Internal and external uses of a second language Vivian Cook

People who can speak more than one language always have a choice of which language to use. Approaches to second language learning have looked at this choice in terms of the circumstances at the time of speaking, in terms of the people who are being addressed, and in terms of the ways in which code-switching from one language to another takes place within a conversation. Much research has concentrated on the social aspects of using second languages, on the external use of language for communication or for social intercourse between people.

However language is also used for internal purposes: we use language to keep our diaries, to plan our day, to solve problems - for all the tasks of everyday life. It is important to the study of L2 learning to know how the two languages are used in roles that aim neither at communication nor at relationships with other people but serve the internal purposes of the speaker. A survey of the use of time in 12 countries showed that speakers in the USA for instance use language for conversation for 0.3 of an hour per day, in Yugoslavia for 0.6 of an hour (Robinson et al, 1972); the question is what happens to language the rest of the time? Internal uses may well outweigh external in terms of time. To accommodate both internal and external uses of language, as well as communicative competence, we need the wider concept of pragmatic competence 'which places language in the institutional setting of its use, relating intentions and purposes to the linguistic means at hand' (Chomsky, 1980, p.225). Language is not just a tool for relating to the world outside the speaker; it is a help in all sorts of tasks that are not concerned with other people. There are choices available to the speaker not only for social and public functions of language but also for those functions that are internal and private; looking only at the overt uses of language for communication and socialisation misses crucial aspects. This will be particularly true of educational issues: unless we know how and when children from minority language backgrounds resort to the L1 for their internal processing, we may have a false picture of the actual processes they are using in the classroom and hence of their language needs in the L2 classroom.

The literature of second language learning research however treats the issue of internal uses of the L2 comparatively sparsely, perhaps on account of its bias towards sociolinguistics rather than psycholinguistics. Introductions to L2 learning such as Ellis (1985), McLaughlin (1978), or Klein (1985) talk of communicative needs and social needs but seldom of those that are non-communicative and non-social. Only Mackey (1962) provides a general discussion of 'internal functions', which he lists as counting, reckoning, praying, cursing, dreaming, diarywriting, and notetaking. Mackey points to the extreme individual variation between bilinguals over which language is used for such internal functions and says 'It would be possible to determine these through a well-designed questionnaire.'

The most typical of such internal uses is the language in which the second language user dreams. While there is a popular belief that the bilingual dreams in the dominant language, Harding and Riley (1986) find little support for it; Grosjean (1982) found 64% dreamt in either language; Saunders (1982) feels the language of a dream 'is determined in much the same way as when awake'. A second internal use that is often mentioned is counting: Grosjean (1982, p.275) states that 'To the question "Which language do you count in?", 62 per cent said their first language.' Kolers (1978) claimed that bilinguals mostly did arithmetic in the language in which they had learnt it. A third internal use is the vague idea of 'thinking' itself; 70% of Grosjean's sample claim to think in both languages. However tiredness or extra emotion may increase the amount of L1 use (Grosjean, 1982).

Thinking shades over into so-called private speech - language spoken aloud but not addressed to a real, actually-present, other person, i.e. non-communicative. Saunders (1982) describes children who code-switch while playing by themselves. A further non-communicative use is language addressed to people who cannot understand it; Saunders (1982) found bilingual children who assume that animals speak the language of the environment but that their dolls are bilingual like themselves.

The aims of the current research were to see what sort of factual basis could be established for these internal uses of the two languages, to what extent speakers resorted to one language or another when not using language communicatively, and how much variation there was both between groups and between individuals. The word 'bilingual' has mostly been avoided here because of its variety of meanings; the basic sample of people studied were all proficient users of more than one language.

The overall design consisted of a questionnaire asking Subjects to report uses of the two languages on a scale running from 'always L1' to 'always L2'. Ss were given an example as follows, with the figures 1-7 added below to show the scoring scheme discussed later:

always L1	mostly L1	usually L1	both L1 and L2	usually L2	mostly L2	always L2
(1)	(2)	(3)	(4)	(5)	(6)	(7)

Before doing the main questionnaire Ss were asked to supply background details. They had to assign themselves to one of four age groups - under 18, 18-35, 35-60, and over 60; and to give their occupation. Then they had to state what they regarded as their first language (L1) and their second language (L2); further questions on L1 concerned the country in which they had spent their first five years and the language their parents spoke; further questions on the L2 concerned whether they had learnt it in a classroom or picked it up naturally; if learnt in a classroom, how long for and at what level; and the age when they had started learning their L2. They were asked to rate themselves as a speaker of the L2 on a scale:

     5             4              3            2                  1
like a                                                            like
native                                                     a beginner

They were also asked if they spoke other languages in addition to the first two.

A general question was given at the end of this section: "This questionnaire will be used with several different groups of people. If you think your own situation is not covered properly by these questions or you think any of the questions strange for your situation, could you say some more about it at the end?"

The uses that were covered represented a range of the internal and personal tasks, each of which was of interest in its own right. In addition some uses were asked about that were communicative and social, so that some comparison could be made between internal and external. The complete set of questions and results is given in the Appendix. The internal uses tested fell into several broad groups: self-organisational tasks that showed language used for individual planning, mental tasks that used language for arithmetic, memory tasks in which the user had to remember everyday information, unconscious uses where the user does not have conscious control over which language to use, emotional effects of language on other tasks, non-communicative uses to non-speakers, and praying to oneself. External uses fell into two groups; one receptive personal uses in which the speakers listened or read language, the other social uses in which other people were involved.

The Ss numbered 59 in all, consisting of 23 Francophone Africans studying English in England, 12 Finnish students of English, and 24 'mixed' L2 users. The Africans had a variety of West African L1s; French was given as the L2 in 17 cases, other African languages in 6 and English in 1. The Finns were all Finnish speaking except for one Swedish speaker; they all gave Swedish as the L2 except for the sole Swedish speaker, who gave Finnish. The mixed group spoke as L1 French, Czech, Japanese, German, Finnish, English, Welsh, Punjabi, Spanish, Urdu, and Persian; their L2s were German, French, Italian, English, Spanish, Welsh, Punjabi, and Russian. All but a handful of the mixed group were living in England at the time of answering the questionnaire; as well as their other languages they also knew English. Though the questionnaire was given out in England, it was not restricted to the position of L2 learners of English in England, but had a majority of subjects from other situations such as a multilingual African society, a bilingual European society, and individuals who happened to function at a high level in two languages. Since the questionnaire was given out in English there was also a demand that all the Subjects also spoke English if that was not already the L2 they specified. The Africans reported average L2 proficiency of 3.9 on the scale from 1 to 5, the Finns 3.9, the mixed group also 3.9. The overwhelming majority were concerned with education either as teachers or students, 50 out of 59; most were in the 18-35 age bracket.

We shall start by giving overall results for individual questions within each set of uses. A convenient way of looking at the results is in terms of the percentage of answers showing either an L1 bias (scoring 1-3) or an L2 bias (5-7) or being evenly balanced (4). Results are given as a percentage of all Ss who answered that they practiced this use of language; some uses such as keeping a diary or praying were utilised by only a few of the speakers, 26 and 35 respectively.

The 4 self-organisational questions asked for the language used for shopping lists, cheque stubs, appointments, and diary entries.

                                  N     1-3       4       5-7
making appointments 43      35%   26%   39%
shopping lists              55      31%   29%   40%
keeping a diary           26      25%    17%   61%
cheque stubs              41      29%      5%    66%

Shopping lists and making appointments were fairly equally divided between L1 and L2; cheque stubs and keeping a diary were more at the L2 end.

The 3 questions on mental tasks were concerned with the use of language for counting things, adding up, and working out sums.

                              N       1-3       4       5-7
counting things        46       54%    26%   20%
adding up                45       54%    16%   30%
working out sums    56        61%     9%   30%

All three seemed more at the L1 end, and a comparatively small number answered both L1 and L2; i.e. mental tasks were more uniformly carried out in either L1 or L2 rather than in both L1 and L2. Grosjean's 62% who counted chiefly in their L1 is mirrored by the 54% who did so here.

3 questions concerned the use of language for everyday memory tasks such as days of the week, working out routes, remembering phone numbers, and remembering historical dates.

	N	1-3	4	5-7
phone numbers	53	41%	15%	44%
working out routes	51	45%	33%	22%
days of the week	57	39%	26%	35%
historical dates	52	58%	24%	18%

With these memory tasks there is an evenness of language choice between L1 and L2 across questions, about 75% usually being more towards one end than the other, memory for historical dates being particularly at the L1 end.

Three questions concerned what can be called unconscious uses of language: dreaming, singing to oneself, and talking aloud to oneself.

	N	1-3	4	5-7
dreaming	54	54%	37%	9%
singing to oneself	45	40%	51%	9%
talking to oneself	48	46%	33%	21%

Results show a definite preference for the L1, and a lack of use of the L2; nevertheless between 33% and 51% used both L1 and L2 for these. The figure of 54% for dreaming in the L1 contrasts with Grosjean's figure of 64% using both languages, but the question may have been put in a slightly different way.

4 questions asked about the effects of emotions - happiness, pain or sadness, things going wrong, and tiredness.

	N	1-3	4	5-7
feeling pain or sad	55	55%	35%	10%
feeling tired	53	53%	32%	15%
things going wrong	51	27%	47%	26%
feeling happy	56	32%	55%	22%

Again a low set of scores for the L2 end, a variable set for L1 with pain and tiredness invoking the L1 more than happiness or things going wrong. A staple of spy fiction is the secret agent who is caught out when he swears in his first language when the interrogator stubs a cigarette out on his hand; this would not apparently be an accurate way of detecting spies in our sample.

Two questions asked about the use of L1 or L2 in circumstances where the listener would not be able to understand either language, here small babies and animals.

	N	1-3	4	5-7
talking to small babies	51	43%	31%	8%
talking to animals	58	38%	28%	34%

The answers divide fairly equally between L1 and L2 for this group, the only exception being the low scores for talking to babies in the second language.

A question on its own asked about the use of L1 or L2 for praying by oneself (as opposed to public rituals). A problem with this was the use of other languages than L1 or L2 for religious purposes, e.g. the use of Arabic by Muslims regardless of L1.

Moving away from the purely internal and non-communicative uses, four questions tested personal receptive use - listening to the radio, reading for pleasure and for work\study, and reading newspapers.

	N	1-3	4	5-7
listening to radio & TV	52	12%	46%	42%
reading for pleasure	52	35%	29%	36%
reading newspapers	48	12%	38%	50%
reading for work/study	50	18%	36%	46%

These results probably say more about the variation in the situations in which the L2 users find themselves than about the users themselves. If there is no radio and newspaper in the L1 available, perforce the use has to come in the L2. Similarly the high use of the L2 for reading for work or study may represent the background of the respondents, many of whom were teachers or students.

The final group of 8 questions looked at social uses of language, i.e. situations where there was a person to communicate with: writing letters, talking at home, talking to friends, shopping, working and studying, visiting the doctor or dentist, speaking on the phone, and writing down phone messages. The purpose was to be able to contrast and compare the internal uses with standard external uses.

	N	1-3	4	5-7
phoning	54	26%	52%	22%
writing letters	48	15%	65%	20%
phone messages	48	25%	25%	50%
talking to friends	54	17%	70%	13%
shopping	46	41%	26%	33%
medical treatment	53	34%	17%	49%
talking at home	56	43%	38%	19%
work\study	46	15%	35%	50%

The replies for these social uses showed a spread between L1 and L2, again presumably largely dependent on the social situation of the respondents. Particularly low for L1 bias were talking to friends and work\study; particularly low for L2 bias were talking at home and talking to friends. Particularly high for L1 were talking at home and shopping; for L2 work\study, medical treatment, and phone messages; for both languages writing letters, talking to friends, and phoning were particularly high. One point that teachers might note is the extent to which the everyday social situations were still carried out through the L1.

A second way of looking at the scores is to combine the questions for each group of uses, displayed in Table 1. As might be expected the average score tends to fall firmly in the middle of the range between 1 and 7, the average for all nine uses being indeed 3.5.

ranked L1 both L1 L2
arith closest bias and L2 bias
ave to L1 1-3 4 5-7
1. self-organisation 3.9 7 31% 20% 49%
2. mental tasks 3.3 4= 55% 17% 28%
3. memory tasks 3.6 6 48% 23% 29%
4. unconscious uses 3.2 3 49% 38% 13%
5. emotional effects 3.3 4= 44% 39% 17%
6. noncommunicative uses 3.1 2 44% 45% 10%
7. praying 2.8 1 60% 20% 20%
8. receptive uses 4.6 9 20% 38% 42%
9. social uses 4.0 8 27% 41% 32%

Ranked in order between L1 and L2, the two external uses overall are most towards the L2 end; praying and noncommunicative uses are most at the L1 end. But at best such figures prevent gross generalisations such as claims that praying and dreaming are only done in the L1.

The groups of uses can also be looked at as showing an L1 bias or an L2 bias or being evenly balanced, using the same method as above. Table 1 shows that over 50% of the answers had an L1 advantage for mental tasks and for praying, over 40% show an L1 advantage for unconscious uses, memory, emotional effects and noncommunicative uses. On the other hand over 40% of answers have an L2 advantage for receptive and self-organisational. And over 40% gave both L1 and L2 (i.e. 4) for noncommunicative and social. Thus in terms of individual answers the preferences are more marked than the arithmetical average suggests; a high proportion of L2 users prefer L1 for mental uses and praying; L2 for unconscious uses, memory tasks, and noncommunicative uses; and are equally torn between L1 and L2 for receptive and social uses.

Let us now look at the differences between the groups of Subjects. These are displayed in table 2. In terms of the relative position of the three groups on a dimension between L1 and L2, the Africans come at the L2 end on 6 uses, at the L1 end on 1 (noncommunicative); Finns come at the L2 end on none and the L1 end on 8; the Mixed group at the L2 end on 3 and the L1 on none. Thus there are some differences between the groups, the Africans tending to L2, the Finns to L1. For self-organisation for example, only 7% of African responses were L1 biassed (1-3); only 3.7% of Finns were L2 biassed (5-7).

arith
average Africans Finns Mixed
1. self-organisation 3.9 5.6 2.2 3.9
2. mental tasks 3.3 4.6 2.0 2.8
3. memory tasks 3.6 4.8 1.9 3.5
4. unconscious uses 3.2 3.2 2.8 3.4
5. emotional effects 3.3 3.2 2.5 4.0
6. noncommunicative uses 3.1 2.4 2.6 4.0
7. praying 2.8 3.1 1.5 3.1
8. receptive uses 4.6 5.1 3.8 4.2
9. social uses 4.0 4.6 2.6 4.4

The results are interesting in two ways. One is that they show a continuum between public and private use. The same person who is using the L2 in public is likely to be using the L2 in private in their heads. A question that follows from this is the extent to which proficiency in the language correlates with either or both of these; do we, as might be expected, learn public L2 uses before private uses, or are the two inseparable throughout? Since the subjects involved here had a narrow range of reported proficiency in the L2, further research is needed into the development of internal uses.

Beyond this lies the question of efficiency. Given that there are both internal and external uses of language, is it more efficient if people use their L1 or L2 for internal purposes? Research reviewed in Cook (1997) across a variety of cognitive processing tasks shows that L2 users have some deficit in their capacity to process information in the L2; for instance they are less able to do mental arithmetic (Marsh & Maki, 1978) or to judge relative numerosity (Dornic, 1977). Given an educational system where the child has to use language for internal processing in many kinds of classroom task, it is a delicate question whether the teachers should endeavour to maintain the internal use of the L1, which may have certain disadvantages. It should be clear however what this argument is not saying. Much recent research with bilingual children has attempted to show the relationship of bilingualism with cognitive development, typically arguing that the bilingual scores over the monolingual in the development of metalinguistic skills (Bialystok, 1987); often this leads to a threshold hypothesis claiming that bilingualism is good for you provided you have enough of it (Cummins, 1983; Skutnabb-Kangas, 1981). But these levels of metacognition or of cognitive development in a Piagetan sense have little or nothing to do with the processes seen in language use, internal or external. It is the working memory used in speech processing or in mental arithmetic that affects efficiency of processing, not the level of cognitive operation. To say there is a cognitive deficit in processing information is not to say that the levels of conceptual development attained by second language learners are in any way inferior to those of monolinguals.

Having refrained from using the word "bilingual" for most of this paper, what does the argument imply for the concept of bilingualism? Let us take the three forms of definition of bilingualism analysed in Skutnabb-Kangas (1981); definitions by competence ('native competence in more than one language', Haugen, 1969); definitions by function ('the alternate use of two or more languages by the same individual', Mackey, 1970); and definitions by attitude ('The speaker ... must be accepted as a native speaker', Malmberg, 1977). As normally interpreted, these all apply to external use of language - competence to communicate to others, use of more than one language as seen by others, ability to use the language as seen by others: use of a second language for internal uses is not considered bilingualism. Skutnabb-Kangas (1981) develops a thoughtful extended definition:

Again the definition concentrates explicitly on the external uses of language rather than on the internal uses, on the use of language in the community rather than in the mind. Useful as such definitions may be, they need rephrasing to take account of the ability of some individuals to use a second language for such everyday human activities as making shopping-lists, talking to pets, and adding up numbers. Definition by function as in Mackey (1970) or based on contemporary usage,as in the COBUILD dictionary 'involving or using two languages', is only adequate if such use covers internal as well as external function.