However, not all language switches are lexically motivated in this way. An alternative approach is to
consider the role of syntactic 'rules' in determining when language switching may take place. The use of such rules was pointed out by Ervin-Tripp (1973), although her examples mainly involve code-switching within a language rather than switching between languages as in the bilingual case. She mentions, in particular, the importance of rules of
co-occurrence.
She illustrated these by the following imaginary episode.
"How's it going, Your Eminence? Centrifuging OK? Also have you been analyzin' whatch' unnertook t'achieve?" This utterance is unacceptable because it includes both formal and casual speech. Similar rules of co-occurrence may also determine when two different languages can be combined, as well as different monolingual codes.
If it is established that language switches are in all probability constrained by certain linguistic rules, the next task is to establish what these rules might be. One method for doing this used by Gingras (1974) and Timm (1975) is that of 'acceptability judgements.' This method involves getting bilingual subjects to make acceptability judgements for a range of sentences that have different language
switches in them. Gingras found that some language switches were far more acceptable than others. For instance,
switches at the sentence's major constituent boundaries were more acceptable than sentences where the switches were
random.
It was also found that bilinguals varied in their judgements of acceptability, depending on whether they had acquired their second language during childhood, or as adults.
However there are problems with the use of
acceptability judgements. Bilingual subjects are being asked to judge the relative acceptability of what are, by most standards, all unacceptable utterances. Certainly the data suggests that they are able to do the task, but there must always be some doubt as to whether the judgements
necessarily accord with the actual utterances of bilinguals themselves.
This particular problem is overcome by the studies of Pfaff (1979), Poplak (1979, 1980), and Poplak, Sankoff and Miller (1988) who used real data. Pfaff carried out
analyses in order to determine which parts of speech were likely to be switched in which types of sentences. From these she deduced that speakers who switched from Spanish to English did so according to a number of constraints. One of the constraints she mentions is that of 'triggering', as discussed above. Others were perhaps specific to the Spanish/English pairing, for instance switches to English verbs were only permitted when preceded by an inflected Spanish verb.
The speakers Pfaff recorded spoke a somewhat
Mixtureado, and, not surprisingly, there were many language switches in her data. But despite appearances to the
contrary, there were many constraints on the switches into English.
A large amount of evidence in support of the claim that language switching is not linguistically random is produced in this study. It would however be a mistake to suppose that the constraints extablished by Pfaff are followed 100% by all speakers. What she finds are no more than very
strong tendencies. Some of the constraints she documented do not accord with those suggested from acceptability
studies. In such cases of doubt, her data would appear to be the more reliable, but the huge variability in findings do cast some doubt on the whole enterprise. Other studies have also produced conflicting data. Some have found
constraints similar to those put forward by Pfaff, for
example Huerta-Marcias (1981), and Zentella (1981), whereas others have not (Sobin (1984), and Singh (1985).
There are several possible explanations for this kind of discrepancy. Perhaps the linguistic rules are very
complex indeed, and the rules developed so far too simple to account for all possibilities. Perhaps there are very large differences from sub-group to sub-group, so that a slightly younger sample, or one from a different region will produce different results.
In view of the difficulties involved in the specifying of many linguistic rules, Poplak and her co-workers put
forward suggestions of more universal constraints, involving only three major linguistic constraints on language
switching.
These are
al the free morpheme constraint - languages may be switched after any constituent in discourse provided that constituent is not a bound morpheme.
b) the equivalence constraint - switches will tend to occur at points in discourse where juxtaposition of L1 and L2 elements does not violate a syntactic rule of either language.
c) the size-of-constituent constraint.
According to this, switches occur more frequently at major constituents such as clauses, than at smaller
boundaries such as single words. The one exception to this is nouns which are frequently switched.
Poplak and Sankoff's extensive data corroborates the hypotheses of the free morpheme and equivalence constraints with less than 1 % of the 1 ,835 switches going against them. These findings are impressive, and the relative simplicity of these rules, makes it likely that they may apply generally to language switching. They have however been criticised for not being specific enough by, for example, Sridhar and Sridhar (1980), and significantly, studies from different parts of the world have failed to find even these very general constraints adhered to (Berk-Seligson 1986, Bentahila and Davis 1983, Clyne 1987).
Poplak and her co-workers did not limit themselves to looking at linguistic rules but made many other observations about language switching. Those that are most relevant to the empirical data of chapters 6 and 7 will be discussed
here. Firstly, they claimed that language switching is a very skilled activity, and is therefore most frequently found amongst the most balanced bilinguals. Language switches are generally made smoothly, and rarely include pauses, hesitation or editing phenomena.
On the basis of their studies of Spanish/English Puerto-Rican bilinguals, they say
"These findings...provide strong evidence that code-switching is a verbal skill requiring a large degree of linguistic
competence in more than one language, rather than a defect arising from insufficent
knowledge of one or the other. ...It is also striking that precisely those switch-types which have traditionally been considered most deviant by investigators and educators, those which occur within a single sentence, are the ones which require the most skill. They tend to be produced by the 'true1 bilinguals in the sample: speakers who learned both
languages in early childhood and who have the most on-going contact with the monolingual English-speaking world. Code-switching, then, rather than representing deviant behaviour, is actually a suggestive indicator of degree of bilingual competence."
In case this seems relatively obvious, it may be worth quoting the diametrically opposite view of Weinreich (1953), a noted authority in the field. He says that the ideal
bilingual is one who
"switches from one language to the other according to appropriate changes in the
speech situation (interlocutor, topic, etc.) but not in an unchanged speech situation, and certainly not within a single sentence"
The view taken in this study, and put forward in the Introduction is much more like Poplak's than Weinreich's.
The second point made by Poplak is apparent in the second part of her quotation above. She studied both
Spanish/English bilinguals in New York and French/English bilinguals in various Canadian communities, and came to the conclusion that there were differences in the language
switching patterns of different communities. In the
Puerto-Rican community, language switching was considered to be emblematic of a dual identity and people were proud of their ability with both languages. Here skillful language switching was common. Conversely, in some of the
French-Canadian communities, bilinguals down-graded their abilities, and rarely switched between languages.
A third claim made by Poplak, based on her Puerto-Rican study, was that the very generalised use of intra-sentential switching found in some speakers could represent an overall discourse mode. She contrasts this discourse mode with what she calls a discourse strategy that is used to achieve
certain effects, which she also found, particularly in her Canadian sample.
She asked the question what sort of speakers were most likely to use this discourse mode, and came up with some useful suggestions. She found, for instance that females, balanced bilinguals, and those who acquired L2 early rather than late were particularly likely to use it. Attitudes towards ethnicity were also considered but showed no significant findings. Some of the points made in the previous chapter in the discussion of Gal's (1979), and Giles' work may be relevant here. In particular, the question of individual choice (choice here need not necessarily imply a conscious process) - why do some individuals choose to language switch intersententially
while others do not and why is it chosen more frequently by individuals in some bilingual communities than in others? Questions such as these remain unanswered by Poplak's work, as she does not move beyond the demographic indicators to consider more psychological factors.
This finding is of particular interest here as several of the children whose language is described in chapter 6 did seem to be using a discourse mode very similar to that
described by Poplak.
The main trend of the work on linguistic explanations of language switching has been to move away from a search for specific linguistic constraints. As empirical findings threw doubt on many of these, rather more general
constraints were put forward. These too have been found to be problematic, although it is felt by Poplak, for instance, that such rules can be found. However, the alternative
point of view, namely that there are no linguistic
constraints to language switching has also been put forward (Lance 1975). In order to find such rules however, it is important to consider the individuals repertoire within the speech community, and to know what the norms are, if only to see if an individual's switching pattern is specific to him or her, or occurs widely within the community.
This makes the point that it does not seem possible to look at linguistic reasons for switching without also
considering psychological and sociological factors. It is to these we turn in the next section.
3.2 INTK/V-SENTENTIAL LANGUAGE SWITCHING - SOCIOLOGICAL AND .