A Meta-Analytic Review of Naming-Speed Deficits in Developmental Dyslexia

ABSTRACT

This study presents a meta-analytic review of serial rapid automatized naming (RAN) deficits in individuals with dyslexia relative to typical readers (based on 216 effect sizes comprising 8335 dyslexic readers, 14,083 age-matched controls, and 921 reading-matched controls). A random-effects model analysis indicated a large impairment in speeded RAN in individuals with dyslexia compared with age-matched controls (d = 1.19) but a similar performance when compared with reading-matched controls (d = 0.13). In addition, dyslexic readers presented a deficit in discrete-naming formats (d = 0.74), although the deficit in serial RAN was notably larger; hence, adding seriality is particularly detrimental for these readers. The deficit appears to span all stimulus types (alphanumeric and nonalphanumeric), indicating that processes beyond letter processing are responsible for the delays and are independent of set size. Poor RAN is a long-term and universal symptom of dyslexia, and the transparency of the writing system does not influence its severity.

Acknowledgments

The first author and this work were supported by IF 2015 Program of the Portuguese Foundation for Science and Technology, FCT (ref. IF/00533/2015) and by the Research Center for Psychological Science at Universidade de Lisboa (CICPSI). This work was also supported by a scientific project supported by FCT, ref. EXPL/MHC-PCN/0299/2013, UID/BIM/04773/2013 CBMR, PTDC/PSI-GER/32602/2017.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary data

Supplemental data for this article can be accessed here.

Notes

¹ We avoid the term discrete RAN because it usually refers to variants of the standard RAN test (Denckla & Rudel, 1976), which has been occasionally used in recent studies (e.g., Jones et al., 2009, in which individual items were presented discretely but in positions analogous to the continuous-matrix version). More often, discrete/confrontation naming studies have used isolated naming tests that do not have the same nature as RAN (using, e.g., a larger set of unrepeated items presented in a single location, e.g., as in Araújo et al., 2011).

² Defined as the number of items that the eyes are ahead of the voice at the onset of vocal response.

³ As can be determined from the supplementary material, in some of the studies that were included based on the first criteria and that explicitly described their samples as dyslexic/with a specific reading disability, either this criterion could not be verified or the cutoff point was lower than −1.25 SD. These studies were left in the analysis to avoid a massive loss of data (it may well be that for those considering a less stringent criterion than −1.25 SD, the dyslexic participants were somehow already compensated). Regarding those studies that were included based on the second criteria, we determined whether the pattern of results held constant when those using a less stringent criterion of −1 SD were also included. The overall mean effect size was likewise large and significant (k = 194, d = 1.19), 95% CI [1.11, 1.30].

⁴ Typically developing European-Portuguese children show a reading accuracy quite similar to that of children from deeper orthographies (i.e. orthographies with less grapheme–phoneme consistency), but reading is typically faster in Portuguese (Seymour et al., 2003) because it is an orthography of intermediate depth (for details, see, e.g., Lima & Castro, 2010; Sucena, Castro, & Seymour, 2009), that is, more transparent than French or English but deeper than Italian, Spanish, or Dutch.

⁵ Hebrew and Arabic have two forms of script, a transparent—the pointed script (with diacritics)—and an opaque, or the unpointed script (without diacritics). Reading skills are initially trained with the pointed script, but by the fourth grade, children are expected to have mastered reading of the unpointed script, and this script is used almost exclusively thereafter.

Additional information

Funding

This work was supported by the Fundação para a Ciência e a Tecnologia [IF/00533/2015, UID/BIM/04773/2013 CBMR and PTDC/MH].