Independent effects of collocation strength and contextual predictability on eye movements in reading

Figures & data

Table 1. Summary of stimulus characteristics.

Stimulus Characteristic	Strong Collocation	Weak Collocation
Adjective length (letters)	5.9 (.2)	6.1 (.3)
Adjective length (syllables)	1.9 (.1)	2.1 (.1)
Adjective frequency (counts/million)	4.5 (.1)	4.7 (.1)
Phrasal frequency (counts/million)	296.7 (59.8)	16.2 (2.9)*
MI score	8.6 (.3)	4.4 (.2)*
Naturalness score (Predictable Context)	4.2 (.4)	4.1 (.5)
Naturalness score (Neutral Context)	4.1 (.4)	4.1 (.6)

Note: The Standard Error of the Mean, calculated across both participant and item variance, is shown in parentheses. Asterisks indicate statistically significant differences, p < .05.

Figure 1. An example stimulus. Collocations are shown underlined with the alternative weak and strong collocations separated using a slash. Note that sentence stimuli were shown normally and including either the strong or weak collocation in the experiment.

Table 2. Eye movements for the collocation.

	Predictable Context		Neutral Context
	Strong Collocation	Weak Collocation	Strong Collocation	Weak Collocation
FPRT (ms)	379 (6)	398 (7)	391 (6)	407 (7)
RO (%)	7 (1)	13 (1)	14 (1)	16 (1)
RPD (ms)	427 (8)	466 (9)	477 (10)	504 (10)
TRT (ms)	473 (9)	539 (11)	536 (11)	602 (12)
RI (%)	19 (1)	24 (2)	28 (2)	31 (2)

Note: The Standard Error of the Mean is shown in parentheses. FPRT = first-pass reading time, RO = regressions-out, RPD = regression path duration, TRT = total reading time, RI = regressions-in.

Table 3. Summary statistics for the collocation phrase.

Factor	Statistic	FPRT	RPD	TRT	RI	RO
Intercept (global mean)	β	407.42	487.46	560.65	−1.4	−2.82
	SE	8.49	7.03	7.85	0.21	0.31
	t/z	47.99	69.3	71.42	−6.79	−9.15
Context (predictable-neutral)	β	−10.53	−26.6	−43.18	−0.49	−0.54
	SE	4.68	4.8	6.37	0.09	0.13
	t/z	−2.25*	−5.54*	−6.78*	−5.21*	−4.09*
Collocation (weak-strong)	β	19.36	35.17	70.6	0.28	0.55
	SE	5.8	9.26	7.46	0.09	0.13
	t/z	3.34*	3.8*	9.47*	2.99*	4.19*
Context  × Collocation	β	−1.45	4.43	−7.78	0.01	0.34
	SE	7.15	7.83	7.24	0.19	0.26
	t/z	−0.20	0.56	−1.07	0.03	1.28

Note: Asterisks indicate statistically significant effects, p < .05. FPRT = first-pass reading time, RPD = regression path duration, TRT = total reading time, RO = regressions-out, RI = regressions-in.

Table 4. Eye movement measures for the collocation noun.

	Predictable Context		Neutral Context
	Strong Collocation	Weak Collocation	Strong Collocation	Weak Collocation
SKIP (%)	21 (1)	18 (1)	17 (1)	16 (1)
FFD (ms)	213 (2)	224 (3)	221 (3)	226 (3)
SFD (ms)	212 (3)	227 (4)	222 (3)	234 (4)
GD (ms)	228 (3)	241 (4)	241 (4)	241 (3)
RO (%)	6 (1)	11 (1)	8 (1)	11 (1)
RPD (ms)	256 (6)	290 (8)	276 (6)	293 (7)
RI (%)	17 (2)	19 (2)	22 (2)	23 (2)
TRT (ms)	270 (5)	300 (6)	301 (6)	333 (7)

Note. The Standard Error of the Mean, calculated across both participant and item variance, is shown in parentheses. All values are in ms unless otherwise stated. SKIP = word-skipping rate, FFD = first-fixation duration, SFD = single-fixation duration, GD = gaze duration, RO = regressions-out, RPD = regression-path duration, RI = regressions-in, TRT = total reading time.

Table 5. Summary statistics for the collocation noun.

Factor	Statistic	SKIP	FFD	SFD	GD	RPD	TRT	RI	RO
Intercept (global mean)	β	−2.59	220.61	225.32	237.87	282.93	302.3	−1.61	−2.59
	SE	0.39	6.83	7.26	6.44	8.47	8.79	0.17	0.38
	t/z	−6.71	32.3	31.03	36.84	33.41	34.38	−9.74	−6.71
Context (predictable-neutral)	β	0.37	−4.97	−8.32	−7.52	−11.33	−23.51	−0.31	0.37
	SE	0.11	2.19	2.66	2.69	3.95	4.18	0.11	0.11
	t/z	3.33*	−2.26*	−3.12*	−2.79*	−2.87*	−5.63*	−2.81*	3.33*
Collocation (weak-strong)	β	−0.16	9.99	15.5	7.95	27.68	32.08	0.14	−0.16
	SE	0.11	4.74	5.96	5.51	8.84	8.39	0.11	0.11
	t/z	−1.48	2.11*	2.6*	1.44	3.13*	3.82*	1.3	−1.48
Context  × Collocation	β	−0.19	4.38	−0.12	9.48	10.31	−6.31	0.01	−0.19
	SE	0.22	4.22	5.01	4.99	6.75	7.57	0.22	0.22
	t/z	−0.85	−0.02	−0.02	1.89	1.53	−0.83	0.03	−0.85

Note: Asterisks indicate statistically significant effects, p < .05. SKIP = word-skipping rate, FFD = first-fixation duration, SFD = single-fixation duration, GD = gaze duration, RO = regressions-out, RPD = regression-path duration, RI = regressions-in, TRT = total reading time.