Table 1.

Results of fixed effects in the linear mixed model, predicting the P(F&M) order from the relative accessibility of female and male terms, the log-frequency of M&F order, and the cluster strength of related cases as weighted by semantic and morphological similarity.

	Estimate	SE	df	t-value	Pr(>|t|)
P(F&M) ~ Competition * Accessibility * ClusterStrength + decade + (1+ decade | item)
Fixed effects
(Intercept)	0.010	0.108	41	0.090	0.929
Competition	−0.631	0.063	225	10.025	0.001
Accessibility(F-M)	0.191	0.064	306	2.967	0.003
ClusterStrength	0.067	0.033	395	2.052	0.041
Decade	0.178	0.037	65	4.792	0.001
Competition:acc	−0.029	0.043	458	−0.684	0.494
Competition: cluster	−0.062	0.027	241	−2.325	0.021
acc:cluster	0.058	0.036	248	1.601	0.111
Competition:acc:cluster	0.122	0.034	280	3.643	0.001
	Random effects
	Variance	SD	Correlation
Item (Intercept)	0.51863	0.7202
Decade	0.03758	0.1939	0.28
Residual	0.04422	0.2103
		Model fit
	Marginal	Conditional
R2	0.438	0.959

Number of obs: 488, groups: item, 45. Boldface indicates statistical significance.