Predictions of Cognitive Individual Differences in Language Acquisition: Commentary on Hulstijn (2024)

Abstract

Hulstijn’s BLC Theory proposes a dissociation between cognitive individual differences and two types of cognition—the cognition of oral language and the cognition of written language. Specifically, cognitive IDs are expected to affect the acquisition of reading and writing skills in both native and non-native speakers, but not the acquisition of speech comprehension and speech production in either native or non-native speakers. This commentary will discuss the potential and the limitations of these predictions and will suggest directions for future empirical research in the context of BLC Theory.

1. Introduction

Hulstijn (2024) presented an update of his Basic Language Cognition (BLC) Theory (Hulstijn, 2011, 2015). Since the essence of the theory remains intact, and due to space limitations, this commentary will focus on the explicit predictions that the update offers regarding the role of cognitive individual differences (IDs) in language acquisition by native speakers (NSs) and non-native speakers (NNSs). These predictions propose a dissociation between cognitive IDs and two types of language cognition: Basic Language Cognition (BLC) and Extended Language Cognition (ELC). BLC refers to the ability to comprehend and produce language in common situations of oral communication (listening and speaking), while ELC refers to literacy skills (reading and writing). BLC involves frequent lexical items and frequent grammatical structures, while ELC involves lexically and grammatically more complex language, connected to language learning in one’s school years and later in life. BLC is shared language knowledge, while differences emerge in ELC with school education. In the context of BLC Theory, Hulstijn predicts that cognitive IDs will affect the acquisition of reading and writing skills in both NSs and NNSs, but they will not affect the acquisition of speech comprehension and speech production in either NSs or NNSs.

2. Cognitive IDs and Receptive and Productive Speech Processing

Several points can be made about Hulstijn’s predictions. With respect to the acquisition of oral language by NSs, Hulstijn’s theory predicts that IDs will only play a role in the speed, or rate, of acquisition, but not in long-term achievement, or ultimate attainment. This prediction can be tested via longitudinal designs that investigate whether cognitive factors of interest predict subsequent speech and language abilities assessed several years later in typically developing children. This type of design could provide support for the causal role of IDs in language development. However, as Hulstijn points out, there is virtually no research to date on this issue. It was, in part, investigated in the 1980s, in the context of the Bristol Language Project (Wells, 1985). This project studied the first language (L1) development of two cohorts of children over several years and revealed large variation in rate of acquisition, which was already evident by age 3–4. In a follow-up study by Skehan (1986), some of the children in the project were given language aptitude tests when they were 13 to 15 years old. Most of these tests (e.g., a verbal intelligence test, a sound–symbol association test or a rote memory test) required thinking about how language works and, therefore, they involved language analytic abilities. The results showed correlations between some of these cognitive abilities and the children’s L1 development prior to age 5. For example, correlations were found in the case of mean length of utterance, range of adjectives and determiners, range of nominal phrase complexity, comprehension, and vocabulary. However, there were more significant correlations which made the role of environmental factors difficult to disentangle. Specifically, family background variables, such as parents’ level of education and parents’ interest in literacy, were significantly related to scores on measures such as verbal intelligence and grammatical sensitivity tests, which, in turn, were related to linguistic indices, such as mean length of utterance and range of adjectives and determiners. Perhaps not surprisingly, only one of the cognitive measures, a sound discrimination test, was unrelated to biographical factors. This ability correlated with two of the comprehension indices in the study and with one of the vocabulary indices, suggesting a distinct dimension of cognitive ability in L1 development. Longitudinal cohort research is as challenging as necessary to understand the predictive power of cognitive IDs in language development. Part of the challenge stems from the fact that, unlike in Skehan (1986), cognitive abilities should be measured before language development takes place or, at least, as early as possible in the longitudinal design and prior to the measurement of language acquisition. In the case of NSs, this requires the use of age-appropriate and child-friendly cognitive measures, which are already available for children as young as five.

The prediction that IDs will not play a significant role in NSs’ ultimate attainment of receptive and productive speech processing is the corollary of the premise that BLC (i.e., oral language) is used and shared by all NSs. Despite the high inter-individual homogeneity in lexical, morphological and syntactic elements claimed by BLC Theory, there is evidence in the literature of variability in spoken language processing among NSs, as well as evidence of an association between this variability and cognitive IDs. For example, Conway et al. (2010) found differences among NSs in their ability to use knowledge of word predictability to aid speech perception under degraded listening conditions. Participants were all undergraduate students, and the sentences that were used as stimuli were simple (e.g., “Bicycles have two wheels”). The study further found an association between the sensitivity to word predictability shown by NSs and IDs in implicit learning ability. This association remained strong even after controlling for vocabulary knowledge and performance on other cognitive tasks, such as general intelligence and short-term memory capacity. This suggests 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. It could be argued that testing conditions included an extralinguistic feature (noise) that interfered with normal processing. However, many situations of everyday life require processing speech signals that are perceptually degraded. Overall, the results of Conway et al. suggest that implicit learning ability can moderate sensitivity to probabilistic relations among language units and ultimately facilitate (or hamper) how language is used in real-time comprehension.

Regarding the acquisition of oral language by NNSs, BLC Theory does not predict differences caused by cognitive factors. This prediction is somehow surprising, given all the research on IDs in implicit cognitive abilities in second language acquisition (SLA) and the evidence in support of a link between these IDs and L2 learning (e.g., Granena, 2013a, 2020; Linck et al., 2013; Suzuki & DeKeyser, 2015). In this sense, BLC Theory fails 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. The theory predicts individual variability in language acquisition, at least in the domains of pronunciation and with respect to the production of some grammatical features in spontaneous speech, which cannot be fully attained by NNSs, but cognitive factors will not account for such variability. This clearly conflicts with studies that have shown that implicit abilities can be a significant predictor of attainment among very advanced NNSs. For example, Granena (2013b) investigated the role of implicit sequence learning ability in early and late second language (L2) learners’ morphosyntactic attainment in Spanish. Language attainment was measured, among other tasks, by means of an auditory word monitoring task tapping automatic processing of incoming speech. The results showed that implicit sequence learning ability moderated both early and late learners’ attainment in the case of structures involving grammatical agreement (noun-adjective gender agreement, noun-adjective number agreement, and subject-verb agreement). These are all structures that would qualify as belonging to BLC since they are highly frequent in Spanish, occur in any communicative situation, and are acquired early by NSs (Montrul, 2004).

3. Cognitive IDs and Reading and Writing

In the case of the acquisition of ELC (i.e., reading and writing), BLC Theory predicts large variation in both speed and ultimate attainment among both NSs and NNSs. In the case of NNSs, and unlike oral language, the prediction is that late L2 learners will be able to become as proficient as NSs in reading and writing, provided that NSs are matched on specific features such as cognitive profile, education, and profession. BLC Theory also predicts that at least part of the hypothesized variation in the acquisition of ELC by NSs and NNSs will be accounted for by cognitive IDs. The question here is which cognitive abilities will explain such variation. Based on the examples that Hulstijn provides (i.e., executive function, memory, and intelligence), he seems to be thinking of cognitive abilities that depend on executive attentional resources and that belong to the explicit cognitive domain. The prediction that cognitive abilities such as executive function or psychometric intelligence can explain differences in reading and writing is uncontroversial. In fact, there is ample evidence that it is so. For example, in the case of research with very advanced late L2 learners, two studies by Granena (2013b, 2021) investigated whether explicit cognitive abilities were differentially related to late L2 learners’ performance depending on whether the language measure involved reading. Granena (2021) found that cognitive ability, operationalized as a combination of rote memory, explicit inductive learning, and ability to explicitly learn sound–symbol associations, was differentially related to the pronunciation ratings obtained by late L2 learners on read-aloud tasks and spontaneous production tasks. Explicit cognitive abilities predicted pronunciation ratings on word reading and paragraph reading, but not on two spontaneous production tasks, controlling for the common variance associated with age of onset and length of residence. This result suggested that late L2 learners with higher explicit cognitive abilities were more successful at consciously monitoring (or attending to) their speech and at relying on metalinguistic abilities while reading aloud.

Granena (2013b) found similar results in the morphosyntactic domain. In this case, the language measures compared were an auditory and a written grammaticality judgment test. The study revealed a significant interaction between explicit cognitive abilities and test modality. While the correlation with auditory grammaticality judgment scores was weak and non-significant, the correlation with written scores was moderate and significant. The study concluded that untimed written tests allow for controlled use of knowledge and monitoring. Learners with high explicit cognitive ability have a greater ability to learn and think analytically, problem-solve, and, more broadly, greater ability to control attention. The fact that L2 learners delayed their responses on the written test by about 4 seconds on average compared to NSs suggests that, when given the opportunity to do so, L2 learners engaged in conscious reflection about sentence grammaticality.

4. Conclusions and Further Research

BLC Theory attributes a role for cognitive IDs in the acquisition of reading and writing skills, but not in the acquisition of oral language (except, perhaps, in the case of speed of language acquisition). What BLC Theory is lacking regarding its predictions of cognitive IDs in language acquisition is a more precise definition and operationalization of “cognitive factors.” There are a myriad of cognitive factors, and not all of them are related to a latent cognitive ability indexing general intelligence. As Hulstijn correctly argues, “one does not have to be particularly intelligent to learn an additional language, with respect to listening and speaking” (p. 6). As he also correctly claims, this applies to both NSs and NNSs. However, the point this argument is missing is that there are cognitive abilities, such as implicit sequence learning and primability, which have minimal overlap with cognitive abilities in the domain of explicit cognition. This means that IDs in implicit learning or priming are either weakly related, or unrelated, to IDs in general intelligence, working memory or intentional associative learning. The same widespread erroneous folk wisdom saying that one must be clever to learn another language also has it that one learns a language by systematically studying its grammar. In order to gain a fuller understanding of the role of cognitive IDs in language acquisition, research is needed that explores the effect of learners’ implicit and explicit cognitive abilities on language learning under conditions of massive aural input or under instructional interventions which give priority to listening and speaking.

Moving on to further directions for future research, a relevant suggestion Hulstijn makes is recommending that researchers carefully choose and justify the language elements which NSs or NNSs will be tested on (words, constructions, grammatical patterns). Hulstijn suggests the use of high-frequency elements that are representative of everyday oral communication when testing listening/speaking, and constructions and grammatical patterns that are characteristic of written discourse when testing reading/writing. As a guide to the selection of target structures, researchers may also consider the L1 acquisition literature and research on the ages by which certain structures are typically acquired. For example, in the case of Spanish, children acquire structures such as gender, number, and subject-verb agreement early (i.e., by age 3), whereas structures such as the subjunctive, the passive, and aspect contrasts are acquired later (i.e., at least age 7 or later) (Montrul, 2004). The late acquisition of the subjunctive, the passive, and aspect contrasts suggests that these structures are more linguistically complex, and that children are not developmentally ready for them until they are equipped with the necessary cognitive resources to process them. For example, in the case of the subjunctive (mood selection), children lack mental representations of “events that are independent or even incompatible with the reality of physical events” (Pérez-Leroux, 1998, p. 589). Structures such as the passive and the subjunctive are also explicitly taught at schools, which indicates that they are frequently used in written language and formal registers, so their acquisition will be influenced by education and literacy level.

BLC Theory provides a useful framework to test predictions concerning cognitive IDs that can be further refined bearing in mind more recent conceptualizations of cognitive abilities. Researchers using the theory for new empirical research on cognitive IDs should also bear in mind that falsifiability in IDs research crucially depends on variability, both on the predictor and criterion variables. Cognitive differences are expected to explain language learning differences. Falsifiability may be compromised if there is no variability in language acquisition, as in the case of ultimate attainment of oral language by NSs, according to BLC Theory. In this case, using the lack of evidence to conclude that cognitive IDs do not play a role in language acquisition would be falling into the absence of evidence fallacy.