Beautiful creak

The above is a sound wave (made in Praat) of an interviewee slowly answering "yes, yes, yes, yes, yes," to a question. The last two yesses (circled) are creaked. This is an excellent example of typical creaky syllables which are often quieter than modal ("normally-voiced") syllables.

Here are two close-ups to compare the waveforms of a modal yes and a creaky yes; the creaky yes has a jagged waveform and irregular glottal pulses (squared).

Modal Yes.

Creaky Yes.

I'm still not sure what social meaning it is accomplishing for this speaker in this situation (I'm still coding my data), but luckily it's easy to identify both auditorily and visually!

So "creakin'" annoying!

This SNL bad news commercial skit, besides being hilarious, is highly relevant for linguists interested in the social meaning of a popular form of phonation used by young American women: creaky voice. The star of the skit uses serious creak at the end of phrases and on words like, "apologize," "that," "is," etc. SNL's view of women using creaky voice? OBNOXIOUS.

Glamour magazine also joins in the conversation on creak in their April 2011 issue with an article called, "You're a bright woman! So...What did you just say?" Among its advice to women interviewing for jobs is to:

• avoid "upspeak" - ending non-interrogative sentences with high intonation
• get "like-osuction" (i.e. stop using "like" so much)
• and avoid creaky voice

The article makes frequent reference to a speech therapist by the last name, Chwat (seriously?), who says this type of voice sounds like, "crackling in a frying pan." Among the many attributes creak can index, according to Glamour and Chwat, are: sexy, uncertain, annoying and smugness.

Finally, another very modern example of female creak in the media is Mary-Kate Olsen's interview voice. Sexy? Uncertain? Annoying? Smug? (perhaps all of the above?)

How we say what we mean

Recently, linguists have begun to study a fascinating intersection of language: discourse and the 'paralinguistic' phenomena that help accomplish social meaning. Conversations are potentially messy social situations, yet we manage to communicate in groups without cue cards telling us when to come in, when someone's turn is complete, etc. One of the ways talk is managed is by the production and perception of paralinguistic phenomena such as pitch, intonation, tempo, and voice quality that mark important discursive events. Apparently how we say something is every bit as important as what we say.

Ogden, R. (2006). Phonation and social action in agreements and disagreements. Journal of Pragmatics (38): 1752-1775.

Ogden uses data from corpora of natural conversations and identifies similar stretches of talk based on linguistic and pragmatic aspects. Specifically, he identifies instances of agreement and disagreement based on the concept of “assessments” and their structure from the field of Conversation Analysis. Ogden uses a pitch track created in Praat to compare the utterances. He finds that the assessments that agree with a previous opinion by a different speaker use a large pitch span (including some creaky voice), “dynamic pitch contours,” slower tempo, and “tenser articulations” (1762). Overt disagreements also share these aspects but are timed directly on the heels of the first speaker’s utterance. Indirect disagreements padded with an initial weak agreement were produced more quickly and quietly (1767), the weak agreement ending in a low-high contour (1768).

A recent study of creaky voice

Yuasa. (2010). Creaky voice: A new feminine voice quality for young urban oriented upwardly mobile American women? American Speech. 85(3): 315-337.

Yuasa conducts a two-part study of creak: an auditory and acoustic analysis of creaky voice in young speakers and a perception survey completed by college students about the social significance of women’s use of creak in natural conversation. The conversations average ten minutes in length and focus on the topic of food, what she sees to be a neutral topic (but is this actually neutral, and how is it "natural" to control for topic?)

Yuasa counts creaky voice at the word level, in a 401-word sample taken randomly from each recording. She locates creak auditorily and confirms her assessment using the program Praat to look at the waveforms and spectrograms of the words. For her statistical assessment of the data, Yuasa uses a t-test. She finds that the differences between the American female speakers and the American men (p = 0.0133) and the American female speakers and the Japanese female speakers (p = 0.0267) is significant; American women have a higher rate of use of creak.

The perception study includes 175 college students in both Northern California and Eastern Iowa who listened to and responded to questions about audio files online (one modal voice recording and one creaky voice recording from the same female speaker). Then they were asked 1.) if they heard the second type of voice (creak) a lot where they live and 2.) what traits the women in the recordings project just with their voice and 3) to rate the women's characteristics based on their voice quality with predetermined binary choices such as formal versus informal.

Many of the participants in the survey remarked that they often heard creak where they lived and that the voice type suggests the following attributes: hesitant, informal, non-aggressive, as well as educated, professional and urban. Yuasa suggests that young American women may be using creaky voice more to continue the trend of “lowering” their voices to reappropriate creak as portraying the contemporary, urban and upwardly-mobile woman.

While the above study is thought-provoking and Yuasa's analysis is intriguing, saying that creaky voice indexes numerous characteristics at once (based on survey questions that gave participants binary choices, perhaps forcing some responses) for American women in general is an ambitious claim that requires much more research, especially since the social meaning of creaky voice and women has only recently been a topic of study.