Which, if any, semantic properties would the utterances of a community of language users have, even if we assumed that the language users had no internal semantic states? My answer will come in multiple posts. Note that by ‘semantic properties’ I mean things like reference, truth, aboutness, and usability-in-an-inference. I will ultimately argue, with a couple of caveats, that their expressions would have a full suite of semantic properties.
In this, the first post in the series, I summarize, defend, and clarify the Standard View of the relationship between the semantic properties of internal states and public linguistic expressions. I’d be interested in comments, as these are ideas I’m slowly developing, and injections of criticism at this early juncture would be most welcome.
The Standard View
Since the welcome demise of behaviorism and interpretivism, the Standard View of the relationship between language and thought is that the meaning of public linguistic expressions is parasitic on the content of representational states in the mind of the speaker (e.g., Searle, Churchlands, Fodor; Robert Brandom is a notable contemporary exception). That is, the semantic properties of public linguistic expressions are inherited from the semantic properties of the internal vehicles of content. If a public linguistic expression were not generated, in the right way, by internal vehicles with semantic properties, that statement would be no more intrinsically meaningful than the shape ‘Snow is white’ that was accidentally carved into the sand by an ant looking for food.
Note that the Standard View isn’t necessarily internalist. It is a theory of how one vehicle relates to another, and how their corresponding contents relate to one another. It does not imply that the contents of those internal vehicles are fixed solely by events in the brain.
Let’s consider four possible arguments for the Standard View. One silly argument would be that for any system with meaningful states, those meanings must derive from some other meaningful states. This argument invites a regress, but it is something like what the interpretivists hold. Thankfully, the focus in philosophy of mind has shifted to providing an account of semantic properties that doesn’t advert to other semantic properties. This is the essential project of naturalizing cognition, or trying to “bake a mental cake using only physical yeast and flour” (Dretske, Knowledge and the Flow of Information, xi).
It could be argued that the meaning of linguistic expressions is derivative because they are public. That is, the first argument is really an abortive attempt to claim that publicly observable items, if meaningful, must derive their meaning from some other unobserved source of intrinsic semantic properties. This argument, as stated, is just another straw man. Nobody wants to make the incredible claim that the vehicles that carry semantic contents cannot be directly observable (for instance, this would make the vehicles depend on technology we use to extend our observation base). Perhaps this argument could be construed as a poorly articulated version of the following (third) argument for the Standard View.
A third, and more interesting, argument for the Standard View is that the semantic properties of all representational states derive from the semantic properties of conscious states. Perhaps conscious contents are the only type of nonderived content (as argued by Uriah Kriegel in The Primacy of Narrow Content). From this perspective, since internal brain states are the vehicle of consciousness, the Standard View seems reasonable: the intrinsic contents of conscious states percolate forth into certain public expressions. I will have more to say about this view in a later post.
The fourth, mainstream, argument for the Standard View is that it is the best psychological theory of the relationship between the semantic properties of thoughts and linguistic expressions. That is, the Standard View is a (contingently) true theory of how we happen to operate. There are multiple lines of evidence which can be marshaled, of which I’ll mention two. For one, we can express our thoughts more or less accurately. We often say things like “I said X, but Y would more accurately reflect my thinking”. We also often disambiguate what we say by clarifying our thoughts (e.g., “When I said I went to the bank I meant the bank of the river”). Second, neuropsychological patients show quite specific patterns of semantic deficits (e.g., the inability to comprehend or generate claims about animals). Conversely, someone with paralyzed vocal chords could still have the intention of speaking, a semantically rich mental life that they unfortunately can’t express (and if they knew sign language or could write they would be able to do so).
I take the fourth suite of arguments to be fairly compelling, so people are right to endorse the Standard View. But note that the Standard View, then, can’t be that it is necessary that public linguistic expressions have semantic properties in virtue of the semantic properties of internal states. Rather, this is a contingent fact about human communications systems. This fact of contingency will become much more important in future posts.
Eric,
The following article recently appeared in Nous:
Speaks, J. (2006). Is mental content prior to linguistic meaning? Nous, 40, 428-467.
In making the case for mental content first, what about the idea that non-linguistic animals appear to think?
An “emotor” theory (“e” for emotive,”motor” for active) of language developed from recent evidence in motor neuroscience with “mirror neurons” playing a important role (Rizzolatt and Arbib 1998), could explain why and how internal states are related to public linguistic expressions given them their semantic content and resolving the “contingency problem”: how linguistic expressions have semantic properties in virtue of the semantic properties of internal states?; and hereby prolonging the cogent appeal of the standard view.
It seems to me that the fourth suite of arguments doesn’t show that the semantic properties of public language are (even contingently) parasitic on mental semantic properties. At most, it shows that both types of semantic properties exist. For example, the fact that we can express our thoughts more or less accurately is perfectly consistent with the view that the mental and linguistic semantic properties are relatively independent. (An example of a theory that maintains such relative independence, at least in some cases, is Ruth Millikan’s.)
Thanks for the reference.
An excellent point about nonlinguistic animals, which I should have included.
I am not familiar with this aspect of Millikan’s work, and without details I can’t say anything specifically about her arguments.
Note don’t think any of the evidence I cited is conclusive, but taken as a whole, the data tend to mutually reinforce the Standard View, perhaps providing grounds for an abductive inference. I left out a lot of possible evidence, and Kenneth rightly pointed out one line. I’d be interested in others’ thoughts about the sorts of evidence that are taken as evidence for and against the Standard View. I will also try to be more thorough with that evidence, as I assume most people don’t need much convincing.
I realize this is a huge topic, and I have not addressed specific claims by people like Brandom. This is partly because I am not primarily interested in defending the SV. As I said in the intro to the post, I will argue that linguistic semantic properties can exist in the absence of internal semantic states (even though in our case it doesn’t work that way), and explore implications of this fact that might be philosophically interesting. It could be that I don’t go far enough, that someone like Millikan is even further removed from the SV, and perhaps closer to the truth.
Thanks for the comments, Dan. Unfortunately, I am not familiar with this aspect of Millikan’s work, so I can’t say anything specific about her arguments.
Note I don’t think that any single bit of evidence I cited is conclusive, but taken as a whole, the data tend to mutually reinforce the Standard View, perhaps providing grounds for an abductive inference. I left out a lot of possible evidence, and Kenneth rightly pointed out one line. I’d be interested in what others’ take as good evidence (or arguments) for and against the Standard View.
Frankly I wasn’t all that thorough in marshalling evidence. This is partly because I am not primarily interested in defending the SV. As I said in the intro to the post, I will argue that linguistic semantic properties can exist in the absence of internal semantic states (even though in our case it doesn’t work that way), and explore implications of this fact that might be philosophically interesting.
It could be that I don’t go far enough, that someone like Millikan is even further removed from the SV, and perhaps closer to the truth.
Here is another question that, to my mind at least, touches on Dan’s comment. What does the natural language first approach say about the idea that we can make visual discriminations of color that exceed those in our natural language? (There is kind of a mismatch between the linguistic and the cognitive.) Is it that language comes first, then thought and thought “expands upon” linguistic discriminations? Maybe I’ve answered my own question.
Even monkeys can make finer color discriminations than we do in natural language.
I think one of Brandom’s big weaknesses is that he has little (if anything) to say about animals that clearly do a lot of information processing without language, animals which (demonstrably) have internal information bearing states that they use to guide behavior. Millikan, on the other hand, is much more ethologically sensitive. She clearly acknowledges the importance of internal intentional states in the absence of language.
A couple of other lines of evidence.
1. The whole suite of studies on basic knowledge of number, animate/animate categories, etc. in prelinguistic children.
One particularly brilliant study, by Jean Mandler at UCSD, cleverly utilized the skills of such children (they are hard to work with on tasks that work fine with adults). Such kids are very good imitators. To test the thesis that their early categories are purely perceptual, she did things like the following.
Play with a toy airplane. Then, give the child two objects, one a toy bird that looks like the toy airlplane (the plane had a face painted on it), and the other a toy helicopter that looks very different from the plane and bird. Statistically, the children were much more likely to play with the helicopter than the bird! This suprising result suggests that prelinguistic children are already using knowledge of general categories to guide their behavior, rather than mere perceptual similarities (as Quine hypothesized).
I don’t know what someone like Brandom would make of such results.
Now, it is clearly a separate question whether learning a language is merely mapping the prelinguistic semantic structures to the public ones (the view favored by modern linguistics)
The next lines of evidence address it, but I am not as familiar with that literature.
2. The first expressions of children are always one-word utterances (unless the data have changed since I last studied this). But they don’t use them to refer to individual things, but seemingly to express more full propositional contents (e.g., ‘Milk!’, while reaching for the bottle seems to express a desire for milk).
This at least suggests that language has a primarily expressive role: it allows us to make public (to express) what is going on in our heads.
As children acquire more nuanced linguistic competence (increased use of grammar and vocabulary expansions) at about a year of age, are they merely learning to express what they already know, learning the right translation rules from the inside to the outside, so to speak?
In favor of this, there seems to be a period when they realize that they can use individual words for individual things (there are whole books devoted to teaching them naming skills), things which they could already individuate perceptually and probably conceptually (based on the Mandler-type studies). At the same time, they come to use terms in the correct order (grammar), so they can individuate ‘the dog on the car’ from ‘the car on the dog’, things they could presumably individuate cognitively before such grammatical competence.
At any rate, even if the standard, expressive, view of language is wrong, it is certainly understandable that it became standard. My arguments might point to some concrete limitations of the Standard View.
While certainly no expert, this article was very interesting. I think that I would agree with the Standard View as a psychological theory between thoughts and linguistic expressions. The author’s note to neuroscience is also interesting, and teases to learn more of the relationship. More detail would be nice, additional reference appreciated. Moreover, while I fully agree with the standard view, I do not entirely reject the others, particularly the second view that meaning is derived from public use. The social element is critical. For example, if I were to make a typo and use poor grammar on my first blog at “Brains,” though it might be clearer if I used better grammar, the meaning is certainly clear to me as the author. However, the meaning might not be evident for other readers as grammar is socially constructed and learned so that we might communicate. It is also the “public,” social element that would make me self-conscious to either make better use of accepted grammar, or to be more nervous and make more mistakes.
Acknowledging that the article concerns primarily verbal speech, the relationship between reading and writing is nonetheless relevant (and poignant to consider with those commentators that discussed learning theory and child development). In much the same way, I do not entirely reject the argument for consciousness, as it seems to me that semantics and the learning of language is part of the developmental consciousness, particularly in relation to the child. Furthermore, I do not reject the first argument of the relationship between semantic meanings derived from other systems of meaning (rather, states of meaning as the article suggests). To expand upon the learning relationship, learning to speak, learning to write, and learning to read are all interrelated activities as states of meaning that develop one’s cognizant awareness and consciousness. The activities are socially constructed, enabling social development and functioning. Hence, they seem interrelated, but the fourth clearly provides best explanation alone.
Here’s a shameless plug for my favourite theory: I wonder if you have had a chance to look at this“>https://www.bbsonline.org/Preprints/Jackendoff-07252002/Referees/”>this Behavior and Brain Sciences article? I guess it still counts as a deviant version of the Standard View? I’m not sure. In any case, it offers a really different conception of semantics from that typically entertained by (non-linguist) philosophers. There’s a lot less emphasis on truth conditions and reference. The rigour is still there (you just need a multi-sorted predicate calculus). Most importantly, unlike some old-fashioned philosophers, the author is at pains to make sure that his work can integrate (and be integrated with) results from cognitive psychology and neuroscience.
Can anybody commenting right here refer me to an alternate source for further research. This blog has some info but I would like a dependable source for more info.
I went over this internet site and I believe you have a lot of wonderful info, saved to fav (:.
Some really nice and useful info on this site, too I think the layout has got wonderful features.
Perfectly indited articles, Really enjoyed looking at.
I like this blog so much, bookmarked.
Some genuinely nice stuff on this site, I love it.