Polinsky and Scontras (Polinsky & Scontras) will be a welcome addition to the literature on heritage language development. As an up-to-date and thorough synthesis of both theoretical and empirical research that explores the most vulnerable grammatical domains, it will definitely pave the way for a comprehensive model of heritage language grammar.
Before scrutinizing their proposal, it is important to note that, although Polinsky and Scontras acknowledge the problems in establishing a baseline for comparison for heritage speakers, they rightfully suggest that what makes the study of heritage language development interesting is mainly the (potential) differences between heritage language grammar and the so-called native grammar. It is true that heritage language grammar does not always differ from the baseline grammar, but the mere observation that sometimes it does, albeit selectively, is what has actually stimulated extensive research on heritage language acquisition over the years. In my view, it is praiseworthy that Polinsky and Scontras recognize this driving force behind heritage language research and do not refrain from comparing heritage speakers to a particular baseline. Furthermore, their linguistic approach to heritage language development leaves aside the unfruitful question of whether heritage speakers should be considered native speakers (cf. Rothman & Treffers-Daller, Reference Rothman and Treffers-Daller2014; see also Ortega, Reference Ortega2019 for further discussion) and instead focuses on predictions about what human language competence might look like under unbalanced bilingualism. Polinsky and Scontras's approach to heritage language data, I argue, is what will further heritage language research and enable us to identify its resilient and vulnerable properties that may otherwise go unnoticed. As is also made clear in this keynote article, a model of heritage language development cannot possibly be developed before establishing where heritage speakers converge on the baseline, alongside considering the phenomena where they do not.
Having said that, I will briefly highlight some of the insightful propositions put forth in the article and the aspects where they need refinement. First of all, Polinsky and Scontras posit transfer, attrition, and divergent attainment as three reasons why heritage languages differ from the baseline. Nevertheless, this typology encounters circularity because transfer from the dominant language into the heritage language is commonly treated as a manifestation of attrition, not a separate phenomenon (e.g., Gürel & Yılmaz, Reference Gürel and Yılmaz2011; Schmid, Köpke, Keijzer & Weilemar, Reference Schmid, Köpke, Keijzer and Weilemar2004). Furthermore, the assumption that heritage speakers demonstrate divergent linguistic representations because they receive input that differs both qualitatively and quantitatively from the one that monolingual learners receive inevitably implies that the major source of heritage language input (i.e., input from first-generation immigrant parents) is also divergent. It is true that the input heritage speakers receive is reduced in quantity. However, most work on adult bilingual immigrants shows relatively stable and qualitatively native-like first language (L1) competence in this group (e.g., Montrul, Reference Montrul2008; cf. Schmid & Köpke, Reference Schmid and Köpke2017), casting doubt on the broad generalization that the linguistic properties of the parents lead to changes in the grammar of heritage language speakers. If, indeed, the language of first-generation immigrant parents, who serve as the main input source for heritage language, remains intact but the grammar of child or adult heritage language speakers appears to be divergent, then factors other than the quality of the parental input should be questioned more closely (cf. Montrul, Reference Montrul2008).
Having proposed input conditions and the economy of online resources as two triggers for deviation from the baseline in heritage speakers, Polinsky and Scontras argue that these triggers lead to an avoidance of ambiguity, a resistance to irregularity, and a shrinking of structure in the less dominant language. Although the empirical evidence they present is, to a large extent, convincing, they do not seem to address the question of how the basic characteristics of economy of language processing resources differ in monolinguals and heritage speakers: that is, if economizing in online resources is a basic trait of a human processor, (how) do heritage speakers differ from monolinguals in terms of what counts as a processing pressure and how it is resolved in two populations? Related to this is the question of whether or not the particularity of the heritage-language-dominant language dyad makes a difference in triggering the economy principle. In other words, what is not clear in Polinsky and Scontras's proposal is whether (and if so, how) a particular language pair as the heritage language and the dominant language comes into the equation in our estimate of the features that will undergo restructuring due to above-mentioned triggers.
On this note, it is important to emphasize that although seeking a unified language-independent model of heritage language development may be desirable, we cannot disregard the possibility that the mental representation and processing patterns of individual speakers may be tuned by the linguistic characteristics of the languages at their disposal (e.g., Frost & Grainger, Reference Frost and Grainger2000). Therefore, the operation of the proposed principles as the initiators of change/divergence (i.e., avoidance of ambiguity, resistance to irregularity, shrinking of structure) may appear only in certain language dyads, depending on the intricate interaction and interference between them.
Polinsky and Scontras (Polinsky & Scontras) will be a welcome addition to the literature on heritage language development. As an up-to-date and thorough synthesis of both theoretical and empirical research that explores the most vulnerable grammatical domains, it will definitely pave the way for a comprehensive model of heritage language grammar.
Before scrutinizing their proposal, it is important to note that, although Polinsky and Scontras acknowledge the problems in establishing a baseline for comparison for heritage speakers, they rightfully suggest that what makes the study of heritage language development interesting is mainly the (potential) differences between heritage language grammar and the so-called native grammar. It is true that heritage language grammar does not always differ from the baseline grammar, but the mere observation that sometimes it does, albeit selectively, is what has actually stimulated extensive research on heritage language acquisition over the years. In my view, it is praiseworthy that Polinsky and Scontras recognize this driving force behind heritage language research and do not refrain from comparing heritage speakers to a particular baseline. Furthermore, their linguistic approach to heritage language development leaves aside the unfruitful question of whether heritage speakers should be considered native speakers (cf. Rothman & Treffers-Daller, Reference Rothman and Treffers-Daller2014; see also Ortega, Reference Ortega2019 for further discussion) and instead focuses on predictions about what human language competence might look like under unbalanced bilingualism. Polinsky and Scontras's approach to heritage language data, I argue, is what will further heritage language research and enable us to identify its resilient and vulnerable properties that may otherwise go unnoticed. As is also made clear in this keynote article, a model of heritage language development cannot possibly be developed before establishing where heritage speakers converge on the baseline, alongside considering the phenomena where they do not.
Having said that, I will briefly highlight some of the insightful propositions put forth in the article and the aspects where they need refinement. First of all, Polinsky and Scontras posit transfer, attrition, and divergent attainment as three reasons why heritage languages differ from the baseline. Nevertheless, this typology encounters circularity because transfer from the dominant language into the heritage language is commonly treated as a manifestation of attrition, not a separate phenomenon (e.g., Gürel & Yılmaz, Reference Gürel and Yılmaz2011; Schmid, Köpke, Keijzer & Weilemar, Reference Schmid, Köpke, Keijzer and Weilemar2004). Furthermore, the assumption that heritage speakers demonstrate divergent linguistic representations because they receive input that differs both qualitatively and quantitatively from the one that monolingual learners receive inevitably implies that the major source of heritage language input (i.e., input from first-generation immigrant parents) is also divergent. It is true that the input heritage speakers receive is reduced in quantity. However, most work on adult bilingual immigrants shows relatively stable and qualitatively native-like first language (L1) competence in this group (e.g., Montrul, Reference Montrul2008; cf. Schmid & Köpke, Reference Schmid and Köpke2017), casting doubt on the broad generalization that the linguistic properties of the parents lead to changes in the grammar of heritage language speakers. If, indeed, the language of first-generation immigrant parents, who serve as the main input source for heritage language, remains intact but the grammar of child or adult heritage language speakers appears to be divergent, then factors other than the quality of the parental input should be questioned more closely (cf. Montrul, Reference Montrul2008).
Having proposed input conditions and the economy of online resources as two triggers for deviation from the baseline in heritage speakers, Polinsky and Scontras argue that these triggers lead to an avoidance of ambiguity, a resistance to irregularity, and a shrinking of structure in the less dominant language. Although the empirical evidence they present is, to a large extent, convincing, they do not seem to address the question of how the basic characteristics of economy of language processing resources differ in monolinguals and heritage speakers: that is, if economizing in online resources is a basic trait of a human processor, (how) do heritage speakers differ from monolinguals in terms of what counts as a processing pressure and how it is resolved in two populations? Related to this is the question of whether or not the particularity of the heritage-language-dominant language dyad makes a difference in triggering the economy principle. In other words, what is not clear in Polinsky and Scontras's proposal is whether (and if so, how) a particular language pair as the heritage language and the dominant language comes into the equation in our estimate of the features that will undergo restructuring due to above-mentioned triggers.
On this note, it is important to emphasize that although seeking a unified language-independent model of heritage language development may be desirable, we cannot disregard the possibility that the mental representation and processing patterns of individual speakers may be tuned by the linguistic characteristics of the languages at their disposal (e.g., Frost & Grainger, Reference Frost and Grainger2000). Therefore, the operation of the proposed principles as the initiators of change/divergence (i.e., avoidance of ambiguity, resistance to irregularity, shrinking of structure) may appear only in certain language dyads, depending on the intricate interaction and interference between them.
Author ORCIDs
Ayşe Gürel, 0000-0001-6162-0252.