In a commendable and sincere self-portrait, linguist Gillian Ramchand explains what it means for her to be a generativist linguist. Among the many things that she thinks you can accept while being Chomskyan is having no reason to think that Jackendoff is crazy. I completely agree. Contrary to what other (quite orthodox) generativists seem to think, I believe that Ray Jackendoff, besides being as generativist as anyone, has done (and is doing) a great service to the cause of building an authentic cognitive science of language, which is the goal of generativism. Incidentally, the same can be said of Steven Pinker, so I do not tremble when I write that Jackendoff and Pinker are as Chomskyan as Chomsky himself, because the framework of Chomskyan linguistics undoubtedly transcends the flesh-and-blood pioneer genius that gives it its name.
But in proclaiming that Jackendoff is not crazy, I do not want to get lost in this minor plot of the sociology of language science, but to focus on Jackendoff’s enigmatic and attractive Unconscious Meaning Hypothesis (UMH, for short). The excuse for this is that I’ve recently re-read Jackendoff’s book A User’s Guide to Thought and Meaning, which recapitulates in a pleasant and (more) intelligible way his more than 30 years of publications on this issue. In fact, the UMH already appears in his essential Consciousness and the Computational Mind (1987), but there (or so it seemed to me at the time) the hypothesis was formulated in a darker and less convincing way (or maybe I was too immature to understand it).
A User’s Guide to Thought and Meaning is an excellent example of how the adoption of the cognitive perspective can represent an advance in the study of language and the mind with respect to the ordinary perspective. In fact, I think that the ordinary perspective, ironically, is the usual one in so-called cognitive linguistics, but that issue deserves a separate discussion.
As its name suggests, the UMH assumes that meaning is unconscious, that is, it is hidden from awareness. There can be no clearer explanation of why it is so easy for us to understand words yet so difficult to explain what they mean.
Surprising as this proposal may seem, the most intriguing and original aspect of the UMH is, in my opinion, the role it gives to what Saussure called the ‘plane of the signifier’, that is, the phonological form of words and sentences: sound is a ‘handle’ so that we can experience the sensation of meaningfulness. In Jackendoff’s own words, quoting from A User’s Guide: “the meaning side of a sound-meaning pair is unconscious, except for producing the feeling that the attached piece of sound is meaningful” (p. 49).
Of course, the fact that meanings are hidden from awareness does not imply that they do not have relevant effects: thanks to them we can identify and categorize objects in the world, we can follow instructions, make inferences and issue statements that other people seem to understand. All of these processes, Jackendoff says, depend on “grasping” meanings. The important thing is that we grasp meanings even though we are not aware of them. The UMH has, then, three essential parts: (i) the ‘pronunciation’ perceived is conscious, (ii) the ‘pronunciation’ is accompanied by a conscious sensation of meaningfulness, and (iii) it is associated with an unconscious meaning (the concept or thought that it expresses).
What we have in our heads are unconscious concepts and thoughts; when some of these concepts and thoughts can be associated with a phonological form, then these concepts or thoughts are the meaning of that phonological form. Therefore, there may be concepts or thoughts that are not the meaning of any phonological form (although we cannot be aware of them), and there may be phonological forms that have no meaning (because they do not connect to any concept or thought).
What Jackendoff calls thought is actually the sum of the concepts and the computational system or syntax that structures them. From this point of view, thought is the same as the ‘internal language’ of the well-known Chomskyan model (formed by the conceptual-intentional system plus the computational system, but disconnected from the sensorimotor system). What we usually call language (and which we group in historical languages such as Spanish or Chinese) necessarily implies a connection of that ‘internal language’ with the systems of externalization (pronunciation, for simplicity). The conventional story is that the externalization of the internal language is at the service of communication. But Jackendoff goes further and suggests that the linking of meanings to sounds not only improves our ability to communicate, but also provides the means for the conscious mind to operate with internal derivations that, otherwise, could only be unconscious. In other words, that phonology is at the base of consciousness and, therefore, of so-called rational thought.
Jackendoff may not be crazy, but many might think so. It seems to me that he is right, and I will try to explain why (although it would be better if you read the book).
As I said, the most striking and attractive part of Jackendoff’s hypothesis is the role that the model gives to the phonological form. Meaning is the basis of our knowledge, but it is unconscious. We are aware of the phonological form of words and sentences, but not of their meaning. Thus, according to Jackendoff, the old intuition that thinking consciously is the same as talking to oneself is correct, since the only way that meanings have access to consciousness is through a handle, typically, the pronunciation. Of course, we can also be conscious through visual images or other perceptions, but the logic is the same: the conceptual and spatial structure that underlies all thought (and which is the basis of our cognition) is totally unconscious, it is hidden in the brain except for the episodic and transitory flashes that we have of them when we use those handles (phonological forms, visual images, tactile sensations, etc.) to experience our own inner discourse.
Of course, this does not mean that our thoughts are exactly the phrases we hear internally. Recall that the phonological form of our inner discourse is only the handle of the cognitive structures confined in our unconscious mind. According to Jackendoff, when we perceive a meaningful phonological form (because someone pronounces it or because it resonates by itself in our brain) we cannot perceive the meaning directly (remember that it is unconscious!). What we are aware of is that such a phonological form is meaningful, but not of the meaning itself: “meaning is unconscious, aside from the simple feeling of meaningfulness” (p. 111). Or in other words: when someone tells us something and we understand what it means, we are not really aware of the meaning, but only that those sounds have meaning.
What we call a ‘conscious thought’ actually has three components: the verbal image, the sense that it is meaningful, and the meaning itself. But, according to Jackendoff, only the first two components access consciousness. The third, the meaning associated with pronunciation, remains unconscious, although it is the one that does all the heavy work in internal thinking, such as establishing inferences, collating assumptions and drawing reference relationships, that is, providing an interpretation of what we are told, and of the world.
According to this model, then, not all human thought is conscious. In fact, almost all the use we make in our life of the internal language, that is, the language of thought (even when we think we are reasoning) belongs to unconscious thought, and this type of thought does not need handholds. Only the thought we are aware of, that is, the one we are capable of experiencing, requires phonological handholds (or other perceptual modalities of them).
It is true that we tend to think that the internal pronunciation is the thought itself, but this is a simplification motivated precisely by the feeling of meaningfulness, of which we are aware. A string of random sounds like clepnodra, for example, does not produce that feeling. Another common error that the UMH clarifies is the confusion between thought and consciousness. Thus, a being without language (understood as internal language plus phonology) could think, and quite well. What it could not do is to be aware of its thoughts. In fact, human beings make habitual and very efficient use of unconscious thinking. The difference is that, compared to other higher primates, not only can we have more complex thoughts (thanks to the computational system we call syntax), but we can also be aware that we have them (thanks to the connection of the computational and conceptual systems with the sensorimotor system).
The same works for vision. When we are consciously thinking, we ‘hear’ the words within the mind (in fact, this perception is the only thing that we are aware of), and we know that nobody is talking to us because there is no connection between the auditory stimuli processed by the ear and the ‘voice’ in our head. When we see something, like the computer screen, we are aware of the visual image and we know that it is out there, in front of us, because there is a connection between the visual image within our brain and the stimuli received in the retina. When there is no such connection, we are imagining a screen (or we have a hallucination, which would be explained as a malfunction of the detection of the connection between the visual image in the brain and the visual stimuli from the retina).
We have seen that, according to Jackendoff’s UMH, the phonological form is the handle that makes consciousness possible. But note that language per se would not be a requirement for consciousness, since also visual images and tactile and olfactory stimuli can function as handles of unconscious internal concepts. In fact, we have no reason to think that we are the only organisms with some kind of self-awareness. Something different is what you can be conscious of.
The human computational system (which is also unconscious) seems to make a difference with respect to its analogs in other species in terms of the complexity of thought (both conscious and unconscious). As Chomsky and followers have shown since the late 1950s, the computational system, with its ability to recursively produce unlimited derivations and to combine all kinds of conceptual elements together, produces a conceptual structure (now using Jackendoff’s typical terminology) that distinguishes us cognitively from other organisms. Everything that we know, we know because the brain has built it, and syntax is the main tool our brain uses. To put it simply: even if we did not have consciousness, we would also be smarter (although we would not know it).
Therefore, such a complex conceptual structure produced by syntax does not imply or necessarily presuppose consciousness. In fact, if Jackendoff is right and thought (meaning) is really unconscious, human consciousness is not a consequence of the internal language or of the sophistication of our conceptual structure, but is a consequence of the connection of our internal language with arbitrary chains of phonemes, which can access consciousness. We cannot represent concepts such as ‘despair’, ‘jealousy’, ‘entropy’ or ‘past’ with a visual image, but we can create them with the internal language and make them present in our conscious experience through sounds associated with them (or with spellings, as I have just done right now).
Yet there is more. We human beings are not only aware of ourselves, but, thanks to language, we are also aware of our thoughts, that is, we are aware that we have them, even though they are in themselves unconscious. Thanks to their linguistic handles, we can store and retrieve our own unconscious thoughts. And here, according to Jackendoff, is the basis of our most notorious attribute as a species: rational thought, that is, conscious thought.
According to Jackendoff, research in cognitive psychology distinguishes between two modes of reasoning, System 1 and System 2. System 1 is fast, automatic and unconscious (it’s what we commonly call intuitive or unconscious thinking), while System 2 is slow, effortful, linear and conscious (which is what we usually identify with rational thinking). Much of the philosophical tradition has insisted that System 2 is specifically human and it has been related to language. Jackendoff’s model does not deny this, but explains it much more coherently. What our author proposes is that System 2 is not different from System 1 but, as he says, that “it ‘rides on top of’ System 1” (p. 214). System 2 would be thought that is linked to a cognitive correlate of consciousness, that is, a phonological form. Since the phonological form is linear, Jackendoff says, conscious thought is linear; since pronunciation is slow (as compared to the speed of thought), conscious thought is slow. And since thought is unconscious, we can only have conscious access to it if it has a conscious handle, the phonological form. Considering that only humans have language, Jackendoff concludes that only humans have rational thinking (System 2). Therefore, System 2 basically corresponds to System 1 plus language: “rationality is intuition enhanced by language” (p. 243).
But at this point we should be very careful with terminology. It would be better to say that System 2 is equivalent to System 1 plus language externalization. The idea is that language (internal language) is already part of System 1. Jackendoff suggests that if a chimpanzee had language it would not be like a human, since human System 1 is also very different from chimpanzee System 1 (it is “more sophisticated” in Jackendoff’s terms). But then we lose the interesting intuition that language (the internal language) is also the cause of the difference between human and chimpanzee Systems 1. Jackendoff is using here ‘language’ as equivalent to a phonological dimension, but language is much more than that, since language (syntax more specifically) differentiates our System 1 from the System 1 of the rest of organisms capable of thought (perhaps making our conceptual structures ‘more sophisticated’). The computational system of human language (syntax) is also unconscious and is also at the service of the construction of thought, and not of communication.
Regarding the question of whether language arose in our ancestors as an enhancement of communication or as an enhancement of thought, Jackendoff concludes, surprisingly, that Chomsky’s minority stance (that the primary innovation in language evolution was structured thought) is erroneous. Jackendoff’s argument is based on the fact that in Chomsky’s model the connection of the ‘internal language’ with externalization systems is secondary or ancillary. Given that Jackendoff’s hypothesis implies that pronunciation is crucial to understanding the contribution of language to the development of rational thought, it would seem that the main contribution of language to thought could not be prior to the development of language externalization. But note that this conclusion seems to ignore the fact that rational thought (that is, conscious thought) is not the only thing that cognitively differentiates humans from their evolutionary relatives. As Jackendoff himself admits, the human System 1 is different from that of other species without the need to appeal to System 2. Therefore, there is nothing contradictory in assuming that part of language (the computational system) evolved at the service of thought (modifying qualitatively the human System 1), while another part of language (the connection with the externalization systems), evolved at the service of communication and, simultaneously (if Jackendoff’s theory is correct), could have a decisive effect on the development of conscious thinking, as a “huge side benefit”, as Jackendoff himself says (p. 222).
In any case, what language (as a whole) adds to the human System 1 to turn it into an (also human) System 2 is not computational complexity (which explains the sophistication of System 1), but enhanced access to consciousness. The possibility that language gives us to refer to our thoughts and manipulate them through their conscious correlates (phonological forms) makes us able to draw conclusions and obtain a kind of knowledge that we could not obtain only with unconscious thought. The wonderful thing is that this knowledge is still unconscious, even if it makes us human.
[Author’s English version of a blogpost published in Spanish (5 July, 2015) in zaragozalinguistica.wordpress.com]