Table 1. Results from mixed effects logistic regression model on Japanese adult productions of /s/ and /ʃ/.

The dependent variables are the two target fricative categories (/s/ versus /ʃ/) and the independent variables are z-normalized values of the three acoustic parameters (fixed effect) and individual speakers (random effect).

	Estimate	Standard error	z-value	p value
Intercept	23.172	4.486	5.165	p < 0.001
M1	−0.003	0.001	−6.213	p < 0.001
M2	−0.003	0.001	−3.072	0.002
Onset F2 frequency	0.003	0.001	2.845	0.004