Bilingual education as a starting point to preserve Welsh: a comparison with EU policy and other local situations

In the previous sections, focus has been on the reasons pushing a bilingual to choose one language instead of another and on the implications of this choice. However, it is common that during a conversation between bilinguals the speakers use both languages, switching from one to another. This phenomenon is called code-switching and Grosjean (1982: 145) defines it as "the alternate use of two or more languages in the same utterance or conversation".
He later points out that there is an important difference between code-switching and borrowing: the former concerns words, phrases and sentences that are inserted in a conversation without being integrated phonologically and morphologically into the other language, unlike borrowing (Grosjean, 1982: 146).
Therefore "çam' étonnerait qu'on ait code-switched autant que ça" is an example of code-switching from French into English, since the expression in italics has maintained its English phonological and morphological structure within the French sentence, while "ça m' étonnerait qu'on ait code-switché autant que ça" contains a borrowing (code-switché) whose English structure has been adapted to the French orthographic and grammatical system3. Code-switching is also compared to other cases of language contact: unlike language mixing and interference, in code-switching the two languages involved still keep their own structure and do not influence each other on a syntactic or grammatical level (Milroy and Muysken, 1995: 74). However there is a type of code-switching in which some syntactic changes may take place, as happens in creolisation and pidginisation: in such contexts, it represents a way to create new forms of words by combining elements of the first language with others of the second; the resulting words will be later included in the creole language.

Milroy and Muysken give an example of the way this type of code-switching gives rise to new words by mentioning the Greek Cypriot community in London. Within the community, it is more and more common to use the Cypriot verb "kamno" (do, make) in combination with English verbs (i.e. "kamno use" meaning "to use"), but in Cypriot kamno can be followed only by a noun (Milroy and Muysken, 1995: 78). The explanation for these structural changes is the outcome of this particular type of code-switching: a new creole or pidgin language is different from bilingual discourse, where the two languages are simply used alternatively, it is a real new language originated by a mixing of the two, with a new, original result4. Code-switching may also have an alternational nature, when the two languages are juxtaposed in the same sentence and the elements do not form a constituent, or an insertional nature, when there is a base-language in which the second is inserted. There are several ways to determine which language is the base: from a discourse-oriented point of view, it is the one of the conversation; as far as statistics is concerned, it is the language the majority of the words are uttered; in psycholinguistics, it is the one the speaker is most proficient in (Milroy and Muysken, 1995:180-182). Grosjean (1982: 146-149) analyses also attitudes towards code-switching: it is a common opinion that this phenomenon should be avoided, since mixing two or more languages into the same utterance is often considered a symptom that the speaker does not have a full competence in any of them. Hence expressions like "Franglais", "Tex-Mex" or "Spanglish" have a more or less explicit negative meaning, showing how code-switching is often perceived as an equivalent of semilingualism or nonlingualism, although research has shown that most bilinguals are still able to use the two languages separately and correctly, when the context requires it. The reasons for this attitude are found in the widespread opinion that code-switching reflects an interference in bilinguals' speech, taking place when they are not able to continue the conversation in the language being used at that moment (Myers Scotton, 1993: 48).
In the past such undervaluation of the phenomenon prevented many linguists from even considering code-switching as a research topic, and Myers Scotton herself claims that the prejudice about code-switching led her to consider language mixing just as the habit of using different languages with different interlocutors, excluding the possibility of using more languages in the same utterance (Myers Scotton, 1993: 49). A binary language choice model actually pointed that only one language could be the outcome of a specific interaction; this restrictive interpretation did not take into consideration that code-switching is a natural response to the need to communicate, with the speaker being aware that their interlocutor will understand the message in one language or another.
Moreover according to Heller (1988:1-3) code-switching is a strategy for defining roles and role relationships in the processes of social negotiation, provided that the interlocutors share the same communicative resources: since a language carries a shared, deep social meaning, such as closeness or distance, the switch from one to another also highlights the relationship between social processes and linguistic phenomena. There are even contexts in which code-switching is considered normal both for ingroup communication, as for the Puerto Ricans in Harlem, who consider code-switching the natural expression of their dual identity (Heller, 1988:222), and intergroup communication, as for French Canadians, whose code-switching originated from contact with English speakers throughout the centuries (Heller, 1988:232).