Language mixing in Palasa


 We explore here several kinds of language mixing to be found in the Greek-Albanian bilingual speech community of the village of Palasa in southern Albania. Palasa is of particular interest for Greek dialect studies because it offers a window in the present day into highly localized dynamics of language contact. Among the mixing observed in Palasa is code-switching, motivated by various factors as identified by Myslín & Levy 2015, borrowing, both lexical and structural, and hybridization, at a number of levels of analysis, including phonology, morphology, and semantics. Our findings indicate that language contact is still alive and well in the Balkans at least at the level of village dialects.


Introduction
The small village of Palasa, located in the vicinity of Himara in southern Albania, provides a fascinating and important look at language contact. The villagers are Greek-speaking but are fully fluent in Albanian as well, so that it is possible to see the effects that contact between the two languages has had for these speakers. Especially interesting is the extent to which these bilinguals show a mixing of Greek and Albanian at various levels. Our presentation of this language mixing here is based on fieldwork in Southern Albania, specifically on work in Palasa that Brian Joseph did individually or together with Aristotle Spiro, Majlinda Spiro, Andrey Sobolev, and Alexander Novik in 2014, 2016, 2017, and 2018; the material is thus on-site naturalistic data, involving recordings of conversations with and between several older Greek-Albanian bilinguals. Rexhina Ndoci provided most of the analysis, based on these recordings, assisted by Carly Dickerson. The notion of "language mixing" we intend here should be understood in its broadest sense, i.e. referring to ways in which, in a language contact situation, elements of both language A and language B co-occur in the discourse of speakers of A and B. Language mixing for us thus takes in such well-known contact-related phenomena as code-switching and borrowing, both of a lexical nature and a structural nature, but also what we call "hybridization", in which phrases and even words show mixing of elements at different levels, especially phonology, morphology, and semantics. In what follows, we present the relevant data showing the mixing and offer our analysis of the individual instances and the more general situation they shed light on.

Code-switching
We start with code-switching, an outcome of language mixing with a long tradition of research (cf. e.g. Poplack 1980). We adopt the definition here of Myslín & Levy 2015:873, namely the "alternation of multiple languages within a single discourse, sentence, or constituent … by fully proficient multilinguals", and we draw on the general framework for the causes of code-switching that they articulate.

2.1
Pragmatic/discourse-functional factors Two different types of discourse-based factors for code-switching can be found in the Palasa data.

2.1.1
Clarification/emphasis/qualification An example showing a speaker switching languages for clarification is given in (1), where the issue was the grammatical gender of a place referred to as [liθerí], where based on its sound shape, ending in [i], the form could in principle be neuter (with definite article το) or feminine (with definite article η): after debating for a little while with the interviewers trying to elicit the gender of the name for the area using definite articles (το Λιθερί ή η Λιθερή 'the(NTR) Litheri(NTR) or the(FEM) Litheri(FEM)') and locative prepositions (στο ή στη 'to.the(NTR) or to.the(FEM)') marked for either neuter or feminine gender, the informant resolves the ambiguity by using the feminine locative preposition στη and then immediately switching to Albanian (n' gjinin femërore 'in feminine gender') to clarify that it is a noun with feminine grammatical gender.

(n') (in)
gjinin gender femërore3 feminine '"to the" in the feminine gender' In (2), the woman is talking in Greek and gives her answer to the interviewer's question on what they used to cut the object of the discussion with (με το μαχαίρι 'with the knife') and then follows it immediately by the exact same answer in Albanian (me thika 'with the knife'). In this case it is not as easy to identify the precise pragmatics behind the switch. The speaker might be trying to make sure that the answer is clear and the researcher has understood what she has said or she might be emphasizing the object of the cutting action being discussed.
(2) I: In the next example, (3), we can see an instance where code switching seems to have a clarifying and qualifying function. The speaker is talking at first in Albanian (Gjithmonë andartin kanë qenë Palasikot 'the Palasikotes had always been rebels') about how rebellious and strong the people from Palasa have been and then switches to Greek (Ελέγανε οι Χειμαρραίοι, λέγανε για τους Παλασκινούς "σέρνει η βάρκα το βαπόρι" 'the Himariotes would say, would say about the Palasikotes "the boat drags the ship"') to qualify this by providing more information about the claim.  βαπόρι" ship The Palasikotes had always been rebels. The Himariotes would say about the Palasikotes that "the boat drags the ship." 2.1.2 Topic An example in which the topic seems to motivate a code-switch can be seen in (4). In this instance, the interviewer is trying to clarify whether one of the informants is from Palasa or from Narta, another Greek village, and then moved to Palasa when she married a local. This negotiation takes place in Greek. This conversation triggers a response from Man 1 who goes on talking about the experiences of the woman's family during the Greco-Italian war. He starts his first turn in Greek (Όχι, έχουμ πάει 'No, we have gone'), as the language of the discussion so far has been Greek, but soon switches to Albanian in the same turn admitting that he feels more comfortable talking about this topic in Albanian (Ta them në shqip më mirë se nuk e [?] 'I'll tell it to you better in Albanian because I don't [?]'). Thus, this seems to be a case of topic-based shift. Between this utterance and his next utterance there is a short discussion between two other informants about how the female informant is related to another man, a passage omitted here to save space. The example continues when Man 1 gets another turn in the conversation and resumes narrating the history of the family in Albanian. 2.2 Accommodation Very common in cases of code-switching, and also found in Palasa, are instances in which a speaker switches in order to accommodate to the language of one's interlocutor. This is shown twice in (5) where the woman accommodates to Interviewer 1 by responding to him in Albanian (Soi burrit. Jo babai im 'Husband's family. Not my father') when he addresses her in Albanian (Domethënë [soi] i burrit është prej Skrapari, ose [soi] babait?5 'That is to say, the husband's family is from Skrapar, or the father's family?'). However, when Interviewer 2 enters the conversation using Greek (Από ποιο σόι είσαι εσύ; 'Which family are you from?'), the woman accommodates again to the language of the second interviewer and responds to him by switching to Greek (Εγώ είμαι από το Παπά. Είχαμε τον παπάν 'I am from the Papa family. We had the priest').

I:
Ήταν παπάς ο αυτός ο παλιός; Was priest the this the elder Was this elder a priest?

2.3
Triggering Psycholinguistic factors also seem to be relevant in the code-switching of the bilinguals in Palasa. Instances of switching seem to be triggered often by lexical items that sound the same in both Albanian and Greek, "trigger words", in the terminology of Clyne 1991, that serve as a bridge between the two languages. Examples of this type are given in (6) and (7) where the utterances begin in Greek (Με παίραν πάλι 'they took me again' and Εσύ γι' αυτή κ´ευτή, για σένα, αλλά 'you for her and she for you, but' , respectively) and are completed in Albanian (ekzekutiv me të gjitha 'the executive, with all' and mirëkuptimi 'understanding' , respectively). The use of the preposition with the pronunciation /mε/ 'with' , which has the same meaning and also sounds the same in Greek (orthographically με) and in Albanian (orthographically me), might be what activates Albanian in the bilinguals' minds and thus triggers the switch from Greek to Albanian.6 a borrowing from Greek into Albanian, or even a borrowing from Turkish. Hence, we give it in a phonetic form in square brackets, except for the one time it occurs in a fully Greek context (I2's question), in order to reflect this analytic ambiguity. 6 It must be noted, of course, that not every instance of [mε] triggers a switch, suggesting that more may well be at work in the switches and the non-switches.

2.4
Seemingly unprovoked So far we have presented instances of code-switching motivated by what we see as the sociocultural dynamics between interlocutors, discourse-functional factors, and psycholinguistic factors. There are, however, instances that seem to be unprovoked so that the motivation behind the switch is unclear. Such cases can be found in examples (8) and (9) where the speakers switch momentarilyperhaps one-word switches if these are not to be treated as loans-into Albanian while speaking Greek, using the Albanian lexical items guzhina 'kitchen' and pika 'drop' instead of the corresponding Greek κουζίνα and σταγόνα/στάλα; these are common, everyday words that might be unlikely candidates for borrowing.8

Borrowing
We use the notion of borrowing in a broad sense here, referring not only to the introduction of lexical items, but also to the introduction of semantic, grammatical, and phonological material. Thus, our approach is closer to Aikhen from one language to another as a result of contact". Let us then proceed by looking at these types of borrowing more closely.

3.1
Lexical borrowing Lexical borrowing in the Greek-Albanian contact situation of Palasa shows that certain lexical domains have been especially conducive to borrowing.

3.1.1
Communist-era terminology One such domain is terminology that refers to the Albanian communist era. In (10) partia 'party' and in (11) spi9 kulture 'house of culture' , both mentioned in reference to the past communist regime, are borrowed from Albanian and embedded into Greek utterances. Official-level, academic, or bureaucratic terminology Official terminology, involving words referring to any governmental dealings, to education, or to bureaucratic affairs, and the like, is another domain where borrowing is very common in the data. For example, biologjia 'biology' in (12), gjykata administrative 'administrative court' in (13), ushtria 'army' in (14), and marrëdhënie 'relations' in (15) belong, respectively, to academic, legal, military, diplomatic fields and are all borrowed from Albanian into the Greek discourse. Terms related to village life Another interesting domain where lexical borrowing is not unusual is terminology related to village life. We find it interesting because we would not expect Albanian being used here by the Greek villagers unless the village culture was originally Albanian and these borrowings are an indication that the older language in the village prevails in domains pertaining to village life. In (16) and in (17), where the discussion concerns weddings and wedding traditions in the village, we find ftesa 'invitations' and vello 'wedding dress' , both borrowed from Albanian and embedded in the Greek utterances. Conversationally based loans Finally, Friedman & Joseph (2014, To appear 2020 propose and exemplify a type of borrowing which they refer to as "ERIC" loans, an acronym standing for those "Essentially Rooted In Conversation". These are loans that do not necessarily fill a need or gap of any sort, are not associated with any particular cultural item, and do not reflect any prestige on the part of the donor language;10 rather, they seem to crucially derive directly from conversational interactions between speakers of the two languages. Such loans include discourse particles, address terms, affirmation, negation, and closed-class grammatical elements like pronouns, among other types of lexemes. Examples of ERIC loans in Palasa are conversational elements like the affirmative element ναι 'yes' , used in mocking agreement in (18) as ναι ναι 'yes yes' , and the terms of endearment xhan 'soul ' and jeta 'life' in (19) and (20), respectively, that are borrowed from Albanian (the former ultimately from Turkish) and embedded into Greek usage. you.try 'Ah yes, yes. You will try, when will you try?' 10 These factors are mentioned as they are the ones that form the basis for two particularly influential typologies of loans, those in Bloomfield 1933 ("cultural" versus "intimate" loans) and Hockett 1958 ("need" versus "prestige" loans).

3.2
Grammatical and structural borrowing In addition to lexical borrowing, a few cases of borrowing of grammatical items as well as of structure can be seen in the contact between Greek and Albanian in Palasa.

3.2.1
Use of Albanian ma ("more") for πιο in Greek Very common in the Greek of the villagers was the use, in cases where the language of communication was Greek, of the Albanian ma 'more' to signal comparative degree in adjectives, instead of its Greek correspondent πιο 'more' . Example (21) is interesting as it showcases this borrowing twice in the speech of the same speaker. At first the speaker uses, in speaking Greek, ma with καλύτερα, an adverb in the comparative degree form, and a little later with μεγάλο, a plain (positive degree) adjective. The ma + comparative constructions are the most frequent in our data, but as (21) shows, ma + plain adjective is a possible construction in the speech of these bilinguals as well.  (22) shows an instance of the borrowing of structure involving a matter of syntax, specifically noun-adjective word order. The adjective κακέα11 'bad' appears after the noun it modifies, αρρώστια 'illness,' and not before it, which would be the default word order in Greek. We can attribute this to structural borrowing from Albanian if we consider the Albanian word order. This construction would have been sëmundje të keqe 'bad illness'12 in Albanian, where the adjective keqe 'bad' follows the noun sëmundje 'illness' that it modifies, thus a model for what we see happening in (22)

Hybridization
In some cases, the language contact effects observed go beyond borrowing in the more usual sense and instead involve what we call "hybridization" in that one and the same linguistic sign shows elements from both languages coming together to form single constituents. The resulting forms are Greek-Albanian hybrids, and they can be seen at a variety of levels of analysis.

4.1
Phonological hybridization An example of hybridization at the phonological level can be seen in (23). While speaking Albanian, the speaker is aiming to say the Albanian word diplomë 'diploma,' which has a voiced alveolar plosive /d/ in initial position in Albanian, but instead uses /ð/ (the voiced dental fricative), which is the initial segment in the equivalent Greek word δίπλωμα 'diploma'; the word is in all other respects Albanian as at the same time it shows the Albanian stress placement, on the penultimate syllable, and not the antepenultimate stress that the corresponding Greek word shows.13 11 Κακέα here is a regional variant of what in Standard Modern Greek is the feminine form κακιά 'bad' . It is transcribed as given by the speaker. 12 This is a very common expression in Albanian essentially referring to 'cancer' . A similar expression can be found in the region of Epirus in Greece where the locals use the compound παλιαρρώστια 'bad-illness' . Here the first part of the compound παλι-literally means 'old' but it is often used to mean 'bad' too as in παλιάνθρωπος 'bad-person' . In (26), one of the speakers is narrating how he got to be a school teacher and in his speech we find another such semantic hybrid. First, we have a momentary switch from Greek θα δώσεις to Albanian gjuhë-letërsi and back to Greek και εσύ. What is at issue here, however, is the use of δώσεις, literally 'you give' , to mean 'you will teach a course' (in this particular case, gjuhë-letërsi 'language and literature'). In Greek, teaching a class, a course, or a lesson would be formulated with διδάξεις 'teach' and not with δώσεις 'give' . This construction in (26) makes sense, however, if we take into account the Albanian verb japesh, which can have the meaning of 'you give' , but also 'you drive' when the topic of discussion concerns cars and driving, and 'you teach' when the topic of discussion concerns courses and classes. Thus, the speaker in (26) appears to have transferred this meaning of 'to teach' of the Albanian japesh, which has 'give' as its basic meaning, to the Greek word δώσεις 'to give' , which does not generally have such a meaning.14 Interestingly, when speakers use a construction such as δίνω μάθημα, literally 'give a course' , in standard Greek, they are conveying that they will take an exam or a test in that course and not teach it. It must also be emphasized that not all types of language mixing are isolated events. The reality is that often the different factors converge to result in multiple types of language mixing in one utterance. For example, we see hybridization at many levels, including the somewhat underappreciated phonological hybrids.
Some questions that remain to be answered concern the extent of shared knowledge of linguistic forms. One consideration is whether what we have identified as borrowings are indeed the local standard forms. That is, the Greek-Albanian bilinguals in Palasa may not know the more standard Greek form, or at least do not use it in their local dialect. In addition, it is possible that some of the examples in this paper reflect individual usages or practices that are not shared linguistic norms among the Greek-Albanian bilingual members of the community. We leave these as open questions at this point, to be checked against further data from Palasa and environs.