Reading in Arabic: the impact of vowellization uncertainty on word recognition

Reading in Arabic: the impact of vowellization uncertainty on word recognition

First Author: Rob Davies -- Lancaster University
Additional authors/chairs: 
Ahmed Alhussein; Gert Westermann
Keywords: Arabic, orthographic processing, Reading, visual word recognition, Phonological Recoding
Abstract / Summary: 

Words in Modern Standard Arabic (MSA) are usually printed without short vowels so that a single unvowellized letter-string can be associated with two or more distinct pronunciations. It has been assumed that MSA incorporates a high level of homography, and that this presents a challenge to the reader. However, this challenge obtains only to the extent that readers know alternate pronunciations for any letter-string. We conducted three studies with adult Arabic speakers to record what readers know about the different pronunciations that can be associated with (potentially) homographic MSA letter-strings, and to identify the effects on word recognition of variation in homography.

(1.) We presented 1,474 unvowellized MSA letter-strings to 445 participants who were asked to produce the one or more word forms (with short vowels) evoked by each letter-string. We asked (2.) 38 participants to read aloud 1,474 letter-strings corresponding to MSA words, and (3.) 40 participants to distinguish 1,352 letter-strings (corresponding to MSA words) from 1,352 matched non-word letter-strings in a lexical decision task.

We estimated the extent to which MSA letter-strings normally evoke alternate pronunciations (vowellization uncertainty). Mixed-effects analyses showed response latencies were faster in lexical decision and oral reading for more frequent, earlier-acquired, words. Greater vowellization uncertainty slowed oral reading latencies but did not affect decision latencies.

Oral reading requires identification of a unique phonological representation so the effect of vowellization uncertainty may reflect resolution of a competition between candidate pronunciations. Word recognition in Arabic is otherwise informed by shared orthographic structure.