Response to Open Peer Commentaries of Hulstijn’s (2024) Update of BLC Theory

Abstract

In this article, I respond to six peer commentaries of my essay in Languages 2024, containing an update of BLC Theory. I address the following issues: embedding BLC Theory in more than one metatheory; the match between the theory and usage-based models; the importance of processing pressures; the distinction between BLC and ELC; the study of heritage languages; the roles of cognitive factors in L1 and L2 attainment.

1. Introduction

To be involved in scientific inquiry (in a free academic world: a huge privilege!) implies listening to others and critically examining one’s own views. Scholarly dialogue—particularly enjoyable during live encounters at workshops and conferences—is crucial for one’s own intellectual journey and for the advancement of the field. I was therefore thrilled when the chief editor of Languages, Jeanine Treffers-Daller, told me that she would like to see the article in which I presented an update of BLC Theory (Hulstijn, 2024) be reflected upon in several commentaries. Six commentaries were written by, in alphabetical order, Granena (2025), O’Grady (2025), Ortega and Wu (2025), Rothman et al. (2025), Schmid (2025), and Treffers-Daller (2025).

I am grateful to all authors for having written a commentary in a constructive spirit. Given a word limit of 3000 words for this response, I must restrict myself to matters which I believe should be interesting for readers1.

Before I respond to the commentaries, let me briefly remind readers of the origin of BLC Theory. As I argued in Chapter 2 of Hulstijn (2015), the first task of research on L1 acquisition, L2 acquisition, and multilingualism is to resolve the miraculous puzzle of language acquisition in the human species. In that chapter, I juxtaposed the generative and usage-based solutions to this puzzle. I proposed BLC Theory to make more concrete (for empirical research) what it might be that makes L1 acquisition universally successful, as claimed by generativists. This goal led me to propose the notion of Basic Language Cognition (BLC). A theory of the human language-learning ability must give an account of what people in reality say (usage) in everyday oral communication, not of what Chomsky’s ideal native speaker could say. We must then ascertain the necessary conditions and the typical conditions (that apply typically but not necessarily) in the acquisition of this shared oral language: input quantity and quality, variability of speakers and contexts (home, wider community, kindergarten, elementary school), distribution of input over time, from birth into adulthood. See Hulstijn (2015, pp. 15–16) for a more elaborate argumentation.

This was, and still is, the origin of BLC Theory. I formulated hypotheses about the domains in which we can expect L1- and L2-learners to differ. I deemed it best to withhold my own stance in the 2015 book, even though I had earlier expressed my inclinations. By the end of the 1990s, having read (i) the ground-breaking studies on parallel-distributed processing conducted by Rumelhart and McClelland (1986), (ii) Bates and MacWhinney’s (1989) inspiring book about their Competition Model, and (iii) the landmark paper of Landauer and Dumais (1997), boldly entitled “A solution to Plato’s problem”2, I was convinced that, what was then called ‘connectionism’ showed a more promising path to the explanation of the miracle than the path of generative linguistics (see e.g., Hulstijn, 2002). I motivated this conviction more extensively in my 2019 article in Language Learning (Hulstijn, 2019, pp. 170–174). This brings me straight to the following question.

2. Can BLC Theory Be Neutral Between the Generative and the Usage-Based/Complex System Metatheories?

Rothman et al. (2025) ask (p. 3): “why is/must the BLC be predicated on a particular (…) understanding of how language is acquired/learned? In other words, why is BLC Theory framed exclusively from a usage-based perspective of language acquisition/cognition?” I acknowledge that, indeed, one can try to work with BLC Theory from a generative perspective (Hulstijn, 2015, pp. 55–56). I have no grounds to object, when anyone uses BLC Theory from the viewpoint of a generative framework (metatheory). Perhaps Rothman cum suis could make a chart informed by a generative theory, as a counterpart of the chart on page 4 of Schmid’s commentary. What I would definitely not recommend is that researchers engage in testing BLC Theory’s hypotheses in a framework-free space. Researchers must make clear what their metatheoretical stance is and show how this choice has motivated or influenced their research questions and their method. [A different scenario enfolds when researchers conduct a study specifically designed to test conflicting theories (e.g., a generative versus a usage-based metatheory). But such a study would still be theory-guided. To call such a study ‘theory-neutral’ would be confusing, if not misleading.]

3. BLC Theory Embedded in (or Under) Usage-Based Linguistics

In his commentary, Schmid (2025) reframes the definitions of BLC and ELC “in a distinctly usage-based way” (p. 2), using his Entrenchment-and-Conventionalization Model. This double-feedback model consists of three elements: usage, conventions (that are highly usualized and widely diffused), and cognition (in individuals). I am delighted to see how well this “exercise” (p. 2) succeeded (see p. 4) and to read that “the points of convergence (…) definitely invite a more intensive exchange of ideas in the future” (p. 6).

4. The Role of Processing Pressures

In his commentary, O’Grady (2025) provides us, as in earlier work, with a nice example of the role of what he calls processing pressures that play a role in shaping language and learning, in addition to frequency and usage. O’Grady shows how so-called negative concord (as in I didn’t do nothing), involving a double (redundant) marking of negation, can result from processing factors. The case of negative concord is a fine illustration, I believe, of a fascinating paradox in all human languages, characteristic of complex adaptive systems. Their vocabularies and grammars cannot solely be transparent. They must also contain opaqueness, fuzziness, redundancy, ambiguity, homophony and syncretism. The meaning of many words can vary depending on their verbal and nonverbal context. This ‘ambiguity’ does not exist in industrial terminology regulated by the International Organization for Standardization. In grammars of natural languages, transparent form–function mappings are rare, in contrast to the transparency in languages invented for computer programming. Opaqueness results from two forces: the “damping of form complexity” and “damping processing complexity” (Steels & Beuls, 2017, p. 32). On the one hand, there is a pressure to compress information, which often leads to ambiguity. However, to combat an overdose of ambiguity, people invent new, explicit expressions which can make a language more structurally complex (e.g., redundant subject–verb agreement, redundant double marking of plurality on nouns, or negative concord). The never-ending dynamics of these opposite forces are characteristic of complex adaptive systems.

5. The Distinction Between BLC and ELC

All commentaries appear to accept the distinction between BLC and ELC as a useful guide for research on individual differences (or the lack thereof) in the acquisition of native and non-native languages. At the same time—not surprising to me—some authors raise concerns with respect to the exact definitions of BLC and ELC. In their (favorable) commentary, Ortega and Wu (2025) raise a few critical issues “in a Popperian spirit” (p. 1). They wonder whether the separation between oral and literate language use is “too dichotomous to be workable empirically” (p. 3); “frequent use of texting and posting on social media, growing reliance on AI (…) can all influence what constitutes BLC and where the distinctions should be between BLC and ELC” (p. 4). Schmid (2025), comparing the earlier definition of ELC, including some forms of oral language use (Hulstijn, 2015), with the later definition, which limits ELC to “the control of the written standard language, as taught in school” (Hulstijn, 2024, p. 4), remarks that he is “not sure whether this revised definition is really an improvement” (p. 2).

I have been struggling with the definitions of BLC and ELC for many years. Surely, there are forms of oral language that resemble or reflect written language, such as oral literature, songs, and rituals. Similarly, there are forms of written language that resemble or reflect oral language, such as texting on social media. My main motivation for strategically restricting ELC to written language use was the purpose of my 2024 essay, namely to propose that the study of individual differences (IDs) might be more fruitful when we separate the investigation of the use and attainment of oral language (the modality of all 7000 languages of the world) from investigations of the use and attainment of written standard languages (less than 200 in number), rather than studying IDs from the premise of a single language-proficiency scale. It is my (admittedly anecdotal, subjective) impression that (even) children, whose oral home language is close to the standard language taught in school (a minority of all children!), have to ‘discover’ this language ‘again’, as it were, realizing that their language can be rendered by means of a script, and that this is carried out with ‘sentences’, obeying conventions that do not necessarily apply to the ‘utterances’ of spoken discourse.

With respect to the term “standard language”, Ortega and Wu “wonder if invoking standard language in ELC is conceptually necessary or profitable. It clashes with mounting critiques against negative ideologies that penalize minoritized speakers of vernacular varieties as lacking in linguistic competence, simply because they do not use the variety arbitrarily chosen by the powerful as correct or appropriate” (p. 3). Arguably, BLC Theory does not use the term “written standard language, as taught in school” (Hulstijn, 2024, p. 4) with the purpose of giving a penalizing stick to people who want to discriminate students speaking a home language distinct of the written standard language. But it is a fact that written standard languages, taught in school, obey certain conventions and that these conventions must be learned. BLC Theory predicts that being clever gives students an advantage, not only in learning the writing system but also in learning these conventions. Evidently, the misuse of scientific terms (in this case, the possibility of this happening to standard language) outside the academic community, disregarding their original context, must be condemned. For a discussion of Bernstein’s much-criticized term elaborated code and Cummins’ cognitive-academic language proficiency (CALP), see Hulstijn (2015, pp. 29–32) and the references there.

6. The Study of Child Bilingualism in Heritage Speakers

At the bottom of page 4 of their commentary, Rothman et al. (2025) point to a possible contradiction in my use of the expression “typical adult native speakers” in combination with the assertation that heritage speakers “may or may not” attain BLC. In a long passage, they then describe scenarios of unwanted consequences of my definition of the term typical adult native speaker.

In Hulstijn (2015, pp. 28, 53), I wrote that it is an empirical matter at which age native speakers have acquired BLC and whether BLC age differs between societies and languages. I still stand by these propositions. The purpose of including “adult” in the definition of native speakers (Hulstijn, 2024, p. 3) was to make researchers aware of the possibility that acquisition of a shared morphosyntax may not be completed at age 6, or even at age 10 or 14. This gives room for my contention that HSs “may or may not attain BLC”, depending on the quality and quantity of the input they receive. BLC Theory defines the construct of BLC not with the purpose of setting a normative standard for who shall be called a native speaker, but as the object of inquiry for those who want to resolve the puzzle of language acquisition. I was keen to avoid the use of the terms dominance and monolingualism (as a yardstick). Note that a “typical adult native speaker” (of any one of the 7000 languages still spoken in the world) is not likely to be someone with a diploma in higher education. Thus, BLC Theory clearly rejects ideologies of normative, monolingual interpretations of the terms native speaker and typical adult native speaker.

In my essay (p. 3), I wrote that theories, constructs, and construct definitions are tools. It is totally acceptable to play with definitions in an attempt to obtain a better understanding of the miraculous puzzle of human language acquisition. If other researchers want to work with another definition of BLC, I can only encourage them to do so. In short, I do not see a contradiction and I trust that Rothman and his co-authors and I are on the same page concerning the goals of studying child bilingualism.

In her commentary, Treffers-Daller (2025) provides readers with fruitful ideas for the study of language acquisition in heritage speakers (HSs), thereby presenting research questions that are inspired (to a larger or smaller extent) by the research agenda of BLC Theory. Treffers-Daller argues that corpora are needed to answer the key question of “what exactly the input to HSs consists of, and what the role of literacy is in the development of vocabulary” (p. 3). For instance, when HSs of Turkish in the U.S. do not use certain types of relative clauses, we need to know whether such constructions are part of their input. “Children cannot be expected to acquire structures that are not in the input” (p. 3). I welcome Treffers-Daller’s call for corpus studies, which resonates with a similar call in my essay.

7. The Study of Cognitive Factors

In Section 2 of her commentary, Granena (2025) reviews several studies showing evidence of individual differences (IDs) in the performance of native speakers (NSs) in speech-processing tasks. These studies gave evidence of an association between word predictability and implicit learning (even after controlling for vocabulary, intelligence, and short-term memory). “This suggest that NSs may share the elements and constructions that occur frequently in the language but may not have the same ability to fluently process them in listening and speaking” (p. 2). Granena is right. However, note that the prediction is not that all NSs process speech equally fluently (or that NSs are identical in neurophysiological terms), but that they process speech “correctly and quickly (…) under normal acoustical conditions” (Hulstijn, 2015, p. 25)3. There is robust evidence for a decrease in information-processing speed in old age but BLC Theory claims that “the vast majority of older people remain capable of processing linguistic information fast enough to allow for relatively unimpaired functional language use, provided that they continue to practice their language skills on a daily basis” (ibidem).

At the bottom of p. 2, Granena is surprised that “BLC Theory does not predict differences caused by cognitive factors”. I acknowledge that this statement was incomplete and therefore conveyed a reading not intended. I wrote (Hulstijn, 2024, p. 6) “BLC Theory does not predict differences caused by cognitive factors in neurotypical populations of non-native speakers.” What I should have written is that moderate to low scores on cognitive-ability tests (statistical-learning ability, intelligence, executive functions, working-memory) do not stand in the way of non-native speakers acquiring BLC in a L2 (i.e., being able to process oral everyday language correctly and quickly)—with the exception of pronunciation and “the production of some grammatical features in spontaneous, unmonitored speech” (Hulstijn, 2024, p. 6).

Granena mentions evidence from her own research showing a link between IDs in implicit-learning ability “and L2 learning”, in L2 learners (top of p. 3). I acknowledge that BLC Theory needs more detail “to consider or, at least, to discuss some potentially relevant cognitive abilities that have been shown to explain variability in L2 learning in the SLA literature” (Granena, 2025, p. 3).

In Section 3, Granena reviews studies on the roles of several cognitive IDs in reading and writing. There is no room here for a detailed discussion of these findings, but studies like those conducted by Granena are needed for a further specification of my hypotheses. I therefore agree with Granena’s conclusion that BLC Theory “is lacking (…) a more precise definition and operationalization of ‘cognitive factors’” (p. 4).

The study of cognitive factors as predictor variables (e.g., foreign-language learning aptitude, statistical-learning ability, intelligence, executive functions, working memory) in language use and language acquisition is fraught with methodological difficulties see (Hulstijn, 2015, pp. 87–90) and the literature referenced there). The challenge is to show (i) support for the validity of the task (often lasting less than, say, 30 min) testing a presumed cognitive factor (the underlying trait) and (ii) a reliable association with individual differences in language use or language acquisition in real life (dependent variables within the domains of BLC vs. ELC)4.

8. Conclusions

As I mentioned in the Introduction, our first errand is to explain the miracle of language evolution and human language acquisition. Two main explanations exist, proposed by the generative school on the one hand, and the school of usage-based linguistics and neural-network science on the other hand. The main proposal of BLC Theory is that it is BLC, not ELC, that metatheories must focus on in explaining these miracles. I regard the predictions about the absence or presence of individual differences (Hulstijn, 2015, pp. 52–53; Hulstijn, 2024, pp. 5–6) as corollaries of the BLC-ELC distinction, formulated with the goal of fostering theory-driven research.