Table 1 Results from multiple regressions predicting L2 change based on standardized L2 performance before the training, standardized MST leaf fraction before the training and change in MST leaf fraction over the training.

Frequency Band	L2 Measure		ΔR2	β	SE β	Bootstrap CI (95%)		p
Frequency Band	L2 Measure		ΔR2	β	SE β	lower	upper	p
Alpha	Complexity		.547
		Intercept		0.976*	0.166	0.176	0.970	.000
		L2 pre-training		− 0.815	0.172	− 1.261	− 0.360	.000
		Leaf fraction pre-train		− 0.148	0.186	− 0.658	0.317	.434
		Leaf fraction change		0.201	0.182	− 0.116	0.632	.282
	Accuracy		.122
		Intercept		1.471*	0.093	1.142	1.682	.000
		L2 pre-training		− 0.119	0.101	− 0.513	0.212	.253
		Leaf fraction pre-train		− 0.027	0.110	− 0.231	0.196	.811
		Leaf fraction change		0.091	0.102	− 0.111	0.303	.380
	Fluency		.118
		Intercept		1.351*	0.117	1.005	1.877	.000
		L2 pre-training		0.088	0.121	− 0.352	0.622	.476
		Leaf fraction pre-train		0.128	0.131	− 0.112	0.527	.338
		Leaf fraction change		0.178	0.127	− 0.026	0.421	.175
Beta	Complexity		.583
		Intercept		0.944*	0.162	0.286	0.879	.000
		L2 pre-training		− 0.792*	0.162	− 1.233	− 0.534	.000
		Leaf fraction pre-train		0.440*	0.192	0.070	0.994	.032
		Leaf fraction change		0.241	0.174	− 0.014	0.549	.179
	Accuracy		.308
		Intercept		1.431*	0.084	1.103	1.611	.000
		L2 pre-training		− 0.121	0.088	− 0.472	0.133	.183
		Leaf fraction pre-train		0.222*	0.101	0.017	0.496	.039
		Leaf fraction change		0.229*	0.091	0.022	0.440	.019
	Fluency		.206
		Intercept		1.333*	0.113	1.103	1.922	.000
		L2 pre-training		0.115	0.117	− 0.230	0.848	.337
		Leaf fraction pre-train		0.295*	0.134	− 0.041	0.687	.039
		Leaf fraction change		0.164	0.123	− 0.170	0.399	.197

Bootstrapped confidence intervals were calculated on the basis of 2000 bootstrap samples. Coefficients marked * are significant (p < .05), those marked bold are robust in that they do not contain zero in the bootstrapped confidence intervals. None of the reported correlations between L2 development and MST leaf fraction remained significant when FDR correction for multiple comparisons was applied to the p-values of MST predictors.

Search