Now that Alias is winding down to its series finale, I’m remembering back to a time a few years ago, a time when Alias was still good, and my wife and I still watched it every week. When the series premiered, I wasn’t planning on watching it at all. I was in the middle of writing a dissertation, and I had even dropped my longstanding Sunday night date with The Simpsons, so I wasn’t about to start watching some new show, no matter how much the critics liked it. But that was before Glen told me that our friend Bob Orci was one of the writers. So then I tuned in out of curiosity, and was hooked for the next couple of seasons.

One of the episodes in season 1 provided the basis for an essay question in an exam I gave later. It went like this:

In an episode of Alias last season, secret agent Sydney Bristow must get past bad guy Arvin Sloan’s electronic security system. She knows the password; the trouble is that the voice-recognition system will only recognize it when it is spoken by Sloan. To get a recording of Sloan saying the password, Sydney engages him in a conversation while wearing a wiretap. She is told that Sloan doesn’t have to say the actual password, because “there are only 40-some phonemes in English,” and if Sloan will just utter the phonemes that are in the password, those phonemes can be digitally strung together to get a recording of the password in Sloan’s voice.

The scriptwriters are right about the number of phonemes in English, but the task would be considerably harder than just picking out the right phonemes from the recording of Sloan. What complication have the scriptwriters left out? (Hint: Suppose, for example, that the password is tickle. The phoneme /t/ is needed, but what if the only word containing /t/ that Sloan says is kitten?)

The answer I was looking for was that each phoneme could be realized as several different sounds (referred to as allophones) that the speakers of the language perceive as being the same. To use my hint for an example, the /t/ in tickle would be pronounced as [tʰ] (that is, a [t] followed by a puff of air), while the /t/ in kitten would probably be pronounced as [ʔ] (a glottal stop, the sound separating the syllables of uh-oh). Pronounce tickle with [ʔ] at the beginning, and all you’ll have is ickle, which isn’t a word. (Or if it is, it’s a different word than tickle.) For each phoneme, Sydney would need to collect the particular allophone(s) that would be needed in the password, or risk the computer rejecting the digitally spliced-together password.

Of course, that was a minor quibble compared to some of the other stuff the show asked us to suspend our disbelief for. I didn’t mind doing so for the cool gadgets the characters used, or for Sydney being able to fluently speak whatever langauge she needed to in order to accomplish a mission, but I did eventually get tired of wondering how many times some character would be killed only to turn out to be alive after all: Sydney’s mom (twice), Sloan’s wife, evil Francie, and from what I’m hearing now, Vaughn, too.

By the time I gave up on Alias, Bob had long since moved on to other projects. But at some point I must have mentioned the phonemes-vs.-allophones technicality to him, because a couple of years later, he sent me an email asking… ah, well, that’s another story.

This weekend, I want to see Mission: Impossible 3, in spite of Tom Cruise. Wait, no. Not in spite of Tom Cruise. That sounds like Mr. Cruise doesn’t want me to go see this movie, and I want to go and see it anyway, just so he’ll make a little bit more money. I’m not too enthusiastic about doing that for this increasingly creepy, couch-jumping, not-content-to-
keep-his-cult-religion-discreetly-to-himself-instead-of-infecting-young-
women-who-fantasized-about-marrying-him-when-they-were-little-
girls-with-it celebrity. What I should say is that I want to see the movie in spite of the fact that Tom Cruise is in it. (Interesting that this ambiguity only arises when the object of in spite of is animate: I lived there in spite of the polluted air isn’t ambiguous.)

Why, you may ask, do I want to see Mission: Impossible 3 in spite of the fact that Tom Cruise is in it? That goes back to the “other story” I mentioned at the end of my last post.
MILD SPOILER AHEAD
In December 2004, Bob Orci sent me an email, saying:

I’m dropping you a line to ask for your help on something. We’re in the middle of Mission Impossible 3 and I was wondering if you could write us a poem?

So here’s the scene: Super agent Ethan Hunt is leading his team to kidnap a well guarded attorney, This attorney has a professional, secret service-like security detail protecting him. At one point, two of the guards begin making their security rounds through the hotel floor where the lawyer is staying. Meanwhile, Ethan is having a mask applied to look like one of the guards. When the guard he is going to double goes off on his own, Ethan’s team grabs him and forces him to read a POEM into their computer mics. This poem should contain all the (what did you call them?) all the bits of phrases and vowels that you would need to digitally recrate someone’s voice.

I couldn’t pass up a challenge like that. Plus, I owed him a favor from when he translated some multiple-wh and coordinated-wh questions into Spanish for me. So I set about finding or creating a poem containing all the allophones of all the phonemes of English, and in January reported back to him:

OK, I have a poem for you. I tried for a while to see if some already-existing poems would do the trick, checking out “There once was a man from Nantucket,” and the Yankee Doodle-tuned Barney theme song. But they didn’t, so I had to become a poet myself. Here it is (ahem):

The pleasure of Shawn’s company
Is what I most enjoy.
He put a tack on Ms. Yancey’s chair
When she called him a horrible boy.
At the end of the month he was flinging two kittens
Across the width of the room.
I count on his schemes to show me a way now
Of getting away from my gloom.

I’m assuming the speaker has a typical midwestern American accent. If you need to move or take out a word, let me know, and I’ll check to make sure the phones that it contained are somewhere else in the poem.

I seem to have started it in iambic heptameter (same meter as in “Casey at the Bat”) and then slipped from iambs to anapests in the second line, with a few caesuras here and there. As for the allophones of each phoneme, I didn’t quite succeed in getting them all in there. I gave up on trying to squeeze in both long and short versions of each vowel (vowels are pronounced for a longer duration before voiced consonants, shorter before voiceless ones). I didn’t bother with the distinction between released and unreleased stops (e.g., the difference between the /p/s in spot and top). Furthermore, upon reviewing the poem, I’ve found that some phones I thought I’d included aren’t in there after all. And there are probably a few here and there that I didn’t think of or know about. (If you’re curious, you can check the listing of phones with words they appear in here.) Oh,well. If someone wanted to generate a phony voice recording, this probably isn’t how they’d do it, anyway. Or maybe it is. Perhaps Mark Liberman or Liz Strand could give a better picture of the state of the art in voice generation.

The last I heard about the poem was in July, when Bob wrote,

I just got back from Rome and watched Philip Seymour Hoffman reciting your poem in a Vatican bathroom. The poem made it to the film with only a few words changed, so thank you again.

“Only a few words changed”? Uh-oh. Some of those allophones appear in only one word. In any case, I’m interested to see whether the poem made it to the final cut, or got left in the cutting room garbage can, as the expression goes. The PSH character in the movie review is a “black market trafficker”, which doesn’t sound much like the security guard that Bob described, so I wouldn’t be surprised if it got cut as a result of script changes.

In the meantime, if any of the phonologically-inclined out there know of, or have made, poems with every allophone of every English phoneme in them, feel free to share! The closest thing I could find was this sentence, which is said to contain almost all of them:

Please call Stella. Ask her to bring these things with her from the store: Six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob. We also need a small plastic snake and a big toy frog for the kids. She can scoop these things into three red bags, and we will go meet her Wednesday at the train station. (link)

And is there a name for this kind of thing? I’d guess panphone (by analogy with pangram), and it looks like this guy had the same idea.

Update: The poem’s in there, and the use they put it to isn’t just to pre-record something in the bad guy’s voice, but to actually make Ethan’s voice sound like the bad guy’s. So how would that work? The throat sensor would have to detect the gestures made by the vocal folds, tongue, and even lips and soft palate somehow, then retrieve the appropriate phone from the memory, and play it out, at the same time playing the exact waveform that would cancel out any noise made by Ethan’s own voice as he talked. The changes I caught were the replacement of Shawn with something like Busby or Thisbe; show me with reveal the; of gettting away from with to escape my.

Anyway, it was a fun movie overall. Giving the Alias treatment to Mission: Impossible 3 made it much more entertaining than the first one. Probably more than the second one, too, though I’m just guessing because I never got interested enough to see it.

Another update: a linguist named Jay, who works doing text-to-speech at Microsoft, offers his thoughts on exactly how implausible this scene is here.

Polar_Bear_-_Alaska by
rubyblossom.

Last September, a reader named Richard Gunton left a comment on my panphonemic poem post with the following panphone that he’d composed:

Catching weary waterfowl on thin ice gives surly polar bears huge pleasure and ensures they enjoy good meat unharmed.

I commented back:

By George, I believe this is panphonemic! How did you come to write it? And if you don’t mind, could you give your own vowel inventory in IPA, and show which word(s) go(es) with which vowel? The low backs are hard enough for me to keep straight in my own dialect, let alone a different accent. If there’s an interesting story behind this, I’d be happy to put it up here as a guest post.

Richard responded back with a list of every phoneme with the words it appeared in, with this commentary and backstory:

It’s interesting that my list of vowels distinguishes a much greater number than the one in your table above. I took it from my dictionary, and I do believe these all represent distinct phonemes for standard British English.

As to how I came up with it – well, back at the end of 2008 I had just moved to France, so had the subtleties of exotic phonemes on my mind, and a bilingual dictionary to hand. My French colleagues and I had been comparing French and English pangrams, so I thought a sentence with every sound of a language would be the next challenge. I realised that /ʒ/ was one of the rarest phonemes in English, so I started with “pleasure”, prefaced that with “huge” since /dʒ/ seemed a bit uncommon too, and built it up around that. That’s all I can remember now – it did take me quite a few idle nights to get there!

Well, that’s interesting enough that I really should have turned his work into a guest post by now. Better late than never. But I’ve made just a couple of adjustments to his panphone (with adjustments made accordingly to the list of words and phonemes), to give it some topicality:

Catching weary dolphins on thin ice gives surly polar bears huge pleasure and ensures they now enjoy good meat unharmed.

Thanks, Richard!

Consonants
/p/ polar
/b/ bears
/m/ meat; unharmed
/f/ dolphin
/v/ gives
/θ/ thin
/ð/ they
/t/ meat; dolphins
/d/ and; good; unharmed; dolphin
/n/ dolphins; on; thin; and; ensures; enjoy; unharmed
̣/s/ ice; surly
/z/ gives; bears; ensures
/l/ dolphin; surly; polar; pleasure
/r/ weary
/ʃ/ ensures
/ʒ/ pleasure
/tʃ/ catching
/dʒ/ huge; enjoy
/j/ huge
/k/ catching
/g/ gives; good
/ŋ/ catching
/w/ weary; waterfowl
/h/ huge; unharmed

Vowels and Diphthongs (‘ follows vowel being referred to)
/i/ weary’; meat
/I/ catchi’ng; thin; gives; e’nsures; e’njoy; ‘dolphin
/ɛ/ plea’sure
/æ/ ca’tching; and
/ɑ:/ unha’rmed
/ɔ/ on
/ɔ:/ dolphin
/ʊ/ good
/u:/ huge
/ʌ/ u’nharmed
/ə/ pola’r; pleasu’re
/ə:/ su’rly
/Iə/ wea’ry
/ɛə/ bears
/eI/ they
/aI/ ice
/au/ now
/əu/ po’lar
/ɔI/ enjoy’
/uə/ ensu’res

Limericks have been on my mind fhttps://literalminded.wordpress.com/?p=6794&preview=trueor the last couple of months. It started when I discovered a Twitter account called @Limericking, which puts out a constant stream of limericks based on the news, usually better than the ones featured each week on “Wait, Wait, Don’t Tell Me”. Here’s the limerick that showed up on my timeline in January:

So clever, and such a good illustration of the cot/caught merger, which I just wrote about in a piece I just did for Grammar Girl on vowel mergers. For me, cause and flaws both have the mid back round vowel /ɔ/, but Oz has the low back unround vowel /ɑ/. It could just be that the writer of this limerick was settling for an imperfect rhyme, but I see that @Limericking is based in Canada, one of the places where the merger is widespread, so it’s probably a perfect rhyme for them.

Then, at the end of the month, Merriam-Webster started tweeting out limericks about English usage. I particularly liked this one:

At the beginning of March, of course, it was National Grammar Day once again, with its annual limerick contest. This was the winner, and deservedly so:

I didn’t write a grammar limerick, but after I read the limericks from Limericking and Merriam-Webster, I decided to take another crack at writing a panphonic poem, within the constraints of five short lines. The first time I tried putting all the sounds of English into a single poem, I tried to work in not only all the sounds that English speakers perceive as separate sounds (in other words, all the phonemes), but also all the variant pronunciations of each phoneme (i.e. all the allophones). For example, I didn’t want to put in just the vowel [i] as in she, but also the nasalized vowel [ĩ] as in scheme. Ultimately, I didn’t succeed, so I set my sights a bit lower this time. Here’s what I ended up with:

In normal spelling, it’s

Hear in this short limerick’s strains
Every sound which my language contains.
Could it be an illusion?
Panphonic profusion?
Something linguists enjoy as a game?

I would rather have said panphonemic profusion because it’s more specific, and because the meter works better, but panphonic was the only word I had with the vowel /ɑ/. And I’d prefer sound that to sound which, but I needed a /tʃ/. Maybe I’ll try again someday, without such a meta topic.

One of the courses I teach is individual pronunciation tutoring for international students who are going to be teaching assistants here at Ohio State University. One of the resources I use a lot is this clickable IPA chart. Click on any of the sounds in this chart, and you’ll hear a recording of someone uttering the sounds.

Sometimes, though, I wished that it was possible to reduce the visual clutter by having the chart show just the sounds of English, or just the sounds of Chinese, or Korean, or whatever other language a student spoke. I could toggle between the different languages’ phonemic inventories, allowing us to quickly view the phonemes common to multiple languages, and those that are in one phonemic inventory but not another.

At the same time as the chart had too many sounds, it also didn’t have enough of them. Some sounds, like the affricates /tʃ/ (as in chump)and /dʒ/ (as in jump) are displayed on a supplement to the chart (not shown in the screenshot here). There are even bigger gaps for Chinese, since it has three times as many affricates as English, and some of them aren’t displayed on the chart anywhere at all. This is because they’re versions of some affricates that are already shown in the chart, but they’re aspirated (i.e., pronounced with a short puff of air after them). It makes sense not to show these, because if you recorded aspirated versions of all the consonants, it would double the size of the consonant chart. And if you’re going to have separate recordings for the aspirated consonants, why not for the glottalized ones, or the pharyngealized ones, or the nasalized vowels, or the creaky vowels? But still, when I’m working with a Chinese student, and want to show them exactly how the set of sounds they’re used to matches up with what we have in English, I’d like to have all the affricates, aspirated and unaspirated, up there in the main chart with everything else.

A more elaborate clickable IPA chart that I recently learned about and have been using is this rtMRI IPA chart. This one was created by the Signal Analysis and Interpretation Laboratory at the University of Southern California. When you click on the IPA symbols in this chart, you not only hear them pronounced, you also see them pronounced with a real-time MRI (rtMRI) video clip. It is incredibly useful that someone took the trouble to do one of these rtMRIs for each of these sounds, and as a bonus, there are also clickable rtMRI recordings of some minimal vowel sets, some short sentences, and a couple of longer passages that I suspect are panphonemic, though I haven’t checked to be sure.

However, as with the other chart, you need to already know what sounds are in a language in order to know which ones you’re interested in clicking. And like the other chart, this one sidelines the affricates, and shows even fewer of them than the other chart. It wasn’t the customized tool that I sometimes wished were available to me and my students.

A few months ago, I was telling the ESL Programs’ curriculum director, Karen Macbeth, about the kind of chart I wished existed somewhere. As it happens, she was (and is) working on creating an e-textbook for all our Spoken English courses to use, and she said a chart like this one would go well in this kind of digital resource. She put me in touch with one Mike Shiflet, who works for Ohio State University’s Office of Distance Education and E-Learning and who has been helping Karen with her project. I gave Mike some printed IPA charts with different languages’ phonemic inventories highlighted on each one: English, Chinese, Korean, Turkish, Hindi, and Spanish. I showed him the clickable IPA chart that inspired this project. I provided him an audio clip of me pronouncing each of the sounds I wanted. From there, Mike produced the chart I had been dreaming of, and it’s now on OSU’s ESL Programs Spoken English web page for anyone to use! Me, I’m going to start using it tomorrow.

Below is a screenshot of just the (Mandarin) Chinese version of the chart.

I hope this chart proves to be as useful to some ESL/EFL teachers and students as the other clickable IPA charts have been for me.