Dan Silverman’s Critical Introduction to Phonology is full of things that I wish I’d seen explicitly stated all in one place long ago. He puts down informally and with the greatest of ease concepts and positions which you find little sustained support for elsewhere in the phonology literature. It’s a critical introduction, which means a lot, but it presents ‘real phonology’ that’s recognisable to serious phonologists in a way that is sensitive to real concerns and problems which otherwise don’t inform how phonology is taught.
Here’s his critique of introspection as a guide to what matters in the phonological system of a language. He’s started off with the example of how the nonword ‘blick’ somehow “sounds better” than the nonword ‘bnick.’ He says,
“Many phonologists – though not I – think that ‘blick’ is a possible word because it doesn’t violate any sound-sequencing constraints of English, except that it just happens to be missing, and so it feels okay. These phonologists propose that ‘bnick,’ by contrast, involves a genuine violation of an unconscious sound-sequencing constraint, and so it sounds awful to English ears. Such a constraint might strictly prohibit English words starting with the sound sequence ‘bn.'”
But, he says, this is a flawed approach to take.
“When I speak English, every sound substitution is always one word or another. [I’m not sure if this might mean ‘results in’ one word or another?] There are no relevant feelings on my part about whether the sound substitution is good or not. They are all good, because they are all English. So if ‘blick’ feels good, and ‘bnick’ doesn’t, parallel sorts of feelings are nowhere to be found when we compare real words. Does it even make sense to ask whether the word ‘brick’ feels better than the word ‘trick’? Even if some people have an intuition on the matter, would their feelings somehow teach us anything about linguistic sound structure? I maintain that we can’t determine the structural properties of linguistic sound systems based on how people feel about the sounds they use. This has been stated quite emphatically by the scholars Bernard Bloch and George L Trager. Writing in 1942, they assert that, ‘The ordinary speaker of English, we are told, … “feels” or “conceives of” the two [l]s in “little” as “the same sound.” This may or may not be true; if true, it is an interesting fact, but it can never be used by the linguist as a criterion for his classifications, or even as a proof that he has classified correctly. … The linguist is concerned solely with the facts of speech. The psychological correlates of these facts are undoubtedly important; but the linguist has no means – as a linguist – of analysing them.'”
I did discover recently, incidentally, that blick is a real word, thus spelling doom for innumerably many introductory phonology texts, but the main point still stands: people’s metalinguistic reflection on language is not a good guide to their actual linguistic behaviour or whatever implicit ‘mental representations’ of language they might have. Reports of introspection can be useful in some contexts, but at the end of the day, they are “opinions. They are high-level meta linguistic performances that are highly malleable. They do not represent any kind of direct tap into competence, but are rather prone to many types of artefacts, such as social expectations, experimenter bias, response bias, and undersampling” (Pierrehumbert et al 2000: 189-190).
* Silverman, D (2006), A Critical Introduction to Phonology. London: Continuum
* Pierrehumbert, JB, Beckman, M, and Ladd, DR (2000), ‘Conceptual foundations of phonology as a laboratory science.’ In N Burton-Roberts, P Carr, and G Docherty (eds), Phonological Knowledge: Conceptual and Empirical Issues. Oxford; OUP