Contact Linguistics. Chap

Exercise:
Examine and compare at least two contact situations, one involving stable bilingualism (e.g., French/Flemish contact in Brussels) and the other unstable bilingualism (e.g., French/English contact in Prince Edward Island). What differences do you find in the social settings and patterns of use? To what extent are these reflected in differences between the types of contact-induced changes in the minority language?


5. Sprachbünde: Contact across contiguous speech communities.

Another kind of setting which can lead to structural diffusion is prolonged contact across geographically contiguous language communities. The groups involved may develop close links and patterns of interaction for purposes of trade, or because of cultural practices such as exogamy, or because they are subsumed through conquest within a larger political conglomerate. A situation of this type, involving the diffusion of linguistic features across geographically adjacent languages, is referred to as a Sprachbund or linguistic area. The term "Sprachbund" was coined by Troubetzkoy (1928) who apparently saw it as a counterpart to the notion of "language family." It has been translated roughly as "language association", "language league" and "union of languages." Other terms that have been used include "convergence area", "diffusion area", and "affinité linguistique." But the term “Sprachbund” is now the generally accepted choice.

Sprachbünde or linguistic areas have been proposed for various parts of the world, on the basis of a variety of criteria, ranging from sharing of a few structural features to substantial similarities in several subsystems of the grammar. For instance, South East Asian languages like Chinese, Thai, Vietnamese, etc. have been treated as a Sprachbund since they share the feature of phonemic tone as well as certain morphological characteristics (Henderson 1965). Similarly, the diffusion of a series of click consonants from Bush-Hottentot languages into neighboring Bantu languages is the basis for treating this area as a Sprachbund (Guthrie 1967 - 71).
There are other more complex linguistic areas characterized by far greater degrees of convergence at more than one level of structure. Among the best-known are the Balkan Sprachbund, Meso-America (Campbell et al. 1986), South Asia (Emeneau 1980; Masica 1976) and the Pacific Northwest (Sherzer 1976).


5.1. The Balkan Sprachbund.

The Balkan Sprachbund is perhaps the best known and most widely-researched convergence situation in the field of Areal Linguistics, its study dating back to the 19th century. The primary languages of the Sprachbund include Albanian (whose dialects include Geg in the north and Tosk in the south); Greek; Romanian (a Romance language), as well as the Slavic languages Bulgarian, Macedonian and Serbo-Croatian (especially the Torlak dialects of the South-east). Other languages more marginally involved include Judezmo (also known as Ladino or Judeo-Espagnol), Romany and Turkish. In keeping with practice in the field, the languages of the Balkan Sprachbund will be referred to as the Balkan languages (as distinct from the geographically based term “languages of the Balkans” in general).

The socio-historical background to this situation involved prolonged contact among the above language groups during the period roughly from 800 to 1700 AD. Contact was due to a variety of causes, including war and conquest, trade, animal herding, etc. Invasions by different groups (Southern Slavs, Bulgars, etc.) led to a long period of migration across language boundaries, leading to the emergence of multi-lingual communities. One important factor in the areal diffusion of linguistic features appears to have been the widespread use of Greek as a High Language across these communities. This was related to the spread of Byzantine civilization and in particular the unifying role played by the Greek Orthodox Church. Hence Greek seems to have been the source of, or the vehicle for, many of the diffused features. However, Greek was also the recipient in some cases, so the picture is not that clear.
The full details of this contact situation are still unknown. However, the linguistic consequences can be seen in various types of convergence at all linguistic levels (Sandfeld 1930; Schaller 1975). In phonology, the Balkan languages share the absence of suprasegmental features such as length and nasalization in vowel articulation, as well as the presence of a mid-to-high central vowel /I/ or /´/ (not present in Greek or Standard Macedonian, though it occurs in some Macedonian dialects). Schaller also pointed out that the vowels systems of the languages had merged to some extent, all having at least the vowels i – e – a – o – u (Joseph 1986:106)
In morphology, we find the following convergence features (among others):

1. A post-posed enclitic definite article (all languages except Greek and Turkish) as in the following examples.

Noun Noun + Article Meaning.

Bulgarian voda voda-ta "water"
Romanian lup lup-ul "wolf"
Albanian shok shok-u "comrade"

2. A periphrastic future marker derived from a verb meaning "want".

Greek: Qa Prafo (cf. Qa < Qeli na 'he wants that')

Romanian: o sa& scriu (cf. va “he wants”)
Albanian: do të% shkruaj (cf. do “s/he wants”)
Serbo-Croatian pisa-c!u (cf. hoc!u “I want”)
"I will write"

3. A periphrastic perfect formed with an auxiliary verb corresponding to "have" (except for Bulgarian).

4. The merger of dative and genitive cases in Albanian, Bulgarian, Greek and Romanian.

5. The merger of locative and directional expressions in Bulgarian, Greek and Romanian. Thus Greek stin eláDa "in Greece" or "into Greece".

In syntax, we find the following shared features:

1. Loss of infinitival structures in favor of finite complements. This is found in all of the Balkan languages. Joseph (1983) provides a thorough account of this development.

Greek: Qelo na Prafo

Serbian: hoc!u da pis&am
I-want that I-write "I want to write"

2. The use of pleonastic personal pronouns to cross-index direct or indirect animate objects.

Romanian: I-am scris lui Ion.

to-him I wrote him John. "I wrote to John".

In addition, there has been a great deal of diffusion of lexical items across the Balkan languages. Some examples include:

1. The borrowing of single lexical items such as Greek Dromos > Alb. dhrom, Bulg. drum, S.Cr. drum, Rom. drum "drum". Also, Turkish boya "paint, color" > Alb. bojë%, Bulg. boja, Greek boyá, Rum. boia, S.Cr. boja.

2. Numerals 11-19 are expressed by combinations whose literal meanings are "one on ten, two on ten" etc (not found in Modern Greek, though occurring in Medieval Greek).¹⁰

Albanian: një% -mbë%- dhjetë [ny´mb´Dyet]

Bulgarian: edi(n)-na-deset
Rumanian: un-spre-zece (< *unu-supre-dece)
one on ten "eleven"

3. Various calques on idiomatic expressions. For example,

(a) "without a doubt" is expressed by "without other".

Albanian: pa tjetër

Bulgarian: bez drugo.
Greek: xorís álo

(b) The expression "whether one VERBS or not" is conveyed by the structure "VERB-not-VERB. Joseph (2000:147) suggests that Greek may have been the source of this pattern.

Bulgarian: pie ne pie "whether one drinks or not"

Greek: fíji De fíji "whether one leaves or not"
Romanian: vrea nu vrea "whether one wants or not"
Turkish ister istemez "willingly or not"

It has been suggested that both borrowing and substratum influence were involved in the creation of the Balkan Sprachbund. However, it is difficult to determine which type of change applies in a given instance. Sandfeld (1930) proposed that the diffusion was due to borrowing from a single language (Greek) into the others. But some of the evidence points to asymmetrical patterns of diffusion. For instance, it has been claimed that Bulgarian is genarally the recipient language in grammar, but the source of phonological innovations, especially in Romanian. But this claim remains controversial. As for the role of substratum influence, this has been interpreted to mean the influence of a pre-existing speech community on the linguistic groups that came to settle in the Balkans. But as Joseph (1983) suggests, it is equally likely that mutual accommodation and shift among the immigrant groups themselves promoted the spread of features. On the whole, the Balkan Sprachbund remains problematic for any attempt to explain the directions, agency and mechanisms of convergence in a principled way. Moreover, as Joseph (1986:111) points out, “there is an entire sociolinguistic dimension to the Balkan Sprachbund which must be confronted” before we can fully understand how it came into being. A better understanding of both the mechanisms of change and the factors that constrained them can be gained by examining situations for which more socio-historical and linguistic information is available.

Despite all the scholarly attention that the Balkan Sprachbund has attracted, the fact is that the degree of lexical and structural diffusion found in most of the languages involved is not that extensive. Heath (1984:378) in fact claims that “ongoing mixture involving European languages and native vernaculars in former colonies is at least as extensive as in the Balkans.” This is perhaps what we might expect in situations of language maintenance, particularly when the groups involved belong to quite distinct communities. Resistance to extreme structural change remains strong in such cases. Even when there is intimate inter-community contact leading to massive lexical diffusion, structure seems to resist externally motivated change.


6. A case of intimate inter-community contact: Arnhem Land.

A case in point is the situation in Arnhem Land, Australia, where several Aboriginal languages belonging to quite distinct language families have converged remarkably over the centuries (Heath 1978). The major language families involved are the Yuulgnu languages, including Ritharngnu, Dhay?yi and others, and the "Prefixing" languages, divided into a northern subgroup containing Ngandi, Nunggubuyu and others, and a southern subgroup containing Warndarang, Alawa and others. These languages came together in the distant past in Arnhem Land as a result of migration. The clans affiliated to each language group congregated during the dry season to hold major ceremonies, to which clans from other language groups were invited. All groups practiced strict exogamy, so that marriage across language boundaries has long been common. For instance, a typical group consisting of four clans and 200 persons would typically have about 50% inter-group marriages (Heath 1981:359). The wife joined her husband's group, and children would acquire both their father's language (as their primary vernacular) and their mother's language, which was used in ceremonies and other rituals. The groups that were closest became most similar culturally and linguistically, and those with the longest history of close association now exhibit the highest degree of convergence. It’s important to note that groups in Arnhem Land spent long periods without close contact; hence individuals might not use their second language for several months during the wet season. This may have helped preserve language boundaries despite the widespread diffusion.

There is clear evidence of convergence among the Arnhem Land languages in the following areas (Heath 1978):


Morphosyntax:
• Pronominal systems.
• Demonstrative determiners and adverbs.


Syntax:
• Free word order in all languages.
• Juxtaposition as the primary strategy of clause linking, with occasional formal subordination.
• Lack of agentive or passive participial formation.
• Similar uses of potential verb forms (e.g., in counterfactual conditions).
• Similar structure for causal sentences, etc.

In addition, there has been massive lexical diffusion across the languages, often involving significant portions of basic vocabulary. Many of these instances of convergence developed in the distant past, and are difficult to demonstrate now. However, using a combination of historical reconstruction and comparison of synchronic structures, Heath was able to document several instances of more recent structural and lexical diffusion across several of the Arnhem Land languages. Two pairs of language groups in particular have had close recent contact - Ritharngu (Yuulnu family) and Ngandi (North Prefixing) on the one hand, and Nunggubuyu (North Prefixing) and Warndarang (South Prefixing) on the other. Both pairs have developed very strong connections through intermarriage and joint ceremonies. The Ritharngu (Ri), Nunggubuyu (Nu) and Warndarang (Wa) groups appear to have been of fairly large size prior to contact, numbering approximately 200 to 300. The Ngandi (Ng) group was fairly small, numbering perhaps 60 to 70 persons. The disparity in size between the Ri and Ng groups has some implications for the kinds of convergence that occurred, as we shall see. The Ng group is gradually shifting to Ri, and the Ngandi language is becoming extinct, being now restricted to a few fluent speakers (Heath 1981:358). The close association between the Nu and Wa groups has led to a gradual absorption of the latter by the former, and in fact the Wa language has recently become extinct. This too is a clue to the types of convergence that took place between Nu and Wa.

6.1. Ritharngu and Ngandi:
Convergence between Ri and Ng involved heavy lexical as well as structural diffusion. There has been massive exchange of lexicon across the two languages, including significant portions of core vocabulary. In lexicon, Ng has borrowed more from Ri than vice-versa, probably because of the larger size of the latter group. According to Heath (1981), the borrowing extends over a variety of lexical fields such as trees and shrubs, insects, physical feature nouns, body part nouns, kinship terms and so on. Some idea of the extent of diffusion is given by the fact that Ri and Ng share some 65% of a set of 106 tree/shrub terms, 35% of 40 insect names, and 28% of 69 body part nouns (Heath 1981:355). Borrowed items include nouns and verbs, as well as other categories such as directional terms ("north, south, up, down") and numerals. On the whole Ri and Ng share about 50% of all the noun and verb stems (about 1300) examined by Heath (1978:29). The borrowing of so-called "thematic" verbs, especially from Ri into Ng, was accomplished in spite of significant differences in verb morphology between the two languages. Thematic verbs have isolatable root forms to which a thematizing suffix -d™u- can be attached, with inflectional endings placed last, for example:

(Ng) d¡ak = root for "cut". d¡ak-d™u-Ni = past continuous "was cutting".

To facilitate the transfer of such verbs, Ng also borrowed the thematizing suffix -d™u- from Ri, along with the root forms themselves. This meant that roots in this class could be borrowed back and forth, and indeed this class of verbs is the largest and most productive in both languages now (Heath 1981:350).

Another interesting aspect of the diffusion between the two languages is the (mediated?) borrowing of grammatical morphemes, especially from Ri into Ng. These included the following:

• Ergative/instrumental suffix -t™u- (< Ri -d™u- ), attached to nouns to express ergative or instrumental case. E.g. Ri Naya-t™u (I-ergative)

• Inchoative verbalizer -t™i- (a derivational suffix attached to nouns to form intransitive verbs.
E.g.: Ri d¡a:l "strong, firm" > d¡a:l-t™i "to become strong, firm".
• Genitive/dative/purposive suffix -ku- (< Ri -gu-).

Ri for its part acquired negative suffix -?may? and the kin-term dual suffix -ka? from Ng (1978:143). In addition, there are several other shared affixes and postpositions for which the directionality of diffusion is unclear. It seems reasonable to assume that the spread of these structural features was facilitated, at least in part, by heavy lexical exchange, as in other situations we have examined.

There is also clear evidence of structural diffusion from Ng into Ri involving "pattern transfer" of the type associated with substratum influence. Interestingly, there appears to be no such influence in the opposite direction. In phonology, Ri has become more similar to Ng in the distribution of glottal stops. Other than this, there has been very little phonological diffusion between the two, perhaps because their phonological systems were already quite similar, though not entirely so in the case of vowels. Heath suggests that the present congruence in the consonant systems was probably due to a long history of convergence whose details cannot now be uncovered. In morphosyntax, several innovations in Ri can be traced to Ng influence. For instance, Ri developed a series of enclitic pronouns marking subjects and objects on the pattern of Ng, which, like the other Prefixing languages, has obligatory pronominals indexed to subjects and objects in all clauses. This feature is not found in the other Yuulngu languages to which Ri is closely related genetically. An example of this use of pronominal clitics follows (Heath 1978:126):

(7) Ri. n™a:-wala d™ali-ña Nay d™in?-wac&-n™a d¡aramu-y

saw them he the women (acc.) man (erg.)
"The man saw the women"

Note that Ri did not imitate the word order of Ng clauses, only the pattern of a nuclear clause consisting of verb + pronominals, fleshed out by NPs identifying the subject and object, as in the example above.

Under Ng influence, Ri has also developed a strategy for creating relative clauses and gerunds by attaching a subordinating suffix -Nu to a clause, as in the following (Heath 1978:128):

(8) wa:ni-na--Nu ra baNgul¡?

having gone I return
"Having gone, I returned" OR "I, who had gone, returned"

This pattern is quite different from that in other Yuulgnu languages, where infinitives are used in relative clauses. It seems clear that the new Ri pattern has its source in Ng, which uses a subordinating prefix ga- in the same function as Ri suffix -Nu.

Finally, Ri has made changes in certain verbal (TMA) categories under Ng influence. These include the following:
• Merger of imperative and future in the same verb form. E.g.: bu-Nu "Kill!" or "will kill".
• A distinction between present (bu-ma) and future (bu-Nu).
• The emergence of a new potential category (bu-w-a "would kill") for use in counterfactual conditionals.

None of these features is found in Dhuwal, a close relative of Ri, yet they have clear equivalents in Ng. The effect of all these changes was to make the Ri verbal inflectional system more similar to that of Ng than that of Dhuwal. Again, it seems reasonable to assume that these innovations were introduced to Ri by Ng speakers learning the former language, and are therefore instances of substratum influence.

6.2. Nunggubuyu and Warndarang:

The patterns of diffusion just described for Ri and Ng are quite similar to those for Nu and Wa. Briefly, they involve significant direct borrowing of lexicon and (indirectly) structural features from Nu into Wa, and some degree of substratum influence from Wa on Nu. Morphemes adopted from Nu into Wa include instrumental suffix -miri, elements of the noun class prefix system, and possibly ablative case marking suffix -wala (Heath 1978:144). There is evidence of significant substratum influence from Wa on the phonology of Nu. Innovations in the latter include the loss of a fortis/lenis opposition in stops, loss of glottal stops, and the restructuring of the vowel system into 3 basic sounds (/i, a, u/) with length distinctions, by contrast with Nu's close relative Ngandi which has five vowel qualities. Heath (1978:64) suggests that these changes in Nu were brought about by Wa speakers who pronounced Nu sounds with their own substituted native phones. This is reminiscent of Weinreich's "phone substitution" by learners shifting to a target language, a classic example of substratum influence.

6.3. Factors regulating convergence in Arnhem Land.

The situation in Arnhem Land is highly instructive with regard to the conditions under which significant structural diffusion takes place. The assymetrical patterns of diffusion we have found in these two language pairs are in keeping with the kinds of dominance relationship that seem to hold between their members. In both cases we have a numerically superior group (Ri and Nu) exerting influence on a smaller group (Ng and Wa respectively) which is increasingly adopting the language and culture of its larger neighbor. In both cases, language shift has occurred, with the smaller group gradually giving up its language in favor of that spoken by the larger group. Indeed, as noted earlier, Wa is now extinct, and Ng is fast becoming so. Yet both of these languages have maintained their distinctiveness and their full complexity to the end. This would provide an explanation for the patterns of diffusion - heavy borrowing from the dominant into the subordinate languages, and substratum influence in the other direction, consistent with the pattern of shift and language death. From this perspective, the situation in Arnhem Land has much in common with other known cases of language attrition and death.

However, our focus here is on language maintenance, and on the innovations that take place in maintained languages as a result of contact. In all the cases we have considered in this chapter, the communities concerned have continued to maintain their languages, despite varying degrees of external influence. In the cases involving extreme diffusion of both lexical and structural features, the vehicles of change include both borrowing and substratum influence, and in some cases it is difficult to separate the two. For this reason, it is not practical to equate language maintenance only with borrowing situations. For instance, no one would deny that both Ritharngu and Ngandi continued be maintained for generations, yet the former changed primarily under substratum influence, while the latter changed primarily as a result of structural borrowing. The agents of change in a maintained language can include native speakers of that language, speakers of some other language, or speakers bilingual in both. If that is the case, then it calls into question the conventional wisdom concerning several other cases of convergence that have been attributed to borrowing alone.


7. Heavy to extreme structural diffusion: Borrowing or Substratum Influence?

Thomason & Kaufman (1988:66) argue that “the traditional prerequisite for structural borrowing …. Is the existence of a bilingual group within the borrowing language speaker population.” This seems to be supported by all the cases examined so far. However, their definition of structural borrowing as “structural interference initiated by native speakers of the recipient language” (1988:66) is rather vague as to the actual mechanisms of change involved in these cases. It seems to imply, at first glance, that structural features can be incorporated and integrated into an rl in the same way as “direct borrowings, e.g., of lexicon, are. But this is clearly not the case. The “native speakers” who initiate the structural changes are in fact also proficient, to varying degrees, in the sl. This familiarity with the sl allows them to change the rl via imposition or transfer from the former. This would mean that the actual mechanisms or processes by which such structural diffusion occurs are similar to those found in cases of shift or second language acquisition. The only distinction would be that shifting speakers are less familiar with the rl (their target language) than bilinguals, at least in the earlier stages of shift. If that is the case, then the boundaries between “structural borrowing” and “interference through shift” are at best fuzzy. In this instance, as in all cases of contact-induced change, we need to distinguish the mechanisms of change (e.g., importation vs imposifrom their results (the “borrowings”).

In addition to the cases we have considered, there are many other maintained languages that have experienced significant structural change under pressure from an external source. Thomason & Kaufman (1988) explain several cases of this type as instances of heavy to extreme structural borrowing, placing them in category 4 or 5 of their “borrowing scale.” In the light of what we have discussed so far, there is good reason to ask whether these kinds of convergence may involve, not (just) borrowing under rl agentivity, but rather change under sl agentivity, via mechanisms of transfer or imposition. Among the situations that raise such questions is the contact between Old Norse and Old English, which has been traditionally regarded as a case of “structural borrowing.”


7.1. Old Norse influence on Old English.

The invasion and settlement of northern and midland areas of England by Norse-speaking Norwegians and Danes lasted from roughly 865 to 955. The size and duration of the settlement led to a situation of relatively intense contact between Norse and English speakers, resulting in considerable lexical and some structural influence from Norse on the English spoken in the Danelaw. Norse speakers gradually shifted to English over the next hundred years or so, but their language left its mark on Northern and Midland varieties of English. T & K (1988:292-98) demonstrate that these dialects, and especially Northern ME, adopted a large number of Norse function words, derivational and inflectional affixes as well as several hundred lexical items. Among the grammatical traits adopted were:

• Pronouns: they, them, their, replacing Old English equivalents hie/he:o, him/hira and heom/heora respectively.

• Quantifiers: minne "less", seer "various" (replacing OE lœ:ssa, syndrig/sundrij respectively.
• A variety of strong forms of verbs, e.g., give(n), gaf, geeven "to give"; vs OE jefa(n)/jifan, jœf/je:fon, jefen/jifen.
• Locatives such as whedhen "whence" and hedhen "hence"
• Prepositions such as til "to", fraa/froa "from", etc.

Table 1 compares a few grammatical features in (Northumbrian) Old English, Viking Norse and Northern Middle English. It shows the close correspondence between the latter two languages, by contrast with O.E. (Source: Thomason & Kaufman 1988:293, Table 6).

Table 1: A comparison of grammatical features in OE, ON and Northern ME.

Old English Viking Norse North. ME

Pronouns hi:e, he:o “they” Qei-r they