This week’s
reading explores the details of language acquisition in children, focusing specifically
on how the productive
control over the suffix -er in coining agent and instrument nouns in English.
What
I thought to be most interesting in the reading was semantic transparency. I
think that this phase is something that occurs for many years and is something
that children, maybe even early teens, must learn through experience. To me, semantic
transparency is the factor that is most commonly seen through kids. For
example, I always find myself asking, “what’s that word that means this?” and I
will use compound words to explain what word I am trying to find. I think that
children can quickly and effectively learn how to advance with their vocabulary and use conventional adult
words through experience, asking questions, and immersing themselves in language filled environments
(i.e. school, home).
Although
Clark and Hecht talk more about the English language, from personal experience,
I believe that semantic transparency and conventionality is evident in other
languages. Growing up speaking Cantonese, there were always times where I would
use multiple compound words to try and explain one word and my relatives would
laugh (not in an insulting way) at how hard and complex I was making a sentence;
however, they knew the difficulty in learning and knowing new words so they
would say, “do you mean this word?” and correct/teach me the conventional adult
word that I was trying to say. This again proves that children acquire language
not through explicit teaching, but through reinforcement and exposure to new
words.
Hey Aleeza -- it's so cool that you tie in reinforcement learning to linguistic acquisition! In my SymSys class, we talk about the importance of supervised learning and reinforcement learning early-on, and I think both your experiences and Clark and Hecht's discussion of principles of transparency/conventionality/productivity assert that many linguistic abilities are definitely learned through social interaction with our environment rather than being learned in a classroom or existing innately when we're born. Probably could lead to some cool applications in AI/machine learning as well!
ReplyDelete