'Is there a direct relation between the development of vocabulary and grammar?' (Brinchmann et al, 2018) - a discussion
Cat Silvey (British Academy Post-doctoral Research Fellow and member of the Language Learning Lab) shares our thoughts from the latest journal club. We discussed: the recent Developmental Science paper ‘Is there a direct relation between the development of vocabulary and grammar?’ by Ellen Irén Brinchmann, Johan Braeken, and Solveig-Alma Halaas Lyster.
The discussion was lively and positive: the group found the paper an exciting contribution, particularly in terms of the statistical modelling. However, the framing of the question and some of the analytical choices left us slightly perplexed! In this post, I will outline the background of the research and summarise the main talking points that came out of our discussion.
The developmental relationship between vocabulary and grammar is a vexed question in language acquisition. At stake are two accounts of the nature of language. On one side is a ‘words and rules’ account, where vocabulary and grammar are fundamentally separate and are learned in different ways. On the other is a ‘single mechanism’ or ‘lexicalist’ account, where vocabulary and grammar are learned in the same way: to paraphrase Elizabeth Bates and Judith Goodman, they are different kinds of animals kept together in one zoo. These authors found a tight linkage between vocabulary and grammar across development, and interpreted this as support for the lexicalist account (Bates & Goodman, 2001). Recent work with larger samples has painted a more nuanced picture. Braginsky, Yurovsky, Marchman, & Frank (2015) analysed parent checklist data tracking from the development of vocabulary and grammar in around 20,000 children across four languages. They found that, while vocabulary size does explain a large part of grammatical knowledge, a child’s age still makes a statistical contribution to explaining variance in grammar after vocabulary is accounted for. This suggests that there is more to grammar acquisition than simply accumulating vocabulary, whether this additional component is domain-specific, domain-general, or environmental.
Brinchmann et al. position their work as evidence against a lexicalist account. They find that once stable individual differences in language between children are controlled for, there is no evidence for a causal influence of vocabulary on grammar over time, and only limited evidence for an influence in the opposite direction. The authors conclude that vocabulary and grammar are best characterised as stable traits that are highly correlated, suggesting the existence of a common source of influence on both. They further find that home literacy factors (such as parents’ educational level and how often they read to their children) explain 16% and 11% of the variance in vocabulary and grammar traits respectively.
To sum up our discussion of the paper, we weren’t sure the findings were quite as damning for the lexicalist account as the authors made out. Part of this was due to how the lexicalist account was characterised. Brinchmann et al. describe lexicalist theories as predicting ‘a direct and causal dependence of grammar on lexical growth’: i.e. influences should run only from vocabulary to grammar, and not in the opposite direction. While this fits some versions of a lexicalist account (e.g., Michael Tomasello’s verb island hypothesis), the specific papers cited — Bates & Goodman (2001) and Marchman, Martínez-Sussmann & Dale (2004) — do not make this claim. Rather, the claim is that the same learning mechanisms are at work in the acquisition of vocabulary and grammar. This doesn’t mean that vocabulary always runs ahead of grammar: instead, the implication is that tests of these two skills are effectively measuring the same thing. As such, it wasn’t clear to us that anyone would predict a unidirectional influence from vocabulary to grammar, except perhaps at the very earliest stages of language acquisition. Particularly, it would seem odd to predict this at the age of the participants in Brinchmann et al.’s study (4-6 years), given the well-established presence of mechanisms such as syntactic bootstrapping (where children use their growing knowledge of grammar to infer the meanings of novel verbs).
This discussion led us on to the authors’ analytical choices — in particular, the decision to model vocabulary and grammar as two time-invariant traits. Two main questions about this arose:
1) If the ‘trait’ of vocabulary is estimated from vocabulary data at all timepoints, how does the model disentangle stable aspects of vocabulary from aspects that might have been influenced by grammar at the previous timepoint (and vice versa)? We found it hard to understand how this approach avoids mistaking causal influences for part of a stable trait, in the case of parallel increases in two highly correlated variables.
2) Did the authors try testing an alternative model with only one underlying trait, to evaluate whether this accounted better for the data than the model with a trait for vocabulary and a trait for grammar? Given the high correlation (.72) between the stable traits for vocabulary and grammar, we weren’t convinced that the results actually conflicted with a lexicalist account. This seemed like one potential way to test whether or not they really do.
We appreciated the authors’ nuanced discussion of their findings, and their suggestions for future work geared towards disentangling the potential causal effects of input from other factors that might influence vocabulary and grammar together. In particular, we agreed that the possible genetic confound inherent in the home literacy/parent education measure would be important to follow up in future studies. Overall, this was an interesting paper that prompted a lot of discussion.
Bates, E., & Goodman, J. C. (2001). On the inseparability of grammar and the lexicon: Evidence from acquisition. In M. Tomasello & E. Bates (Eds.), Essential readings in developmental psychology. Language development: The essential readings (pp. 134-162). Malden, MA: Blackwell Publishing.
Braginsky, M., Yurovsky, D., Marchman, V. A., Frank, M. C. (2015). Developmental changes in the relationship between grammar and the lexicon. In D.C. Noelle, R. Dale, A. S. Warlaumont, T. Matlock, C. D. Jennings, P. P. Maglio (Eds.) Proceedings of the 37th Annual Conference of the Cognitive Science Society (pp. 256–261). Austin, TX: Cognitive Science Society. https://mindmodeling.org/cogsci2015/papers/0054/index.html
Brinchmann, E. I., Braeken, J., & Lyster, S. H. (2019). Is there a direct relation between the development of vocabulary and grammar? Developmental Science, 22(1), e12709. https://doi.org/10.1111/desc.12709
Marchman, V. A., Martínez-Sussmann, C., & Dale, P. S. (2004). The language-specific nature of grammatical development: Evidence from bilingual language learners. Developmental Science, 7(2), 212–224. https://doi.org/10.1111/j.1467-7687.2004.00340.x