Extract from the continuation of that paper:
"
The role of language in cognition
There’s really no way to discuss the long and complicated philosophical tradition of discussing the relation between language and cognition without being glib and superficial, but, happily, I’m pretty adept at glib and superficial, so that won’t stop us. A number of philosophers, including Michael Dummet, have offered ‘strong’ theories of language’s role in thought.
Their ‘language-first’ approaches argue to varying degrees that certain kinds of thought, or even reflective thought as a whole, is only possible once a community-wide practice of communication through language occurs. We can find strong and weak variants in the work of theorists like William Calvin, Merlin Donald and Daniel Dennett.
Language-first models predict that thought is more or less limited by the absence of language, the strongest suggesting that most of thought would be disrupted, and posit a definitive break in the forms of cognition available once human had produced language. The language-first approach also generally suggests that cognitive capacities vary with one’s language ability, meaning that not all linguistic communities likely have the same cognitive capacities. One noteworthy example is work on the Pirahã, a Brazilian Native American group whose language lacks numbers according to many researchers (see Frank et al. 2008, or a
popular press version at The Independent or see
the collection of Pirahã-related links at Language Log).
In contrast, opposing ‘thought-first’ arguments suggest that language expresses thought rather than being a precondition for thought occurring. For example, Jerry Fodor has argued that a prior ‘language-of-thought,’ sometimes referred to as ‘mentalese,’ underlies language ability, and partially explains similarities among languages. The thought-first model, however, can develop a problem of infinite regress, as it’s unclear how the ‘language-of-thought’ itself arises except from a prior set of symbols.
In the corner of the ‘thought-first’ argument, we could site a range of empirical evidence, such as the work of psychologists
Susan Hespos and
Elizabeth Spelke. Hespos and Spelke (2004), for example, found that five-month-old infants born to English-speaking parents perceived object relations concepts that were not highlighted in English, and that their parents did not see as perceptually salient (a relationship of ‘tight-’ and ‘loose-fitting’ that their research had shown to be salient to Korean speakers, whose language does highlight this distinction). That is,
the infants in the English-language environment seemed to develop a pre-linguistic concept that was not supported by their first language, and thus the distinction atrophied and disappeared from their perceptions (much as sounds that are not featured in one’s language become less perceptually vivid after six months of age, eventually becoming hard to perceive).
In anthropology, Edward Sapir and Benjamin Whorf are frequently credited with bringing into sharp focus the role of language in shaping perception and cognition, although they arguably offered a less deterministic account of the relationship than some language-first philosophers (see our posts,
Sapir-Whorf hypothesis is right… sort of? and
Sapir-Whorf hypothesis was right… about adults, for more of a discussion of the Sapir-Whorf hypothesis).
Their approach suggests that language biases perception, affecting how people are capable of perceiving, making some ideas or even qualities of the phenomenal world, more or less difficult to perceive. Coupled with work like that of Hespos and Spelke, the work on language biasing perception suggests that pre-linguistic perception is actually more attuned to sensory discrimination that may later disappear if not buttressed by language; that is, the pre-linguistic conceptual world is perhaps more attuned to certain gradations, less likely to overlook intermediate or uncategorized sensations.
When we actually look at the evidence in Schaller’s account, we find that neither a ‘language-first’ nor a ‘thought-first’ model seems to capture the inconsistency of Ildefonso’s conceptual capacities. Schaller suggests that his ‘brain was kept alive with problem solving,’ figuring out how to get money, whether by begging or working, find food and shelter, and interact with people who were unable to communicate with him.
Ironically, he seemed to understand certain sorts of symbolic processes, such as performative identity. Schaller says he apparently understood, for example, ‘macho behavior’ because he ‘could see that.’ But other sorts of processes – she says things Ildefonso ‘couldn’t see’ – they remained a mystery; she offers ‘history’ and immigration patrols in the US as two examples. In fact, of course, the division is not really visible-invisible (after all, border police are quite visible when they arrest a person), nor is it symbolic-non-symbolic (macho behaviour, after all, is a symbolically rich performance). Rather,
Ildefonso’s difficulties and his successful abilities suggest to me that our own category of ‘symbol’ glosses cognitive capacities that are not all identically difficult, nor are they all dependent upon either shared symbol or language. That is, our concept of ‘symbol’ may, in fact, blind us to the very divisions that Ildefonso’s disability sketches out; not all symbols are equally symbolic, we might say. The degree of arbitrariness, for example, or the hierarchical nature of some symbols — premised on other symbols — might make them particularly opaque to the language-less. . . . "
Ezequiel A. Di Paolo1,2,3*
Marc D. Hauser1*, 
Charles Yang2, 
Jeffrey Watumull6, 
