “The Hurts” Serbian speakers seem to be aware of this aspect of the standard language. When asked to nativise the words workshop and feature the way they use them in an SMS and the way they would use them in an academic article, they choose vrkšop/fičr in the former and vorkšop/fičer in the latter case3. It is important to note that this procedure is not entirely accessible to the speakers - they are not able to formulate a generalization - and that adaptation strategies are not taught at any level of formal education. Still, the speakers are able to support their choice with other examples of standard adaptations. This example shows that even if the speakers have intuitions about certain correspondences (which then get invested into substandard nativisations), they can acquire a different, prescribed pattern and apply it productively. Again, Serbian speakers pick up on the available patterns and apply them in further adaptations, rather than developing their own intuitions on optimal correspondents in two languages. The goal of this section has been to illustrate the proposed new model of language contact in a number of examples from various contacts and show how the new approach requires moving away from the adaptation bias. Rather than assimilating foreign material, the inter-language tissue has the function of enabling the transfer of linguistic material in a straightforward way. Throughout the section, the insight has been emerging that for different contacts/entanglements different representations are to be taken into account, because aspects 3 However, in the case of a text message the speakers point out that maintaining the English spelling feels much more natural. 16 like written form, normative interventions etc. are as real as any other part of the entangled representations. 4. By way of conclusion This intervention can be read as an attempt to entertain a new discourse on language contact/entanglement, which, rather than extending the logic from the regularities which hold within single languages to contact situations, focuses on what is specific to language contact situations: the inter-language mappings. In this sense, this intervention conceptually divorces the discussion of language contact from that of processes within languages or at least allows for such a separation. The emergent agenda of loanword research would have to focus on alignment between languages without thinking in terms of adaptation, but rather in terms of matching multiple systems in order to make material transferable. Importantly for linguistic theories, this means that the link between theories of contact phenomena and theories of grammar should be argued for rather than simply presumed. One important issue in such an agenda is what are the possible representations and levels of analysis accessible to interlanguage mappings and how they relate to representations within the language. The issues discussed in this contribution also invite a reconsideration of the boundaries of prototypical languages. When something becomes a word of Dutch/Serbo-Croatian/Basque is not obvious at all and I am crucially claiming that it shouldn’t be. What to make of the potential presence of massive subsets of English/Spanish in Dutch/Serbo-Croatian and even of the fact that we are already able to predict the stress pattern of any incoming Japanese word without knowing a single one of them? Inter-language mappings are as real as anything and keep destabilising the “separate” existence of codes we share. Importantly, the fact that there are no absolute and eternal boundaries does not mean that there is no reality of linguistic difference to speak of. Rather, the business of boundary formation/research becomes more responsible and endlessly more interesting, because we are not reduced to describers of facts, but invited to explicate the apparatuses within which certain boundaries matter. It is my hope that I have opened some perspectives for this kind of scientific involvement with linguistic boundaries. 5. Bibliography MY_NAME. 2009. MY_NAME 2012. 17 MY_NAME in preparation Barad, Karen (2007). Meeting the universe halfway: Quantum physics of the entanglement of matter and meaning. London: Duke Universal Press. Benveniste, Emile (1971). Problems in General Linguistics, trans. M. E. Meek. Coral Gables, FL: University of Miami Press. Bogawac, Miwena (2008) “WRAJTUJ KAO SHTO SPIKUJESH, RIDUJ KO SHTO JE WRAJTNUTO” Blog entry at B92. Available online at <http://blog.b92.net/text/6039/%22WRAJTUJ%20KAO%20SHTO%20SPIKUJESH %2C%20RIDUJ%20KO%20SHTO%20JE%20WRAJTNUTO%22/>. Accessed on 9.12.2012. Boersma, Paul & Silke Hamann (2009) "Loanword adaptation as first-language phonological perception." In: Andrea Calabrese & Leo Wetzels, eds. Loan Phonology. Amsterdam: Benjamins. Calabrese, Andrea & W. Leo Wetzels, eds. (2009). Loan phonology. Amsterdam & Philadelphia: John Benjamins. Haugen, Einar (1950) The Analysis of Linguistic Borrowings. Language 26: 21-231. Haugen, Einar (1956) Bilingualism in the Americas: A Bibliography and Research Guide. American Dialect Society. Alabama: University of Alabama Press Haugen, Einar (1969) Norwegian Language in America: A Study in Bilingual Behavior. Philadelphia: Universitity of Pennsylavania. Hualde, Juan (2000) “Patterns of correspondence in the adaptation of Spanish borrowings in Basque”. BLS 25. Janda, Richard, Brian Joseph, and Neil Jacobs (1994) “Systematic hyperforeignisms as maximally external evidence for linguistic rules”. In S. Lima, R. Corrigan, & G. Iverson, eds. The reality of linguistic rules. John Benjamins Publishing Co., 67-92. Jørgensen, J. Normann (2004) “Languaging and languagers”. In C.B. Dabelsteen and J.N. Jørgensen, eds. Languaging and Language Practices. Copenhagen Studies in Bilingualism 36: 5–23. 18 Kang, Yoonjung (2011) Loanword Phonology. In Marc van Oostendorp, Colin Ewen, Elizabeth Hume and Keren Rice, eds. Companion to Phonology. Wiley: WileyBlackwell: 2258 – 2282. Kirby, Vicki (1997) Telling Flesh: The Substance of the Corporeal. New York: Routledge. Matras, Yaron (2009) Language Contact. Cambridge: Cambridge University Press. Taalnet (2000) “De gehandicapte fan” Available online at <http://www.vrt.be/taal/degehandicapte-fan>. Accessed on 9.12.2012. Yip, Moira (2002) “Necessary but not sufficient: perceptual influences in loanword phonology” Journal of the Phonetic Society of Japan 6:1. 4–21. Yip, Moira (2006) “The symbiosis between perception and grammar in loanword phonology” Lingua 116. 19