Table 6. Fixed effects of optimal LMM for initial landing positions in the test words.

	Estimate	Std. Error	t value
(Intercept)	-0.56344	0.11509	-4.89566
FREQ	0.02252	0.01820	1.23699
PRED	0.01517	0.01128	1.34415
LENGTH	-0.20296	0.02242	-9.05257
LAUNCH	0.43429	0.02497	17.39223
FREQ:LENGTH	0.02822	0.00931	3.03042
FREQ:LAUNCH	0.00907	0.00575	1.57741
PRED:LENGTH	0.01479	0.00635	2.32793
LENGTH:LAUNCH	0.04709	0.00517	9.10263
FREQ:LENGTH:LAUNCH	0.00517	0.00298	1.73603

Initial eye landing positions were expressed in letters relative to the center of the test words. The fixed structure included the effects of word length (“LENGTH”; 3–11 letters), word frequency (“FREQ”; between -1.97 and 7.16 log units), word predictability (“PRED”; between -2.60 and 2.60 logit units), and saccadic launch-site distance (“LAUNCH”; between -8.00 and -0.002 letters from the space in front of the test words), as well as the two-way interaction between word predictability and word length, the three-way interaction between word frequency, word length and launch site and all corresponding two-way interactions; the random structure included a random intercept by participant and sentence pair, as well as by-participant random effects of word length, word predictability and saccadic launch-site distance (see S1 Table). The intercept estimate gives the initial landing position when all variables were at their reference, mean, value (Word Length: 6.20 letters; Launch Site: -4.39 letters; Word Frequency: 2.91 log units; Word Predictability: -0.97 logit units). Colon stands for interaction. See S5 Table for the corresponding minimalist optimal GLMM.