I Got Laboved

Bill Labov came to visit Ohio State University this week. This is the guy who, 50 years ago, began to answer what was then a 100-year-old question: What is the origin of the sound changes that run through a language, changing entire vowel systems, collapsing two phonemes into one, splitting one phoneme into two? More specifically, who starts these changes, and how, and why? With just a couple of well-known studies which are now standardly cited in historical linguistics textbooks, he changed how linguists went about researching these questions.

One of those early studies involved listening to how clerks in higher- and lower-end New York department stores pronounced the phrase fourth floor, in order to hear whether they were pronouncing or omitting the /r/ in those words. (This study was recently the subject of a two-part episode of Lexicon Valley.) The method consisted of asking a clerk where to find some item that the researcher knew to be on the fourth floor. When the clerk said, “Fourth floor,” the researcher would pretend not to have heard properly, and the clerk would say it again. In this way, Labov obtained a pair of utterances of the same phrase, said casually (the first time) and more carefully (the second time). Comparing the percentages of speakers who omitted the /r/ both times, pronounced it both times, or omitted it and then pronounced it provided interesting insights when put together with the demographics of the speakers; for a fuller presentation, listen to the Lexicon Valley podcast.

During his visit to OSU, Labov made several presentations, and tonight he and his colleague Gillian Sankoff were the guests of honor at a party at a professor’s house (his daughter’s, in fact). When I got to the party, I saw Labov talking with Brian Joseph, who introduced me.

“Neal Whitman,” I said, shaking hands.

“What was that?” Labov asked.

“Neal,” I said. It was a bit noisy, so I did my visual aid of making as if to kneel. (Get it?)

“No, your last name.”

“Oh!” I said. “Whitman.”

“Ah, you aspirate your W!” he said.

I burst into a grin. “Yes, I do!”

After that we talked for a few minutes about where I grew up, the “Cool Whip” Family Guy clip on YouTube, vowel mergers, and about sounds that persist in a language long after their reported death.

Driving home, I realized: One of Labov’s oldest tricks had taken me completely unawares.

Christmas Codas

During some of the Advent church services in the past month, and the Christmas Eve service earlier this week, I’ve had occasion to be reminded of a phonotactic constraint that, evidently, wasn’t so hard and fast when a lot of our classic Christmas music was written. Specifically, I’m talking about syllables that end with [vn], as in heav’n and giv’n, which come up a lot in these songs. Often they come up very close to each other in order to make a close-enough rhyme. For example, there’s this pair of lines in “O Little Town of Bethlehem”:

How silently, how silently, the wondrous Gift is giv’n;
So God imparts to human hearts the blessings of His Heav’n.

It also happens with [zn] in the ris’n that I ran across in another song. So to generalize, these songs allow a syllable to end with a voiced fricative (i.e. [z] or [v]) followed by an [n]. The other voiced fricatives in English are [ð] (as in thy) and [ʒ] (as in genre). As far as I know, there are no English words that end in [ʒən], so there’s no chance of finding such a word shortened to end in just [ʒn]. English words that end in [ʒən] include words like vision and fusion, but those tend to turn up in hymns so much. As for words that end in [ðən], there’s heathen, so I’d predict that if any of these songs had the word heathen in them, we could expect to see it written heath’n. But I checked, and heathen isn’t such a popular word in hymns.

As I struggle to sing heav’n and giv’n as single syllables, I have to wonder why it’s so difficult. After all, the consonant clusters [vn] and [zn] aren’t so different from other consonant clusters that form easily pronounceable syllable codas in other English words. (A syllable’s coda is the string of whatever consonants occur at its end.) Fricatives in a syllable coda can combine with certain non-nasal stops, provided the voicing is the same. Here are the admissible and inadmissible combinations of voiceless fricatives with voiceless stops:

• *[fp]
• [ft] lift
• *[fk]
• *[θp]
• [θt] frothed (for some speakers)
• *[θk]
• [sp] asp
• [st] mist
• [sk] ask
• *[ʃp]
• [ʃt] mashed
• *[ʃk]

Summing up the voiceless fricative-stop combinations, it looks like [s] can combine with any of [p], [t], or [k], but the other fricatives can only go with [t]. Now here are the admissible and inadmissible combinations of voiced fricatives and voiced stops:

• *[vb]
• [vd] lived
• *[vg]
• *[ðb]
• [ðd] breathed
• *[ðg]
• *[zb]
• [zd] raised
• *[zg]
• *[ʒb]
• [ʒd] massaged
• *[ʒg]

These are even more restricted than the voiceless combinations: Now, only three out of the four eligible fricatives ([v], [ð], and [z]) can combine with a stop, and even then only with [d]. However, the fact is that these voiced fricatives can combine with [d] to form a syllable coda. Furthermore, the only difference between [d] and [n] is that for [d], your nasal passage is blocked, whereas for [n], air is coming out through your nose. So why are [vd] and [zd] so easy for English speakers to say, while [vn] and [zn] aren’t?

One possibility that occurred to me was to blame it on the fact that [n] is a continuant. That is, because the airstream can escape through your nose, you can stretch out an [n] as long as you have breath, whereas a [d] is over in an instant. For that reason, the [n] after another consonant feels like another syllable. But that won’t work, because fricatives are continuants, too, and fricative-fricative codas are perceived as one syllable: buffs, lives, writhes, fifth.

Instead, the rule seems to be that a sonorant sound can’t come after a fricative in a syllable coda. Sonorants consist of vowels, liquids (that is, [r] and [l]), glides ([j] as in yet and [w]), and nasals, so this rule also explains why words that end in [zm] or [ðm], such as chasm or rhythm have two syllables instead of one. (I imagine that this rule has been long known, and written up in some article or textbook somewhere, but I haven’t found it. References or corrections are welcome in the comments.) Sonorants after sonorants are OK, as in kiln, barn, and film (though I understand that in some dialects, film is pronounced with two syllables: “fill-em”). For another phonotactic constraint involving codas and sonorants, see this handout for a UMass linguistics class taught by Kyle Johnson.

All that’s well and good for present-day English, but I still wonder: When did it stop being OK for English codas to end in [zn] and [vn]? Was it ever part of everyday language, or just for poetry and songs?

The Oral-Aural Merger?

I sent a message to the American Dialect Society email list earlier this month, about a pronunciation that I’ve begun to wonder about recently. Here’s what I wrote, but with more accurate IPA symbols inserted:

I’m sure this has been analyzed somewhere at some point, but I don’t know where. What is the dialect that has [ɔ] lowering to [ɑ] in a stressed vowel preceding /ɹ/ and an unstressed vowel? In other words, the dialect that pronounces forest as “farrest,” Florida as “Flarrida”,Oregon as “Ahregun,” horrible etc. as “harrible” etc., authority as “autharity”, but still has [ɔ] in fort, lore, etc.? What is this realization called?
I’ve been vaguely aware of it for many years, but have begun to notice it more, especially among certain NPR speakers. I even heard one guy on Planet Money talk about a “flarrist” (florist), which is right in line with the phonetic environment I described, but was still a new pronunciation to me.

Actually, this question is complicated by the fact that various historically distinct vowels have merged in various combinations in various dialects of English when they appear before /ɹ/. These include the so-called Mary-merry-marry merger, the steer-stir merger, the fir-fur merger, and others, which you can read about in this Wikipedia post. I was even surprised to learn about a horse-hoarse merger, which made me realize that my father was not joking or deluding himself when he once claimed that for him horse and hoarse were not homophones. I pronounce them both [hɔɹs], but speakers without this merger pronounce hoarse as [hoɹs]. I have a hard time even imagining this pronunciation, with [o] coming before an [ɹ] at the end of a syllable (or in coda position, as phoneticians say), and have never perceived it in Dad’s speech. However, I can definitely hear it when it comes before [ɹ] at the beginning of a syllable (that is, in onset position). If you know my father, you can hear it when he calls someone a moron, which he has always pronounced [moɹɑn]: “Mo-ron!” When he does that, I find myself imagining a Southern pair of twin boys, named Jim Bob and Mo Ron. (For more on vowels before [ɹ], see this post.)

Anyway, I got some interesting responses. Kate Svoboda-Spanbock wrote, “It is a longstanding source of amusement to my L.A.-bred children, who laugh when I say AH-rinj but who nonetheless say that they are SAH-rry.” Her post jolted me into looking at my own pronunciation, because I definitely say “SAH-rry” (i.e. [sɑɹI]), and for that matter “to-MAH-row” and SAH-row,” and find the [ɔɹ] pronunciations of these words unusual.

In fact, my “SAH-rry” might not even match that of Kate’s children, because phonetically, there is more than one “ah” sound. There’s the low back unround [ɑ] that I’ve been using in the IPA notations, but there’s also the low back round [ɒ], which might appear in cough, depending on your dialect. To tell you the truth, I’m not very good at distinguishing the low back vowels, and as far as I can tell, I might be using either of them.

Ben Zimmer wrote that [ɑɹ] instead of [ɔɹ] was common in New York City, as well as in Philadelphia and the Carolinas, and gave a link to the Wikipedia page I linked to above. Paul Johnston corroborated, citing his NYC parents’ consistent [ɑɹ] pronunciation, while also noting his own nearly universal shift to [ɔɹ] in his adult life.

Larry Horn wrote that the change is almost certainly happening via lexical diffusion–that is, somewhat haphazardly on a word-by-word basis. He recalled social pressure he experienced to change some of his pronunciations in college:

[T]ypically, whatever the shibboleths are may be under the most pressure to change, which is why I switched [to the [ɔɹ] pronunciation] on corridor and moral earlier–and more consistently than–Florida or florist.

Joel Berson confirmed the somewhat unpredictable nature of this change, writing:

[M]y vacillations and shifts are different from Larry’s…. For example, I’m sure I seldom
say “florist” but mostly “flarrist”. But I say “floral”, not “flarral”.

Eventually, the discussion wandered to some of those other pre-R mergers that I mentioned earlier. Although I excluded from my query words that had [ɹ] in coda position, some of them came up in the discussion anyway. Matt Wilson mentioned the cord-card merger, which Wilson Gray (recalling his youth in Saint Louis) might have called the fort-fart merger. In elementary school, he and his classmates preferred to avoid saying any number between 39 and 50 for this reason. I also hear this kind of merger in the speech of Jessica Lange’s character in American Horror Story: Asylum.

As the discussion petered out, Charlie Doyle brought up the knock-knock joke that depends on the [ɑɹ] pronunciation of orange, whose punch line is “[ɑɹə̃nʤ] you glad I didn’t say ‘banana’?” That reminded me of a poem composed by Tom Lehrer in response to the challenge of finding a word that rhymed with orange:

Eating an orange
While making love
would make for bizarre enj-
oyment thereof.

A couple of respondents to my post noted that there wasn’t a nice, convenient name for this particular phonetic phenomenon. Larry Horn proposed and quickly rejected “East Coast Ah-ringe”. My humble proposal is in the title of this post. If any dialectologists are reading this, what do you say? Is there a name? If not, what do you propose? Ben Trawick-Smith, and Rick Aschmann, I’m looking at you!

Twenty Wung Guns

Glen once noted that the trouble with being able to put all your favorite songs on one convenient device is that you have to consciously decide to listen to new music. Motivation to listen to the radio plummets: “Why listen to someone else’s lousy mix plus advertisements, when you can listen to the best mix ever without advertisements?”

So true, so true, as I concurred once before in this space. These days I get exposed to unfamiliar music only when something unusual is going on. As I noted in that earlier blog post, in 2010 I heard a few new songs I liked only because I made a point of listening to the pop station every day for two weeks while I was writing a column on the use of the word <I>Im(m)a</i> in popular songs. I’ve heard a new song here and there in the bowling alley with Doug and Adam, or in the movie theatre while I’m waiting for the previews to begin.

And, as it happens, in the roller skating rink, too. That’s where, at Doug’s birthday party later in 2010, I heard a song that I identified with my song-identifying phone app as “21 Guns” by Green Day. I added it to my iPod, and now, two years later, “21 Guns” has become just one more piece of music that I listen to to the exclusion of new stuff.

After a couple of years of listening to it, I’ve gradually become interested in the chorus:

One
Twenty-one guns
Lay down your arms
Give up the fight.

One
Twenty-one guns
Throw up your arms
Into the sky.

Two things are interesting about the chorus, one of them because of the way Green Day sing it, and the other because it brought back memories of writing Chapter 1 of my dissertation.

In English, the phoneme /n/ might be pronounced not only as [n], as in Neal, but also as [ɲ], as in In your face!, or as [ŋ], as in drink. That last assimilation is consciously known to most literate English speakers, some of whom had to be taught that ng was actually pronounced [ŋ], and not [ng] “nuh-guh”, as it was spelled.

In fast, or even normally paced speech, these assimilations can cross word boundaries, as happens in my example of In your face! Speaking carefully, I would pronounce 21 guns as “twenty [wʌ̃n] guns”. But speaking freely and easily, I would (and do) pronounce it as “twenty [wʌ̃ŋ] guns”. (The ~ is supposed to go over the ʌ in those transcriptions

In fact, Green Day sing it this way, too, as you can hear in the video. What I find unusual, though, is that they do this even though the song is somewhat slow (about 80BPM, the low end of “andante”, according to my metronome). Nevertheless, every time they sing that chorus, it’s a very carefully enunciated “twenty wung guns”. Why?

Yokult

A couple of weeks ago, I was pleased to discover that a new episode of The Tobolowsky Files had come out. (You may recall my blogging about this podcast last year.) This one was about a time in Stephen Tobolowsky’s life when he had an Icelandic horse. I never knew there was a breed of horse called an Icelandic, but I guess there is. The horse’s name, Tobo said, was something that sounded like Yokult. He explained that the name was Icelandic for glacier.

Wait a minute–Icelandic for glacier? Didn’t I already know the Icelandic word for glacier? Hadn’t I learned it somewhere? And it wasn’t yokult, it was…

Ah, right! It was jökull, as in Eyjafjallajökull, the volcano that preempted so much trans-Atlantic air travel back in 2010. People made fun of the name–The Oatmeal’s take on it was hilarious–but the news stories explained that it meant “island mountain glacier” (or more literally, “glacier of the mountains of the islands”). On Language Log, Mark Liberman explained the pronunciation, and on his Phonetic Blog, John Wells gave some additional details.

The parts I was interested in were the ll sequences. As I’ve learned from the blog posts, in Icelandic represents a “pre-stopped lateral”. The lateral part means basically that the sound is a kind of /l/. In phonetic terms, lateral refers to the sides of the tongue. To get the full picture, you have to know what the tongue is doing for other kinds of consonants, in particular the stops (or plosives) and fricatives. For stops that involve the tongue, the tongue blocks the airflow from the lungs completely. For example, put your tongue in position to say a [t] or [d] and you’ll feel it form a seal all around the edges of your palate, from your top left molars to the area behind your top incisors to your top right molars.

For fricatives that involve the tongue, the tongue obstructs the airflow enough to create turbulence, resulting in the hissing or buzzing sound of, for example, [s] or [z]. The air that does get out passes over the top of the tongue. To see how, put your tongue into position for a [t] again, and now turn that [t] into an [s]. You’ll notice that the sides of your tongue are still touching your top molars. The part of your tongue that’s making way for the air to escape is the tip.

What if instead of lowering the tip of your tongue and leaving the sides in place, you do the opposite? What if you lower the sides and leave the top in place? In that case, what you end up with is an /l/, or to be more precise, a whispered (voiceless) /l/, written in IPA as [l̥ ]. If you turn on your voice, you end up with the ordinary voiced [l].

[l̥ ] and [l] are said to be lateral approximants (or sometimes liquids), which means that the tongue causes the airflow to take a different path out of the mouth than it would if you were just saying a vowel, but doesn’t obstruct it enough to result in a fricative or a stop. But if you want to, you can turn your lateral approximant into a lateral fricative. Just stiffen up your tongue and close the space between the sides of the tongue and the teeth above, just enough to get that turbulent airflow. If you’re doing this without voicing, you’ll get the hiss of the voiceless lateral fricative [ɬ]. With voice, you’ll get the buzz of the voiced lateral fricative [ɮ].

So much for lateral. What about pre-stopped? If you guessed that it has something to do with stop consonants, you’re right. As a reminder, the primary stop consonants in English are [p, b, t, d, k, g]. Sometimes you’ll get a stop right before a fricative. You can probably identify the stop-fricative neighbors in dipshit, ribs, cat sitter, red zone, suck face, and beg the question. A couple of stop-fricative pairs have even achieved the status of phoneme in English; that is, they’re perceived as a single sound. Those pairs are [ʧ] and [ʤ], as in cheer and jeer, respectively. These consonants might have been called pre-stopped fricatives, except that another name had already been established for these: affricates.

So instead of thinking about stops coming right before fricatives, think about stops coming right before other continuant consonants (i.e. consonants that you can keep saying until you run out of breath, as opposed to stops, which are done the moment you allow airflow to resume). Those are the nasals /m, n, ŋ/ and the approximants /l, r, w, j/. Put a stop consonant before any of these sounds, and it’ll be a pre-stopped version of that sound, right?

Not quite. To count as pre-stopping, there are two additional requirements. First, the stop and the continuant have to be homorganic (that is, made with the same parts of the mouth). So, for example, [bm] would count, because both [b] and [m] are made with the lips. [ps] would not count, because [p] is made with the lips, while [s] is made with the teeth and tongue. The second requirement–and this is where English gives up any hope of having pre-stopped consonants–is that the pair of sounds be considered a single sound by speakers of the language we’re discussing, just the way that the affricates [ʧ] and [ʤ] are considered individual sounds by English speakers. The closest English comes to having pre-stopping is in words like hidden, assuming you’re pronouncing it with no vowel between the [d] and the [n], and are just keeping your tongue tip in place and letting the blocked airflow suddenly escape through your nose. But if you ask an English speaker how many syllables hidden has, they’ll say two, not one. For [dn] to be a pre-stopped /n/, the speaker would have to consider hidden to be just as good a monosyllable as catch, or for that matter, lets, fifth, ghosts, and sixths.

Now I can get back to the Icelandic ll. This orthography represents a pre-stopped lateral, i.e. /tl/. This is easy to hear in the slower recordings of fjalla; it sounds like “fyatla”. The difficulty comes in jökull. Icelandic has final devoicing, which means that voiced consonants at the end of a word are devoiced. So /tl/ would be realized as [tl̥ ]. Supposedly. In fact, as John Wells notes and Mark Liberman agrees, that final [l̥ ] sounds more like a voiceless fricative than an approximant, so that the final consonant of jökull is actually [tɬ]. In other words, it’s actually an affricate, not a pre-stopped lateral.

It really blows my mind to force myself to think about [jœːkʏtl̥ ] as two syllables instead of three. I hear it as “yokoot” followed by static as the transmission is suddenly cut off. If I insist on interpreting that static as a speech sound, the same as I do with the staticky sound at the end of catch, the best I can do is hear it as three syllables, taking the [kʏtl̥ ] part as something like “kootle”, but with the /t/ actually pronounced as [t] instead of tapped as in poodle, and the /l/ whispered.

That accumulation of phonetic unfamiliarities–/t/ not turned into a tap, a voiceless lateral fricative that doesn’t exist in English, plus the necessity of interpreting these two sounds as a single phoneme–is too much for most English speakers, as we learned during the season of Eyjafjallajökull. In a collection of clips of newscasters pronouncing the word (which Wells links to), the most common adaptation was to metathesize the [t], and put it before the [k]: “Ayafyatlayotkul”. The adaptation I’d probably use would be to ignore the final devoicing and pronounce it to rhyme with poodle. And coming back to Stephen Tobolowsky, his adaptation is a different metathesis, namely swapping the [tl] to get [lt], as well as not trying to make a [ɬ]. That was an adaptation I hadn’t heard before, but let me ask now: How do you pronounce jökull when you’re not perfecting your Icelandic pronunciation?

The Witch Mary

Grammar Girl is running a guest script I wrote today (that is, she’s running it today; I wrote it some time ago), on difficult syntax in Christmas carols in general, and in particular in “What Child Is This?” The script was inspired by a real-life misunderstanding that Doug had seven years ago, and which I blogged about at the time. I’ve also been thinking about that song because Adam has been practicing playing it on the piano, and he sounds really good!

As I wrote in that blog post and in today’s Grammar Girl podcast, part of the difficulty is due to the perennial confusion between lie and lay (which I also wrote about in this post about the song “If I Just Lay Here”). For a while, I considered concluding the podcast with a sentence or two about how other traditional Christmas carols can serve as good models of for using lie and lay in the way that is currently considered the standard:

• Where the snow lay round about, deep and crisp and even
• the little lord Jesus laid down his sweet head
• the stars in the sky looked down where he lay
• how still we see thee lie
• …certain poor shepherds in fields as they lay

I decided against it, because I didn’t want to give the impression that the whole episode was just about lie vs. lay. But as my wife and I were thinking about other Christmas songs, she started running through “God Rest You Merry, Gentlemen” (which I wrote about last year). The second verse goes like this:

In Bethlehem in Israel this blessed babe was born,
And laid within a manger upon this blessed morn;
The which his mother Mary did nothing take in scorn.

This one isn’t so good for helping you remember the difference between lie and lay. Sure, you could parse it as was [born and laid], the standard way, but if you don’t already know that’s how it’s supposed to be, you could easily just parse it as [was born] and [laid], with laid used nonstandardly as an intransitive verb.

However, that wasn’t the part that grabbed my attention. Before my wife could move to the third verse, I was interrupting with, “Mary, a witch?!” Then: “Oh, which!”

Two changes in English created this misunderstanding. First is the simplification of the consonant cluster [hw] to [w] for many speakers, as highlighted in this Family Guy clip that I learned about from Language Log a few years ago.

Having the last name I do, I think I still have the [hw] cluster in my language. Sometimes when I give my name over the phone, the person on the other end will hear it as “Quitman”, because they don’t have [hw] in their speech and figure that I must have been saying [kʰw] instead of [hw]. On the other hand, other times they’ll simply not hear the [h] at all, and think my name is “Wittman”, which makes me wonder if I actually pronounce [hw] as consistently as I think I do.

The second change is the loss of the which as a relative pronoun. I never knew about it until I listened to this verse. The which is in the Oxford English Dictionary, though. It’s sure enough archaic now, but was showing up in the 1300s, as in this OED citation:

How god bigan þe law hym gyfe Þe quilk the Iuus in suld life.

Their last citation is from 1884, from Tennyson:

He holp the King to break down our castles, for the which I hate him.

There have to be kids who got all confused when they learned Jesus’s mother was a witch. Any of you know of any?

Trick or Treat!

In the course of writing a Visual Thesaurus column on aspects of the word Halloween, I looked into the history of trick or treat. Some of the questions I had about it were:

• When did it become a verb, as in trick-or-treating?
• If its origin is indeed a threat, why is the threat said first and the demand second? That is, why isn’t it Treat or trick, following the same demand-punishment template as Your money or your life or Truth or consequences?
• What’s with the kids in It’s the Great Pumpkin, Charlie Brown saying “Tricks or treats”? Is that a 1950s/60s thing, or a regional thing?

In the book Death Makes a Holiday: A Cultural History of Halloween, by David J. Skal, I learned that trick-or-treating in the United States began only in the 1920s, or possibly slightly earlier, on a regional basis. Skal adds that it “became widely known and adopted as a distinct property-protection strategy during the late Depression” (54). The sugar rationing of World War II put a damper on it, but trick-or-treating really took off in the post-war years.

The earliest attestation of trick or treat in the OED is from right after the war, in a 1947 article in American Home:

The household larder needs to be well stocked on October 31, because, from dusk on, the doorbell rings, bright eyes peer through crazy-looking masks, and childish voices in ghostlike tones squeal, croak, or whisper, “Trick or Treat!”

However, Skal has the phrase eight years earlier, in a 1939 article in the same magazine. It’s not talking about trick-or-treating as we know it, but as sort of a password for a Halloween party, put on for the same purpose of allaying Halloween vandalism. Skal writes that this attestation is “apparently the first time ‘trick or treat’ is used in a mass-circulation periodical in the United States” (p. 53):

…they found our front door open and a jolly Jack o’lantern grinning from a window at them. Seeing me, they summoned nerve to speak the age-old salutation of “Trick-or-Treat!”

Skal notes that even though the article refers to Trick or treat as an “age-old” greeting, it gives no support for this claim.

Returning to the post-war years, Skal writes that the Donald Duck cartoon “Trick or Treat” in the early 1950s helped popularize trick-or-treating on a national scale.

All this agrees with the picture you get from the Google Ngram viewer:

So how soon did trick or treat become a verb? The earliest example in the OED is from 1950:

So let the kids go out tonight and have a grand time with their masquerading and trick-or-treating.

As for the order trick or treat instead of treat or trick, as far as I can tell, the trick part has always come first. I wondered if it was some kind of phonetic thing going on, like roly poly or knick knack, but it doesn’t seem to fit the patterns. Unlike ping-pong or see-saw, the phrase trick or treat doesn’t have a front vowel followed by a back vowel: [I] adn [i] are both front vowels. And the initial consonants are the same, so whatever explanation you have for hanky panky instead of *panky hanky won’t apply. I tried to think if other common words or phrases had the [I]-[i] sequence, and didn’t come up with much: snickersnee (a kind of sword) striptease, and Mister T, but that’s about it.

Tricks or treats actually antedates trick or treat, as far as I’ve been able to determine. In Google Books, I found it in a 1938 issue of The Alpha Phi Quarterly:

Yes, it is Hallowe’en — the time for “tricks or treats.” But as far as Alpha Phi life is concerned, we know it holds only treats.

In an archive of Peanuts comics, I found that Charles Schulz had his characters saying “Tricks or Treats” all through the 1950s (sometimes with the addendum “Money or eats!”), though once he introduces storylines involving Linus and the Great Pumpkin in the 1960s, you don’t see it so much. Jumping forward to 1993, though, there’s a Sunday strip with Linus and Sally in the pumpkin patch, with Snoopy making an appearance at the end. In Snoopy’s thought balloon is “Trick or Treat!”, so somewhere along the way Schulz fell into line with the rest of the country. You can see in the Ngram View above that tricks or treats peaked in the mid-1950s.

One last item for those who read this far: Trick or treat! Smell my feet! Give me something good to eat! is noted as early as 1966 in the Keystone Folklore Quarterly. As for the further extension involving the pulling down of underwear, I can only date that back to my childhood in the 1970s.

How Do You Say Hubert?

In a post at Lingua Franca, Geoff Pullum writes about reading a novel and being pleasantly surprised when the protagonist referred to the “th” sound in that as a voiced dental fricative, which, in fact, it is. (Interdental, more specifically, but still.) But his admiration turned to disgust when he read another novel in the same series, and the protagonist tells the Secret Service that from their recording of a bad guy saying, “You won’t get that lucky again” and “Hey, I want to talk to you,” they have all the phonetic information they need to identify the guy: “All the vowel sounds, most of the consonants. You got the sibilant characteristics, and some of the fricatives.”

A panphonic set of unscripted utterances consisting of only 13 words? Pullum sets the record straight in his usual style. I already knew firsthand how difficult it would be to round up all the English phonemes in one utterance, having tried doing it in the Mission: Impossible poem, which Ben Zimmer kindly linked to in a comment. For panphonic passages written by other people, check the other posts in the Panphonic Phun category.

As it happens, I was thinking about my panphonic poem just the yesterday. I had just read a post at Grammarphobia about the pronunciation of h before [ju], as in Hubert or Houston (the city in Texas, that is, not the street in Manhattan). Here’s Patricia O’Conner’s description of it when it is pronounced (instead of dropped, as some speakers do):

Phonetically, the letter “h” in these words is a voiceless palatal fricative (a consonant produced by narrowing the air passages, arching the tongue toward the hard palate, and not vibrating the vocal cords).

I was surprised for a moment, since I’m used to thinking of [h] as a voiceless glottal fricative, made simply by opening your vocal folds wide and letting air from the lungs pass through the opening between them (i.e. the glottis). But then I realized that I do pronounce Hubert and Houston with a palatal fricative at the beginning. I started to say Hubert, but quickly switched to home after saying the /h/, and the pronunciation sounded off.

This phonetic realization makes sense, since [j] (that is, the “y” sound) is a palatal consonant, and turning the glottal fricative [h] into the palatal fricative [ç] before [j] is a typical assimilation. Alternatively, instead of producing a fully palatal fricative, a speaker might get the back of the tongue only as far forward as the velum (aka soft palate) before making the /h/ sound, in which case it would come out as the voiceless velar fricative [x]. If you speak German, you’ll recognize [ç] as the sound at the end of Ich, and if you listen to Bill Cosby comedy routines, you may recognize [x] as the way he often pronounces /k/, but that’s about as but English doesn’t have /ç/ or /x/ as phonemes in their own right, so using them for /h/ here and there doesn’t cause confusion.

The significance for my poem, in which I had attempted to use not only every phoneme but also every allophone (way of pronouncing) every phoneme, was that I had learned about one more allophone that I hadn’t managed to squeeze in. I had /h/ in the words he, him, and horrible, and in all those words I think it’s realized as simply [h] and not [ç] or [x]. Some speakers might have it as [ç] in he, but not as reliably as they would in Hubert.

What about you? Do you use a glottal, velar, or palatal /h/ before the “you” sound?

Srimp and Jritos at the Groshery Store

In my second post on the pronunciation of “tr” as [ʧr] (i.e. as “chr”), my question was this: If the /ʧ/ and /ʤ/ (that is, the “ch” and “j” sounds) are phonemes in English, then why don’t English speakers think of words like trick and drape as chrick and jrape? (At least, why don’t the English speakers who pronounce them that way think of them as chrick and jrape? Some speakers do pronounce /tr/ and /dr/ as [tʰr] and [dr].) To put it in phonological terms, why would someone who didn’t know the alphabet perceive [ʧrIk] as /trIk/ and not /ʧrIk/? Or [ʤreip] as /dreip/ and not /ʤreip/? In fact, children who are just learning to spell sometimes do spell [ʧr] as , and [ʤr] as . However, English speakers eventually come around to perceiving [ʧr] and [ʤr] as /tr/ and /dr/. One reason is that as they learn the spelling system, they see that that’s how [ʧr] and [ʤr] are spelled. Another reason is that if English allowed the affricates /ʧ/ and /ʤ/ to form consonant clusters with /r/, we’d have a strange phonological system on our hands. In it, all the plosive consonants other than /t/ and /d/ could form clusters with /r/, while /t/ and /d/ for mysterious reasons could not. Meanwhile, we have /ʧ/ and /ʤ/, which do not normally form consonant clusters, able for some reason to form them with just the consonant /r/.

With that in mind, consider the consonant cluster [ʃr], in words like shrimp, shriek, shred, shroud, shrew, and shrike. I hadn’t given it much thought before, but comments from Herb Stahlke in some of the posts linked to this one have got me to thinking about it. Like the affricate /ʧ/, the sibilant /ʃ/ forms clusters only with one consonant: /r/. You do get [ʃt] if it’s followed by an /r/, as I discussed in a recent post, but speakers generally perceive that as /str/. And you don’t get words like shkop, shtame, or shpoonkle (oh, wait…). German or Yiddish borrowings like schlep, Schwinn, Schmidt, and schnitzel are acceptable, but you don’t find many new words created that begin with /ʃl/, /ʃw/, /ʃm/, or /ʃn/. On the other hand, the sibilant /s/ can form a cluster with several other consonants. It can form them with voiceless plosives: spit, stick, sky. It can form them with nasals: smack, snoot. It can form them with glides: swoop, and in some dialects, words like suit. (See this post on Dialect Blog for more on American English “yod-dropping”.) It can form them with liquids: slide and … Oops. It can form clusters with lateral liquids, i.e. /l/. It can’t form them with retroflex liquids, i.e. /r/. How many of you pronounce the Sri in Sri Lanka as [sri], and not [ʃri]? I try to, but it feels weird.

So by the same phonological reasoning that leads us to perceive [ʧr] and [ʤr] as /tr/ and /dr/, why don’t we perceive [ʃr] as /sr/? In other words, why don’t we have a system in which /s/ can form clusters with both kinds of lateral liquids, and note that before /r/, /s/ is realized as [ʃ], instead of having a mysterious gap where /sr/ should be? Well, in this case, the spelling points toward hearing it the way it actually sounds: Words like shrimp, shriek, shred, shroud, shrew, and shrike are actually spelled with . But if it weren’t for the spelling, how would speakers perceive it? (Stahlke observes that some Southern American English speakers actually do say “srimp”, but what about other words beginning with “shr”?)

There is at least one word where speakers may perceive something pronounced as [ʃ] as an /s/. Listen to this classic Sesame Street video:

Did you hear it? “Ten tiny turtles on the telephone, talking to the groshery men”? That’s how I heard it as a kid, but gradually wrote it off to my imagination, as I grew up in a family that pronounced it gro[s]ery. Years later, though, I learned that many speakers unquestionably do pronounce grocery with [ʃ]. On her blog, Jan Freeman wrote:

But ever since I started reading similar criticisms of my native Ohio speech oddities, I’ve been wary of ascribing motives to people’s pronunciations. I grew up with “mirror” pronounced MERE and grocery as GROSHERY. But my parents didn’t use those pronunciations because they were uneducated; they used them because everyone did.

As I wrote this post, I realized that I had an explanation for this pronunciation: If you elide the unstressed schwa in the middle syllable, you’re left with an /s/ right next to an /r/. (Linguists call such a deletion syncope.) Looking at it that way, I see that gro[ʃ]ry is no more unusual than C’lumbus, Ohio, or Web’los. But if you keep the unstressed syllable, then both gro[ʃ]ry and C’lumbus may strike you as a bit odd.

Now Freeman may or may not have recognized that her pronunciation of grocery contained a [ʃ] (feel free to chime in, Jan), but here’s a speaker for whom [ʃ] is just how you pronounce /s/ before an /r/. A commenter going by the handle embolini9 responded to a query on seriouseats.com, “How do you pronounce ‘grocery’?” , writing, “I’m from New England, and I’ve never heard the ‘sh’ sound. I’ve always said ‘gross-ree.’” But a few comments later, embolini9 returned to write, “Oh wait! I just said it out loud, and I guess sometimes I do say ‘groh-shree.’ Maybe more often than not… yup, I definitely say ‘sh.’ Now I’m the crazy girl sitting at her desk saying ‘grocery’ to herself.” (The rest of the comments are fun,too, ranging over a lot of regional pronunciations, an dsurprisingly little peeving.)

This case of syncope feeding a phonetic alteration brings me back to the posts on “shtr” and “chr/jr” that got me onto this subject. I was listening to the Sept. 7, 2011 “Radium Girls” episode of the Stuff You Missed in History Class podcast, and one of the hosts pretty consistently pronounced str as [ʃtr]. There were one or two occasions when she didn’t, but one of the words that got a [ʃtr] was history. She pronounced the word historic with an [s], but history with a [ʃ]. Why? In historic, the middle syllable is stressed, so the /st/ is separated from the /r/ by a vowel. But in history, the host syncopated the unstressed medial vowel, leaving the /st/ right next to the /r/, eligible for the [ʃtr] pronunciation. As for “chr” and “jr”, I remembered way back to when Doug was three or four years old, and his favorite lunch was a turkey sandwich with Doritos. He tended to syncopate that initial unstressed syllable, leaving the /d/ next to the /r/, eligible for the [ʤr] affrication. As a result, he would ask for a turkey sandwich and “Jritos”.

Shtraight Talk

When Adam’s Cub Scout den planned a trip to go horseback riding early last summer, I signed up to ride, too. I wondered why only one other parent in the den was going to ride. What were they going to do while the boys all saddled up and went out on the trail?

At the stable, all the kids and parents stood along the wall of a big room with a dirt floor while the horse handlers did a 15-minute lecture on safety around horses. Then they had the boys come up one by one to receive a Post-It with a piece of a horse’s anatomy written on it, which they were then to stick on a cooperative model horse named Jet. That part was interesting; I finally learned what a horse’s withers were, although I forgot later.

Then it was time for the riding. Each boy stepped up onto a platform, where an adult volunteer (me), helped him onto the horse. The handler then led the horse away, walking with it to the far wall, around to the side wall, along the side wall to the near wall, and from there back to the platform, where the one boy got off and another one got on. And that was the horseback ride I had paid for. I went ahead and chased that sunk cost (as Glen would say) by taking the ride when it was my turn.

After the excitement of the ride, the scouts and their parents relaxed with a tour of the stable. In one room, the handler showed us the hay and the straw, and asked if anyone knew the difference between them. I didn’t, so I listened carefully. She began by mentioning a practical difference:

Horses eat hay; they sleep on shtraw.

What? What was that? Did she say “shtraw”? Maybe I hadn’t heard right. The handler went on to explain the essential difference between hay and straw:

Hay is grass; shtraw is the stalks of oats and things like that.

She did it again! Oh, and of course, oats are a kind of grass, too, but I got the idea. But back to the phonetic point: The handler had substituted [ʃ] for [s] twice. She didn’t do it for all /s/s; she pronounced grass, stalks, and oats with [s]. Did she do it for any /s/ before a /t/? No: stalks. How about for any /s/ before /tr/? During the rest of the talk, I listened for more [ʃ]-[s] substitutions, and heard her use the words “stronger” and “street”, pronouncing each with [ʃtr]. No other /str/ Word came up, although the handler did utter an interdental /l/ when she said, “Horses eat a LOT of food.” Otherwise, her /l/’s were alveolar, so she might have been one of the speakers who pronounce their /l/’s interdentally for emphasis in a word that begins with /l/.

But back to the [ʃtr]-for-[str] substitution: I first learned about it in a paper called “Getting [ʃ]tronger Every Day?: More on Urbanization and the Socio-geographic Diffusion of (str) in Columbus, OH,” by David Durian. He notes that in this area, it’s more common among younger speakers, working class speakers, and speakers who grew up in the city of Columbus rather than its suburbs; and this last set of speakers is spreading the change to the suburbs they’ve moved to as adults. He also cites a 1984 study by Bill Labov which documents widespread [ʃtr] in Philadelphia.

Patricia O’Conner wrote about the [ʃtr]-for-[str] substitution in a Grammarphobia post in May 2008. Three months later, the topic came up on the American Dialect Society mailing list in August 2008, when Herb Stahlke reported hearing it in a speech by Michelle Obama. (More on that at the end of this post.) Since becoming aware of this sound change, and since that visit to the stables, I’ve been hearing [ʃ] in place of [s] in /str/ clusters in other places, too…

• When my wife and sons and I were watching the movie Independence Day (1996), I heard Harry Connick Jr.’s character say to Will Smith’s character, “You’ll never get a chance to fly the space shuttle if you marry a shtripper.” I made everyone wait while I rewound twice to make sure I’d heard right.
• A month later, we were watching Beverly Hills Cop (1984), and I heard Eddie Murphy’s character utter this other sentence about stripping: “The only reason these officers were in a shtrip club….”
• A couple of weeks into the school year, I overheard a conversation among a couple of Adam’s fellow fourth graders as they picked up their “Grab n Go” breakfast in the school hallway on the way to their classroom. Apparently the school can’t count on parents actually giving their kids breakfast every morning, so they provide snacks before school for any kids who want them, so they can start off the day with something nutritious and be able to concentrate better in class. This morning, it was Pop Tarts. One girl said to another, “It was funny, because you said brown sugar and I said shtrawberry!” It really must have been funny, because the girl said it again, and again pronounced strawberry as shtrawberry.
• At about 7:51 into episode 414 of This American Life, the producer of the first story, Ben Calhoun, says, “These weren’t regular uniformed cops. They were the guys in shtreet clothes.”
• In the past year, I’ve heard one of each of Doug’s and Adam’s friends pronounce /str/ as [ʃtr], usually in the word destroy.
• During a family trip to New York City last month, a bus tour guide consistently pronounced /str/ as [ʃtr].
• In a subsequent whale-watching trip that departed from Long Island, a guy from Madison, Wisconsin consistently pronounced /str/ as [ʃtr]. I later learned he’d grown up in Long Island.
• One of the audiobooks we listened to in the car on our trip was Judy Blume’s Blubber. The reader has [ʃtr] for /str/ about 80% of the time, I’d guess offhand. I’ve heard it in street, strip, stripe, and elsewhere. The occasional [str] pronunciations that come up make me imagine the reader in the studio, with the engineer making her go back and re-read those words, but giving up because the reader’s [ʃtr] is just too consistent to fight.

At this point, I’m starting to forget all the places I’m hearing [ʃtr] for /str/. But my question is why it would occur in the first place. Summarizing previous research, Durian mentions three possibilities. One is that it’s a case of the /s/ assimilating to become more like the /r/; specifically, it’s pronounced with the tongue pulled further back toward where the /r/ is pronounced. That’s a little unusual, because it would be a case of “long-distance” assimilation: The /s/ is taking after not the /t/ right next to it, but the /r/ after that. I’ll add that for some speakers, this could actually be a more typical case of assimilation. Speakers who produce a retroflex [r], by curling their tongue tip backwards, might well retroflect the /t/ before it as well, and if that /t/ is retroflected, the /s/ before it is liable to be retroflected, too. When that happens, it sounds like “sh,” but not quite like the [ʃ] version I’ve been talking about. In the IPA, this retroflex sibilant is written [ʂ]. Under this scenario, the “shtr” pronunciation is [ʂʈr] instead of [ʃtr]. (Most English speakers, including me, cannot hear the difference [ʂ] and [ʃ].)

A second possibility is restricted to a subset of those speakers who, like me, turn /t/ into an affricate before /r/, pronuncing trap as “chrap”. In particular it’s limited to those speakers who (unlike me), even affricate their /t/ when an /s/ comes before it. That is, some speakers (including me), pronounce trap beginning with [ʧr] (“chrap”). Within that group, some (including me) pronounce the trap part of strap with a [tr], while others pronounce it with [ʧr]. Within that smaller group, some speakers pronounce the /s/ as [s], to produce “s-chrap”, while others assimilate the /s/ to the [ʧ] by making it palatal: “sh-chrap”. I imagined a scenario like this near the end of one of my posts about /t/ affrication. But I can’t really tell if I’ve been hearing, say, “shtreet” or “sh-chreet”. In this paper (note 9), Brian Joseph and Rich Janda profess not to have found any reports of [ʃʧr] in the literature.

The third possibility, and the one Durian favors, is proposed by Joseph and Janda. It so happens that when [ʃtr] occurs in the middle of words, the preceding vowel is almost always a high vowel such as [i], as in restructure. Therefore, it may be a case of the tongue not lowering fast enough after the high vowel, resulting in the [s] turning into [ʃ]. Then, once the [ʃtr] cluster became familiar, speakers started using it at the beginnings of words, too. This would account for why in his data, [ʃtr] occurs more in the middle of words than at the beginning.

Let’s hear from some of the /s/-retractors out there. Do you pronounce str as “shtr” sometimes? All the time? Does it depend on the word? On the social context? Give it to us shtraight.