A Language Acquisition Device Without Universal Grammar

17 January, 2017
Considering a Language Acquisition Device based on Phonological Universals.

Language Acquisition Device

The Language Acquisition Device is a hypothetical mechanism or module within the human brain that is specialized for the acquisition of an infant's first language, which contains some kind of "pre-learned" information about what form a language can take.

The term was invented by Noam Chomsky, the famous and somewhat controversial linguist.

Chomsky's hypothesis about the Language Acquisition Device centres around Universal Grammar. The Universal Grammar is assumed to contain some rules that apply to all languages, and forearmed with this information about grammar, the infant learns their first language more quickly and easily.

A more specific hypothesis, advanced by Chomsky, is that Recursion is found in all human languages, because it is part of the Universal Grammar.

This specific hypothesis has encountered a specific objection, from linguist Dan Everett, who has made a career out of the studying the Pirahã language, which, apparently, has no recursion.

Linguistic Universals

A hypothetical Universal Grammar is one type of Linguistic Universal, ie something that might be true of all, or almost all, natural human languages.

But grammatical universals are not the only possible type of linguistic universal. For example, we can consider Phonological Universals.

Phonological universals tell us something about what a language sounds like, even if we have no insight into the meaning or grammatical structure of the language.

For example, the following rule applies to all known languages which are learned as a first language by infants with normal hearing:

(Obvious exceptions to this rule are the "whistling" languages and sign languages, but I believe these are excluded by the conditions of "learned as a first language" and "with normal hearing".)

The Human Brain Module for "What Language Sounds Like"

If Chomsky's theory about Universal Grammar is correct, then there must be some specific "module" in the human brain responsible for representing and applying the Universal Grammar (whatever the details might be) to the learning of an infant's first language.

The interesting thing about Phonological Universals is that scientists have already discovered a brain module for "what language sounds like". This module consists of a set of neurons that respond to the sounds of spoken language, whether the language be familiar to the listener, or not.

This discovery was published in Distinct Cortical Pathways for Music and Speech Revealed by Hypothesis-Free Voxel Decomposition, authors Sam Norman-Haignere, Nancy G. Kanwisher and Josh H. McDermott

It would be speculative at this point to assume that the cortical pathway for speech sounds found by Norman-Haignere et al encodes any specific phonological universals, but given that "what speech sounds like" and "phonological universals" are more-or-less the same thing, it's not an implausible assumption.

Phonological Universals in the Language Acquisition Device

I hereby propose an alternative hypothesis about the Language Acquisition Device:

This version of the Language Acquisition Device speeds up the process of learning language, because the infant learner does not have to waste time learning from scratch that:

How I Came Up with this Hypothesis

You might think I formulated this hypothesis because I had some strong opinion about Noam Chomsky and Universal Grammar.

But you would be wrong.

I developed this hypothesis as an essential component of my latest theory about music.

Very briefly, my theory about music is:

One question that arises, given the assumption that the sole function of music is to trigger a response in a builtin "music recognizer", is why the rules for music have to be as complex as they are, and what evolutionary pressures would have caused them to develop.

One possible answer to this question is that the music recogniser evolved as a mutated copy of some other existing recogniser that already had a similar degree of complexity. And given the many similarities between music and speech, a speech recognizer is the most plausible candidate for being the evolutionary precursor of the music recognizer.

Hence the hypothesis given above.

It should further be noted that the groups of neurons analysed by Norman-Haignere et al include both a speech recogniser and a music recogniser. Which is consistent with a hypothesis that one of those recognizers has evolved from the other.

Appendix: What Speech "Looks Like"

One of the many mysteries of music is that we can have an emotional response to dance, as observers, just as we have an emotional response to music, as listeners.

If it is the case that music has some relationship to speech, then it seems reasonable to speculate that dance has the same relationship to the visual aspects of speech.

And if music is specifically related not just to speech, but to a speech recognizer that facilitates the learning of a first language, then it follows by analogy that the speech recogniser includes information about the visual aspects of speech, ie what speech "looks like", and that this visual information contributes to the language learning process.

In other words, if Mummy is making sounds that sound like speech, and Mummy looks like she's speaking, then I should pay special attention to those sounds she's making, I should make an effort to understand their structure and meaning, and I should treat them as belonging to a special category of stimulus that is very separate in nature from all the other sounds I hear.