An technicality at worst

I'm not sure I want to write this post. Ben Zimmer's study of (and commentary on) language is impressive reasonable thorough precise entertaining and...good.

And he has been kind enough to leave a few comments here. For all--our thanks. So I don't want to protest too much.

In his wholly reliable style he recently posted over at Language Log about Victor Washington's case against the NFL. Washington seeks fuller retirement benefits because of his injuries. The NFL doesn't want to pay out so much. Washington's plan offered higher benefits if he suffered "a football injury" and the NFL is sticking to the argument that his several injuries don't entitle him to "Level 1" benefits. (Follow this link to read the story.)

The sentence that caught my attention was the following observation by Zimmer regarding arbitrator Sam Kagel's use of "a" instead of "an" in the phrase "a injury."

"The legal emphasis on the word a would apparently be lost if it underwent the regular addition of the epenthetic consonant /n/ to create an before a word beginning with a vowel like injury."

His suggestion makes sense regarding the rhetoric. I'll buy it. I'm not sure about his analysis of the a/an alternation as epenthesis: the /n/ being an additional letter. It is additional if we say the underlying form of the article is a. But historically the form was an which was realized as a before most consonants. In that case "an" alternates with "a" by deletion of the /n/ coda. So at most this is a pure technicality.

But let's say that for the sake of a clear point Zimmer is using the wording of the retirement/disability plan as the underlying form. Since the plan uses the phrase "a football injury" we'll say our input is "a"; and from that form to its reflex in the phrase "a(n) injury" we would expect that /n/ would be added because the new phrase places the article before [i] instead of [f]. But I'm still not sure I'd call it epenthesis in that case.

And I'm not sure that I wouldn't.

_

Not your coal. Charcoal.

Earlier today Buffy finished reading The Scarlett Letter. She loved it. And she's not a bit embarrassed by her late introduction to the classic. Honestly. How many of you have read every book that's listed on the back cover of Cliff's Notes? I've never read Walden.

Halfway through she mentioned something about Hester "Prine" and I corrected her without teasing. "Prynne...like thin" I said. Then I had to add "at least that's how I've always heard it." Because names really are all about lexical pronunciation. Read this amusing post about a dissertation defense and Pepys the diarist.

There's no English spelling rule that should make the "peeps" pronunciation obvious. It's only from hearing it said by someone (whom I believed) that I learned not to say peppis peepis pep-eyes peppees peppeez or anything else that would get a laugh. So why should Buffy have known that Prynne doesn't rhyme with line? Some might argue that Prynne looks enough like Lynne/Lynn to make the pronunciation clear. But we all know that analogy is a fallible rule.

So late this evening Buffy interjected a non sequitur into a conversation about politics...or religion...or ice cream...(I can't remember)

"Chester Prynne right?" she asked.

"What?" I collapsed laughing. "Who's that? Hester's twin brother? Did Hawthorne write a sequel called The Azure Letter about Chester and his daughter Gem?"

Buffy insisted that she had said "Hester."

"I said It's Hester Prynne! I said Hester!"

She wasn't sure how I could have misheard so when I started to explain she told me to put it in a post.

When it's begins a phrase the vowel is often elided. Especially when the phrase opens an utterance. This same [ɪ] has been elided from it for hundreds of years. Even Shakespeare saved time by ignoring the poor high front lax vowel. Tis true; tis true tis pity, And pity tis tis true.

So instead of saying "it's Hester" Buffy said "ts Hester." More like one word: tsester. That [h] is chronically truant. And following a fricative it's easily hidden even if present.

This is one of those not so common cases in English pronunciation when the [ʦ] affricate can't be broken by syllabification. As an initial sound there was no coda available to claim the [t]. And in a coda position the /t/ would more likely be a glottal stop [ʔ]. So I was obviously unprepared for an onset affricate with alveolar release [ʦ] and I interpreted it as a postalveolar release [ʧ].

But I do think Chester could be a fascinating character.

Where have all the good glides gone?

I was watching the end of a House MD rerun and I got lost regarding the plot. So I looked online for the script and found it pretty easily. Reading through it I came across the following lines. A young girl (Kama) is asking the doctor (Foreman) about her brother's prognosis:

KAMA: Is he gonna die?

[Foreman looks at her.]

FOREMAN: No. No one's gonna die.

KAMA: [smart-ass] In the whole world ever? That's so great.

FOREMAN: [chuckles] I meant...

KAMA: I know what you meant. But I also know bad things do happen. My dad always had a few drinks whenever he went out. Always said it'd be okay to drive. [shrugs sadly] Until it wasn't. I would just like some mourning this time.

[Foreman looks at Kama, sympathetically.]

FOREMAN: We're nowhere near anything like that happening right now.

It's an odd wish that Kama has for "some mourning." I though maybe the story was about having to rush through a loss and the poor girl never had a chance to mourn her father's death. So she wanted to know ahead of time if her brother was going to die so that she'd...

Oh. Yeah. That makes more sense. Some warning.

The two phrases are often identical. After some consonants it's common for a [w] to be deleted. As evidenced by the some warning/some mourning confusion the deletion can even occur across a word boundary. Between the labial [m] and rounded [ɔ] is an especially suitable environment for w-deletion. It can be hard to distinguish between some more and some wore (except that the former is usually an iamb and the latter is often a trochee). And we sometimes find this deletion after other consonants in words like towards [tɔrdz] and quarter [kɔɻɾɻ]. These words can still be pronounced either with or without the [w], while sword lost it permanently long ago (except in the not-so-common spelling pronunciation).

Of course there are several words in which the deletion does not occur. Before a front unrounded vowel the [w] stays put in tweed switch square and quack. If a speaker pronounces quartz with a distinctly low and unrounded [a] my guess is that the [w] is regularly preserved. I've never heard the mineral pronounced identically to carts.

I'm not prepared to give an answer to the question that I now pose: why did sword lose the [w] while swore has kept it? Usually we look to the original language or the time of adoption to help with an explanation. Because both words have been with the language since it was Old English that doesn't give us much help. It may have something to do with the preservation of a front vowel in swear. By analogy with a related form the [w] could have been preserved by kinship.

But that's only a primary suggestion and I'm happy to hear a more informed explanation.

I never said that!

***Please forgive the increasingly dry and technical cant of my posts. School is coming up and I have to get back in the mode of writing like a textbook. My contributions will likely become reflections on my studies - I certainly hope that doesn't shatter the interest of those few of you who do me the favour of checking in to see if you care about what I write.***

Abstract forms are a controversial topic in linguistics. In syntax there are schools of thought that claim our minds work with un-pronounced (un-expressed? un-said?) items all the time.

One simple example of this is the phrase: "I want to eat." According to the minimalist camp there are no fewer than 2 words that are nullified in this phrase. The underlying representation (or the grammatical form that our mind knows but our mouth doesn't express) is closer to this: "I want (for me) to eat."

One argument is that every verb needs to be done by a noun. So the first "I" does the "wanting." Something else needs to do the "eating." So we put a form of "I" in there to do the eating (but because it's an object of the first clause it takes the "me" form). So apparently it's true - "You" can't have your cake and eat it too...

The suggestion is that we know these things and they shape the grammar of the words we utter - even if we never pronounce the null constituents or realize that they're in that crazy little brain of ours. The inevitable argument against such abstractions is countered with the following: Being proficient does not mean that we are able to analyse our proficiency.

In phonology there are some proposed abstract forms that may change our phonemic inventory if we believe they exist. It's a very technical argument and I apologize that I don't have the skill to explain it in an interesting manner. Let's start with a question and just hope that you care to find the answer. Here's the setup to the question. Of all the sounds in the English inventory there is only one that never occurs at the beginning of a word: the eng. This is that nasal sound that we put at the end of sing thing ring wing and just before the /k/ in sink think rink and wink. Every other consonant is possible at the beginning of a word. Why does this never occur initially? Some propose an abstract representation of all those -ing words.

The argument goes like this. The nasal in those words is underlyingly the same as the /n/ in sin thin or win. But because it comes before a /g/ it moves back to assimilate in place (just like the /n/ moves forward to assimilate to a /p/ in a phrase such as "ten people." Trust me - in normal speech you say "tem people." No one ever believes us when we tell them this). After the assimilation the /g/ is deleted - perhaps due to some special rule that deletes a voiced velar /g/ after a nasal /n/. And we see some evidence that this rule is not applied all the time. You may have heard someone from New York talk about "Long-Gisland." This is not proof - just evidence. So perhaps the eng never occurs at the beginning of a word because it's not an actual phoneme of the English language - it's just an allophone. There are no words that begin with ng- or nk-. Those would be the only clusters that would create the eng consonant.

(The real controversy in this is the distinction between dynamic and static phonological data - are we talking about language just as it is today or are we talking about language as it changes over years and years?)

And then we come to the /h/. This sound never occurs at the end of a word.¹ But this isn't so strange because the /h/ is a fickle sound that has disappeared in many places.² take for instance the common pronunciations of vehicle and vehement. Consider also the /h/ sound that used to be found at the beginning of which and whether. And then we have the fast-speech habit of dropping initially.

Tell 'im we're leaving?
Does 'e care?
Well 'e said 'e wanted to see 'er...

(Calling attention to such habits is why Eliza Doolittle was trained with the phrase "In Hartford, Hereford, and Hampshire Hurricanes hardly happen.")

Here is the most startling discovery about English I've made in the last several years (I didn't actually discover this - I merely learned it while doing some research). When we say "'em" for the plural third person (e.g. "Those people look hungry. Give 'em some pickles.") we are not deleting the initial th- of them. The evidence? There is no other word that begins with th- that we pronounce without that onset. We never say -

Where are 'ey going?
Is 'is the right place?
I'd like some of 'ose over 'ere?

By observation of the lack of this process we conclude that the 'em we say is not a shortened form of them. We are in fact deleting the initial /h/ of an older form hem that was more common in Middle English. So for most of our lives we've been using the shortened form of a word we had no idea we knew.

Again - analysis must reveal those things that proficiency hides even from the proficient.

---

An allophone is an alternation of a single phoneme that is not recognized as a distinctive part of the phonemic inventory - i.e. the hear interprets the two sounds as the same sound. The clearest class of allophonic alternation is the difference change in the /t/ in tack and stack. In tack there is a puff of air after the /t/ that we call aspiration. In stack there is no similar puff of air. But the ear does not consider these two different sounds because they are merely allophonic. Two distinct phonemes would change the meaning of the word and be heard as different. So "lock" and "rock" begin with different phonemes in English. But not every language hears the difference. In some Asian languages these would merely be allophonic and be heard as the same word. --Go back to reading--

1. Well this isn't completely true. Loch is sometimes pronounced with the final /h/. This word comes from the Old Irish and into the Scottish Gaelic and has preserved a phoneme that is for all intents and purposes obsolete in the U.S. We can see this effect in the American use of the word - homophonic with lock. To our phonological ear the heavy /h/ is merely the product of a brogue. We would no more likely use that sound than we would rrroll our arrrs. Most Americans only know these sounds as the tricks to sounding like groundskeeper Willie.

"Eef elected merrr, mi fairst act wull be tuh kill theh whole lot of yuh - and boorrn yer toon to cindairs!"
--Go back to reading--

2. And sometimes we throw it in somewhere just to clarify borders. I remember in high school choir one kid had to sing the phrase "ever more and ever more." He always took a breath before the "and." This left him thinking that he had to accentuate the beginning of "and." So it always sounded like "hand." With a little choral training he would have learned that the proper syllabification could best be written "ever mo -ran -dever more."--Go back to reading--