The neural underpinnings of reading skill in deaf adults

Highlights

• Skilled readers had greater activation in left IFG and MTG when reading for meaning.

• Reading skill correlated with activity in left IT (orthographic-semantic interface).

• Semantic task accuracy correlated with activity in left ATL (conceptual processing).

• Phonological task accuracy correlated with activity in left IFG (articulatory coding).

• Reading skill did not correlate with neural activity during the phonological task.

Abstract

We investigated word-level reading circuits in skilled deaf readers (N = 14; mean reading age = 19.5 years) and less skilled deaf readers (N = 14; mean reading age = 12 years) who were all highly proficient users of American Sign Language. During fMRI scanning, participants performed a semantic decision (concrete concept?), a phonological decision (two syllables?), and a false-font control task (string underlined?). No significant group differences were observed with the full participant set. However, an analysis with the 10 most and 10 least skilled readers revealed that for the semantic task (vs. control task), proficient deaf readers exhibited greater activation in left inferior frontal and middle temporal gyri than less proficient readers. No group differences were observed for the phonological task. Whole-brain correlation analyses (all participants) revealed that for the semantic task, reading ability correlated positively with neural activity in the right inferior frontal gyrus and in a region associated with the orthography-semantics interface, located anterior to the visual word form area. Reading ability did not correlate with neural activity during the phonological task. Accuracy on the semantic task correlated positively with neural activity in left anterior temporal lobe (a region linked to conceptual processing), while accuracy on the phonological task correlated positively with neural activity in left posterior inferior frontal gyrus (a region linked to syllabification processes during speech production). Finally, reading comprehension scores correlated positively with vocabulary and print exposure measures, but not with phonological awareness scores.

Introduction

Very little is known about the neural adaptations that support skilled reading when the process of learning to read is altered by deafness rather by a specific reading disability. Reading presents a challenge for the majority of deaf children because they do not receive the same everyday exposure as hearing children to the language that is encoded by print. Hence, most deaf children begin reading instruction without the benefit of the strong language foundation and well-established pedagogical methods available to their hearing peers. Although reading outcomes are generally poor for deaf people, some deaf individuals do, nonetheless, achieve high levels of reading proficiency (Qi & Mitchell, 2012). However, how such skilled readers differ from less-skilled deaf readers is poorly understood because most investigations simply compare groups of deaf and hearing readers without regard to the variability of reading skill or language background within the deaf group. The goal of the present study was to identify possible neural signatures that differentiate between skilled and less-skilled reading in deaf adults.

This study focuses on prelingually deaf adults with severe-to-profound hearing loss who acquired American Sign Language (ASL) in early childhood. We target this population because they form a relatively homogeneous group and they are less likely to access phonological codes when reading, compared to “oral” deaf readers who acquire only a spoken language (e.g., Hirshorn et al., 2015, Koo et al., 2008). One question of interest is whether and how impoverished (e.g., less precise) phonological representations of speech affect the reading system in deaf readers. Emerging evidence suggests that phonological abilities may be only weakly related to reading skill for deaf people (Mayberry et al., 2010, Miller and Clark, 2011), although this idea is controversial (Paul, Wang, Trezek, & Luckner, 2009). Here we used fMRI to examine both phonological and semantic processing of English words by deaf readers whose reading levels ranged from the fourth grade (9 years) to college level (22 years). One aim of the study was to determine whether the less-skilled readers rely on a distinct neural circuit when reading (for meaning or when extracting speech-based phonological information) compared to skilled deaf readers.

In a recent fMRI study of implicit word reading (detect an ascending letter/symbol in words vs. false-font strings), Corina, Lawyer, Hauser, and Hirshorn (2013) reported that relatively skilled deaf readers (mean reading age = 14 years) engaged several brain regions (left superior temporal gyrus (STG), bilateral fusiform gyri, and bilateral inferior parietal lobules) to a greater extent than less skilled deaf readers (mean reading age = 9.9 years). Separate examination of the less-skilled readers (p < 0.005, uncorrected) revealed bilateral activation in the middle frontal gyrus (BA 46/9), which was not observed for the more skilled readers. Based on these activation patterns, Corina et al. (2013) suggested that less skilled deaf readers may “process English word forms as non-decomposable logographic-like forms analogous to Chinese” (pg. 6). However, the between-group contrast did not reveal greater activation in the middle frontal gyri (or in any brain area) for the less-skilled deaf readers, even at uncorrected thresholds, and thus this characterization must be considered tentative.

Using a lexical decision task (and a feature detection task with unreadable symbols as the baseline), Aparicio, Gounot, Demont, and Metz-Lutz (2007) found that less skilled deaf readers (all bilingual in French and French Sign Language) exhibited greater activation in the right inferior frontal gyrus (IFG) and left STG, as well as reduced activation in left IFG compared to more skilled hearing readers. However, these differences were only evident in region of interest (ROI) analyses – not in the whole-brain analysis. Li, Peng, Lui, Booth, and Ding (2014) found a similar (and statistically more robust) pattern for deaf Chinese readers making semantic relatedness judgments to pairs of Chinese characters. The deaf readers (fluent users of Chinese Sign Language) showed greater activation in right frontal cortex (as well as in the right inferior parietal lobule) and reduced activation in left IFG compared to hearing readers (reading ability was not assessed). Li et al. (2014) suggested that greater recruitment of right hemisphere regions reflect a reliance on sign language processing during reading because sign language engages right frontal and parietal regions to a greater extent than text processing (e.g., Neville et al., 1998). The two groups of deaf participants in the current study differed in reading ability but were equally proficient in ASL, allowing us to determine whether recruitment of the right hemisphere reflects a compensatory mechanism related signed language use.

In a previous fMRI study, Emmorey, Weisberg, McCullough, and Petrich (2013) compared the same highly skilled deaf readers from the current study with hearing peers who were matched for reading ability. Participants made semantic decisions (concrete/abstract?) or phonological decisions (two syllables?) to English words, and during a control task, made decisions to false font strings (underlined?). For the semantic processing task, Emmorey et al. (2013) reported strongly left-lateralized neural activity for both groups and found no regions where activation significantly differed between skilled deaf and hearing readers. These skilled deaf readers did not differentially recruit right hemisphere regions nor did they exhibit distinct activation patterns within the left hemisphere when reading words for meaning – the deaf and hearing readers both engaged left IFG, left superior temporal sulcus, and left inferior temporal gyrus (including the putative Visual Word Form Area, VWFA; Cohen et al., 2002). In the study reported here, we compared the neural responses of these skilled deaf readers with a group of less-skilled deaf readers performing the same tasks to investigate how reading skill modulates activation within the reading circuit for deaf adults who are fluent in a sign language.

Although Emmorey et al. (2013) found no differences between skilled deaf and hearing readers for semantic decisions, they observed robust differences in neural activity when participants judged the number of syllables within a word. This phonological task was significantly more difficult for the deaf readers, and they exhibited increased activation (relative to hearing readers) in the intraparietal sulcus bilaterally and in the left precentral gyrus. Emmorey et al. (2013) suggested that these findings reflect greater processing demands for deaf, compared to hearing individuals, when making speech-based phonological decisions (parietal increases), and greater reliance on articulatory representations of speech for deaf compared to hearing people (left frontal increase) (see also MacSweeney, Brammer, Waters, & Goswami, 2009). Other studies have also found greater recruitment of inferior parietal cortex for deaf compared to hearing readers when performing phonological tasks, e.g., rhyme judgments (Aparicio et al., 2007, Li et al., 2014). In the present study, we investigated whether neural activity differs for less skilled versus skilled deaf readers when performing the syllable judgment task. Given that a meta-analysis by Mayberry et al. (2010) indicated that phonological ability was not strongly predictive of reading level for deaf individuals, we may find that reading skill does not modulate neural activity during phonological processing tasks for deaf readers.

In addition, we take advantage of the variability in reading skill across this large group of deaf participants (N = 28) to examine brain-behavior correlations between neural activity and reading ability, as well as between neural activity and performance on the semantic and phonological tasks. One question of interest is whether activation within the VWFA increases with reading skill for deaf readers, as has been found for hearing readers (McCandliss, Cohen, & Dehaene, 2003). Corina et al. (2013) reported greater activation in the mid fusiform gyrus bilaterally (within the range of locations reported for the VWFA) for skilled relative to less skilled deaf readers. In contrast, Wang, Caramazza, Peelen, Han, and Bi (2014) found that the anatomical location and activation strength of the VWFA were highly similar between deaf and hearing Chinese readers despite the fact that the deaf readers (all signers) had lower reading levels than the hearing readers. Similarly, Aparicio et al., 2007, Waters et al., 2007 reported no activation differences in this region when deaf readers were compared to hearing readers with better reading ability.

The brain-behavior correlation analyses for task performance also provide clues as to the linguistic or cognitive operations that support word level semantic and phonological processing for deaf readers. For example, a positive correlation between syllable judgment accuracy and neural activity in left prefrontal cortex will suggest that success at phonological tasks is associated with greater use of articulatory codes (see MacSweeney et al., 2009), while a correlation with parietal activity will suggest that success is associated with mapping orthographic to phonological codes (see Pugh et al., 2000).