Contact Linguistics. Chap

Morphological constraint 1 (Borrowing):
The greater the congruence between morphological structures across languages in contact, the greater the ease of borrowing.

When there is both functional and structural congruence between morphological elements, even highly bound morphemes can be transferred. In addition to the example given earlier (Meglenite Rumanian borrowing of Bulgarian verb inflections), Thomason & Kaufman (1988:57) cite the example of case and number suffixes from Standard Serbo-Croatian replacing older suffixes in Chakavian dialects of that language. Changes like these, involving the importation of morphemes to express functional categories already present in the recipient language, seem to be among the most common types of morphological borrowing (Thomason & Kaufman 1988:54), and are especially frequent in dialect contact situations. The Viking Norse influence on Northern ME may be a case in point. As noted earlier, Thomason & Kaufman treat this as a case of “typologically favored borrowing” (1988:97), a claim which may have validity to the extent that native speakers of OE imitated Norse speakers. It is certainly reasonable to argue that the diffusion of Norse features into Northern ME was facilitated by the close typological fit between the languages.

In cases involving contact between less closely related languages, typological similarities or congruence in certain subsystems of the grammar can also favor structural borrowing. The changes in Kupwar Urdu under Kannada influence (discussed earlier) seem to fall into this category, though it isn’t clear that they were borrowings in the strict sense. The various changes in gender marking and agreement patterns were facilitated by the close similarity in morphosyntactic structure between the languages. The same is true of the diffusion of some morphemes from Marathi into Kupwar Urdu, a pattern which may be compared with the spread of certain Marathi morphemes into Standard Hindi (Thomason & Kaufman 1988:98).
Most of the borrowings from Ritharngu into Ngandi in Arnhem Land also fit the pattern described here. The morphological structures of the two languages are quite similar, particularly in the case of verbal derivation and inflection. For instance, the borrowing of thematizing suffix d™u by Ngandi was facilitated by the fact that the two languages shared a verb-inflectional pattern of the form: Vroot + thematizing suffix + inflection - as well as the fact that the favored shape of a thematizing suffix in Ngandi is CV (Heath 1978:110). Similarly, the sharing of a verb-forming pattern of the form: Noun stem + Verb-forming suffix + inflection made it easier for Ngandi to borrow the derivational suffix which converts nouns to verbs. Other instances of borrowing in Arnhem Land seem to involve the replacement of a native morpheme by one from the external language - a process once more facilitated by a close typological fit between the languages. Examples include the borrowing of inchoative verbalizer t™i from Ritharngu into Ngandi, apparently replacing an older verbalizer *man (found in Ngandi's close relatives Nunggubuyu and Rembarrnga (Heath 1978:117). The same may be true of thematizing suffix d™u, discussed above, as well as several other morphemes that were borrowed to replace older native ones in the recipient languages Ngandi and Warndarang, from the source languages Ritharngu and Nunggubuyu respectively. Heath does not place enough emphasis on congruence or typological similarity as a factor favoring morphological borrowing in these cases.


9.3. Constraints based on transparency/markedness.

Perhaps the most important structural constraints on morphological borrowing are related to general principles based on markedness, and particularly on the notion of transparency. Several of Weinreich's observations on the diffusion of morphemes seem to appeal to this notion in one way or another, for instance:

• The fuller the integration of the morpheme, the less the likelihood of its transfer (p. 35).

• Morphemes with complex grammatical functions seem less likely to be transferred by the bilingual than those with simpler functions. (p. 34).

• A relatively unbound morpheme is more likely to replace its counterpart in another language if the latter is more bound and is involved in a greater variation of alternants in fulfilling corresponding functions. (p. 34).

• In the interference of two grammatical patterns, it is ordinarily the one which uses relatively free and invariant morphemes in its paradigm - one might say the more explicit pattern - which serves as the model for imitation, (p. 41).

As we saw above, Heath attempted a reformulation of these criteria in terms of factors reflecting the degree of transparency of a potential morphemic loan. As Thomason & Kaufman (1988:) note, all of these factors "fit well with the notion of markedness as connected with ease of learning." All fall out from a very general constraint on morphological borrowing which we might formulate as follows:

Morphological Constraint 2 (Borrowing):
The greater the degree of transparency of a morpheme, the greater the likelihood of its diffusion. By contrast, the more opaque (complex, bound, phonologically reduced) a morpheme is, the less likely it is to be borrowed.

Again, we find evidence of constraints based on transparency and markedness in many instances of morphological borrowing surveyed in this chapter. The clearest examples come from situations involving contact between typologically similar languages or dialects, where congruence and transparency operate simultaneously. But we find equally compelling examples in other situations. In Arnhem Land, Heath noted that most borrowed morphemes in fact satisfied all of his criteria favoring diffusability, while those resistant to borrowing did not. Among the latter were bound pronominals and verbal inflectional affixes which were typically non-syllabic (e.g., expressed by a single consonant), and in the case of verbal inflections, multi-functional (portmanteau morphs). Interestingly, not all the morphemes that met the criteria were in fact borrowed - a fact that might be explained partly by functional factors, discussed below.

Scholars working on other contact situations have also appealed to "transparency" as an explanation for morphological borrowing. For instance, as noted in Chapter 2, Dalton-Puffer (1996) used concepts such as semantic and morphotactic transparency to explain why certain French-derived derivational morphemes became productive in Middle English. She found that borrowed morphemes that were most transparent in both senses were the ones that tended to become most productive in later Middle English.
Dalton-Puffer's approach can easily be extended to cases involving the borrowing of inflectional morphology. Her concept of transparency has the advantage of appealing to both semantic and formal criteria, which are not clearly distinguished in Heath's taxonomy of factors favoring morphological borrowing. Heath's factors may be subsumed under Dalton-Puffer's two dimensions as follows:


Semantic Transparency Morphotactic Transparency
unifunctionality syllabicity
categorical clarity sharp boundaries
<---analogical freedom--->

As noted before, constraints based on structural notions such as transparency and naturalness are related to more general cognitive principles that regulate the transfer of morphemes across linguistic systems. Weinreich himself related his structural constraints to "psychological" factors such as "economy" (p.24) and the preference for a more explicit pattern (p. 41). Thomason & Kaufman (1988:56) also note that structural constraints "fit well with the notion of markedness as connected with ease of learning." Approaches which treat markedness values as relative to the typological distance between the systems in contact (e.g. Mufwene 1990) would seem to be especially useful in further research on morphological borrowing (and other types). As Thomason & Kaufman (1988:57) suggest, such an approach, has the advantage of recognizing that

"it is not morphology itself that is marked and unlikely to be transferred from one language to another; rather it is certain common features of morphological structure that often, but not always, make morphology hard to learn."


9.4. Functionally-based constraints on morphological borrowing.

There is some evidence that functional constraints also play a role in structural borrowing. The addition and loss of morphological categories is often dependent on such constraints. There are cases where a category present in one of the languages in contact is absent in the other(s). Depending on the direction of influence, such mismatches may lead either to the addition of a new category to the language that previously lacked it, or its loss in the other. For cases of the former type, the constraint may be worded as follows:

Morphological Constraint 3 (Borrowing).
The existence of gaps in the morphemic inventory of a recipient language facilitates the importation of new morphemes and functional categories from a source language.

For cases involving loss of categories, the following constraint applies:

Morphological Constraint 4 (Borrowing):
The lack of a functional category in a source language may lead to loss of a similar category in a recipient language.

Evidence of functionally motivated borrowing of morphemes comes once more from Arnhem Land. For instance, the borrowing of ergative/instrumental suffix -t™u- by Ngandi (from Ritharngu) and the borrowing of instrumental suffix -miri by Nu (from Ritharngu) seems to have been encouraged by the lack of these categories in the recipient languages. In several other cases, however, borrowed morphemes simply replaced functionally equivalent ones in the recipient languages. In general, functional factors played a minor role, and fail to account for most cases of morphological borrowing in Arnhem Land. The Kupwar situation also provides few examples of functionally motivated borrowings, such as the introduction of an inclusive/exclusive "we" distinction into Kupwar Urdu on the model of Kannada.

Dalton-Puffer also notes that certain French derivational suffixes introduced into Middle English became productive because they filled functional gaps and conveyed distinctions not previously made in the language. These included adjective-forming suffix -able and feminine agentive suffix -esse (1996:221).
It is clear, however, that linguistic explanations for morphological borrowing must appeal to all of the constraints discussed in this chapter, whether based on typological distance, transparency, or functional factors.


10. Constraints on syntactic diffusion.

Comparatively little attention has been paid to the study of cross-linguistic influence at the level of syntax, and consequently we know little about the constraints that apply to it. Throughout this chapter, we have suggested that syntactic structure very rarely, if ever, gets borrowed. In stable bilingual situations, there are very strong constraints against such change, even in languages subjected to intense pressure from a dominant external source. As we saw in section 4.1, Silva-Corvalán (1998) argued that apparent syntactic changes in Spanish under English influence are relatively slight adjustments that do not affect the abstract syntactic structure of Spanish, but only its surface properties. For example, the use of jugar rather than tocar to refer to the playing of music simply introduces a surface modification of the subcategorization properties of jugar. This led Silva-Corvalán to propose that this kind of syntactic “borrowing” can only take place when the changes conform to the structure of the rl (1998:226) This of course is what other scholars such as Meillet and Jakobson have argued for in the case of morphological borrowing.

When extensive syntactic diffusion does occur, as it clearly does in cases of extreme convergence, it appears to be due to mechanisms of change associated with language shift. Understanding the constraints on this kind of syntactic diffusion means, first, understanding the vehicles of such change. The chief of these is substratum influence in learner versions of a recipient language, leading to innovations that may subsequently be imitated by native rl speakers. Another closely related factor may be substitution/switching of construction types across languages in the kinds of language mixture practiced by more or less skilled bilinguals. Bilinguals who are dominant in the external language being switched to, appear to play a key role in importing foreign features into an ancestral language, though these may not always be imitated by native speakers of the latter. For instance, Silva-Corvalán (1998:233) points out that bilinguals with less proficiency in LA Spanish (that is, those who are in a more advanced stage of shift and are English-dominant) introduce more significant English-influenced changes into Spanish than more proficient speakers. Thus examples of exact calquing such as shown in (11) were produced by two speakers with extremely reduced proficiency in Spanish.

(11) LA Span Yo gusto eso

I like-1s that
Gen Sp. A mi me gusta eso
To me pro please-3s that
“I like that”

Here gustar ‘to please’ is reanalyzed as a transitive verb with an Experiencer subject and an accusative theme on the model of English like, whereas in Gen. Spanish it is a verb with a “theme” or “patient” subject and an indirect experiencer object.

In addition, bilinguals have the ability to substitute entire phrasal and clausal constituents for each other, and in many cases, to switch the “matrix” language into which they incorporate structures from the other language. As we will see in Chapter 5, these strategies of language mixing are quite common in code-switching situations generally.
Hence it is not clear in what sense we can talk about constraints on syntactic diffusion in situations of convergence. We clearly need to distinguish constraints on substitution or incorporation of external features into a speaker’s primary language from constraints on L1 influence on an acquired L2. Since these issues are treated in chapters 5 and 7 respectively, we can postpone our consideration of them. For the moment, we can say that, in both cases, the greater the congruence between syntactic structures in two languages in contact, the greater the likelihood that one will replace the other. For instance, Thomason & Kaufman explain the relative ease with which basic word order is borrowed or acquired by noting that patterns like SVO and SOV are functionally congruent. In other words, they "typically perform the same basic syntactic function - identification of subject and object by their position relative to each other and to the verb" (ibid.). They cite several examples of such word order changes, such as the change from SOV to SVO in Finnish under Indo-European influence, and from SVO to SOV in Austronesian languages of New Guinea under Papuan influence. However, functional congruence by itself cannot explain what factors favor or inhibit these kinds of replacement. It is not clear, for instance, whether replacement of SOV by SVO is more likely than the reverse, and if so, why.