1. Carruthers's proposals are at least problematic: How does an account based solely on domain-specific modules and LFs deal with the complexities of “language production,” for which it has been necessary to postulate non-verbal processes such as “macroplanning” and “microplanning”? (Molina 2002). On the other hand, how does this account deal with the fact that we can have bare or wordless concepts (i.e., concepts that do not have a word associated with them), such as “the pathetic strands of hair that some men drape carefully but ineffectively over their bald spots” (Dennett 1998, p. 286) or “the moist residue left on a window after a dog presses its nose to it” (Murphy 2003, p. 389)?

2. For Jackendoff's concept of consciousness see Jackendoff (1987; 1997, Ch. 8).

3. I am, however, somewhat uncomfortable with the idea that in language production, “feedback and attention [are] not possible until there is a conscious phonological structure available” (Jackendoff 1996b, p. 205). This is because it is stated that in language production, besides being capable of monitoring the phonology, syntax, and semantics of the sentences that reach our inner speech, it also appears to be possible to monitor the construction of the preverbal message, for which no overt conscious clues are still available. In other words, it appears that the speaker can directly monitor the preverbal messages he is preparing to express, and he may reject a message before its formulation has started. As Levelt puts it:

What kind of “unconscious” monitoring would this be? Would it be part of what could be called the “dynamic of thought”?