Charles E. Osgood (Lectures on Language Performance. Springer-Verlag. 1980. p. 25) describes the following phenomenon:semantic satiation - where rapid seeing/saying repetition of a word, like canoe-canoe-canoe... produces a loss of meaningfulness, but repetition of a nonsense overt response having the same shape, nuka-nuka-nuka...does not.The phenomenon was discovered by Titchner and his co-workers (see http://www.soc.hawaii.edu/leonj/499s99/yamauchi/semantic.htm#further> and has aroused intense curiosity among a few; I am one. In addition, I have been interested for some time in the notion of "inner speech," noting that a number of philosophers, such as Wilfred Sellars, have made use of this notion. I would like to record the results of a couple of "thought experiments" combining these two phenomena.
1. In my own case, if I repeat the word 'canoe' in the way Osgood suggests, BUT do it subvocally, then I fail to achieve semantic satiation.
2. I have been able under special circumstances to satiate the meaning of entire sentences, but not subvocally.
3. I have observed that I can repeat (subvocally) sentences more rapidly than words.
4. I have, further, observed that the number of "speech errors" decreases during subvocalization.
It is perhaps worth mentioning that the failure of satiation may suggest that inner speech is in some sense acoustically based; rather than based on articulatory gestures. This may have implications for the debates surrounding acoustic (Fant G.) vs. motor theories of speech (Liberman A. M.)
I do not believe these results have been recorded elsewhere. I noticed these facts many years ago. Chisholm thought they were interesting but suggested I submit them to a psychology journal. I was encouraged in my research by Jack Adams of the Dept. of Psychology at the University of Illinois at Urbana - a consummate experimental psychologist.
I believe these results pertain to certain aspects relevant to the "modular" approach to language advocated by Chomsky and others, notably Jerry Fodor. One conclusion I have arrived at, tentatively, is that subvocal speech is not merely withholding vocal speech. Note, also, that word morphology is preserved, suggesting satiation does not rely entirely on morphology. I am sure others have different experiences; some may fail to arrive at these results. Still, I have and a number of others.
[Added Jan. 25, 2009] In addition to the above I now add a couple of other observations. [
5. Over a period of two years, I repeated three sentences, OUTWARDLY, in the form of a dialogue. "Hi X. How are you doing? Fine." After about 18 months of doing this, roughly, twice a day, all three were satiated in meaning. No experiment was performed as a control; that is, I did not repeat three sentences, or one - even - without the dialogue form. Still, the satiating rehearsals were discontinuous over a long period of time.
6. The time it takes to vocalize inwardly is not a function of the the complexity of the word morphology. In other words, if I rehearse 'canoe', inwardly, it takes - about the same time (about a second) as it take to rehearse more complex words.
7. The speech errors which disappear in inner speech are grammatical, but there are other forms of error that disappear. More on this later.
8. Notice that the "voice" of inner speech is that of the overt speech.
[ADDED JULY 16, 2009] I notice that I have nowhere in this brief note made mention of one technique for increasing the rate of subvocalized single word repetition. Here is how to do it. Begin to tap your finger at the rate of about once per second. This is, roughly, the rate at which I am able to subvocalize, virtually, any word regardless of morphology. Now increase the rate of the finger tapping. As you do this try to increase the rate of subvocalization, accordingly. I have found that in my case this increases the rate at which subvocalization can take place substantially. What I take from this is that subvocalization is closely related to motor action; but there is a fixed value associated with subvocalization when no motor action is in the offing. It would be interesting to perform the following experiment: examine the rate of subvocalization over varying complexity of word morphology. I doubt it makes a difference. If it did this would be significant, since this does not enter into the values in the "null case."