Abstract:

According to the "Simple View of Reading" (SVR; Gough & Tunmer, 1986), reading ability is the product of word identification and listening comprehension processes. Word identification is described as a process that activates phonological, orthographic and semantic representations. But, the latter is not sufficient to read. Reading also uses listening comprehension processes to extract meaning from words, sentences and discourse. In this context, vocabulary is identified as part of comprehension processes. Alternatively, vocabulary could be an external component that would directly affect reading comprehension and word identification (Perfetti, 2010). Indeed, two theories of reading summon vocabulary. 1/ According to the lexical restructuration, vocabulary is strongly related to the development of phonological awareness (Metsala & Walley, 1998). 2/ In the lexical quality hypothesis, semantic representation of word is critical to reading comprehension (Perfetti, 2007; Perfetti & Stafura, 2014). Numerous authors presented data suggesting an extension of the SVR in which knowledge of vocabulary affected the reading comprehension (Braze et al., 2007; Tunmer & Chapman, 2012; Protopapas et al., 2013).

The use of computer-based assessment (CBA) have many advantages, including the standardization of items, automatic data collection and especially the speed of processing which is a good indicator of reading ability in all components. In this study, word identification have been examined at three levels, phonological in a decoding task (children had to decide whether two pseudo-words can be pronounced in the same way (e.g. baccai/baquai: yes; rozan/rossan: no), orthographic in a discrimination task (children had to decide whether two pairs of items are identical or not (e.g. tente-tente: yes; rubis/rudis: no) and semantic in a semantic categorization task (children had to decide whether two words are semantically related or not ("yes" response: pomme – poire; apple – pear; "no" responses: pomme-salon; apple – living room). For comprehension, three processes were assessed in both oral and written comprehension: literal, text-connecting inferences and gap-filling inferences. For vocabulary, children had to decide whether propositions about a target word were right or wrong: eight words were selected from comprehension texts with three questions for each word: also, one version of vocabulary task comprises 24 items with 12 “yes” responses (e.g. “un ballon est-il rond ?” - Is a ball round?) and 12 “no” responses (e.g. “un chat aboie-t-il?” – Does a cat bark?).

We expect 1/ that scores taking into account accuracy and speed should increase with grade levels and 2/ that the weight of vocabulary should increase with age and reading expertise. Thus, vocabulary should be more strongly related to reading comprehension and less related to word identification processes.

Four hundred children from grade 2 to grade 9 have been tested on all types of process involved during reading through the four computerized tasks: word identification, reading comprehension, listening comprehension and vocabulary. Data analyses are in progress. First to examine the expected grade level effect, Anovas will be conducted for the four components of reading (global scores). Then regression analyses will be realized in two groups, children in G2 to G5 and children in G6 to G9.

Keywords:

Reading, Comprehension, Vocabulary, Computer-based assessment.