Question:
Has anyone read the research making the media circuit lately by Peter
Gordon of Columbia University regarding the Piraha and their apparent
inability to count and/or represent numbers beyond 1, 2 and "many?"
I immediately take issue with a couple aspects of the work. I feel that
perception comes first, language second. It seems that the commentary
on this research too often implies the reverse. But here's my
reasoning: infants, who speak no language, clearly perceive things
around them, and are able to assimilate them into categories beyond
language. I have a 5 year-old son who'd been speaking in complete
sentences for a long time, but also a 1 year-old. His word for
everything is "da!". But he obviously behaves differently when pointing
at the cat (I now occasionally hear "da.....t" with a "t" at the end),
at me, food, or something else. Categories result from familiarity,
which may or may not require the structural representation of language.
There is some other supra-linguistic categorization going on. Infants
are able to do things that baffle us -- they can think or process
information (to a degree at least) beyond all symbolic language.
Representation of quantity in abstract means (a numerical label) is only
useful for later retrieval or processing. I'd be willing to bet that a
member of the Piraha tribe who may have lost a finger could tell,
instantly, that he was minus a finger. This isn't something you need to
count to 10 to verify (nor subtract 1 from X). It's experiential, and
therefore transcends mere language.
Last, I studied Whorf's hypothesis a decade ago. There's one important
caveat I've yet to see anyoint point out -- namely, that (except for
infants) language is the primary tool we have for learning other
language. So while the Hopi may express a completely different concept
of time with their language, anyone can learn Hopi, live awhile in that
part of the Southwestern US for additional familiarity (experience) and
get an in-depth understanding of what some have over-read into Whorf: a
sort of linguistic determinism. Language doesn't merely constrict
thought, it can open it wide.
The essential problem facing AI it seems is that computers do not have
the "experience" component. At least, not in the emotional sense. A
computer capable of learning language would need some other sort of
feedback mechanism to make sense of a symbol + the experience/stimuli it
references to marry the two together, and of course it would need a
mentor (as a human infant does) in the process.
Answer:
You might want to read some of the work of Patricia Kuhl and others. For example
Kuhl et al., "Linguistic Experience Alters Phonetic
Perception in Infants by 6 Months of Age", Science 255 (1992)
606-608.
This provides persuasive evidence that experience in early language
acquisition can change perception.
Submit Your
Own Answer!
