Re: Tomasello: Cultural Origins of Human Cognition

From: Krisztian GABRIS (
Date: Sun Jun 10 2001 - 21:35:08 BST

I rewrote the majority of the original skywriting still
unfinished at the time of the course, so this text is under the
strong influence of my interpretation of the discussion.

> Perhabs the most significant outcome of acquiring natural
> language is that the language user partitions her world into
> discrete units of particular kinds. This partitioning process
> does not create new conceptual material, of course, but it serves
> to package existing conceptual material in special ways - often
> in ways that an individual would not need if she was not engaged
> in linguistic communication.

This partitioning process is categorisation, which helps deciding
in choosing ways to survive. Language use extends the
capabilities of categorisation, it no longer just helps deciding
which mushroom to eat, it enables us to share more complex
features of the world around us (this is the basis of culture, of
cultural transmission, if we'd like to say).

> Given that the major function of language is to manipulate the
> attention of other persons - that is, to induce them to take a
> certain perspective on a certain phenomenon - we can think of
> linguistic symbols and constructions as nothing other than
> symbolic artifacts that a child's forebears have bequeathed to
> her for this purpose. In learning to use these symbolic
> artifacts, and thus internalizing the perspectives behind them,
> the child comes to conceptualize the world in the way that the
> creators of the artifacts [the linguistic symbols] did.

I'm not sure this is the major function of language. A major
function of language is that it helps ME to manipulate the
attention of other persons, for obtaining precious information
from them about the world, reducing uncertainty in deciding ways
to survive (I eat mushrooms, but not red ones, I sow the seeds
after the white stuff fallen from the sky have gone). I don't
want to induce others to take a certain perspective on a certain
phenomenon. Suppose I want to, but why would that be adaptive for
me? I could convince others to work instead of me (but I'm not
sure these kind of strategies can spread to whole populations as
language, just because of their very nature). Or I could explain
others my hunting tactic, resulting in more food. Or inducing
young children to take a certain view on certain phenomena of the
society for example is adaptive for them, and it is adaptive for
our whole culture. But the above involve complex reasoning and
mental skills, they point far beyond the condition when language
emerged. And now we're talking about the emergence, and basis of
language, not further development.
The advantage of language is what's above, that I don't have to
try out everything myself, and this greatly reduces the chances
of wounding up dead in the process. But do I have to know the
other's intentions while learning? Or all I need is the
information, which makes me less uncertain about the way the
world "works"? That which berry is edible, why do birds migrate,
etc? Almost everything can be said about the world with language
using a non-analogue code. People posess certain kind of info
about the world (the red berry is OK, but the puffy green makes
you sick), and I can have their description if I'd like, so I
don't have to try all the berries myself (if they're willing to
help only of course). Language is a description of the world, and
their "carrier" (the girl I'm talking to) has intentions, which
according to Tomasello are so important in this process. But
theoretically I can learn which berry to eat, or why birds
migrate as good from a good book, as I would from someone else,
reading the non-analogous code telling me the secrets of the
world. Books are also descriptions of the world, and they don't
have intentions, only their author has. And books would be the
same regarding the content if they would be sold without the name
of the authors on their covers. I would still learn which berry
to eat, without understanding the intentions, or mental states of
the author. (I don't even know who they are...) One could argue,
that yes, but we are deducting the author's intentions from her
written speech. But then, the way I get to know the author's
intentions is through her language, and something acquired
through language can't be the basis of language... So why would I
need to understand the intentionality of someone else I learn
from the secrets of the world by language use? Once again, why
would understanding someone else as an intentional being be the
basis of language, if I don't need this ability to reap benefits
from language use, to actually use language at all?

> Some linguists and psycholinguists believe that young children
> operate from the beginning with abstract, adult-like, linguistic
> constructions - because they are born with them. But this theory
> can work only if all languages have the same basic linguistic
> constructions, which they do not - not even at some deep
> structure level. The alternative is the view that early in
> ontogeny individual human beings learn to use their species-
> universal cognitive, social - cognitive, and cultural learning
> abilities to comprehend and acquire the linguistic constructions
> their particular cultures have created over historical time by
> sociogenesis.

So this is what it is all about...

> The studies were designed so that none of the well known word
> learning constraints [...] would be helpful. [...] in all cases
> the majority of children learned the novel words in either
> comprehension or production or both [...]
> [...] the most plausible explanation is that [...] children
> developed a deep and flexible understanding of other persons as
> intentional beings, and so they are quite skillful at determining
> the adult's communicative intentions on a wide variety of
> relatively novel communicative situations - assuming that they
> can find some way to understand these situations as joint
> attentional scenes.

> What the child is learning at this point is that a linguistic
> symbol embodies a particular way of construing things - a
> particular perspective - that is tailored to some communicative
> situations, but not to others. [This is] suggested by the fact
> that they can, from soon after they have started using language
> productively (18 to 24 months), refer to the exact same referent
> with different linguistic expressions in different communicative
> circumstances [...]

> [...] the principle that all word contrast with one another in
> meaning in some way is really a principle of rational human
> behavior along the lines of "if someone is using this word,
> rather than that word in the current situation, there must be
> some reason for it." [...] Altough the process has not been
> studied in much detail, being able to contrast word meanings with
> one another in this way almost certainly facilitates children's
> acquisition of new words [...]

I think the weakness is that the possible role of grammar in
language acquisition is underestimated here. Different
grammatical structures employing the same words could have very
different meanings, and the same words could be used for
describing a variety of different situations. Children without
the help of some syntactical knowledge would know the intentions
of the speaker, but perhaps they would be too confused by the
seemingly inconsistent (consistent with grammatical rules) usage
of words. All this is possible, it's just that this rational
human behavior perhaps could be very ineffective without
syntactic knowledge.

> Another [...] process [...], the process of learning new
> linguistic expressions with help from the linguistic context
> within which it is embedded. Some versions of these processes
> have been conceived of as so-called syntactic botstrapping in
> which the child uses everything from the presence of grammatical
> markers such as the[,] to whole syntactic constructions as hints
> to the meaning of a word [...]

It is not fully explained in the book how the embedding context
is exploited with some form of syntactical knowledge, but that is
understandable, that's what Tomasello wants to avoid (building
theories based on the presence of some innate syntax knowledge),
but always keeps bumping in it.

> [...] there are other, more mundane versions of bootstrapping
> that are less syntactically based. That is, if the children hears
> I'm tamming now as the adult bangs her head against the desk, the
> child can infer that the action being referred to by tamming is
> not one that changes the state of the object acted upon, because
> the desk is not even mentioned.

Leaving parts from the sentence structure seems to be a good
solution, but the child still needs to complete the missing
parts, or it will be not clear that tamming is the verb meaning
what it means, or it means the adult's intention to make the
child laugh, so clowns are also tamming onstage?

> More subtle versions of this process might also occur if the
> child hears, for example, a verb with a particular locative
> preposition, as in He is meeking it out of the box, in which case
> she can assume that the "out of" meaning is not part of the
> verb's meaning since it has its own expression in the
> prepositional phrase. This process may be thought of as a kind of
> contrast as well in that the child must apportion the meaning of
> the adult's utterance as a whole into its component parts, each
> of which plays its role in the meaning as a whole; the novel word
> must then be assigned its portion of the whole [...]

So I think we're at grammatical knowledge mediating language
acquisition right now... After all it seems we do need this
knowledge in "some way", in order to "decompose" the sentences.
But then how does a child know what is a prepositional phrase?
How could she "assume", and "assign", and "decompose" a sentence
without syntactical knowledge of the language, aided only by joint
attentional visual experience?

> [...] to have a full understanding of both the construction and
> it's constituent parts, the child must at some point isolate or
> extract the linguistic elements involved from the whole
> expression [...]

> The verb island hypothesis proposes that children's early
> linguistic competence is comprised totally of an inventory of
> linguistic constructions of this type: specific verbs with slots
> for participants whose roles are simbolically marked on an
> individual basis. At this early stage children have made no
> generalizatons about constructional patterns across verbs, and so
> they have no verb-general linguistic categories, schemas or
> marking conventions. [...] there are no other hidden principles,
> parameters, linguistic categories, or schemas that generate
> sentences.

But how can children extract these verb islands from adult's
sentences without constructional patterns, etc, needed to
generate those sentences?

Tomasello's solution for the acquisition of linguistic expressions
may, or may not work, but to me it seems that no matter which, they
don't give a sole explanation for the problem.

And something from the end of the book concerning the effect of
language on human cognition. It's a pity I couldn't find more
problems with that part (besides the ones still resulting from
the previous problems), so there's no real reason to quote the
interesting ideas for those who missed this book.

> [...] the categories and schemas imminent in language enable
> children, among other things, to take multiple perspectives on
> the same entity simultaneously. There's no good evidence that
> nonhuman animals or prelinguistic human infants categorize or
> perspectivize the world in this hierarchically flexible manner.
> Other animals may be able to take different perspectives on
> things in different circumstances, but because they do not have
> available the many perspectives of others as embodied in
> language, they do not understand that there are a multitude of
> ways that a phenomenon may be constructed simultaneously.

Suppose an animal who just escaped from a predator by climbing on
a tree, starts to eat the leaves, while the predator still lurks
under. Now how does this animal (we can rule out infants in this
specific case as participants..) perceive that tree? I know this
is far-fetched, but the point is that there's no good evidence
for the contrary either (that "animals don't perspectivize the
world in a hierarchically flexible manner"). I think this would
need further investigation.

This archive was generated by hypermail 2b30 : Wed Jun 13 2001 - 18:38:04 BST