Language, Consciousness, and Self-Reference
W. Zachary Wolff, 1999

Philosophers and psychologists have posited many theories about the nature of thought language. To a lay person it may appear that thoughts take the forms of words, the inner monologue heard every day. Developments in linguistics and psycholinguistics have shown that this is not the case. In a series of writings over the course of more than a decade, linguist and cognitive scientist Ray Jackendoff has developed a theory of the nature of our thoughts and our language as distinct but related entities. This short paper will outline the motivations for and merits of such a theory.

I would first like to motivate the idea that language and thought are not identical. This first requires a definition of thought that may not be the most intuitive. I will use the word thought to refer to all mental processing, conscious and unconscious. When referring to conscious thought I will make a point of noting it as such with a word such as consciousness or awareness. First evidence for the separation of thought and language comes from translation (Jackendoff, 1996). In translation, one attempts to change the language of a text while still expressing the same thoughts. This would not be possible if language were thought. Even within a language it seems that one thought can be represented by more than one linguistic expression, as in the case of active and passive sentences expressing the same situation (Jackendoff 1994).

Another way of illustrating this point is to look for cases of thought without language. One example comes from the fact that many thoughts are expressed in a non-linguistics manner, or are difficult to express at all (Jackendoff, 1996). While Picasso and Beethoven may not be able to verbalize the meanings of their works, they certainly seem to reflect high level thought. From a more pedestrian point of view, if language were thought, even the most verbally challenged among us would never experience the phenomenon of knowing what we want to say, while being unable to put it into words. Similarly, many animals, in particular primates, seem to exhibit complex reasoning without language (Jackendoff 1996).

Besides these simple arguments for the separation of language and thought, there exists real psycholinguistic evidence. In a series of tests involving in memory in which the subjects were not told that it was their memory being tested (a crucial component of the method), subjects remembered hearing or reading sentences with identical meaning but different linguistic structure than those actually presented (Stevenson 1993). At the level of thought, language was not stored, only meaning. If this meaning is not represented in language, how then, is it represented?

Jackendoff proposes a theory he calls conceptual structures. While this is not identical to Jerry Fodor's idea of a language of thought, the two can certainly be compared. Both Jackendoff (1987) and Fodor (1987) use a model in which thought is compositional. Thus, the thought p is contained within the thought p&q. This characteristic seems to have profound philosophical implications, but I have been unable to grasp them. For a survey of and reply to these implications see Fodor (1994).

On a more practical level, conceptual structures are the building blocks of thought. Critical to Jackendoff's theory, conceptual structures (henceforth CS) are not available to conscious introspection. CS are the processing units of the mind which create consciousness. This relation of unconscious conceptual structure in a larger theory of the mind is shown is in this diagram from Jackendoff (1996):

In this diagram, only the shaded area is available to consciousness. The outer ring of this diagram represents our link with the external world. We are not conscious of this first level of knowledge. For example, we have no conscious awareness of our visual system calculating distance relations, or our auditory system transducing pressure waves into sound. Nor are we conscious of the inner circle of our thoughts. For example, we have no conscious awareness of the complex processing our minds must perform to parse a sentence (Akmajian et. al. 1995). From my experience I would propose one significant refinement to this diagram. While it is often true that information flows in the lines shown, from sensory arrays, to consciousness, to our inner mind and back, it seems that another type of flow is possible, from the inner circle to the outer ring, bypassing the conscious middle. This happens in simple activities such as when our unconscious spatial representations cause us to unconsciously move our feet in exactly the appropriate manner on rough terrain.

As shown in the diagram, Jackendoff's conceptual structures lie within this inner circle. While we have come a long way by distinguishing thought from language and locating it within a larger structure of the mind, we still have not really addressed the original question of how thought is represented. Jackendoff quickly and efficiently rejects many of the same notions we considered and rejected in class: mental pictures, feature lists, necessary and sufficient conditions, and prototypes. Instead he introduces a theory of categorization that seems rather plausible: preference weighting (Jackendoff 1987). The idea works something like this: Each attribute of something infers a preference of a certain categorization. Each preference has a weight attached to it based both on the relevance of the particular attribute to the overall CS and the degree to which the attribute is expressed. I find this theory rather convincing, particularly in that it closely parallels connectionist architectures of the mind with their weights and nodes. One could add an elaboration to the theory whereby each attribute does not simply have a preferred interpretation, but a series of variously preferred (or discouraged) interpretations ranging in weight from -1 to 1. Of course this fact could also be viewed as a liability as many cognitive scientists are in staunch opposition to connectionism.

Now that we have a reasonable (although admittedly far from complete) picture of what thought is and where it resides in the mind, we will examine how thought interacts with language. Jackendoff (1996) makes the bold claim that it is language that makes thought available to consciousness. This explains rather well the popular misconception we dealt with initially, that thought is language. If thought is only conscious when it comes in the form of language, that is if the only way we experience thought is through language, it is perfectly sensible that we would consider them the same. Of course the evidence and arguments in paragraphs 2-4 shows the flaws in this idea. Of course some of this evidence also shows that the bold claim I have attributed to Jackendoff is not entirely true; there are other, severely limited, forms of consciousness. Naturally, Jackendoff acknowledges this.

The position is still very strong with this minor refinement. Consider for a moment the amount of your consciousness that is portrayed in linguistic imagery, a voice in your head. Imagine your life without that. It seems at first that it might be very near to the life of a zombie, but a realistic interpretation reveals that it might actually be very much like the life of monkey. Life would consist only of a continuum of sensory experience interrupted by sudden inexplicable urges, similar to the nonlinguistic urges we experience now, such as hunger. Many of our decision making processes would be the same, but we would not be aware of them.

Besides salvation from this tiresome and rather meaningless existence, there are other ways in which language effects our thoughts. Perhaps the most important rests on a claim Jackendoff makes which I take objection to. Jackendoff (1996) claims that we can only direct our attention to that which we are conscious of. I do not have the counter-data available, but I believe it is true that the eyes, and thereby attention, are drawn to moving objects in the periphery of vision which were outside of consciousness. Jackendoff also overlooks the fact that any connection between intention and consciousness need not be causal. It could be the case that the standards for consciousness of a CS are less than those for attention, such that anything that draws attention will by necessity have already met the requirements for consciousness. I suppose a definition of attention would help clarify this issue, but Jackendoff does not offer one and I will not presume to provide one, as I want to use the word as he meant it. In the end, it seems to at least true that being conscious of something makes it much easier to direct ones attention to it.

What are the ramifications of this? Attention seems to be the mind's way of focusing resources on a particular problem. Attention gets focused on that which is conscious. Language allows the underlying CS to enter consciousness. Therefore, language allows attention to be directed to our underlying CS. Language is what allows metaconsciousness, introspection, and self-reference. This conclusion is huge. These seem to be very closely aligned with the qualities that make us human.

To sum up, we have shown that thought and language are not the same thing. Thought is built up of unconscious conceptual structures organized in a compositional manner. Language does, however, influence thought. In fact, language adds an entire new dimension to thought, it makes it reflexive, recursive, an endless loop that allows us to think about thinking. The complex decision making strategies we have already can be used not just to make decisions about actions, but to make decisions about which decision making strategies are best. What we identify as the human intellect may largely be provided by language in that it makes our thoughts conscious.

References

Akmajian, Adrian; Demers, Richard A.; Farmer, Ann K.; Harnish, Robert M (1995) Linguistics: An Introduction to Language and Communication 4th Edition. MIT Press: Cambridge, Mass

Braine, Martin D. S. and O'Brien, David P. editors (1988) Mental Logic. Lawrence Earlbaum Associates: Mahwah, New Jersey.

Case, Robbie (1991) The Mind's Staircase: Exploring the Conceptual Underpinnings of Children's Thought and Knowledge. Lawrence Earlbaum Associates: Mahwah, New Jersey.

Fetzer, James H. (1990) Epistemology and Cognition. Kluwer Academic Publishers: Boston

Fodor, Jerry A. (1987) Psychosemantics. MIT Press: Cambridge, Mass

Fodor, Jerry A. (1994) The Elm and the Expert: Mentalese and its Semantics. MIT Press: Cambridge, Mass.

Haugeland, John editor (1997) Mind Design II. MIT Press: Cambridge Mass

Jackendoff, Ray S. (1987) Consciousness and the Computational Mind. MIT Press: Cambridge, Mass

Jackendoff, Ray S. (1992) Languages of the Mind. MIT Press: Cambridge, Mass

Jackendoff, Ray S. (1994) 'Language as a Window on Thought' in Patterns in the Mind. Basic Books: New York.

Jackendoff, Ray S. (1996) The Architecture of the Language Faculty. MIT Press: Cambridge, Mass.

Stevenson, Rosemary J. (1993) Language, Thought and Representation. John Wiley & Sons: New York.