March 05, 2008

Je ne regrette rien: phonetics and phonology

Edith Piaf's recording of "Je ne regrette rien" ("I don't regret anything") was on the radio the other day. (It is also frequently on British TV in a commercial spoofing it, with subtitles regretting not having purchased the product... Key sign of a bad advert: I cannot remember the name of the product, or indeed, anything about it.) Listening to Piaf's delivery reminded me that in the first line, after the word "Non", she sings the title in as few as three syllables. I think rien counts as monosyllabic, though it could be bisyllabic (the melody line rather suggests that); but what is quite clear to me is that the part before rien is at most just two syllables, and possibly just one. The vowels of je and ne are completely gone (they are heard loud and strong as separate syllables when repeated to a different melodic phrase in the next line). And the third vowel is hardly clear. In other words, she sings je ne regrette as something very much like [ʒnʁgʁɛt], although the first two consonants are certainly soft and indistinct.

Word-initial consonant sequences like [ʒnʁgʁ] have virtually never been reported in any language (there might be exceptions among certain groups like the Salishan languages of the Pacific North West or the North West Caucasian languages, but I don't think so). French doesn't allow words beginning with [ʒnʁgʁ] — that is, words that would have spellings beginning jnrgr. It's just that remarkable things happen in rapid or casual connected speech.

Sometimes when linguists tell introductory classes about phonotactics, they illustrate by pointing out that although English has words beginning with [sn], none of the other fricative consonants occur before the nasal [n] at the beginning of a word: there are words like snoop, but you simply cannot have words like *znoop, *vnoop, *hnoop, *shnoop, *thnoop, or *fnoop. They are not phonetically inconceivable; they are just phonologically disallowed in English. And sometimes a bright undergraduate will put up a hand and say, "What about the word [fnetIks]?" At that point a bad instructor will say "Shut up. You're just being silly."

But a good instructor, such as one might expect to be taught by at the UC Santa Cruz, or at the University of Edinburgh, will use it as a teachable moment, and explain that the prohibition seems to be phonological but not phonetic. Remarkable things happen in rapid or casual connected speech, as Piaf's first-line pronunciation of je ne regrette shows; but the word phonetic is not a true example of a [fn-] word. People who use it also use words related to it like phonetician and phone and allophone, and in those words there is always a vowel between the [f] sound and the [n] sound. What the pronunciation [fnetIks] illustrates is that when a vowel between two consonants is totally unstressed, the vowel can be completely inaudible. Potato can sometimes be pronounced with [pt] at the beginning. Yet pterodactyl never is. Why not? Because words beginning with [pt] are phonologically impossible in English despite being phonetically possible. (Pterodactyl has pt in the spelling only because of its Greek etymology.) And that is a crucial, and very useful, example of what the difference is between phonology and phonetics. That's all for today. Class dismissed.

Posted by Geoffrey K. Pullum at March 5, 2008 01:21 PM