A phonetic feature whose importance was overlooked in the early days of phonetics, yet it explains a long-standing problem with English vowels, and would have been easily visible in x-ray pictures. It was first noticed in connexion with a puzzling system of vowel harmony in a number of African languages.

In languages such as Igbo, all the vowels in a word must belong to the same set, where there are two sets. Grammatical markers attached to the word take on one of two forms, to ensure that their own vowel harmonizes with those in the stem. A typical system is this:

        i     u
         I    U
          e   o
           E  O
            a
In this diagram they are laid out according to the physical position of the highest point of the tongue. The letters are the SAMPA (IPA) symbols for the sounds. One harmonizing set consists of {i e u o} and the other consists of {I U E O}. You should disregard the fact that some are capitals: this is just a coincidence in SAMPA. Typically, either {a} harmonizes with both sets, or there are two kinds of {a}.

The problem is what is common to the sets. Vowel harmony in Uralic or Altaic languages, like Finnish or Turkish, groups vowels in obvious ways: high {i u} versus low {a o}; or front {i e} versus back {u o}. Neither of these can explain what is common about the interleaving sets {i e u o} versus {I E U O}.

In the 1960s x-ray studies showed that the key difference was that the {i e u o} set all had the root of the tongue advanced, to create a wider cavity in the pharynx. The other set have a narrowed pharyngeal cavity. The low vowel {a} is already close to the pharynx, so isn't as susceptible of having two variants. The feature is commonly abbreviated ATR. So some vowels are +ATR and others -ATR, at least in many African languages.

There was a long-standing problem with how to describe the long vowels of English, such as those in bead, beat, booed, boot, as opposed to short vowels as in bid, bit, book, bet, bed. In English (much more so than in other languages) the length of a vowel is affected by its position, pre-eminently by whether the following consonant is voiced. So the vowel of bead is much longer than that of beat, and that of bid much longer than that of bit. In fact the "short vowel" of bid may be quite as long as, if not longer than, that of beat. It depends on various considerations such as stress, but the basic point is that physical duration is not essential to the difference between "long" and "short" vowels. Yet they form natural sets.

For many years the words "tense" and "lax" were bandied around. These are pretty meaningless and if I had my way I'd ban them from phonetics. What, precisely, is tense or lax? Which muscles are doing what, which bit of the larynx or mouth is differently shaped or at a higher tension? I have never seen an adequate formulation of what tense and lax are supposed to mean. Unfortunately you can still see them in texts.

The correct answer is that the long vowels are +ATR and the short vowels are -ATR. Musculature is complex and subtle, and it is quite possible that widening the pharynx does increase some muscle tension somewhere in a way that also affects the vowel, but it is now known that the main difference is the ATR feature.

I suspect also that the essential difference between the W-like R of young Britons in the Estuary English accent, and the more usual kind of R, is that the new one is +ATR.

Log in or registerto write something here or to contact authors.