TABLE 3.

Results of generalized linear mixed effects regression models predicting the choices of the participants in the visual salience condition.

	võta see klots siit “take this brick from here”	võta too klots siit “take that brick from here”	võta see klots sealt “take this brick from there”	võta too klots sealt “take that brick from there”
Intercept	3.00***	2.38***	0.41	−1.17
Origin SE vs. NE	−1.19	−1.51	−0.32	−1.16
Word order Pron-Adv vs. Adv-Pron	0.21	−0.44	−0.41	−0.29
Picture 8 vs. 7	0.80	0.24	0.18	−0.09
Picture 9 vs. 7	−0.06	−0.63	−2.48***	−2.45**
Origin SE: picture 8 vs. origin NE: picture 7	NA	NA	NA	−0.01
Origin SE: picture 9 vs. origin NE: picture 7	NA	NA	NA	1.94*
Observations	1,662	1,648	1,803	1,806
Model accuracy (%)	92.59	63.30	59.02	72.73

The reference category of the dependent variable is the salient brick. The accuracy of the models was calculated based on the agreement between the bricks actually chosen and the choices suggested by the probabilities predicted by the model. *p < 0.05, **p < 0.01, ***p < 0.001.