From painstakingly memorising German declension tables to trying to come to grips with Japanese word order, language learning has always been an effortful endeavour for me. I’m sure many of you share the same sentiment. I’ve hence always been baffled by how all children (those with specific impairments aside) manage to learn their first language(s) so effortlessly and quickly despite being less cognitively competent and having a limited amount of non-explicit/opaque linguistic data to work. I eventually hope to do a post comparing first language acquisition and second language acquisition, but I think it’s important to first properly breakdown what “first language acquisition” (henceforth L1A) entails and what a theory of L1A should address.

First Language Acquisition

A very standard and simplified definition of language acquisition is “a development of mappings between forms (i.e. words, sentences) and meaning/function (i.e. concepts, propositions)”. For L1A, the child learns the mappings pertaining to their native language in a naturalistic context from birth. I’m not very happy with this definition as it places a heavy emphasis on semantic meaning over other aspects of language and fails to address the compositional nature of meaning. Instead of providing a definition of my own, I think it’d be easier (and more helpful) for me to briefly outline the main tasks children need to ‘complete’ in order to acquire a language.

1. Segmenting the sound stream
   • The only linguistic input children get is from speech (for the sake of simplicity, I won’t be touching on deaf children). Although as adults we appear to hear separate words in a speech stream, the speech signal is actually continuous — we typically don’t pause between words. Children hence have to somehow segment the speech stream into discrete linguistic units at various levels of analysis (phonemes, syllables, words). Acquiring different forms (of units)
   • This requires first recognising and storing the different forms. Speech production (or attempts at speech production) come after.
   • Forms include: phonemes, to some extent prosodic elements (e.g. intonation), words
2. Mapping form to meaning
   • This mainly pertains to words and involves the construction of the lexicon. This is not a straightforward task as there is basically an unlimited range of potential meanings when attempting to connect input to context. Other information such as the argument structure in verbs (e.g. transitive, 2 arguments) also has to be stored.
3. Putting different units together according to the ‘rules’ of their native language and interpreting the result
   • This involves having knowledge of larger linguistic units. Depending on which view you subscribe to (more on this later), this involves constituents or constructions. The generative view also requires children to have knowledge of syntactic categories, word order and more complex operations and computations (i.e. anaphora, wh-movement). For constructivists, this stage involves a lot of analogizing, decomposing and generalising. This is really brief and unclear but it would take pages to explain this properly (which I might do in a few posts).

What does a good theory of L1A need to account for?

• Linguistic input. What kind of input do children get quantitatively and qualitatively? How does this shape their linguistic knowledge? Different views have different takes on the nature of input exposed to children. However, there is some degree of consensus that children get a finite number of positive evidence and very limited negative evidence.
• Learning mechanisms. Do children use general cognitive mechanisms or a special ‘language faculty’? This touches on other issues such as innateness and modularity. Are these mechanisms linguistic-specific? Are they based on innate linguistic constraints and principles?
• Linguistic representations. How is linguistic knowledge represented (what Chomsky terms linguistic competence or I-language)?
• All observed empirical phenomena (i.e. all the errors and non-errors that children make). Note that we can’t just look at the utterances produced by children. Experimental studies have revealed a gap between linguistic comprehension and production in young children. We should also account for the ability of children to generalise and be productive in language (i.e. linguistic creativity) rather than merely imitating what adults say.
• How linguistic knowledge fits in with other general cognitive abilities. Is this knowledge modular or does it interact with other general cognitive abilities? Or perhaps there is no specific linguistic knowledge; children are just using general cognitive skills. This also requires a good understanding of how language acquisition fits into theories of cognitive development.
• Developmental stages (sequence of linguistic developments). The theory should outline why and how a specific phenomenon is acquired from these conditions (e.g. why something is acquired at age 2 and not age 4); ‘conditions’ can pertain to cognitive, physical (e.g. vocal tract development) and social effects. Any cross-linguistic differences should also be addressed. It is also important to explain how the child’s grammar eventually develops into an end-state grammar that is consistent with an adult’s — or if one does not agree that there’s continuity between a child’s grammar and an adult’s, what prompts this change in grammar.

Note: the use of “grammar” refers to its linguistic definition (i.e. the system of rules governing the structure of a language).

L1A theories broadly fit into two main camps: the generativists and constructivists. They share different hypotheses on many of the points above. More on this next time.