January 21, 2005


Mark Liberman's post yesterday about deep-seeded reminded me of a tiny quasi-experiment I conducted years ago, after reading a letter to the editor that used meddle instead of its homophone mettle. Various phonological theories have various ways of dealing with alternations between [t] and a tap (which is a name for the middle consonant in middle, seeded, and seated) in word pairs like write, with a [t] at the end, and writing, with a tap in the middle, and between [d] and a tap in word pairs like ride and riding. But what, I wondered, is the basic, underlying consonant in the middle of single-morpheme words like metal, meddle, mettle, and medal? Is it a tap, or a d, or a t, or sometimes a d and sometimes a t? I didn't expect the tap to be basic, because the English tap is usually assumed to secondary, a phonetic realization in a particular medial context from a basic /d/ or /t/ phoneme. But if it never alternates with anything else, maybe the tap actually is basic in simple words; and if it isn't, is it a realization of a /t/ or a /d/ in such words?

The problem is, since no alternations can be found in English in words that have a tap in the middle in all forms, how can we find out what the basic unit is? I thought that asking for a very slow pronunciation of the word might help, because a tap can only be said fast -- it you slow it down, it's not a tap any more. So in super-slow speech, would speakers still use a tap (in which case that part of the word wouldn't be slowed down), or would they replace the normal-speed tap with a stop [t] or [d]?

But now there's another problem: English speakers are too literate to make this experiment easy to run. In spite of misspellings like deep-seeded for deep-seated and meddle for mettle, speakers are all too likely to have a mental image of how the word is spelled, which will make it hard to discover their actual phonological representation of the word: I expected, but didn't test this hypothesis, that speakers who thought the word was spelled with one or two [t]'s would pronounce it slowly with a [t], and likewise for [d]. So, motivated by mere curiosity and unhampered in those long-ago days by IRB boards with their stern warnings against potential damage to speakers that could be produced by requiring them to (gasp!) pronounce words, I collected a gaggle of illiterate English speakers, namely, half a dozen neighborhood kids under the age of six. I read each of them a short list of relevant words and asked them to say the words very, very slowly. The kids were excellent experimental subjects, and every one of them, independently and without hesitation, pronounced every one of the words with a [t] in the middle.

What does this mean? I haven't the faintest idea. As an experiment, it's laughable -- historical linguists are not noted for expertise as experimental scientists, and even I can think of several confounding factors that could have skewed the responses. But I've always wondered why those kids (who did not hear each other's pronunciations) were so consistent in replacing the tap with [t], which is less phonetically similar to the tap than [d] is: both [d] and the tap are voiced, and the [t] is voiceless.

An afterthought: it occurred to me that Google might shed some light on this, so I googled two phrases with both spellings: on your mettle/on your meddle, and don't meddle/don't mettle. Google has 1140 hits for on your mettle and 0 hits for on your meddle, so there are no replacements of tt by dd. And it has 26,700 hits for don't meddle and 77 hits for don't mettle. The examples of don't mettle clearly mean the same thing as don't meddle, for instance I don't mettle in such things. So there's a small but nontrivial number of replacements of dd by tt. Not much data, though this apparent pattern does fit with what the little kids did in my quasi-experiment. But I still have no idea what, if anything, it means.

Posted by Sally Thomason at January 21, 2005 11:47 AM