Language is a Part of General-Purpose Learning Systems

Megan Zhou ‘21

A posterior view image of the human cerebellum which functions in procedural memory. (Source: Wikimedia Commons)

The theory behind how language is learned is currently based on the declarative/procedural (DP) model of language, a theoretical framework based on the principle that new functions are developed from previous mechanisms. This model explains that since language must be learned, it should heavily depend on the declarative and procedural learning and memory systems (1). Declarative memory refers to explicit memory, which is definitively stored and retrieved; procedural memory is the part of long-term memory that is responsible for motor skills. Grammar, which uses the declarative and procedural memory systems, has thus been used as a point of comparison in recent studies.

Researchers at Georgetown University showed that these general-purpose neurocognitive mechanisms that were formed before humans existed contribute to the process of learning language. Through a meta-analysis of multiple studies using a total of 665 participants, it was determined that children learn their native language and adults learn foreign languages in brain circuits that were formed in ancestors that are evolutionary extremely distant (2). From 16 different studies, ten of which focused on children and the other six examining adults, the researchers synthesized the information by only using two data points from each—the correlation coefficient and its associated sample size (2). The main method of this meta-analysis focused on the systematic search for articles by using an “abstract screening” approach followed by a full text evaluation (1). Abstract screening means that the researchers made the decision of retaining a certain study or not based on any indications from the title, abstract, or keywords that had to do with grammar and declarative or procedural memory. The chosen papers had to examine associations between behavioral measures of language or grammar in addition to behavioral measures of learning in declarative or procedural memory. The original 56 correlations ultimately yielded 40 weighted correlations for which the meta-analysis was computed using Comprehensive Meta-Analysis, a software program developed by Biostatistics Programming Associates (1).

Thus, for children’s first languages, this study has shown that language abilities correlate with declarative memory, but not with procedural memory. The grammar abilities of children were related to the tasks that we learn, such as riding a bicycle or playing a musical instrument. In contrast, for adults, grammar correlated with declarative memory at earlier stages and then with procedural memory at later stages when learning a foreign language (2). Ultimately, the main conclusion from this study is that language relies on the two learning systems—both declarative and procedural memory, which are the general-purpose cognitive systems. Since the correlations were large and consistent across languages and tasks, this study effectively establishes that these links between language and brain systems are statistically significant.

Further research may focus on evaluating the numerous genes identified in the two brain systems in order to obtain further understanding of the genes that underlie language ability. Similarly, by investigating the evolution of these brain systems, scientists can develop insights into the evolution of language. Furthermore, a lead author of the present study suggests that this can lead to approaches that can improve foreign language learning and language-based disorders, such as dyslexia (2).

(1) Phillip Hamrick, Jarrad A. G. Lum, Michael T. Ullman. Child first language and adult second language are both tied to general-purpose learning systems. Proceedings of the National Academy of Sciences, 2018; 201713975 DOI: 10.1073/pnas.1713975115

(2) Georgetown University Medical Center. (2018, January 30). Language is learned in brain circuits that predate humans. ScienceDaily. Retrieved February 4, 2018 from www.sciencedaily.com/releases/2018/01/180130094713.htm