Table 2.

Syntactic and semantic features used for predictive modeling

Description	Example
a. Adjective, comparative	“braver”, “closer”, “cuter”
b. Possessive pronoun	“her”, “his”, “mine”, “my”, “our”, “ours”, “their”, “your”
c. WH‐determiner	“that”, “which”, “what”
d. WH‐pronoun	“that”, “what”, “which”, “who”, “whom”
e. WH‐adverb	“how”, “however”, “whenever”, “why”
f. Minimum coherence at 5‐level, normalized
g. Minimum coherence at 5‐level
h. 90th percentile coherence at 5‐level
i. Maximum coherence at 6‐level
j. Mean coherence at 7‐level, normalized
k. Standard deviation coherence at 7‐level, normalized
l. 90th percentile at 7‐level
m. Standard deviation coherence at 7‐level
n. 90th percentile at 8‐level

A k‐level measure of semantic coherence was used, which computes word‐to‐word variability at “k” inter‐word distances, with k ranging from 5 to 8