Verbal Minds

Language and the Architecture of Cognition

Toni Gomila

• Publisher: Elsevier
• Year: 2011
• ISBN:
  9780123852014
• Topic: Languages & Linguistics
• Subtopic: Psycolinguistics
• Index: Languages & Linguistics

Ten years ago, the hegemonic idea was that language was a kind of independent module within the mind, a sort of "print-out" of whatever cognitive activity was taking place, but without any influence whatsoever in that activity. While this view is still held, evidence amassed in the last 10 years suggests another view of their inter-relationships, even though exactly which one is not clear yet, in part because of the lack of a unified view, and in part because of the inertia of the previous position, in part because all this evidence must be considered together. An increasing number of researchers are paying attention to the issues involved as the human language specificity may provide a clue to understand what makes humans "smart, " to account for the singularities of human cognition.

This book provides a comprehensive review of the multiple developments that have taken place in the last 10 years on the question of the relationships between language and thought and integrates them into a coherent framework. It will be relevant for anyone working in the sciences of languages.

• Synthesizes recent research
• Provides an integrated view of cognitive architecture
• Explains the relationships between language and thought

1. Introduction

Language as the Key Factor to Human Singularity

The issue of the influence of language on thought is introduced in the context of the search for the key to our uniqueness. The main features are flexibility and cognitive control. The hypothesis that these features have something to do with the fact that we have verbal minds deserves attention. The last decade has seen an explosion of empirical research in this direction. Our goal in this monograph, then, is triple: (a) to set apart the different ways that the relationship between language and cognition has been conceived; (b) to review the multiple evidence amassed in recent years on this relationship; and (c) to try to adjudicate which of the multiple ways to conceive of the relationship best accounts for the facts. We will argue for a dual architecture of the human mind, with the higher processes as distinctive of verbal minds.

Keywords

Cognitive architecture, verbal minds, cognitive control, flexibility, language

The natural human interest in self-understanding has traditionally pointed to language as the most distinctive human trait. Many other features are also uniquely human. Some are anatomical: a big brain in proportion to the body, a lack of a tail, and a larynx with special phonetic capabilities. Some traits involve distinctive activities: religion, mathematics, art, and sport, for example. However, when it comes to making sense of human singularity, it seems that many of those specificities are not basic, but they became possible through achievement of more basic ones, such as language.

Of the several features uniquely human, then, language has been most consistently chosen as the key to understanding the human mind and to providing the building blocks necessary for achieving other specificities in human cognition: abstract/propositional thought, recursivity, decoupling of current situation, creativity, and conscious control (Chomsky, 1988; Macphail, 1996). To put it in some distinguished scholars’ words: language is thought to be what makes us “smart” (Gentner, 2003; Spelke, 2003); or, at least, it is an important element of human intelligence, if not the only one (Premack, 2004). Maybe the influence of language depends in its turn on a more basic structural novelty that makes both human language and thought possible (Penn et al., 2007). Human cognition is characterized by its flexibility and creativity, which gives rise to, and is molded by, cultural diversity. Cultural diversity, in its turn, feeds back into cognitive diversity through the socialization process that takes place during the long period of human development. Language, as a symbolic system of communication and also of representation, is thought to play a critical role in making possible this interplay of individuality and sociality.

It is not so clear how language influences human cognition, however. The issue of the kind of role of language on human thinking—from which the cultural and behavioral novelties of human culture are thought to stem—is a polarizing one: while some people take it as obvious, others regard it as of marginal interest. For some thinkers, the constitutivists, the relationship is so intimate that thought is not even conceivable in nonlinguistic creatures: they view language as conceptually necessary for thought (Davidson, 1973 and Davidson, 1975; Dummett, 1981 and Dummett, 1989; McDowell, 1994). At the other extreme, the communicativists contend that language has nothing to do with thought whatsoever, beyond making it explicit (Fodor, 1975 and Fodor, 2008; Pinker, 1994 and Pinker, 2007). Of course, such extreme positions require a great deal of qualification. The first group is really only concerned with “propositional thought”—or the ability to entertain propositional contents—which is characterized by its truth conditions. Nonverbal creatures might be capable of simpler, referential thoughts, but given the conditions of content ascription, propositional thoughts are solely ascribable (by linguistic creatures) to linguistic creatures. The second group willingly accepts that language may be instrumental in the acquisition of many concepts, even most concepts (Devitt & Sterelny, 1987). Pinker, one of the spokesmen for this position, also concedes in passing that for one to be able to speak about reality, one needs to conceive of reality in terms of a particular language’s requirements for communicating contents through one’s speech (thus, languages differ in whether they require marking number, person, aspect, or voice) (Pinker, 1989, p. 360). They conceive of the main relationship between language and thought in the contrary direction, however: it is thought that conforms to language. Language is just the means for expressing thought, which is psychologically and semantically previous to language and is independent of how it is expressed. They adhere to a purely communicative view of language.

There has been a sort of pendulum dynamic in linguistics over the past 30 years or so. The communicative approach became hegemonic in the cognitive sciences in the eighties, but in the last decade there has been a lot of new evidence in support of the constitutivist approach. In 2011, it looks as though constitutivism is becoming mainstream. During the heyday of the communicative view (Gauker, 1992), the question of whether and how language might influence thought fell into disrepute. It was confronted with many central postulates of the cognitivist-computational approach that became dominant a strong nativism, an understanding of psychological processes as logical inference, a language-like view of mental representation, a modularist view of cognitive architecture, and an assumption of semantic–conceptual isomorphy. It became too difficult to fit linguistic effects on cognition into this general view of cognition, to the point that Pinker (1994) included a chapter with a necrological note on Whorf.

It is a well-known phenomenon in psychology, however, that the way in which a situation is linguistically described greatly influences whether it is attended, remembered, and valued. The effects of language on verbal tasks have been proven beyond doubt by the work of many, including that done by: Carmichael, Hogan and Walter (1932) on the effect of the lexical labeling of ambiguous pictures on memory; Glucksberg and Weisberg (1966) on problem solving; Loftus and Palmer (1974) and Schooler and Engstler-Schooler (1991) on explicit memory; Tversky and Kahneman (1981) and Kahneman and Tversky (1982) on the “framing effect” in decision making; Wickens (1972) on the influence of language on short-term memory; Barrett (2007) on emotion perception; and Styles (1994) on voluntary attention. How an experience is linguistically coded deeply influences how it is cognitively processed, and problem solving, in particular, benefits from linguistic formulation. The very diversity and sophistication of verbal tasks bear witness to the important role language plays in cognition.

The same realization can come from social life in: the feminist concern with sexist language; the diplomat’s care with choice of words; the publicity and propaganda efforts of public figures; and the general tendency to use euphemisms. It is easy to find many examples of such behaviors showing that how we describe a situation in linguistic terms has powerful, cognitive effects that may also determine our emotional reactions and valuations. The following anecdote provides an example: one woman said to another: “Thank goodness for the word ‘muffin.’ Otherwise, I’d be eating cake for breakfast every morning.” Similar effects of language have even led to some words being considered taboo and hence forbidden at times in history: advanced democratic societies and learned associations included (Chamizo, 2009). For a simple example: the British Sociological Association, in its “Guidelines for antisexist language” (April 2004), banned such words as “disseminate” and “seminal.”

Of course, these examples amount to demonstrations of linguistic constitutivism. But they foster interest in the cognitive roles of language. After a period of disrepute, then, interest in the question of the relationship between language and thinking slowly returned, and it can be said that a cognitive view of language is currently fashionable, lively, and full of vitality. It is not easy to spot the stimulus, if there was one, of such intellectual changing of gears. Different disciplines have been involved: developmental, cultural, and comparative psychology; cognitive, linguistic, and evolutionary anthropology; cognitive linguistics; and philosophy of mind and language. Critical milestones in this resurgence were collective books such as: Gumperz and Levinson (1996) and Gentner and Goldin-Meadow (2000), which managed to put the discussion on firmer methodological ground and renewed theoretical approaches. The trend has given rise to a wealth of research in the last decade that deserves to be reviewed and synthesized, as we will try to do in this work. New approaches, new experimental paradigms, more stringent standards of evidence, and new ways to conceive of the relationship have been developed, so that it can be asserted that the debate has been moved to a new dimension.

Against the wealth of evidence that has been amassed in recent years, critics of the cognitive view of language tend to react in a paradoxical manner: they contend both that empirically demonstrated effects are trivial, and that they do not really support a cognitive role for language. If they were really trivial, then the shaping role of language would not even be up for discussion! But if they were really trivial, so much effort at experimental control to prove them would not have been required in the first place. In other words, a prerequisite to joining this debate is to show a proper respect for the empirical evidence, so painstakingly amassed. It required ingenuity in experimental design and the application of new statistical techniques, cross-cultural and comparative research, and long-term projects. If such a respect is achieved, the real issue, then, is one of superior explanation: which theory provides the best way to account for the empirical effects uncovered. It is then that the architecture of cognition occupies center stage: it is the source of explanatory concepts, basic processes, and levels of cognitive organization and mental representation. But there is not a single, universally agreed upon, cognitive architecture that can play a touchstone role. Thus, empirical evidence in this area—as in any other—can cast doubt on previous assumptions concerning cognitive architecture. Explanatory coherence requires a sort of cognitive equilibrium that pushes arguments both ways.

That coherence is why a proper account of the cognitive influence of language on human thinking also involves a discussion of cognitive architecture. In this regard, in the final chapter we will argue for a dual theory of thinking, as the best way to accommodate the evidence. This approach brings to the foreground the hypothesis that verbal minds are special because they are verbal, and that it is language that makes human cognition special: flexible, self-conscious, slow, and systematic. The “duality of mind” approach naturally accords with the idea that language has to do with what makes human minds dual in that way. Dual theories are committed to a view of “basic” cognition as independent of language, thus accounting for nonverbal thinking and for language acquisition processes, while being compatible with the hypothesis that higher order thinking comes about with language. I will propose a version consistent with an embodied cognitive science (Calvo & Gomila, 2008). The logical geography of cognitive explanation that it opens up, despite it coming short of forming a unified paradigm (Gomila & Calvo, 2008), provides for an easier accommodation of language as an organizational force of human cognition. In this work, however, I’ll avoid direct discussion of the debate on the ground level of cognition. In particular, I’ll take for granted that speaking of mental representations does not prejudge the outcome of this debate, assuming that postcognitivist cognitive science will also need to honor offline and internal state-mediated processes. I will argue, however, that a “language of thought,” as the representational medium that is to account for the systematicity and productivity of higher cognition, is not basic, but parasitic on natural language (Gomila, 2008 and Gomila, 2010a).

Of course, there is also a trend that focuses on differences among human groups, and finds in language a crucial element for such differences, in the tradition associated with Whorf. While we will also pay attention to the cognitive effects of speaking one language versus speaking another, our emphasis will be on the effects of “being verbal” versus not being verbal as the crucial aspect to take into account for a proper understanding of the “verbal mind” and its architecture. Linguistic differences, while important, seem not to be as divisive a factor among human minds as nineteenth-century Romanticism claimed, in reaction to the Enlightenment’s hierarchical views of human differences and Western supremacy. However, a lot of work has also focused on this dimension. I will follow Gentner & Goldin-Meadow’s way of labeling these two areas of research: “language as lens”, for the effects of speaking one language vs speaking another, and “language as tool kit”, for the shaping effects of language on thinking (Gentner & Goldin-Meadow, 2003). The former approach looks for cognitive differences due to linguistic differences; the latter looks for cognitive surpluses made possible by language.

Our project, then, can be seen as “the case for a role of language in human cognition.” Exactly which role will be proposed as a conclusion to our review, and it will depend on the reviewed evidence. The main dialectical rival, though, will be those views of human cognition that oppose the very possibility of such an influence and that tend to conceive of language as a sort of peripheral of the mind, as an extra “module” that we happen to have, without any remarkable consequence for the way the rest of alleged mental modules work. In its most extreme version, this view claims it is impossible for language to play any cognitive role: a bold contention that we will have to discuss from the start.

Our goal in this monograph, then, is triple-pronged: (a) to analyze the different ways the relationship between language and cognition has been conceived, (b) to review the evidence amassed in recent years on this relationship, and (c) to conclude which of the multiple ways to conceive of the relationship best accounts for the facts. I can already advance that it is not going to be an extreme or a radical conception, but that it will articulate how language makes possible some outstanding properties of human cognition. In addition, given the interdisciplinarity of the project, special attention will also be paid to methodological issues: the type of data required in these matters and how we can improve what we already have available.

2. Clearing the Ground

The hypothesis that language has a cognitive role in verbal minds has found serious opposition in mainstream Cognitive Science. Cognitivism, either in the form of Fodor’s rational nativism, or in the form of the massive modularity of Evolutionary Psychology, cannot accommodate such a role. The assumption that cognition is a matter of logical inference (computation) over logical formulae of an innate language of thought makes language a peripheral to the mind, decoupleable from the rest of the cognitive system. In order to properly assess the empirical evidence, then, such a basic set of assumptions has to be questioned in the first place. We will call into question the arguments for the language of thought, reply to the arguments from this standpoint against the very possibility for language to play a cognitive role; and will show that this perspective fails to provide a proper understanding of human development, which is the period when language acquisition takes place.

Keywords

Cognitivism, language of thought, massive modularity, computationalism, logicism, development, language acquisition, nativism, evolutionary psychology

Before considering the relationship between language and thought, a preliminary dialectical move is required: calling into question the “in principle” arguments that foreclose the very possibility of a cognitive conception of language. Those arguments are grounded in a view of the architecture of the mind that conceives of language as a set of extra modules, added to an already modular cognitive system, working in a language-like representational medium, the “language of thought” (Fodor, 1975 and Fodor, 1983; Shallice, 1988; Smith & Tsimpli, 1995). There are two main versions of such an approach. The first is the “rational nativism” of Fodor, which proposes an innate language of thought and a cognitive architecture of input/output modules, plus a holistic central system for which a cognitivist–computational approach fails (Fodor, 2001a and Fodor, 2008). The second is the “massive modularity” approach, which shares the computational view of mental processes, but hopes to block the holism of the central, cognitive ones, by splitting them into a series of computationally tractable cognitive modules (Barkow, Cosmides & Tooby, 1990; Carruthers, 2006; Pinker, 1997; Samuels, 2000; Sperber, 1996), at the cost of relaxing the notion of “module” to mean little more than a domain-specific system. Both camps agree that the conceptual primitives of the system cannot be learnt, but while Fodor argues that most concepts are primitive and hence innate (Fodor, 1975 and Fodor, 2008), Pinker believes that the primitive set is smaller and that most concepts are structured out of this set of primitives (Pinker, 2007).

According to this general approach, language is conceived as a “peripheral” to the mind, from which it is “decouplable.” Its evolutionary emergence—according to this approach—has had no effect whatsoever on the rest of our cognitive abilities. Such an approach is committed to the view that our thoughts would be the same even if we were not linguistic beings. Language is just a means of expressing—of communicating—these language-independent thoughts. In Fodor’s words: “English inherits its semantics from the contents of the beliefs, desires, intentions and so forth that it’s used to express, as per Grice and its followers. Or, if you prefer (as I think, on balance, I do), English has no semantics” (Fodor, 1998, p. 13). Or in Jackendoff’s words: “The terms semantic structure and conceptual structure denote the same level of representation” (Jackendoff, 1983, p. 24). The semantics of language is dependent on the conceptual contents spec...