Atypical phonological processing impairs written word recognition in children with cochlear implants

Abstract

Children equipped with cochlear implant (CI) do not achieve similar levels of word recognition as typical hearing (TH) children, but it is unclear whether the reading deficit results from less accurate phonological representations, atypical reading procedures, or both. Phonological representations are crucial for reading acquisition in an alphabetic writing system, but CI users learn to read without having achieved the same level of speech perception as TH children. In this behavioural study, we addressed whether word reading in children using a CI (n = 25) is as strongly anchored in phonological operations as in TH children, matched for both chronological age (n = 25) and reading experience (n = 25). Using auditory phoneme perception tasks, we confirmed that children with a CI performed less accurately than TH children. When further tested for visual word recognition, CI users applied the same basic reading procedure as TH children, i.e., they read pseudowords through phonological decoding and irregular words through orthographic coding. Finally, using a visual lexical task where subjects had to decide whether pseudowords were or not real words, we observed that CI users rejected word homophones as accurately as TH children, but performed less well than TH controls for rejecting non-homophones pseudowords. Preserved performance for homophones but not for non-homophones relative to controls suggests that children using a CI compensate for defective phonological processing by relying on lexical representations. Altogether, this series of studies allows us to propose that the reading operations in children with a CI are similar in nature as in TH children, yet constrained by less reliable phonological representations.

Keywords:

• phonology
• reading
• word recognition
• development
• cochlear implant

Acknowledgements

We thank Valérian Chambon and Jennifer Martin for their comments on the manuscript; Marina Yao-N’dré, Stéphanie Riès, Éléonore Ardiale, Pauline Ayora, Kim Uittenhove, Soazig Casteau and Suzanne Hodzik for useful discussions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

SB was supported by a scholarship from the French Ministry of Research; SB and ALG are funded by the European Research Council for the Compuslang Project [Grant agreement 260347] and by the Swiss National Fund [Project Number 320030_149319]. WS was supported by a public grant overseen by the French National Research Agency (ANR) as part of the ‘Investissements d’Avenir’ programme [reference: ANR-10-LABX-0083].