Polinsky and Scontras (Polinsky & Scontras, Reference Polinsky and Scontras2019) convincingly demonstrate that linguistic theorizing and research on bilingualism can mutually benefit from each other. We agree that just as linguistic theory contributes to inform research on heritage languages, so too can research on heritage languages either support or refute linguistic theories. The study of heritage languages represents a particular challenge for linguists, because heritage speakers’ grammars are native grammars that, nevertheless, may differ in several ways from the grammatical knowledge of monolingual native speakers of the same language. We want to highlight, however, that divergence is a possible, but not a necessary, outcome of heritage language development (Putnam, Kupisch & Pascual y Cabo, Reference Putnam, Kupisch, Pascual y Cabo, Bayram, Denhovska, Miller, Rothman and Serratrice2018). In fact, many studies report on (sub)groups of heritage speakers that do not differ from monolingual peers (Łyskawa & Nagy, Reference Łyskawa and Nagy2019). General statements like “the heritage language differs from the baseline language that served as the input for acquisition” should be formulated with caution. Importantly, variability and variation in heritage language grammars cannot be equated with deviance.
A number of different factors have been discussed as potential sources for differences between heritage languages and monolingual grammars. Polinsky and Scontras point out three culprits: transfer, attrition and divergent attainment as a result of differential input. They discuss two triggers of divergent outcomes, namely (quantitatively and qualitatively) reduced input and processing demands which may lead to avoidance of ambiguity, resistance to irregularity and shrinking of structure. In particular, infrequent grammatical properties that impose relatively high cognitive demands are affected.
We want to highlight a fourth factor that should be taken into consideration, when it comes to predicting and accounting for differences between heritage bilinguals and a monolingual baseline and which is often ignored in heritage language studies, namely language-internal phenomena of variation and variability. We argue that heritage speakers may boost and further develop tendencies of language (internal) evolution inherent to variable phenomena. Furthermore, they may also retain variation that is typical of non-standard registers, because they are mainly exposed to colloquial input.
The first interesting case in point relates to what Polinsky and Scontras call the “Silent Problem”. The authors argue that heritage speakers have difficulties with empty categories, leading to an overuse of overt subject pronouns and a reanalysis of interpretative possibilities of null and overt subjects. In the domain of objects, however, the “Silent Problem” may have different consequences, as shown in our recent work on object realization in heritage Portuguese in Germany. In a corpus study including first and second generation migrants and two age-matched generations of monolingual speakers, we demonstrate that heritage speakers (i.e., second generation migrants) tend to produce more null object constructions instead of fewer (Rinke, Flores & Barbosa, Reference Rinke, Flores and Barbosa2018). In particular, they omit more animate objects, which mirrors development that occurred also in other varieties of Portuguese. In the diachronic evolution of Brazilian Portuguese, for instance, null objects have expanded along a referential hierarchy from less to more referential (and animate) arguments (Cyrino, Duarte & Kato, Reference Cyrino, Duarte, Kato, Kato and Negrão2000), whereas null subjects undergo the opposite development: they get first lost in more referential contexts and later in less referential contexts (e.g., with expletive arguments). This shows that bilingual speakers do not have a particular problem with silent elements. Instead, they further develop the variable phenomenon according to a diachronic path, which can be predicted on the basis of a universal referential hierarchy. This evolution may be triggered by reduced input, but importantly it is a language-internal phenomenon.
A further case in point is related to variation that is typical of colloquial language registers, which is preserved and even amplified in heritage speakers’ speech because they are predominantly exposed to colloquial registers (within the family) and have reduced access to formal instruction in the heritage language and to standard registers. One example is the availability of clitic climbing in European Portuguese, as shown by Rinke and Flores (Reference Rinke and Flores2014). With some auxiliaries (e.g., poder/’can’), the clitic pronoun may freely attach either to the non-finite verb (posso fazê-lo /‘(I) can do it’) or to the finite auxiliary (posso-o fazer/‘(I)can-it do’, clitic climbing). Clitic climbing is frequent in colloquial European Portuguese (Barbosa, Paiva & Martins, Reference Barbosa, Paiva, Martins, Barbosa, Paiva and Rodrigues2017): however, it is less used in more formal speech and tends to be rejected by educated monolinguals. Heritage speakers prefer the option that is typical of colloquial speech, precisely because their linguistic competence is less affected by standardization processes. In this case, heritage speakers may differ only from educated monolinguals, but not from non-educated speakers who, like themselves, are mainly exposed to non-standard registers. Hence, predominant reliance on colloquial input does not necessarily lead to a more reduced system; on the contrary, it may contribute to the maintenance of variation, which generally is pinned down by formal education and language standardization.
This also leads to the question of the correct baseline for comparison. Heritage languages should be compared not only to standard, but also to vernacular non-contact varieties. A predictive model for heritage language development has to be able to account for language-internal, variable phenomena and to distinguish variation from deviation.
You have
Access
Polinsky and Scontras (Polinsky & Scontras, Reference Polinsky and Scontras2019) convincingly demonstrate that linguistic theorizing and research on bilingualism can mutually benefit from each other. We agree that just as linguistic theory contributes to inform research on heritage languages, so too can research on heritage languages either support or refute linguistic theories. The study of heritage languages represents a particular challenge for linguists, because heritage speakers’ grammars are native grammars that, nevertheless, may differ in several ways from the grammatical knowledge of monolingual native speakers of the same language. We want to highlight, however, that divergence is a possible, but not a necessary, outcome of heritage language development (Putnam, Kupisch & Pascual y Cabo, Reference Putnam, Kupisch, Pascual y Cabo, Bayram, Denhovska, Miller, Rothman and Serratrice2018). In fact, many studies report on (sub)groups of heritage speakers that do not differ from monolingual peers (Łyskawa & Nagy, Reference Łyskawa and Nagy2019). General statements like “the heritage language differs from the baseline language that served as the input for acquisition” should be formulated with caution. Importantly, variability and variation in heritage language grammars cannot be equated with deviance.
A number of different factors have been discussed as potential sources for differences between heritage languages and monolingual grammars. Polinsky and Scontras point out three culprits: transfer, attrition and divergent attainment as a result of differential input. They discuss two triggers of divergent outcomes, namely (quantitatively and qualitatively) reduced input and processing demands which may lead to avoidance of ambiguity, resistance to irregularity and shrinking of structure. In particular, infrequent grammatical properties that impose relatively high cognitive demands are affected.
We want to highlight a fourth factor that should be taken into consideration, when it comes to predicting and accounting for differences between heritage bilinguals and a monolingual baseline and which is often ignored in heritage language studies, namely language-internal phenomena of variation and variability. We argue that heritage speakers may boost and further develop tendencies of language (internal) evolution inherent to variable phenomena. Furthermore, they may also retain variation that is typical of non-standard registers, because they are mainly exposed to colloquial input.
The first interesting case in point relates to what Polinsky and Scontras call the “Silent Problem”. The authors argue that heritage speakers have difficulties with empty categories, leading to an overuse of overt subject pronouns and a reanalysis of interpretative possibilities of null and overt subjects. In the domain of objects, however, the “Silent Problem” may have different consequences, as shown in our recent work on object realization in heritage Portuguese in Germany. In a corpus study including first and second generation migrants and two age-matched generations of monolingual speakers, we demonstrate that heritage speakers (i.e., second generation migrants) tend to produce more null object constructions instead of fewer (Rinke, Flores & Barbosa, Reference Rinke, Flores and Barbosa2018). In particular, they omit more animate objects, which mirrors development that occurred also in other varieties of Portuguese. In the diachronic evolution of Brazilian Portuguese, for instance, null objects have expanded along a referential hierarchy from less to more referential (and animate) arguments (Cyrino, Duarte & Kato, Reference Cyrino, Duarte, Kato, Kato and Negrão2000), whereas null subjects undergo the opposite development: they get first lost in more referential contexts and later in less referential contexts (e.g., with expletive arguments). This shows that bilingual speakers do not have a particular problem with silent elements. Instead, they further develop the variable phenomenon according to a diachronic path, which can be predicted on the basis of a universal referential hierarchy. This evolution may be triggered by reduced input, but importantly it is a language-internal phenomenon.
A further case in point is related to variation that is typical of colloquial language registers, which is preserved and even amplified in heritage speakers’ speech because they are predominantly exposed to colloquial registers (within the family) and have reduced access to formal instruction in the heritage language and to standard registers. One example is the availability of clitic climbing in European Portuguese, as shown by Rinke and Flores (Reference Rinke and Flores2014). With some auxiliaries (e.g., poder/’can’), the clitic pronoun may freely attach either to the non-finite verb (posso fazê-lo /‘(I) can do it’) or to the finite auxiliary (posso-o fazer/‘(I)can-it do’, clitic climbing). Clitic climbing is frequent in colloquial European Portuguese (Barbosa, Paiva & Martins, Reference Barbosa, Paiva, Martins, Barbosa, Paiva and Rodrigues2017): however, it is less used in more formal speech and tends to be rejected by educated monolinguals. Heritage speakers prefer the option that is typical of colloquial speech, precisely because their linguistic competence is less affected by standardization processes. In this case, heritage speakers may differ only from educated monolinguals, but not from non-educated speakers who, like themselves, are mainly exposed to non-standard registers. Hence, predominant reliance on colloquial input does not necessarily lead to a more reduced system; on the contrary, it may contribute to the maintenance of variation, which generally is pinned down by formal education and language standardization.
This also leads to the question of the correct baseline for comparison. Heritage languages should be compared not only to standard, but also to vernacular non-contact varieties. A predictive model for heritage language development has to be able to account for language-internal, variable phenomena and to distinguish variation from deviation.
Author ORCIDs
Cristina Flores, 0000-0001-5629-9556