Does having ejective consonant phonemes predict having aspirated consonant phonemes?
If yes, should it be explained just so that ejectives are such atypical and hard-to-pronounce sounds and other distinctive features are first utilized? Or is there some deeper connection to do with articulatory phonetics?
The data are
451 languages in UPSID. 119 of them have aspirated stops. (Here, I cheated a bit and supposed that languages with other aspirated sounds have aspirated stops.) 68 of them have ejectives.
I regressed [Aspiration] on [Ejectives] and [Sounds].
[Aspiration] and [Ejectives] are dummies: (1, when the languages has the sounds in question; and 0, when not). [Sounds] is the number of phonemes in the language.
The summary of the statistics is below.
Code: Select all
Call:
lm(formula = Aspiration ~ Ejectives + sounds, data = Totaali)
Residuals:
Min 1Q Median 3Q Max
-1.0618 -0.2579 -0.1360 0.1564 0.9249
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.229422 0.054078 -4.242 2.69e-05 ***
Ejectives 0.144132 0.055248 2.609 0.00939 **
sounds 0.015228 0.001713 8.890 < 2e-16 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.3943 on 448 degrees of freedom
Multiple R-squared: 0.2047, Adjusted R-squared: 0.2012
F-statistic: 57.67 on 2 and 448 DF, p-value: < 2.2e-16
Interpreting estimates of dummy variables is a bit abstract, but having ejectives seems to predict having aspiration b = 0.144132
The result is calculated, controlling for [suonds], i.e. the size of phoneme inventory, so the hypothesis that ejectives appear in big phoneme inventories where aspiration is also utilized as a distinctive feature is not enough to explain the correlation. The estimate is though much bigger without the control, below.
I don't think one can pose causation in phonological typology. Phonemes do not cause other phonemes. But, the correlation is interesting, however.
I would have liked to control for linguistic area, as well, but it was to hard this time.
Code: Select all
Call:
lm(formula = Aspiration ~ Ejectives, data = Totaali)
Residuals:
Min 1Q Median 3Q Max
-0.5294 -0.2167 -0.2167 0.4706 0.7833
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.21671 0.02183 9.927 < 2e-16 ***
Ejectives 0.31270 0.05622 5.562 4.59e-08 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.4272 on 449 degrees of freedom
Multiple R-squared: 0.06446, Adjusted R-squared: 0.06238
F-statistic: 30.94 on 1 and 449 DF, p-value: 4.587e-08