yawwwn vs. yaaawn

Here's another sports & media & language bit.

This morning the NHL announced a deal has been reached "in principle" to end the 10 month labour dispute. I checked the Yahoo sports page and encountered the following poll:

Poll: are you excited?
[ ] Definitely! [ ] Yaw-w-w-n

These polls are annoying because of their false-alternative setup. Nevertheless I picked "excited", because I lean more that way than "yawn". But I also thought, in writing, is that really how you express an exaggeratedly long vowel?

The issue for me is not the hyphens, but the choice of using multiple w's instead of multiple a's. Just the kind of thing to investigate with some GoogleCorpusSearching.

Not expecting much luck with "yawn" extensions, I first tested aaah vs ahhh, with repetitions up to 10. This is kind of a replication of a Language Log study. Here are the results:

string	Gh(K)	string	Gh(K)
ahh	1,910	aah	838
ahhh	1,240	aaah	748
ahhhh	659	aaaah	321
ahhhhh	562	aaaaah	147
ahhhhhh	245	aaaaaah	88.6
ahhhhhhh	155	aaaaaaah	47.2
ahhhhhhhh	115	aaaaaaaah	49.2
ahhhhhhhhh	89.1	aaaaaaaaah	33.7
ahhhhhhhhhh	65.1	aaaaaaaaaah	26.4

I think the appropriate test is a paired-subjects t-test; MeanA-MeanB=319.011, t = 2.8, df = 9, p = 0.023198. It looks like repeating the second chacracter of a digraph is the preferred way to express exaggerated length.

[Incidentally, the high number of [ahh] actually lowers the p-value. I reran the analysis lowering it to 1,410, based on the assumption that lots of [ahh]'s refer to various associations, and p drops quite a bit.]

Intrigued, I decided to go after yaawn vs yawwn anyway. Results are below:

string	Gh(K)	string	Gh(K)
yawwn	2.02	yaawn	6.95
yawwwn	0.770	yaaawn	5.06
yawwwwn	0.674	yaaaawn	3.57
yawwwwwn	2.250	yaaaaawn	0.742
yawwwwwwn	1.280	yaaaaaawn	1.83
yawwwwwwwn	0.720	yaaaaaaawn	2.03
yawwwwwwwwn	0.446	yaaaaaaaawn	1.21
yawwwwwwwwwn	0.301	yaaaaaaaaawn	0.834
yawwwwwwwwwwn	0.302	yaaaaaaaaaawn	0.568

Why these have second peaks at around 6 repetitions is kind of interesting. I think the extra-extra repetition (i.e., 6 rather than 2) makes the string look less like a typo, a problem that ahh and ahhh don't seem do have. (This might also explain the appearance of the dashes in Yahoo's yaw-w-w-n.

But also interesting is how there's more repetition of the first letter in the digraph. This is nearly significant on a paired-subjects t-test: MeanA-MeanB = -1.559, t = - 2.25, df = 8, p = .0545. Who knew?