Domínguez, Hicks, and Slabakova (Reference Domínguez, Hicks and Slabakova2019) provide an in-depth analysis of the controversial term incomplete acquisition as it relates to heritage speakers from the perspective of generative linguistic theory. In their article, they defend the use of the term on theoretical grounds, yet advise caution with respect to use beyond the narrow theoretical domain for which it was originally intended. In other words, they make a distinction between theoretical precision (which, in their view, justifies the term) and impact on public discourse (where the negative connotations and misinterpretations can be damaging).
The problem of impact is a serious one. A cursory glance at how the word incomplete is used in everyday language points to some immediate pitfalls. An incomplete application will likely be rejected. In a university context, an incomplete (grade) is typically given to students who did not do the work required for the course. An incomplete sentence is one that needs to be fixed. In each of these contexts, we infer that what is “incomplete” needs to be resolved, revised, or repaired. Given this everyday usage, it is no surprise that the term might be misinterpreted when applied to the domain of heritage language acquisition. Interestingly, Schmid (2013) notes that the term attrition may also be problematic because of inherently negative collocations (e.g., war of attrition).
In what follows, I won’t focus on the issue of impact on public discourse because it has been addressed in significant detail by Domínguez et al. (Reference Domínguez, Hicks and Slabakova2019) as well as by Kupisch and Rothman (Reference Kupisch and Rothman2018). Instead, I will examine what incomplete acquisition might mean from a usage-based perspective. Specifically, I explore the possibility that for some heritage speakers, and in the context of reduced input in the heritage language, certain constructions may not become fully abstract but rather remain lexically specific (for a detailed discussion of the assumptions of usage-based approaches, see Lieven, Reference Lieven2016; Wulff and Ellis, Reference Wulff, Ellis, Miller, Bayram, Rothman and Serratrice2018). I elaborate on this point in the following text in my discussion of the Spanish subjunctive, but first, I address the issue of control groups and the timing of development.
Domínguez et al. argue in favor of research designs that include a control group, and in particular, defend the monolingual native speaker control: “The control group consists of adult native speakers not because of any inherent value we assign to the speakers themselves, but because their grammars are scientifically useful” (2019, p. 246). The question of who constitutes a valid control group for heritage speaker bilinguals continues to be a major issue in the field. In light of the critiques against monolingual controls, researchers are faced with several options. One is to dispense with a control group altogether and only study heritage speakers in their own right. The other option is to find a control group that may be more appropriate, such as bilingual controls living in a majority-language context (cf. Kupisch & Rothman, Reference Kupisch and Rothman2018), child bilinguals who are “future heritage speakers” (Polinsky, Reference Polinsky2018, p. 548) or adult native speakers who are first-generation immigrants (Montrul & Sánchez Walker, Reference Montrul and Sánchez-Walker2013). Nevertheless, monolingual controls are not always inappropriate provided that they share some extralinguistic characteristics of the heritage speaker group. Hulstijn (Reference Hulstijn2019) proposes defining “native speakers” in terms of shared/basic language cognition and two extralinguistic dimensions: the degree of being bilingual and literacy. He explicitly advocates for a research agenda that investigates differences in the language cognition among types of native speakers (including heritage speakers) as a “function of their membership along the two extralinguistic dimensions” (p. 163). Thus, a monolingual control group can be justified in heritage language research, but researchers must be cognizant of the characteristics of their participants—especially as they relate to formal schooling and literacy.
Another reason to include monolingual controls pertains to research with child heritage speakers and issues surrounding the timing of development (see Shin, 2018, for a comprehensive review). A usage-based approach assumes that acquisition depends on a critical mass of data of any given property of the language. For bilingual children, the reduction in input may result in initial delays in acquiring a particular structure, but with additional exposure, the differences between monolingual and bilingual children may eventually be neutralized. This hypothesis requires an experimental design in which bilingual heritage speakers of different ages are compared to monolingual children (see e.g., Flores & Barbosa, Reference Flores and Barbosa2014; Flores, Santos, Jesus, & Marques, Reference Flores, Santos, Jesus and Marques2017). Furthermore, rates of development depend on the type of structure under consideration; for simpler structures, the child heritage speaker may have been exposed to sufficient input even before starting school (when a drastic reduction in heritage language input tends to occur). However, for more complex structures, a much larger critical mass of input is needed to make the necessary generalizations. And this is precisely when the notion of “incomplete acquisition” comes into focus because, as noted by Gathercole (Reference Gathercole2007), “It may thus be the case that for some of these structures acquisition will be incomplete—‘timed off the map’, so to speak” (pp. 241–242).
The ideas presented in the preceding text about timing seem to align with those presented by Domínguez et al., who state, “[i]t is possible that since monolingual children up to 5 years old still have problems with these structures, heritage speakers may not have sufficient time to acquire them” (2019, p. 250). However, a key clarification is necessary. The more complex structures that are at risk for being “timed off the map” are typically acquired by monolingual children after—not before—age five. A good example of this is the subjunctive mood in Spanish, which is considered to be a late-acquired structure because monolingual children show a protracted period of development (Blake, Reference Blake1983). Consider the phrases in (1) as exemplars heard in the input.
(1) cuando salgas… / cuando sales... / cuando saliste… / cuando salías…
“when you leave…” / “when you leave…” / “when you left…” / “when you were leaving…”
The adverbial clauses in (1) are all possible in Spanish, yet there is a meaning difference encoded by the form of the verb. Only the first clause (cuando salgas) makes reference to a future event of leaving, one that might be followed by a command such as cierra la puerta con llave “lock the door” (Kanwit & Geeslin, Reference Kanwit and Geeslin2014). Children growing up in a Spanish-speaking home will hear all of the constructions in (1) with a variety of verbs, and their task is to map the forms to a meaning. Initially, this process begins with rote-learned strings that appear frequently in the input (cuando seas grande “when you grow up”) and low-scope schemas of cuando followed by a limited inventory of verbs. The more abstract construction (Adv Vsubjunctive) develops only gradually and is dependent on continued exposure to input. Do heritage speakers of Spanish receive enough input to build up the abstract construction? If the input conditions are insufficient, the resulting grammar may end up being lexically specific rather than fully abstract; this means that speakers’ use of the subjunctive in adverbial clauses would appear with only certain verbs.
Another potential result of insufficient input is that the grammar will not be able to retreat from overgeneralizations. I have studied this with respect to derivational morphology in Spanish, an area of language that continues to develop throughout elementary and middle school. In the heritage speaker context, where there is little or no opportunity for formal schooling in Spanish, speakers may create new words by applying their knowledge of derivational suffixes. For example, heritage speakers may produce words such as amarguez (bitterness) and profundez (depth) in contrast to the conventional forms that can be found in the dictionary: amargura and profundidad. The results of an acceptability judgment study (Reference ZyzikZyzik, 2019; Zyzik & Sanchez, Reference Zyzik and Sanchez2019) indicate that heritage speakers were significantly more likely to accept such creative words than monolingual Spanish speakers. In this sense, the heritage speaker grammar is overly productive, allowing heritage speakers to produce linguistic innovations that are considered unacceptable by others. This is important because Domínguez et al. (Reference Domínguez, Hicks and Slabakova2019) emphasize that “incomplete acquisition” is one of four possible outcomes in the heritage speaker context (the other three outcomes include successful acquisition based on monolingual-type input, successful acquisition based on divergent input, and attrition or reanalysis of previously acquired features). We should consider adding innovation to the list of possible outcomes in heritage language acquisition.
One positive outcome of the debate surrounding “incomplete acquisition” is an increased awareness of the significant amount of time it takes to acquire the full spectrum of constructions used by adult native speakers. Understanding acquisition as a slow and piecemeal process has always been a central part of usage-based theory. What linguists see as abstract linguistic representations do not start out that way; they are initially item based and develop only gradually. In a reduced input environment, a particular construction may remain at the lexically specific level rather than becoming (more) abstract. This is one way of understanding “incomplete acquisition” from the perspective of usage-based theory; another complementary idea is that particular structures might be “timed off the map.” Finally, given the narrow scope of the term incomplete acquisition, we might be missing other equally interesting phenomena (e.g., innovation) in the study of heritage languages.
You have
Access
Domínguez, Hicks, and Slabakova (Reference Domínguez, Hicks and Slabakova2019) provide an in-depth analysis of the controversial term incomplete acquisition as it relates to heritage speakers from the perspective of generative linguistic theory. In their article, they defend the use of the term on theoretical grounds, yet advise caution with respect to use beyond the narrow theoretical domain for which it was originally intended. In other words, they make a distinction between theoretical precision (which, in their view, justifies the term) and impact on public discourse (where the negative connotations and misinterpretations can be damaging).
The problem of impact is a serious one. A cursory glance at how the word incomplete is used in everyday language points to some immediate pitfalls. An incomplete application will likely be rejected. In a university context, an incomplete (grade) is typically given to students who did not do the work required for the course. An incomplete sentence is one that needs to be fixed. In each of these contexts, we infer that what is “incomplete” needs to be resolved, revised, or repaired. Given this everyday usage, it is no surprise that the term might be misinterpreted when applied to the domain of heritage language acquisition. Interestingly, Schmid (2013) notes that the term attrition may also be problematic because of inherently negative collocations (e.g., war of attrition).
In what follows, I won’t focus on the issue of impact on public discourse because it has been addressed in significant detail by Domínguez et al. (Reference Domínguez, Hicks and Slabakova2019) as well as by Kupisch and Rothman (Reference Kupisch and Rothman2018). Instead, I will examine what incomplete acquisition might mean from a usage-based perspective. Specifically, I explore the possibility that for some heritage speakers, and in the context of reduced input in the heritage language, certain constructions may not become fully abstract but rather remain lexically specific (for a detailed discussion of the assumptions of usage-based approaches, see Lieven, Reference Lieven2016; Wulff and Ellis, Reference Wulff, Ellis, Miller, Bayram, Rothman and Serratrice2018). I elaborate on this point in the following text in my discussion of the Spanish subjunctive, but first, I address the issue of control groups and the timing of development.
Domínguez et al. argue in favor of research designs that include a control group, and in particular, defend the monolingual native speaker control: “The control group consists of adult native speakers not because of any inherent value we assign to the speakers themselves, but because their grammars are scientifically useful” (2019, p. 246). The question of who constitutes a valid control group for heritage speaker bilinguals continues to be a major issue in the field. In light of the critiques against monolingual controls, researchers are faced with several options. One is to dispense with a control group altogether and only study heritage speakers in their own right. The other option is to find a control group that may be more appropriate, such as bilingual controls living in a majority-language context (cf. Kupisch & Rothman, Reference Kupisch and Rothman2018), child bilinguals who are “future heritage speakers” (Polinsky, Reference Polinsky2018, p. 548) or adult native speakers who are first-generation immigrants (Montrul & Sánchez Walker, Reference Montrul and Sánchez-Walker2013). Nevertheless, monolingual controls are not always inappropriate provided that they share some extralinguistic characteristics of the heritage speaker group. Hulstijn (Reference Hulstijn2019) proposes defining “native speakers” in terms of shared/basic language cognition and two extralinguistic dimensions: the degree of being bilingual and literacy. He explicitly advocates for a research agenda that investigates differences in the language cognition among types of native speakers (including heritage speakers) as a “function of their membership along the two extralinguistic dimensions” (p. 163). Thus, a monolingual control group can be justified in heritage language research, but researchers must be cognizant of the characteristics of their participants—especially as they relate to formal schooling and literacy.
Another reason to include monolingual controls pertains to research with child heritage speakers and issues surrounding the timing of development (see Shin, 2018, for a comprehensive review). A usage-based approach assumes that acquisition depends on a critical mass of data of any given property of the language. For bilingual children, the reduction in input may result in initial delays in acquiring a particular structure, but with additional exposure, the differences between monolingual and bilingual children may eventually be neutralized. This hypothesis requires an experimental design in which bilingual heritage speakers of different ages are compared to monolingual children (see e.g., Flores & Barbosa, Reference Flores and Barbosa2014; Flores, Santos, Jesus, & Marques, Reference Flores, Santos, Jesus and Marques2017). Furthermore, rates of development depend on the type of structure under consideration; for simpler structures, the child heritage speaker may have been exposed to sufficient input even before starting school (when a drastic reduction in heritage language input tends to occur). However, for more complex structures, a much larger critical mass of input is needed to make the necessary generalizations. And this is precisely when the notion of “incomplete acquisition” comes into focus because, as noted by Gathercole (Reference Gathercole2007), “It may thus be the case that for some of these structures acquisition will be incomplete—‘timed off the map’, so to speak” (pp. 241–242).
The ideas presented in the preceding text about timing seem to align with those presented by Domínguez et al., who state, “[i]t is possible that since monolingual children up to 5 years old still have problems with these structures, heritage speakers may not have sufficient time to acquire them” (2019, p. 250). However, a key clarification is necessary. The more complex structures that are at risk for being “timed off the map” are typically acquired by monolingual children after—not before—age five. A good example of this is the subjunctive mood in Spanish, which is considered to be a late-acquired structure because monolingual children show a protracted period of development (Blake, Reference Blake1983). Consider the phrases in (1) as exemplars heard in the input.
(1) cuando salgas… / cuando sales... / cuando saliste… / cuando salías…
“when you leave…” / “when you leave…” / “when you left…” / “when you were leaving…”
The adverbial clauses in (1) are all possible in Spanish, yet there is a meaning difference encoded by the form of the verb. Only the first clause (cuando salgas) makes reference to a future event of leaving, one that might be followed by a command such as cierra la puerta con llave “lock the door” (Kanwit & Geeslin, Reference Kanwit and Geeslin2014). Children growing up in a Spanish-speaking home will hear all of the constructions in (1) with a variety of verbs, and their task is to map the forms to a meaning. Initially, this process begins with rote-learned strings that appear frequently in the input (cuando seas grande “when you grow up”) and low-scope schemas of cuando followed by a limited inventory of verbs. The more abstract construction (Adv Vsubjunctive) develops only gradually and is dependent on continued exposure to input. Do heritage speakers of Spanish receive enough input to build up the abstract construction? If the input conditions are insufficient, the resulting grammar may end up being lexically specific rather than fully abstract; this means that speakers’ use of the subjunctive in adverbial clauses would appear with only certain verbs.
Another potential result of insufficient input is that the grammar will not be able to retreat from overgeneralizations. I have studied this with respect to derivational morphology in Spanish, an area of language that continues to develop throughout elementary and middle school. In the heritage speaker context, where there is little or no opportunity for formal schooling in Spanish, speakers may create new words by applying their knowledge of derivational suffixes. For example, heritage speakers may produce words such as amarguez (bitterness) and profundez (depth) in contrast to the conventional forms that can be found in the dictionary: amargura and profundidad. The results of an acceptability judgment study (Reference ZyzikZyzik, 2019; Zyzik & Sanchez, Reference Zyzik and Sanchez2019) indicate that heritage speakers were significantly more likely to accept such creative words than monolingual Spanish speakers. In this sense, the heritage speaker grammar is overly productive, allowing heritage speakers to produce linguistic innovations that are considered unacceptable by others. This is important because Domínguez et al. (Reference Domínguez, Hicks and Slabakova2019) emphasize that “incomplete acquisition” is one of four possible outcomes in the heritage speaker context (the other three outcomes include successful acquisition based on monolingual-type input, successful acquisition based on divergent input, and attrition or reanalysis of previously acquired features). We should consider adding innovation to the list of possible outcomes in heritage language acquisition.
One positive outcome of the debate surrounding “incomplete acquisition” is an increased awareness of the significant amount of time it takes to acquire the full spectrum of constructions used by adult native speakers. Understanding acquisition as a slow and piecemeal process has always been a central part of usage-based theory. What linguists see as abstract linguistic representations do not start out that way; they are initially item based and develop only gradually. In a reduced input environment, a particular construction may remain at the lexically specific level rather than becoming (more) abstract. This is one way of understanding “incomplete acquisition” from the perspective of usage-based theory; another complementary idea is that particular structures might be “timed off the map.” Finally, given the narrow scope of the term incomplete acquisition, we might be missing other equally interesting phenomena (e.g., innovation) in the study of heritage languages.