Jürgen Meisel's (JM) article is literally thought-provoking, especially for the issues that one can raise out of the central position that he develops, viz., “although bilingual acquisition in situations of language contact can be argued to be of significant importance for explanations of grammatical change, reanalysis affecting parameter settings is much less likely to happen than is commonly assumed in historical linguistics” (p. 142). This is a position that calls for grounding language change, hence historical linguistics, in the pragmatics/ethnography of language practice, a question that linguists can continue to ignore no more than the actuation question (Weinreich, Labov & Herzog, Reference Weinreich, Labov, Herzog, Lehman and Malkiel1968; McMahon, Reference McMahon1994; Labov, Reference Labov2001; Mufwene, Reference Mufwene2008). The latter regards what particular ethnographic factors trigger particular changes at particular places and at particular times but not at others. In other words, do structural changes happen simply because they must happen or because particular agents are involved at specific times under specific ecological conditions of language practice?
In response to JM, I contend that in the first place grammatical change need not be associated with (child) successive bilingualism, in which the dominant language, acquired (a little) earlier or practiced more often, influences structures of the second language. Change can happen any time, even in monolingual populations, when patterns of intra-community language variation change owing to modifications of the structure of the relevant population. I argue that transmission and acquisition are important factors in this process, though we cannot continue to use these concepts without questioning them (Mufwene, Reference Mufwene2001, Reference Mufwene2008, Reference Mufwene2010).
As in much of the linguistics literature, JM uses the terms transmission and acquisition in a Saussurean way, with language being an institution that native speakers would preserve intact if it were not for nonnative speakers who acquire it imperfectly, under the influence of the languages they have acquired previously. Otherwise, native speakers would transmit their linguistic systems to their children intact, more or less like animals are usually believed to transmit intact their biological characteristics to their offspring. Conversely, in ideal conditions, children would acquire their parents’ systems faithfully, just like biological offspring are believed to inherit the biological characteristics of their parents. However, even this characterization of biological inheritance is inaccurate. Biologists speak of gene-recombination. Thus, although the genes inherited by the offspring from the common gene pool availed by their parents remain intact (except in cases of mutations), the recombinations are not identical. This leads to family resemblance (lack of identity despite numerous similarities) among kin. Biologically, offspring certainly do not replicate their parents, especially since the latter have different family genetic histories.
Like any cultural transmission, linguistic transmission is a misnomer without a match in biology, for several reasons: (i) the putative transmission is primarily horizontal, determined by whom the learner interacts with; (ii) it is partly bidirectional, because the model speakers also learn from the learners, though this is more the case if the caregivers are nonnative speakers; (iii)) acquisition is more important than transmission, as the latter amounts only to making the primary linguistic data (PLD) available to the learners, who have to select what features they want to integrate in their systems – in fact, unlike in biology, no speaker/signer would “acquire” the language of their social environment if they did not work hard, however unconsciously, at projecting, indeed by induction, a system that would enable them to communicate. This is what makes possible JM's “reanalysis”, to which I return below; and it is also to it that parameter-setting applies, perhaps in ways different from the received doctrine in generative linguistics, as I also point out below.
The whole transmission/acquisition link becomes more complex once one factors in a population dimension, which brings the following factors to bear, while shedding light on why evolution involves more than multiplying ontogenetic development by the number of speakers and by generations: (i) It is for good reasons that modern linguistics (although one can already detect this insight in Hermann Paul's work in the 1880s; see Paul, Reference Paul1880, Reference Paul and Strong1891) has the notion of idiolect, because every speaker has worked out their own system, so to speak, which is only similar to but not identical with those of other speakers in their communities. (ii) The “acquisition” process is selective, as every speaker's/signer's idiolect is largely determined by their unique interactional history in which they have adopted different features from different sets of speakers/signers guided by principles that remain largely elusive, although one may invoke ecology-specific markedness principles to account for the selections (Mufwene, Reference Mufwene2001).Footnote 1 (iii) Adding to complexity at the communal level is the fact that learners are not equally gifted, as is obvious from classroom experiences, for instance, where students acquiring information from the same teachers, in the same settings, and/or from the same sources do not assimilate it in identical ways (see also Lee et al., Reference Lee, Mikesell, Joaquin, Mates and Schumann2009). (iv) On the other hand, unlike biological organisms, idiolects also integrate the kind of variation that has preoccupied variationist sociolinguists, owing importantly to the fact that language “acquisition” is incremental and subject to polyploidy, accepting alternative variants from different speakers and allocating them to different structural or pragmatic functions. (v) Population structure is a powerful filtering factor which determines who the learner can interact regularly with, learn from, and affiliate with. (vi) Changes in population structure, such as when there is massive emigration to a colony or when a population must assimilate a substantial number of immigrants who bring new variants to the community, affect the feature pool that provides the PLD (i.e., utterances) to the learners, setting things up for change.Footnote 2
Undoubtedly, language contact fosters change, through extensive individual bilingualism within a population. However, language “acquisition” as what Andersen (Reference Andersen1973, cited by JM) characterizes as a reconstruction processFootnote 3 can produce change even in a monolingual or monolectal population, as it leaves a lot of room for some idiolects to diverge significantly from the norm in the communal system. We should wonder why changes do not occur more frequently than may be expected in any language. The reason is that there are only so many innovations or deviations from the current norms that are unique to particular individual speakers, since their mental and anatomical infrastructures, as well as their communicative needs, are already so similar. Chances are that one's innovations have already been produced by some other speaker/signer in the same or another network. This state of affairs endows the communal linguistic systems some sort of stability, while individual idiolects are creative and can in fact change over time (see below).
However, when the structure of a population changes, their feature pool too can change (significantly), as some variants can be lost or decrease in significance, new features can be introduced, and/or the demographic and statistical strengths of particular variants can change. This change of state bears on the language learning process as the PLD change accordingly. In this scenario children play an important role as agents of selection (as also suggested by JM; see also DeGraff, Reference DeGraff and DeGraff1999; Mufwene, Reference Mufwene2008). They select the variants that get the job done for them, as they favor those that are most frequent, the most transparent, the most regular, etc., relative to the PLD. They select the variants that appear to be particularly advantageous to them, including those that help them be integrated in the networks of their choice. Bilingualism just adds another dimension to the competition and selection involved in language “acquisition”, because it brings new features into the pool, but it is not the critical factor.
So, who are the agents of change? Potentially, it is anybody who interacts with others in a language community (DeGraff, Reference DeGraff and DeGraff1999), and who can innovate new strategies (phonetic, lexical, morphological, syntactic, semantic, pragmatic) or spread strategies innovated by other speakers/signers (Mufwene, Reference Mufwene2008), provided, as noted by Matthews (Reference Matthews2003), that these changes at the utterance level modify the underlying linguistic systems, especially at the communal level. I assume that if a communal linguistic system is modified, most idiolectal systems must be too, for the former is an extrapolation from the latter (Mufwene, Reference Mufwene2001).
However, the reader should beware! JM uses the term reanalysis more loosely than in historical linguistics, where it means that a speaker/signer assigns to a particular string of morphemes an analysis that is different from the current one. For instance, the coalescence of the future auxiliary phrase going to into gonna yields for some speakers a monomorphemic marker which loses the semantic transparency that links it etymologically to the motion verb go with the direction preposition to. In some dialects (e.g., African American English) the reanalysis allows its reduction to gon, which is no longer a progressive form and is used without the copula, as he gon sing ‘he will sing’.Footnote 4 JM uses reanalysis primarily in the sense of “process of reorganizing grammars”, also identified in creolistics as “restructuring” (Mufwene, Reference Mufwene2001).
On the other hand, language “acquisition” is incremental and is revised several times in a speaker's/signer's lifetime, though most of the system is in place by puberty (or perhaps even earlier, as claimed by JM). Speakers’/signers’ idiolects undergo several successive adjustments as their interaction networks change. For instance, a native speaker of an English dialect with double negation and negative concord can change these particular features after being drilled at school with standard English grammar and interacting with speakers who avoid these features. Likewise, a speaker of a variety without agentless passives, producing, say, they fired Peter instead of Peter was fired, can adopt the passive alternative in their adult and professional life. I am sure there is no shortage of examples to prove that parameters can be reset several times in a speaker's/signer's lifetime and parameter-setting is not the exclusive capacity of child learners. (See also Newmeyer, Reference Newmeyer2003.) Whether or not parameters are not reset, or whether or not the settings are not adjusted, in cases of language attrition is open to debate, as shown below.
It is imperative that syntacticians address the fundamental question of what it really means to say that the learner has set the parameters to such-and-such language-specific values. Can we assume the language learner to be a folk linguist who collects data and who, after gathering a representative set, proceeds to analyze them and determine the typological values to which the parameters must be set? Or, contrary to Lightfoot (Reference Lightfoot1997, Reference Lightfoot2006), can we assume that the process is not catastrophic, and patterns emerge gradually on their own, subject to several adjustments and readjustments, as the speaker/learner focuses on producing acceptable utterances while relying on partial analogies to previous productions (Mufwene, Reference Mufwene2008)? Although linguists profess to languages being rule-governed systems (which is plausible as the outcome of language “acquisition” as explained above), is it justified to assume that the learner follows only rules that are learned/acquired early during the learning process? When does one know that for a particular parameter they have acquired a sufficient and reliable body of data that is representative of the whole range of complexity that can be attested in the language? Deriving my questioning from a slightly different discussion of the subject matter by McCawley (Reference McCawley1976), how would the learner know that the classes they have identified for the application of particular rules are general enough? I advocate an emergentist approach to language “acquisition”, with self-organization taking place as the speaker's communicative experience increases (see also Lee et al., Reference Lee, Mikesell, Joaquin, Mates and Schumann2009).
There are a host of other issues, such as the distinction between internally- and externally-motivated change. After all, languages do not literally structure themselves on their own. Assuming an emergentist perspective, self-organization simply refers only to how order emerges from the particular ways speakers/signers use the communicative strategies they have developed. Changes can only be initiated by the way speakers/signers manipulate the strategies under pressure residing in themselves or experienced from the outside (Mufwene, Reference Mufwene2003). What are the specific kinds of pressures that qualify as internal to a language?
I also take issue with the claim that adult L2 learners produce deviant grammatical strategies from the target language because they have no, or less, access to Universal Grammar (UG). It is like expecting a PC computer to function when Microsoft operating software, the requisite infrastructure on which every other software runs, is crippled. No one can learn a language, or even continue to speak/sign their mother tongue, once their UG qua biological endowment for language is damaged or is not (fully) accessible. Also, because typological variation is enabled by UG itself (according to the principles-and-parameters approach assumed by JM), having access to UG during L2 acquisition does not prevent interference from the language(s) that the learner already speaks/signs (competently). Using a hybrid grammar is not mutually exclusive with having access to UG.
Overall, JM's article has the merit of prompting us to think harder on the causes, or to use Hickey's (Reference Hickey2003) term, “motives”, of language change qua system reorganization. I believe he has provided only part of the answer but not the fundamental one, which lies in the fact that language acquisition is reconstruction from multiple, variable, and changing inputs, and learners are not equally gifted.
Jürgen Meisel's (JM) article is literally thought-provoking, especially for the issues that one can raise out of the central position that he develops, viz., “although bilingual acquisition in situations of language contact can be argued to be of significant importance for explanations of grammatical change, reanalysis affecting parameter settings is much less likely to happen than is commonly assumed in historical linguistics” (p. 142). This is a position that calls for grounding language change, hence historical linguistics, in the pragmatics/ethnography of language practice, a question that linguists can continue to ignore no more than the actuation question (Weinreich, Labov & Herzog, Reference Weinreich, Labov, Herzog, Lehman and Malkiel1968; McMahon, Reference McMahon1994; Labov, Reference Labov2001; Mufwene, Reference Mufwene2008). The latter regards what particular ethnographic factors trigger particular changes at particular places and at particular times but not at others. In other words, do structural changes happen simply because they must happen or because particular agents are involved at specific times under specific ecological conditions of language practice?
In response to JM, I contend that in the first place grammatical change need not be associated with (child) successive bilingualism, in which the dominant language, acquired (a little) earlier or practiced more often, influences structures of the second language. Change can happen any time, even in monolingual populations, when patterns of intra-community language variation change owing to modifications of the structure of the relevant population. I argue that transmission and acquisition are important factors in this process, though we cannot continue to use these concepts without questioning them (Mufwene, Reference Mufwene2001, Reference Mufwene2008, Reference Mufwene2010).
As in much of the linguistics literature, JM uses the terms transmission and acquisition in a Saussurean way, with language being an institution that native speakers would preserve intact if it were not for nonnative speakers who acquire it imperfectly, under the influence of the languages they have acquired previously. Otherwise, native speakers would transmit their linguistic systems to their children intact, more or less like animals are usually believed to transmit intact their biological characteristics to their offspring. Conversely, in ideal conditions, children would acquire their parents’ systems faithfully, just like biological offspring are believed to inherit the biological characteristics of their parents. However, even this characterization of biological inheritance is inaccurate. Biologists speak of gene-recombination. Thus, although the genes inherited by the offspring from the common gene pool availed by their parents remain intact (except in cases of mutations), the recombinations are not identical. This leads to family resemblance (lack of identity despite numerous similarities) among kin. Biologically, offspring certainly do not replicate their parents, especially since the latter have different family genetic histories.
Like any cultural transmission, linguistic transmission is a misnomer without a match in biology, for several reasons: (i) the putative transmission is primarily horizontal, determined by whom the learner interacts with; (ii) it is partly bidirectional, because the model speakers also learn from the learners, though this is more the case if the caregivers are nonnative speakers; (iii)) acquisition is more important than transmission, as the latter amounts only to making the primary linguistic data (PLD) available to the learners, who have to select what features they want to integrate in their systems – in fact, unlike in biology, no speaker/signer would “acquire” the language of their social environment if they did not work hard, however unconsciously, at projecting, indeed by induction, a system that would enable them to communicate. This is what makes possible JM's “reanalysis”, to which I return below; and it is also to it that parameter-setting applies, perhaps in ways different from the received doctrine in generative linguistics, as I also point out below.
The whole transmission/acquisition link becomes more complex once one factors in a population dimension, which brings the following factors to bear, while shedding light on why evolution involves more than multiplying ontogenetic development by the number of speakers and by generations: (i) It is for good reasons that modern linguistics (although one can already detect this insight in Hermann Paul's work in the 1880s; see Paul, Reference Paul1880, Reference Paul and Strong1891) has the notion of idiolect, because every speaker has worked out their own system, so to speak, which is only similar to but not identical with those of other speakers in their communities. (ii) The “acquisition” process is selective, as every speaker's/signer's idiolect is largely determined by their unique interactional history in which they have adopted different features from different sets of speakers/signers guided by principles that remain largely elusive, although one may invoke ecology-specific markedness principles to account for the selections (Mufwene, Reference Mufwene2001).Footnote 1 (iii) Adding to complexity at the communal level is the fact that learners are not equally gifted, as is obvious from classroom experiences, for instance, where students acquiring information from the same teachers, in the same settings, and/or from the same sources do not assimilate it in identical ways (see also Lee et al., Reference Lee, Mikesell, Joaquin, Mates and Schumann2009). (iv) On the other hand, unlike biological organisms, idiolects also integrate the kind of variation that has preoccupied variationist sociolinguists, owing importantly to the fact that language “acquisition” is incremental and subject to polyploidy, accepting alternative variants from different speakers and allocating them to different structural or pragmatic functions. (v) Population structure is a powerful filtering factor which determines who the learner can interact regularly with, learn from, and affiliate with. (vi) Changes in population structure, such as when there is massive emigration to a colony or when a population must assimilate a substantial number of immigrants who bring new variants to the community, affect the feature pool that provides the PLD (i.e., utterances) to the learners, setting things up for change.Footnote 2
Undoubtedly, language contact fosters change, through extensive individual bilingualism within a population. However, language “acquisition” as what Andersen (Reference Andersen1973, cited by JM) characterizes as a reconstruction processFootnote 3 can produce change even in a monolingual or monolectal population, as it leaves a lot of room for some idiolects to diverge significantly from the norm in the communal system. We should wonder why changes do not occur more frequently than may be expected in any language. The reason is that there are only so many innovations or deviations from the current norms that are unique to particular individual speakers, since their mental and anatomical infrastructures, as well as their communicative needs, are already so similar. Chances are that one's innovations have already been produced by some other speaker/signer in the same or another network. This state of affairs endows the communal linguistic systems some sort of stability, while individual idiolects are creative and can in fact change over time (see below).
However, when the structure of a population changes, their feature pool too can change (significantly), as some variants can be lost or decrease in significance, new features can be introduced, and/or the demographic and statistical strengths of particular variants can change. This change of state bears on the language learning process as the PLD change accordingly. In this scenario children play an important role as agents of selection (as also suggested by JM; see also DeGraff, Reference DeGraff and DeGraff1999; Mufwene, Reference Mufwene2008). They select the variants that get the job done for them, as they favor those that are most frequent, the most transparent, the most regular, etc., relative to the PLD. They select the variants that appear to be particularly advantageous to them, including those that help them be integrated in the networks of their choice. Bilingualism just adds another dimension to the competition and selection involved in language “acquisition”, because it brings new features into the pool, but it is not the critical factor.
So, who are the agents of change? Potentially, it is anybody who interacts with others in a language community (DeGraff, Reference DeGraff and DeGraff1999), and who can innovate new strategies (phonetic, lexical, morphological, syntactic, semantic, pragmatic) or spread strategies innovated by other speakers/signers (Mufwene, Reference Mufwene2008), provided, as noted by Matthews (Reference Matthews2003), that these changes at the utterance level modify the underlying linguistic systems, especially at the communal level. I assume that if a communal linguistic system is modified, most idiolectal systems must be too, for the former is an extrapolation from the latter (Mufwene, Reference Mufwene2001).
However, the reader should beware! JM uses the term reanalysis more loosely than in historical linguistics, where it means that a speaker/signer assigns to a particular string of morphemes an analysis that is different from the current one. For instance, the coalescence of the future auxiliary phrase going to into gonna yields for some speakers a monomorphemic marker which loses the semantic transparency that links it etymologically to the motion verb go with the direction preposition to. In some dialects (e.g., African American English) the reanalysis allows its reduction to gon, which is no longer a progressive form and is used without the copula, as he gon sing ‘he will sing’.Footnote 4 JM uses reanalysis primarily in the sense of “process of reorganizing grammars”, also identified in creolistics as “restructuring” (Mufwene, Reference Mufwene2001).
On the other hand, language “acquisition” is incremental and is revised several times in a speaker's/signer's lifetime, though most of the system is in place by puberty (or perhaps even earlier, as claimed by JM). Speakers’/signers’ idiolects undergo several successive adjustments as their interaction networks change. For instance, a native speaker of an English dialect with double negation and negative concord can change these particular features after being drilled at school with standard English grammar and interacting with speakers who avoid these features. Likewise, a speaker of a variety without agentless passives, producing, say, they fired Peter instead of Peter was fired, can adopt the passive alternative in their adult and professional life. I am sure there is no shortage of examples to prove that parameters can be reset several times in a speaker's/signer's lifetime and parameter-setting is not the exclusive capacity of child learners. (See also Newmeyer, Reference Newmeyer2003.) Whether or not parameters are not reset, or whether or not the settings are not adjusted, in cases of language attrition is open to debate, as shown below.
It is imperative that syntacticians address the fundamental question of what it really means to say that the learner has set the parameters to such-and-such language-specific values. Can we assume the language learner to be a folk linguist who collects data and who, after gathering a representative set, proceeds to analyze them and determine the typological values to which the parameters must be set? Or, contrary to Lightfoot (Reference Lightfoot1997, Reference Lightfoot2006), can we assume that the process is not catastrophic, and patterns emerge gradually on their own, subject to several adjustments and readjustments, as the speaker/learner focuses on producing acceptable utterances while relying on partial analogies to previous productions (Mufwene, Reference Mufwene2008)? Although linguists profess to languages being rule-governed systems (which is plausible as the outcome of language “acquisition” as explained above), is it justified to assume that the learner follows only rules that are learned/acquired early during the learning process? When does one know that for a particular parameter they have acquired a sufficient and reliable body of data that is representative of the whole range of complexity that can be attested in the language? Deriving my questioning from a slightly different discussion of the subject matter by McCawley (Reference McCawley1976), how would the learner know that the classes they have identified for the application of particular rules are general enough? I advocate an emergentist approach to language “acquisition”, with self-organization taking place as the speaker's communicative experience increases (see also Lee et al., Reference Lee, Mikesell, Joaquin, Mates and Schumann2009).
There are a host of other issues, such as the distinction between internally- and externally-motivated change. After all, languages do not literally structure themselves on their own. Assuming an emergentist perspective, self-organization simply refers only to how order emerges from the particular ways speakers/signers use the communicative strategies they have developed. Changes can only be initiated by the way speakers/signers manipulate the strategies under pressure residing in themselves or experienced from the outside (Mufwene, Reference Mufwene2003). What are the specific kinds of pressures that qualify as internal to a language?
I also take issue with the claim that adult L2 learners produce deviant grammatical strategies from the target language because they have no, or less, access to Universal Grammar (UG). It is like expecting a PC computer to function when Microsoft operating software, the requisite infrastructure on which every other software runs, is crippled. No one can learn a language, or even continue to speak/sign their mother tongue, once their UG qua biological endowment for language is damaged or is not (fully) accessible. Also, because typological variation is enabled by UG itself (according to the principles-and-parameters approach assumed by JM), having access to UG during L2 acquisition does not prevent interference from the language(s) that the learner already speaks/signs (competently). Using a hybrid grammar is not mutually exclusive with having access to UG.
Overall, JM's article has the merit of prompting us to think harder on the causes, or to use Hickey's (Reference Hickey2003) term, “motives”, of language change qua system reorganization. I believe he has provided only part of the answer but not the fundamental one, which lies in the fact that language acquisition is reconstruction from multiple, variable, and changing inputs, and learners are not equally gifted.