Bibi and Koka

#Coke #cocacola #starwars #bb8 #droid #forceawakens #digiart #digitalart #cool pic.twitter.com/vuJzYzTkcI

— Fresh Composure (@freshcomposure) March 22, 2016

Every now and then, I’ll come across a mention of the bouba/kiki effect, a classic study of sound symbolism that has been revisited several times over the years. The procedure involves showing presenting experiment participants two shapes and two nonsense words, and asking them which word goes with which shape. One of the shapes is always spiky, the other bulbous, such as those shown here. One of the nonsense words consists of voiceless velar or coronal consonants and unround vowels, such as /kiki/. Variants have included the original /takete/, as well as /keiki, kʌte, kʌtiti,/ and /tite/. The other word consists of round vowels and voiced bilabial (usually) consonants, such as /buba, bamu, mabuma, maluma/, and the original /baluma/. As you have probably correctly guessed by now, most speakers tend to put the kiki-type word with the spiky shape, and the bouba-type words with the bulbous shape. “Right, because of the sounds of the word,” you may be saying. But how, exactly, because of the sounds of the word? Maybe it seems obvious to you, as it does to most people, that kiki just sounds like it belongs with something sharp and angular, and bouba with something balloony, but why, exactly?

I wondered: Are people taking their cue from the voiced or voiceless consonants? From the round or unround vowels? From the combination of unround vowels and voiceless velars, or round vowels with voiced bilabials? What would happen if I took kiki and bouba and just swapped the vowels in the two words, to get bibi and koka? What would people do then?

For the past few weeks, I’ve been finding out, running this experiment on my wife and sons, co-workers, fellow parents of members of the high-school marching band, and anyone else as opportunity arose. I’ve presented them with the shapes, drawn on a scrap of paper, and told them to imagine a previously undiscovered tribe of people, with no prior contact with any other civilization, and a language that seems to be unrelated to any known language. They have these two objects or shapes in their culture, and call one of them a bibi, and one a koka. Which is which? If you suspect that the consonants are the deciding factor, then the spiky shape should be the preferred koka, just as it is for kiki, and the globby shape should be the preferred bibi, just as it is for bouba. On the other hand, if you think it’s all about the vowels, then the spiky shape should be favored for bibi, and the rounded one for koka.

I’ve varied the order in which I presented the words, and the orientation of the paper when I show the shapes, to avoid bias based on order of presentation. Unfortunately, there are other sources of bias. After running my first three subjects (my wife, Doug, and Adam) by presenting the words on slips of paper, I realized that the angles of the letter k and the curves of the letter b might be a source of bias. Even now, this could still be a source of bias for literate participants, since they may imagine these words written down, but I couldn’t do much about that. (For a study involving pre-literate participants, see Maurer, Pathman and Mondloch (2006). I found it in the references for the Wikipedia article I cited above). In addition, by taking the most-popular words for this experiment, kiki and bouba, and making the minimal change of swapping their vowels, I’ve ended up with two words that actually do mean something in English. /koka/ can be a leaf, a carbonated drink, or of course, the Corpus of Contemporary American English. /bibi/ is a famous blues singer, an Israeli prime minister, a lovable droid, or the ammo for an official Red Ryder, carbine action, 200-shot, range model air rifle, with a compass in the stock and this thing that tells time. Any of these could provide associations that lead a subject to choose one shape or the other, but hopefully with a large enough sample size, they won’t matter.

So what happened? So far, my sample size is 27. Of them, 15 participants (55%) mapped bibi to the rounded shape, and koka to the angular one. The other 12 (44%) mapped bibi to the angular shape, and koka to the rounded one. So it looks like the consonants have a slight edge, but not much of one. In fact, if I throw out the data from my wife and sons, who were looking at written representations, only 12 participants mapped bibi to the rounded shape, and it’s now a 50-50 split. So maybe it’s the combination of vowels and consonants that produce the well-defined bouba/kiki effect, since it mostly disappeared when I flip-flopped the vowels.

There are other possibilities, too. Maurer et al., who hypothesized that the roundness or unroundness of the vowels was the important factor, mentioned that it could be that English or other languages have a detectable pattern whereby real words with rounded vowels (or spelled with round letters) denote round things, and real words with unround vowels refer to pointy things. For support, they point out that there has been one study where the kiki/bouba effect did not show up, involving the Songe people of Papua New Guinea. Maybe, they note, this language doesn’t have the same kind of previously existing sound/shape correspondences. This was why Maurer et al. wanted to do their experiment on very young children, who hadn’t had as much exposure to their native languages.

More intriguingly, Maurer et al. suggest that “the effect is stronger for some consonant/vowel pairings than others” (p. 320). They suggest this because in just one out of their four pairs of nonsense words, one word contained voiced velars and round vowels–in other words, the same combination I had in koka that I didn’t think other researchers had looked at. Their word was /goga/, and for the pair /goga, tite/, they did not get the clean mapping that they got with the other words. (They also suggested that it could have been a problem in the story that they told their toddler subjects for this particular pair of words.)

So that’s my bibi/koka experiment. If you try it on your friends or family, let me know how it goes.

Unexpected Glottal Stops

It began a couple of months ago, as I would listen to the morning news on the radio. Whenever this one guy from the Ohio Statehouse News Bureau signs off, he says, “Andy Chow, Statehouse News Bureau,” but he pronounces Statehouse as [steɪʔhɑus], realizing the /t/ as a glottal stop, instead of turning it into a tap, like I do: [steɪɾhɑus]. I thought it was just a one-time pronunciation glitch the first time I heard it, but the next day, he did it again. I started to listen for more of Andy Chow’s unexpected glottal stops, and they were there: whenever a word ended with a stressed syllable followed by /t/, and the following word also began with a stressed syllable, possibly with an /h/ at the front.

This is not where I expect glottal stops in American English. In a post on his now-discontinued but still great Phonetiblog, John Wells quotes himself from his Longman Pronunciation Dictionary on glottal stops in American English:

ʔ is found as an allophone of t only
• at the end of a syllable, and
• if the preceding sound is a vowel or sonorant

Provided these conditions are satisfied, it is widely used in both BrE and AmE where the following sound is an obstruent

football ˈfʊt bɔːl → ˈfʊʔ bɔːl
outside ˌaʊt ˈsaɪd → ˌaʊʔ ˈsaɪd
that faint buzz ˌðæt ˌfeɪnt ˈbʌz → ˌðæʔ ˌfeɪnʔ ˈbʌz

or a nasal

atmospheric ˌæt məs ˈfer ɪk → ˌæʔ məs ˈfer ɪk
button ˈbʌt ən → ˈbʌʔ n
that name ˌðæt ˈneɪm → ˌðæʔ ˈneɪm

or a semivowel or non-syllabic l

Gatwick ˈɡæt wɪk → ˈɡæʔ wɪk
quite well ˌkwaɪt ˈwel → ˌkwaɪʔ ˈwel
brightly ˈbraɪt li → ˈbraɪʔ li…

This has been my understanding of American English glottal stops up until now. I take it to be an indication of the novelty of this pronunciation that even John Wells, who has made a career out of knowing this stuff, doesn’t mention it at all.

The next phase began when I heard Doug refer to that classic 1990s comedy cartoon duo, Beavis and Butthead. He pronounced Butthead as [bʌʔhɛd] instead of [bʌɾhɛd]. Once I heard him say that, I started listening more closely, and now know that he regularly produces a glottal stop in such words as butthole and pothole as well. Just a couple of days ago, he was making spaghetti sauce, and said,

I [heɪʔ] how the brown sugar gets so hard.

(Yes, we put 2 tablespoons of brown sugar in our spaghetti sauce. So what?)

Finally, I drove from Ohio to Northern Virginia a few weekends ago for the funeral of the wife of oldest friend. On the way back, I listened to this episode of This American Life, which was devoted to a single story reported by Susan Zalkind. As I drove, I realized that Zalkind had this pronunciation, too. Every few minutes, she’d do it again, in a string like met Eric or shot Ibragim. But having an entire episode to listen to, I decided to listen closer, to hear if there were places where she had the opportunity to make one of these glottal stops, but realized her /t/ as a tap. It turned out there were, and that they had just been slipping by, undetected because they sounded so normal.

When I got back home, I re-listened to the podcast, and wrote down every example of /t/ that occurred at the end of a word before a word that began with a vowel or /h/ plus a vowel. I kept a list of /t/ realized as [ʔ] and /t/ realized as [ɾ], put them in a table, and was surprised to find that the two columns were just about equal. The glottal stop hadn’t completely taken over this phonetic environment after all.

So then the question was whether Zalkind (and others with this pronunciation) used it randomly, or there was some rule that could predict when she would use it. It didn’t seem to matter whether the following word began with a stressed syllable (e.g. at all) or unstressed (e.g. about it), or what vowel the second word began with. But I was able to make one generalization: When the second word began with /h/–in other words, the very environment that I’d noticed with Andy Chow’s Statehouse and Doug’s butthead–the /t/ was almost certain to be realized as a glottal stop. Out of 17 examples of /t/ at the end of a word before a word beginning with /h/, 15 of them realized /t/ as [ʔ]. Furthermore, if that second word began with a stressed vowel, chances of a glottal stop were 100%. (The /h/ examples appear at the bottoms of their respective columns.) In other words, a phrase like beat him up was likely to contain a glottal stop, and a phrase like got home was certain to.

In thinking about this pronunciation, I’ve begun to wonder why I should consider it such a natural environment for speakers like me to have a tap. The canonical location for [ɾ] is between a stressed and an unstressed vowel. This isn’t the case in a word like statehouse, where the vowels on both sides of /t/ are stressed, and we have an intervening consonant, /h/. In fact, having a glottal stop before /h/ would allow Wells’s rules to be stated more concisely. Instead of referring to “obstruent, nasals, semi-vowels, and syllabic /l/,” it could refer to “all consonants”. Well, make that, “all consonants except /r/”. Even so, this pronunciation that sounds so strange to me can be seen as just a step in the direct of regularity.

If you have encountered this pronunciation or use it yourself, leave a comment! (And not just any comment; a comment on the pronunciation. But of course, you knew that from the Maxim of Relevance.)