Abbreviations are always interesting. Some like HTML prounounced letter by letter as in /eč ti ɛm ɛl/, but others are prounced as a “new word” as in NASA or /næsə/. This pattern is fairly random – for instance I have heard the database query language SQL as both “S-Q-L” and “sequel”.

An interesting class are abbreviations with no vowels that are still pronounced like words. Whenever you have a word, the phonological component has to “organize” it into syllables. For most, languages the rule is generally to cluster consonants around a vowel where each vowel (or diphthong) is the nucleus (or backbone if you will) of a syllable. To take the NASA case again, you can generalize that it has two syllables because it has two vowels.

HOWEVER…in some languages (like American English), you can form syllables around other sounds such as /l,r,n/. If you’re not sure what vowel is in the final syllable of words like table, prism, butter, button and others, it’s because they’re not phonetically not a vowel but syllabic consonants. You may see these transcribed as /n̩,r̩,m̩/ (n,r,l,m, etc with dashes beneath)

The consonant-free abbreviations shows that English can go pretty far. For instance I have heard the term VRML become /vr̩ml ̩/ or “vrmle”. But even more interesting case is the Penn State first year counseling process called FTCAP or more fondly [fɪtkæp] or “fit-cap.” Although a vowel has been inserted, it’s not the usual epenthetic schwa /ə/ (vowels inserted to break up unwieldy consonant clusters (as in the Maryland river Potapsco /pətæpsko/ becoming [ptæpəsko] or “Potapsaco” in some parts of Maryland).

One analysis would be to posit the vowel /ɪ/ as the epenthetic vowel in particular situations (when syllable is stressed?). However, I’m going out on a limb here, and say it’s almost as if the /t/ in FTCAP has become a syllabic /t/. It is definitely the case that a high front like /ɪ/ allows the tongue to remain in /t/ position until the /k/ is pronounced. It would be interesting to compare spectrograms on some of these words.

I should note that are some derviations from the pattern. SQL becomes “sequel” [sikwl ̩] and not “sickle” [sɪkwl ̩]]. And the one that inspired this blog post is wff (well-formed formulation) which becomes not “wiff” [wɪf] but “woof” [wʊf]. Is this a case where speakers are using /ʊ/ as the closes vowel to /w/ or a chance to make an interesting pun from an acronym?

Some time last week, Jeopardy had a “Phonetics”, and we know it’s an obscure topic for most of the educated world because it was in a Double Jeopardy round for a Tournament of Champions (prelim round).

The most interesting question was something like this

Question: The word agency contains the most common sound in the English language. What is this sound?

I admit I missed this one, because the answer was supposed to be schwa /ə/ (say what?). Pondering, this I think I found a quirk in my dialect

First, where did the Jeopardy schwa come from? Probably from actual pronuciatiion. In many dialects of English (certainly U.S. English) many unstressed vowels are changed to schwa. Hence a word like piña kolada which is pronounced as /piña kolada/ in Spanish usually emerges as something like [piñə kəladə] in English (with 3 schwas). This is why schwa is probably the most common vowel. Even in my dialect agency is something like [ejǰɛnsi] or [ejǰɨnsi]

But why did I miss it? Here’s my phonemic transcription of agency – /eiǰɛnsi/. You’ll note that there is no schwa in there because in my mental dictionary, I don’t think of this “e” vowel as a schwa, but as the same vowel as in sense /sɛns/.

Similarly I think that agent rhymes with gent /ǰɛnt/ (or at least pretty close). On the other hand, I don’t think agent rhymes with hunt /hʌnt/ or /hənt/. So even though I know about the proliferation of schwas, I really didn’t think it was in this word. For me, it’s not the same status as the schwa in the [ðə] or a [ə]

However, I may be the quirky one here.

For instance, I did find that the Oxford English Dictionary transcribes agency as [eiǰənsi] with the schwa [ə]. I also found one rhyming dictionary (http://www.rhymer.com/RhymingDictionary/agent.html) which does identify agent and hunt as rhymes.

On the other hand, WikiRhymer is matching agent with magenta (definitely a /ɛ/ in the second syllable), so maybe it’s not just me. Since none of the contestants detected the schwa, I am wondering if they had the same computational problem I did.

Thinking about my dialect, I suspect that my difference in perception partly because because I can distinguish [ə] (first vowel of Alissa) from [ɨ] (first vowel of Ilisa). As far as I can tell though, this happens only in a narrow band in the urban northeast, and other speakers have only [ə]. And the Jeopardy writers are working in California (heavy schwa territory).

Fortunately, descriptive linguistics, unlike Jeopardy, doesn’t always have to have one right answer.