BLC and Subordination in Heritage Speakers—Towards a New Research Agenda: Commentary on Hulstijn (2024)

Abstract

In his update on Basic Language Cognition (BLC), Hulstijn formulates a number of predictions derived from BLC Theory, and explains how BLC differs from Extended Language Cognition (ELC). BLC is used to refer to an individual’s capacity to process spoken language productively and receptively in everyday life, while ELC is defined as control of the written standard language, as taught in school. In the literature on heritage speakers, so far surprisingly little attention has been paid to the differences between BLC and ELC, despite the relevance of the distinction between oral and written language for our understanding of heritage speakers’ language profiles. In this commentary, I argue that BLC Theory can be used to inform studies of heritage languages, and conversely, how insights from heritage languages can be used to develop BLC Theory further. By way of example, I revisit some of the literature on subordination in Turkish as a heritage language. I also point to issues that need to be clarified and future directions in the study of these phenomena.

In his update on Basic Language Cognition (BLC), Hulstijn formulates a number of predictions derived from BLC theory and invites researchers to further test these predictions. As explained in the update, the term BLC is used to refer to an individual’s capacity to process spoken language productively and receptively in everyday life. Importantly, BLC is common to all adult native speakers who belong to a particular speech community. By contrast, Extended Language Cognition (ELC) is defined as “control of the written standard language, as taught in school” (Hulstijn, 2024, p. 11).

It seems to me that evidence obtained from studies into the language profiles of heritage speakers (HSs) is of key importance for testing the predictions of BLC theory. HSs are a specific group of native speakers (Rothman & Treffers-Daller, 2014) who grow up in contexts where their first language (L1) is not the majority language (e.g., speakers of Turkish as a heritage language living in Germany). Many HSs become dominant in their second language (L2) in the course of development (Daller et al., 2011). Because they attend school in which the majority language (e.g., German in Germany) is the language of instruction in schools, they generally develop literacy skills in their L2 to the same extent as their monolingual peers. However, HSs generally do not develop literacy skills in their L1 to the same extent, even when attending complementary schools in their L1.

The key question researchers working with HSs try to answer is which grammatical, phonological and lexical skills this group develop in L1 and L2, and to what extent their language profiles differ from those of monolingual speakers and L2 learners of each language. In addition, a key focus is on understanding what exactly the input to HSs consists of, and how the quantity and quality of input affect developmental outcomes.

In the literature on HSs, so far surprisingly little attention has been paid to the differences between BLC and ELC, despite the relevance of the distinction between oral and written language for our understanding of heritage HSs’ language profiles. This may in part be because in the different versions of BLC theory that I am familiar with, HSs are only briefly mentioned, but not discussed in detail. In footnote 4, Hulstijn notes that “typical adult native speakers attain BLC but heritage native speakers may not do so.” This then leads to the question under which conditions heritage speakers do or do not develop BLC. This small point serves to clarify that evidence from research into HSs could be used for developing BLC theory further, and conversely, predictions derived from BLC theory could be used to inform studies of HSs.

As I see it, the following predictions can be derived from BLC theory for HSs. I will only consider predictions P1 and P2 from Hulstijn (2024), because HSs are native speakers (Rothman & Treffers-Daller, 2014; Wiese et al., 2022), while P3 and P4 focus on non-native speakers.

• HSs will only develop BLC if they receive sufficient and high-quality oral input in their L1 during childhood while living in the host country.
• If the oral input is as specified under (1), with respect to BLC, HSs should achieve proficiency levels in their heritage language that are comparable to that of monolinguals.
• HSs will not acquire Extended Language Cognition (ELC) in their heritage language if they do not receive extensive literacy practice in it by attending education through the medium of their L1.

Clearly, these predictions can only be tested if a distinction can be made between structures that are typically found in oral language and those that are reserved for written language. This distinction would need to be made on the basis of two corpora: first of all, transcripts of a corpus of child-directed speech, and speech to older children and adolescents, reflecting the characteristics of oral input in the specific context in which they grow up, and second, a corpus of written language typically encountered in school in the respective communities.

Here, I would like to illustrate the issues by revisiting some of the literature on Turkish as a heritage language, with a focus on subordination (relative clauses, complement clauses and adverbial clauses), which have been studied extensively in the literature on Turkish grammar, and specifically with respect to heritage Turkish as spoken by HSs in different countries in Europe and the US (Bayram, 2013; Coşkun Kunduz & Montrul, 2024; Özsoy et al., 2022; Onar Valk & Backus, 2013; Şan, 2023; Treffers-Daller et al., 2007). Özsoy et al. show that HSs of Turkish in the US tend to use more finite subordinated clauses by comparison with HSs of Turkish in Germany. In many cases, participants replace non-finite subordinate clauses with paratactic constructions, particularly in informal situations. While there are subtle differences between the results of studies (France, Germany, the Netherlands, Sweden, the UK and the US), in all contexts under study, children who grow up as HSs of Turkish avoid using complex forms of subordination1. However, there are interesting differences between the various types of subordination. In some studies, adverbial subordinate clauses were used much more frequently than relative clauses and complement clauses (Treffers-Daller et al., 2007), and in later studies, adverbial constructions were found to be less vulnerable than other types of subordination among HSs too (Iefremenko et al., 2021). On the one hand, this could be explained on the basis of the fact that adverbial clauses are less complex in terms of their agreement patterns, but on the other hand, it could also be the case that adverbial clauses are more frequent in everyday speech than other types of subordinate clauses, and therefore learned more easily. To ascertain whether the latter explanation is accurate, more information would be needed about the presence of different types of subordination in speech (including child-directed speech) and writing.

For the purposes of the current summary, it may also be relevant to observe that most studies focus on production rather than reception. The few studies that did include a comprehension task (e.g., Coşkun Kunduz & Montrul, 2024) show that HSs were generally able to understand complex embeddings, even though they did not use them productively. Because even the oldest child HS was found to obtain lower scores on the tasks by comparison with age-matched monolingual counterparts, the authors conclude that relative clauses are not completely acquired by Turkish HSs in the US. One of the reasons for these between-group differences, according to the authors, is the absence of these constructions in the input. In addition, the Turkish reading skills of the Turkish HSs were much lower than those of their monolingual counterparts and lower than their reading skills in English. Children cannot be expected to acquire structures that are not in the input, at least not if one adopts an emergentist perspective on language learning2, nor can they learn complex embeddings that are only found in writing if their reading skills are underdeveloped. However, the contrasts between receptive and productive skills are puzzling and deserve to be investigated in more detail.

From the perspective of BLC theory, the question is whether relative clauses, complement clauses and adverbial clauses are part of BLC or if some subtypes of these are more likely to be part of it than other subtypes (e.g., subject relative clauses rather than object relative clauses). As far as I know, we do not currently have sufficient evidence about the frequency of different types of embeddings in speech and writing in Turkish in different media, nor do we know to what extent these are present in the input to heritage speakers to be able to answer this question. However, according to Dąbrowska (2012), there are considerable individual differences in ultimate attainment in the acquisition of complex syntax, in part dependent on educational levels and exposure to reading. Put differently, it is at least possible that some complex embeddings fall outside the purview of BLC and will only be acquired together with other characteristics of ELC as a result of extensive exposure to reading.

For studies of heritage speakers, it is of interest to consider BLC theory in more detail because it can be used to refine hypotheses about the development of heritage speakers’ grammars in that structures which are clearly ELC are unlikely to be acquired by HSs who are mainly exposed to oral language. To test such hypotheses, detailed information about reading practices over the course of development of HSs would be needed. In my view, this issue is not currently addressed in great detail in most studies of HSs.

It may also be of interest to establish to what extent vocabulary knowledge falls within BLC. There is new evidence, for example, about differences in vocabulary development between HSs and monolingual children (Kubota & Rothman, 2025), and about the vulnerability of English L2 verb–noun collocations in child returnees (Alraddadi et al., 2024). Quality and quantity of input in both languages are likely important variables in the consolidation of vocabulary knowledge in HSs and returnees. Again, the key question remains what exactly the input to HSs consists of, and what the role of literacy is in the development of vocabulary knowledge. Of course, it is also relevant to specify further what native speakers are before we can evaluate to what extent heritage speakers fall under this term. The extensive discussions about the meaning and usefulness of the term native speaker (Dewaele et al., 2021; Ortega, 2013) will need to be integrated into this discussion.

One anonymous reviewer notes that bilingualism is ultimately a gradient concept, that does not fit well with binary labels such as native – non-native, or having acquired BLC – yes or no. While I agree, testing strong claims is easier than a watered-down version of such claims. A key question is also whether or not there is a threshold for exposure that needs to be met if BLC is to be acquired. As there are now more sophisticated tools for studying language exposure (Arndt et al., 2023), it may well be possible in future to study threshold levels for exposure in more detail.

In summary, I hope to have shown there is much to be gained by further testing BLC theory against evidence from heritage speakers and returnees, and I hope many researchers will find it interesting to pursue this research agenda.