In the earliest generative work, Chomsky took linguists to be writing grammars for socially defined languages like English, Turkish, or Warlpiri (Chomsky Reference Chomsky1957: 11). Later he postulated ‘an ideal speaker-listener, in a completely homogeneous speech-community, who knows its language perfectly’ (Chomsky Reference Chomsky1965: 3), idealizing away from individual variation.

Twenty years afterwards, in 1986, he abandoned the idealized speaker-listener and the idea of socially defined languages as having any kind of psychological reality. He embraced language variation and distinguished between external E-language and internal, individual, intensional I-languages as the proper focus of linguistic theory. In effect, he took up ideas from nineteenth-century German philologists and adopted Wilhelm von Humboldt's (Reference Humboldt, Buck and Raven1836) distinction between the languages of nations and those of individuals. Hermann Paul (Reference Paul1877: 325) emphasized the individual and biological view of language, noting in an early work ‘dass die reelle Sprache nur im Individuuum existiert’ [real language exists only in individuals]. Later, he attacked the group psychology of Lazarus and Steinthal and wrote that ‘Wir müssen eigentlich so viele Sprachen unterscheiden als es Individuen gibt’ [we must in fact distinguish as many languages as there are individuals] (Paul Reference Paul1880: 31).

Chomsky (Reference Chomsky1986) followed von Humboldt and Paul, distinguishing external E-language and internal, individual I-languages. E-language refers to language out there in the world, the kind of thing that a child might be exposed to, an amorphous, unanalyzed, mass concept. I-languages, on the other hand, refer to biological systems that grow in children's mind/brains in the first few years of life and characterize an individual's linguistic capacity. I-languages consist of structures, categories, morphemes, and phonemes, and computational operations that copy, delete, and assign indices to items, etc. One's I-language is a private object that, along with other mechanisms, permits thought and some communication, often incomplete or inaccurate, although it may differ from the I-languages of speakers communicated with.

After 1986, everything was individualized and linguists were free to postulate different I-languages for different individuals, allowing the possibility of billions of grammars or, now, ‘I-languages’. Indeed, a little later, Kroch (Reference Kroch1989) postulated that individuals may have multiple co-existing grammars, and that idea turned out to be very productive for work on syntactic change and opened the possibility of billions and billions of grammars, which could be studied through the methods of population biology. Under this view, multilingualism is everywhere and does not reflect special ‘contact situations’ (Lightfoot Reference Lightfoot2011).

Jürgen M. Meisel, Martin Elsig & Esther Rinke (henceforth MER) balk at this individualization. Indeed, they think that it ‘borders on the absurd’ (8) and the main driving force for their book is to plead for a return to the ‘speech community’ and a smaller number of available grammars. There are convenient fictions that are adopted for particular purposes, like the setting of the sun, the French liver, a runner's stride; nobody believes that such things exist but it is sometimes convenient to invoke them for expository purposes. MER, however, do not propose the speech community as a convenient fiction but as an actual element of psychological reality; they aspire not to discover individual I-languages but socially defined S-languages (my term), holding of a speech community. They rightly believe that historical linguists must attend to the linguistic community but wrongly assert that ‘if we … pursued a purely biologically oriented research programme, diachronic linguistic analyses would not be feasible’ (7); there is nothing incompatible about a biological approach to grammars and attention to group properties.

Rather than treating children as growing their grammars afresh, MER adopt the view of earlier eras that a grammar is a social object that moves through time and space:

They say nothing about how a speech community would be defined precisely, or a grammar as a social object, which would entail addressing old chestnuts about how to distinguish a language from a dialect or how an S-language might treat phenomena that occur in certain I-languages but not others. Postulating a social grammar poses many problems.

We will return to these difficulties later but let us first welcome the book as a serious attempt to relate morphosyntactic change to language acquisition by children. The authors are surely right to seek to explain diachronic changes in grammars through their acquisition and they have written what is in many ways a well-conceived, valuable, and readable book, relating details of acquisition to details of change. They offer rich discussion of two major changes in French, the (separate) losses of verb-second and of null subjects (Chapter 5). There is also good discussion of particular interrogative constructions that developed in Quebec French (Section 2.2).

They are appropriately skeptical of the claim that there are weightings or biases built into Universal Grammar (UG) to explain historical tendencies for grammaticalization or cycles or that such changes are driven by internal factors of economy (67, 91, 181). Similarly, Chapters 3 and 4 are devoted to showing that structural ambiguity cannot trigger new grammars. The authors are also careful to distinguish properties of grammars from factors of language use; Chapter 2 has helpful discussion of changes taking place in individuals across their lifespan, and argues that they reflect changing usage rather than changing grammars.

However, they get into a tangle in explaining ‘transmission failure’ in an S-language changing from one generation to another (Section 3.2). So it is worth noting that under the individualist perspective there is no perfect or imperfect transmission of I-languages, because nothing is transmitted: I-languages grow afresh in each individual on exposure to ambient E-language and, particularly, to the Primary Linguistic Data (PLD), which are a part of ambient E-language. The only way a person can develop a new I-language is if he/she is exposed to significantly different E-language, including new PLD, and there is no transmission to explain.

This confusion about transmission is complicated further by their formulation of a ‘paradox’ that results

But this is a fundamental error resulting from MER's commitment to social grammars and to the traditional generative view of children evaluating the capacity of grammars to generate a corpus of data. Under the individualist view, the E-language that children are exposed to results not from the output of one group S-language but from the aggregate I-languages of many individuals, modulated by changing factors of language use, as MER showed in their Chapter 2.

Despite MER's overwrought claims, proponents of I-languages do not ignore the role of the linguistic community. Indeed E-language is the product of speakers in the child's environment and that is where triggering experiences are to be found, but MER, like many linguists, pay little attention to properties of E-language. E-language, emerging from many sources, is a critical element in accounts of language acquisition and therefore in accounts of diachronic change explained by acquisition.

If one uses the methods of population biology, the ideas of mass E-language and individual I-languages suffice for the purposes of accounting for language acquisition and we do not need the problematic, conventional, socially defined Spanish or Swedish or S-languages. One can view children as acquiring their individual I-language on exposure to external E-language. However, rather than evaluating systems against sets of data, children should be viewed as paying no attention to what any I-language or grammar generates but instead as growing an I-language by identifying and acquiring its elements (Lightfoot Reference Lightfoot2006). Children parse the E-language they hear and acquire the categories and structures needed to understand what they are exposed to; in that way, they accumulate the elements of their I-language. Specific aspects of unanalyzed E-language trigger specific elements of I-languages, a different approach to language acquisition.

Consider very briefly the well-known emergence of a distinct class of Inflection elements in Middle English, resulting from the loss of rich verb morphology (Lightfoot Reference Lightfoot2006). In early English, items like can, may, must, etc. were merged as verbs, as is shown in (1) below and, like all verbs, could move to a higher functional category, I or T. Because these items had been members of the preterite-present class, the loss of verb morphology rendered them distinct from other verbs in not having the one surviving property of verb morphology, the third person singular -s ending (which the preterite-presents never had). Now preterite-presents, lacking that -s ending, had become morphologically distinct from other verbs. Evidence shows that children exposed to the new E-language grew I-languages with the old preterite-presents categorized as Inflections rather than as verbs, merged into structures like (2), and as a result there was new syntactic distribution (with the obsolescence of forms that could not result from a structure like (2): She has could eat fish, Canning eat fish, She wants to can eat fish, She will can eat fish).

1. (1) … I _VP[_V can eat fish]

2. (2) … _Ican _VP[eat fish]

For a period, both systems co-existed: some speakers had (1) and others had (2), the former becoming rarer over time, the latter more numerous. A large literature is now devoted to this kind of sociological variation, changing over time, and the spread of the new system can be studied through the methods of population biology. As some speakers began to have the new I-languages with (2), that affected the ambient E-language, making it more likely that members of a speech community would acquire the new I-language. Eventually (1) disappeared from I-languages; Sir Thomas More was the last known speaker to have the obsolescent forms manifesting (1), in the early sixteenth century, and they are not manifested in texts after that time. Niyogi (Reference Niyogi2006) provided interesting computational simulations of this approach to language acquisition and diachronic change.

New E-language with impoverished verb morphology triggered new I-languages that failed to generate certain sentence types. We understand the spread of the new system through the methods of population biology. Nothing is transmitted and there is no paradox, whereby speakers with the old and new system were exposed to the same PLD. Approaches using amorphous, mass E-language and individual I-languages provide a good account of the emergence of new systems, while socially defined grammars fail to work, confront major conceptual problems, and are unnecessary.