Usage-Based Models Around and Inside BLC Theory: Commentary on Hulstijn (2024)

Abstract

In his essay entitled “Predictions of Individual Differences in the Acquisition of Native and Non-Native Languages: An Update of BLC Theory”, Jan Hulstijn updates his theory of Basic Language Cognition (Hulstijn 2011, 2015). His central claim is that there is a fundamental difference between the Basic Language Cognition that is sufficient for dealing with oral language and the Extended Language Cognition required for handling reading and writing. In my invited commentary, I argue that used-based models of language provide a wider theoretical background that helps to ground Hulstijn’s theory. I illustrate what this means with reference to the Entrenchment-and-Conventionalization Model (Schmid 2020). Based on this model I discuss Hulstijn’s predictions regarding the extent of interindividual differences to be expected among L1-speakers and L-2 speakers.

1. Introduction

Jan Hulstijn (henceforth JH) introduced his theory of Basic Language Cognition (BLC) in a journal article published in 2011 (Hulstijn, 2011). Four years later, he published a book-length elaboration and extension of his ideas on BLC Theory (Hulstijn, 2015). The most basic tenet of BLC Theory is that there are two fundamentally different types of linguistic cognition or proficiency: “The cognition of oral language (speech reception and speech production), called Basic Language Cognition (BLC), and the cognition of written language (reading and writing proficiency), called Extended Language Cognition” (Hulstijn, 2024, p. 1). In the essay under discussion here, JH provides an updated version of BLC Theory and formulates predictions concerning the extent of individual differences in proficiency with regard to BLC and ELC of native speakers and learners.

As requested by the journal editor, I will comment on JH’s essay from the perspective of usage-based models. I will first argue that usage-based models help to ground BLC Theory. Then I will propose a reformulation of BLC Theory from the perspective of one particular usage-based model, i.e., my own Entrenchment-and-Conventionalization Model (Schmid, 2020), which highlights points of convergence. In the final section, I will briefly, and more critically, discuss JH’s predictions concerning individual differences from the perspective of usage-based models.

2. Usage-Based Models Help to Ground BLC Theory

Usage-based models are mentioned in JH’s essay in a section entitled “Metatheoretical Framework”. What JH essentially seems to do here is put BLC theory on a new and broader footing. In his 2015 book, JH had mentioned and discussed usage-based models, alongside the generative approach, with the detached sympathy of an external bystander. In the present essay, however, JH embraces the family of usage-based approaches as a congenial home for BLC theory. This is very appropriate, I think, because usage-based models spell out a number of conceptual foundations on which BLC was built in the first place (although, as we will see in Section 4 below, there are also points of divergence). The most prominent of these foundations include the following claims:

• Processing in usage leaves traces in cognition that become noticeable in usage.
• The quantity and quality of input a speaker gets affects their linguistic knowledge.
• The type of input a speaker gets, e.g., in terms of medium and style, affects their linguistic knowledge.
• Cognition is affected by usage frequency, but usage frequency is not a linear predictor of cognition.

Overall, it seems to me that JH adds an extra layer of plausibility and coherence to BLC theory by offering it for adoption into a family of like-minded approaches.

What else is there to be gained from potential cross-fertilization in this milieu? May usage-based approaches have even more to offer to their newly arrived relative? And is there anything that BLC Theory might have to contribute to usage-based approaches?

3. Putting the Usage-Based Approach into BLC Theory

JH emphasizes the instrumental and contingent nature of theoretical constructs. He adds, “It is totally acceptable to explore whether a theory would gain in explanatory power if one changed the definition of a construct. In other words, one can conditionally ‘play’ with definitions”.

I want to accept JH’s invitation to play with his constructs by reframing his definitions of BLC and ELC in a distinctly usage-based way. The purpose of this exercise is to probe whether by putting the usage-based approach into BLC theory rather than over or around it as a metatheory, as JH explicitly does, we “might gain a better understanding of the phenomena”. My exploration will be based on my own Entrenchment-and-Conventionalization Model (EC-Model; see Schmid, 2020), which is as usage-based a model as we can currently find.

To be clear about what it is that I am reframing, I first quote the definitions of BLC and ELC from JH’s essay:

In terms of internal language, BLC refers to a person’s ability to comprehend and produce spoken language in situations of everyday life, common to all adult native speakers in a given language community. […]

In terms of external language, BLC refers to frequent lexical items and frequent grammatical structures—that is, to lexical items and morpho-syntactic structures that may occur in any communicative situation, common to all adult native speakers, regardless of age, literacy, or educational level. […]

In psycholinguistic terms, BLC is defined as the largely implicit, unconscious experience-based cognition in the domains of phonetics, prosody, phonology, lexis, morphology, constructions, and syntax, plus the mapping of words, phrases, constructions, and patterns onto meaning (semantics) and use (pragmatics).

(Hulstijn, 2024, p. 3)

As far as ELC is concerned, JH has decided to propose a revised definition, “to avoid confusion between frequent and infrequent language and oral and written discourse, which could arise from the wider definition”. The revised definition reduces ELC to literary skills: “ELC is the control of the written standard language, as taught in school”. To be honest, I am not sure whether this revised definition is really an improvement, in particular since JH keeps referring to frequency in order to distinguish between BLC and ELC in numerous parts of his essay. But that is not my main point here.

3.1. The EC-Model in a Nutshell

In order to give a rough sketch of the Entrenchment-and-Conventionalization Model, it is first important to draw attention to the distinction it makes between communal linguistic conventions and individual linguistic cognition. Conventions are collective regularities of behavior that the members of a community adhere to, mutually expecting each other to do so (Schmid, 2020, p. 88). These regularities come into being and are sustained by two types of conventionalization processes: usualization, i.e., the establishment of regularities of linguistic behavior among the members of a community; and diffusion, i.e., the spread of a regularity of behavior among the members of a group and across groups and communities. Linguistic cognition ‘takes place’ in the minds of individual speakers and is available, according to the model, in the form of an associative network (Schmid, 2020, pp. 43–45). The network is shaped by processing experience and is continually sustained by the process of entrenchment, i.e., the routinization of different types of linguistic associations (e.g., symbolic form-meaning associations, syntagmatic associations between sequentially arranged words, or pragmatic associations between parameters of context and linguistic choices). These associations are activated in processing to handle production and comprehension, and they become routinized by repeated reactivation (Schmid, 2020, pp. 226–228).

Conventionalization as a social process and entrenchment as a cognitive process can only interact via usage, and they do so by means of a double feedback loop (Schmid, 2020, pp. 3–5): What is more conventional is more likely to be used in usage events, because more conventional utterance types facilitate effortless mutual understanding and are therefore used more frequently than less conventional utterance types; what is used more frequently becomes more strongly entrenched in the minds of individual speakers of the community who share the conventions; what is more entrenched in the minds of individual speakers and therefore more likely to be used by them is more likely to become and remain conventionalized in the communities of speakers the individual is part of or interacts with.

3.2. Reformulating BLC Theory in EC-Model Terms: L1-Speakers

Though very simplistic and fragmentary, this sketch should suffice as background to my reformulation of BLC and ELC in the EC-framework. How do the usage-based assumptions underlying the EC-Model square with JH’s definitions quoted above? I will present my answers to this question in two columns to facilitate comparison between BLC and ELC, beginning with L1-speakers.

L1-speakers’ BLC from the perspective of the EC-Model		L1-speakers’ ELC from the perspective of the EC-Model
Usage		Usage
The usage events associated with BLC as conceptualized by JH are everyday situations of more or less casual spoken language use. This type of communication is spontaneous, situated, and synchronous. Therefore, it offers limited options for planning utterances, but allows recourse to shared contextual information and gestures, gaze direction, etc. Utterances are often fragmented (due to limited planning) and may remain incomplete (due to reliance on context; see Biber et al., 1999, pp. 1041–1067). In terms of communicative intentions, usage as associated with BLC by JH is marked by a large proportion of highly recurrent communicative tasks which can be solved by highly recurrent and comparatively simple linguistic solutions. Conceptual complexity tends to be comparatively low.		The usage events associated with ELC as conceptualized by JH are situations of written language use. This type of communication is asynchronous. Therefore, it does not impose any time constraints on planning, editing, and monitoring in production, or on reading and rereading in comprehension. It is not possible to take recourse to the situational context, i.e., to leave information unspecified because it is retrievable from context. In terms of communicative intentions, usage as associated with ELC by JH is marked by a very large variety of communicative tasks, which can sometimes only be solved with the help of comparatively rare and complex linguistic solutions. Conceptual complexity tends to be higher than in BLC contexts.
Conventions		Conventions
The conventions associated with BLC as conceptualized by JH are implicit conventions that have been shaped by usage and keep changing under its influence. They facilitate effortless mutual understanding in everyday situations of casual spoken language use. They are marked by relatively low structural complexity, by frequent repetition of frequent words and chunks, as well as low-level patterns that offer highly conventional solutions for recurrent everyday communicative tasks. Many of these solutions are lexically specific or only partly variable, rather than based on highly schematic constructions. The conventions associated with BLC are both highly usualized (i.e., speakers behave in a very similar way, due to strong mutual expectation) and widely diffused (i.e., the vast majority of speakers, groups, and communities follow these regularities in many different contexts).		The conventions associated with ELC as conceptualized by JH are partly the implicit conventions of BLC and partly explicit norms of the written (and formal spoken) standard. They facilitate mutual understanding in situations of written and formal spoken language use. These conventions include higher demands on structural integrity. Since they have to meet the demands of a larger variety of communicative tasks and of higher conceptual complexity, they offer more diverse and more complex linguistic solutions. These conventions also need to operate on a more abstract and schematic level and have to offer a larger variety of lexical resources, including rare and technical vocabulary. The conventions associated with ELC are highly usualized among those who partake in them (due to their normative character), but less widely diffused than BLC conventions (i.e., not all members of the larger community partake in them, and they are only used under the specific conditions of written and formal language use).
Cognition		Cognition
The associative network corresponding to JH’s BLC is shaped by implicit strengthening of routines, as predicted by the EC-Model. Routines become automatized, unconscious, uncontrolled, and partly uncontrollable (especially in oral comprehension) with increasing entrenchment. Many routines are lexically specific or only partly schematic. Both types become highly entrenched by frequent repetition, supported by the strong correspondence of tasks, contexts, functions, and conventional utterances.		The part of ELC, in JH’s sense of the term, that goes significantly beyond BLC is learnt, not entrenched by repetition. Early learning of ELC is explicit and builds up declarative knowledge, applied under conscious executive control in reading and especially writing. While these skills are not automatic to start with, intensive practice can make them more procedural and more automatized by entrenchment. For reading, this shift happens much earlier, at a faster pace, with much less effort and with more ‘success’ than for writing.

Admittedly, this is a rather dense translation of the basis of BLC Theory into a usage-based framework. What I hope to have shown is that the inner workings of BLC Theory are indeed usage-based in spirit and that it could be possible to refine it by spelling out these claims in more detail. For example, usage-based theory can contribute to providing a solid functional explanation of factors like usage frequency, lexical variety, structural integrity, and structural complexity. In particular, since researchers in the usage-based framework have explored the effects of frequency in great detail, JH’s somewhat crude suggestions for how to operationalize the difference between BLC and ELC based on frequency counts could be improved considerably. One specific contribution of the EC-Model could lie in the distinction between usualization (as degrees of conventionality) and diffusion (as spread of conventions) and the observation that ELC is less widely diffused in the larger speech community. Reversing the perspective, my impression is that usage-based approaches, and the EC-Model in particular, can learn to sharpen their ideas about explicit learning in the context of written language processing.

3.3. Reformulating BLC Theory in EC-Model Terms: L2-Learners

To transfer these ideas to L2-learners, it is essential to distinguish between two idealized types of learning circumstances: on the one hand, the immersed learner who gets large volumes of input in the spoken and written medium, and the classroom learner who can profit from explicit teaching and supervised practice but does not get much immersion-type input, on the other. The immersion learner resembles the L1-speaker in that experience-driven entrenchment can play a leading role in BLC learning. The major differences between L1-speakers and immersed L2-learners, as noted by JH with reference to MacWhinney (2017), is that learners have to break up entrenched L1 routines and create new neural pathways and routines. Success and proficiency depend on factors such as age and corresponding brain plasticity as well as the volume of input and output. For the classroom learner, the situation is very different because they do not get enough input to fuel implicit experience-driven entrenchment. In this respect, classroom learners are on a similar journey as L1-speakers working towards ELC.

4. Individual Differences with Regard to Proficiency from the Perspective of the EC-Model

JH formulates four predictions targeting the extent of individual differences regarding proficiency. These predictions connect BLC and ELC, on the one hand, with L1-speakers and L2-learners, on the other. I summarize his predictions in

Table 1. Summary of JH’s predictions regarding individual differences in proficiency.

	L1 (“Native Speaker”)	L2 (“Non-Native Speaker”)
BLC	P1. Speed of BLC learning by L1-speakers differs, ultimate attainment does not differ.	P3. There are large differences regarding BLC learning by L2-learners caused by input, typological distance between L1 and L2, affecting levels of language and modes of use/skills in different ways.
ELC	P2. There are large differences in reading and writing skills (ELC) among L1-speakers, due to differences in cognitive abilities, level of education, and profession.	P4. There are maximal differences among L2-learners in acquisition speed and ultimate attainment of reading and writing skills (ELC), modulated by typological differences between L1 and L2, differences in writing systems, and differences in cognitive abilities.

.

Plausible as these predictions sound, it must be said that the notions of ‘ultimate attainment’, ‘attainment’ and even ‘proficiency’ are not fully in line with the spirit of usage-based models, which regard variation as an inherent property of language and are more reluctant to assess utterances as being indicative of different degrees of proficiency. Individual differences are not so much conceptualized in terms of attainment but more in terms of individual variation (but see the work by Dąbrowska and colleagues, e.g., Dąbrowska, 2015). Individual differences in usage are a function of many factors, including those mentioned by JH, as well as the volume and variety of linguistic experience and the regularity and complexity of conventions, as long as entrenchment is the dominant learning mechanism. As noted above, it seems reasonable to propose a principled difference between the skills required for oral and written processing and the learning mechanisms involved. Nevertheless, in a usage-based account, a gradient approach seems to make more sense, not only for L1-speakers but also for L2-learners. Ultimately, all would hinge upon the question of what we take “attainment” and “proficiency” to mean here. This in turn would depend on how the nature of conventions is conceptualized, since ultimately attainment and proficiency are measured against some notion of communal conventions rather than as cognitive notions.

5. Conclusions

In this commentary, I have emphasized the points of convergence between BLC Theory and usage-based models. These definitely invite a more intensive exchange of ideas in the future. As noted in the final section, the notions of attainment and proficiency would certainly have to be discussed as part of this exchange. Two fields where I see a lot of potential for developing BLC Theory are refined operationalizations of frequency in usage-based models and a more detailed discussion of linguistic and conceptual complexity in formal spoken language.