Examples of complex real situations.

Lest it appear that this is too finely tuned for any natural language situation, this list may be compared with that shown in Rubin (1968:518-20), where she attempts to distinguish in what contexts citizens of Paraguay use their repertoires of Spanish and Guaraní, or the careful distinctions made by triglossic Luxembourg, whose citizens' differential use of Lëtzebuergesch, German, and French is described by Hoffmann (1981).

To make the contrast between repertoire and register clearer, let us examine what a particular speaker's repertoire might be like if a speaker had a repertoire commanding registers that involved two levels of a diglossic language plus socio-professional registers from another (third) variety, in this case an international language. To be specific, this woman is a Tamil-speaker and has a B.S. in Mathematics and Computer Science.

Figure 2.4: A diglossic and multilingual linguistic repertoire; the speaker is female, Tamil-speaking, with higher degrees in mathematics.

In the schema in Fig. 2.4, the repertoires of this speaker are divided up among two varieties of a diglossic language (Tamil), and English (in fact Indian English, since there are lexical, syntactic, intonational, and phonological differences between IE and American or RP varieties.) [This table doesn't show the English registers.] The speaker controls some repertoires actively and some only passively; the technical/mathematical registers are actively controlled in (Indian) English, and it is safe to say that the speaker would not be able to function in Tamil in these modes except to do low-level arithmetical calculation. In some cases the speaker may actively control a Tamil register and passively control an English register, such as in belles-lettres, meaning that such a person is capable of doing creative writing in Tamil but would probably not exhibit high proficiency in the English register, except passively, i.e. to read and enjoy a novel. For ordinary prose, the speaker would have an active command of both L-variety Tamil and English, but no prose register exists for L-variety Tamil, so Tamil writers have to `wing it' when they write dialogues in Spoken Tamil. (Some people also write letters in ST, but this is also without standard conventions.). A male speaker of Tamil with a similar professional background would probably have a similar repertoire, except that his control of `Motherese' would be passive, rather than active, his control of `Automotive/mechanical' might be active, his control of `folk taxonomies' would more likely be passive and his control of `sports' would perhaps be active.

Useful though these squared-off tables may be, I prefer to propose another type of diagram, one that represents linguistic repertoires as a set of concentric circles rather than of parallel columns. The reason for preferring a diagram such as that shown in Fig. 2.5 is that the nested circles can better demonstrate a number of factors not indicated by the parallel columns.

1. One is that all speakers of a diglossic language share the innermost circle, which also represents early childhood and nuclear-family intimacy. This is the domain of family life, story-telling, jokes, folk wisdom, conversation, food, street life, and intimacy. With schooling, speakers begin to acquire domains associated with education in the L-variety language. As Tamil speakers approach the end of secondary education, where they have mastered the H-variety of their language, proficiency in English increases, and certain registers are reserved for the domain of English---technical subjects, mathematics, social sciences, etc. Higher education expands these repertoires further, but they will only be acquired after the acquisition of H-variety Tamil repertoires. The circles can be understood as a progression from the center outward, with thresholds at the H-variety Tamil `boundary' and at the English `boundary'. Different speakers will have different repertoires, represented by different sectors or segments within each circle, but the sequence of acquisition will be similar for all speakers.The only exception would be someone who attends an English-medium school from an early age, in which case acquisition of H-variety Tamil may be minimal, and if nothing is done to change this, the control of this domain will be passive at best.
2. The concentric-circle model emphasizes that the L-variety language is the basis of and at the center of ones linguistic repertoire---it is the native language, the language one dreams in, the first learned and the last forgotten. All other repertoires are added on, but do not replace the L-variety. This is the repertoire that is taken for granted, and typically is given no protection in language policy---domains and reserves are usually only specified for H-variety languages and/or foreign imports, with exceptions in a few polities such as Luxembourg.
3. The concentric-circle model allows us to account for gradience and variability in proficiency, something that few policies seem to want to deal with. The line that begins at the center and moves toward the edge can represent gradient ability, since proficiency usually increases with time.
4. The concentricity of the circles also shows that if domains shift,That is if one register is replaced or displaced by another, perhaps as a result of a conscious policy decision. it is usually from one adjacent circle to another, not hopping over one band to penetrate a non-adjacent circle. A passive competence is possible within any circle, of course, but the largest number of passive competencies will most likely be in the language learned last, in this case English. The circle might also contain a small `passive' segment for yet another language, since in India it is quite common for speakers who claim no active command of Hindi to watch and enjoy Hindi movies, or to understand a bazaar language associated with marketing and shopping. In the concentric circle model, a niche for passive Hindi (Cinema) is shared with Tamil and English.
5. The circles also show what happens when attempts are made (such as have been in India and Sri Lanka) to restructure the repertoire with demands for reserved segments for another language (such as Hindi or Sinhala, respectively). English then functions as the buffer language, the neutral territory that keeps the invasive languages at bay. Under Indian language policy, the attempt to capture domains of languages not reserved for Hindi has so far been restricted to those registers associated with English, whereas in Sri Lanka, the perception (among Tamils) was that Sinhala intended to invade and replace not only English sectors, but sectors that had been the reserve of H-variety Tamil, such as entrance examinations to university. English, under that policy, was to be stripped away, or reduced severely. It is perhaps the disparity between the levels of penetration that explains the intensity of Tamil resistance in Sri Lanka, compared with that seen in India, where Hindi made claims only to the registers in the outer band of sectors.
6. The concentric-circle model also depicts language registers and repertoires as an integrated, natural continuum---something akin to a living organism, rather than just a list of disparate proficiencies. As such, the threat of another language is seen as an invasive virus seeking to penetrate the cell, displace certain repertoires, alter its makeup so as to replicate itself, and perhaps even kill the original cell. One could carry this virological analogy too far, but it could help explain something of the emotional reaction to language spread and policies that seek to bring it about.
7. Finally, the concentric-circle model allows us to depict diglossia as a feature of registers and repertoires. As such it is not a deficit or a disfunctionality, but a natural part of the repertoire of a speaker. Diglossia is, after all, a kind of language policy, since it is the result of decision-making of some sort, cumulatively over time.
8.