Developmental trajectories for children with dyslexia and low IQ poor readers.

Procedure

An auditory task battery was presented to all children, composed of measures of Amplitude rise time, Duration, Frequency, and Intensity discrimination (see Appendix C Auditory task descriptions for a description of each task). Two tasks were administered to assess discrimination of the rise times of amplitude envelopes. All auditory tasks were delivered using the Dinosaur program, a threshold estimation interface designed to be attractive to children (originally developed by Dorothy Bishop, Oxford University). Tasks were delivered using an AXB paradigm (where X is the standard and either A or B differ from X in one direction) or a two interval forced choice format. Children were asked to select the target by pointing to the screen or by naming the color of the dinosaur producing the target sound. Auditory and visual response feedback was provided to motivate learning and increase interest, while catch trials (presenting the easiest discrimination) were used to assess attention levels in individual participants. All children were given five practice runs for each task in order to ensure task comprehension before beginning. Further detail regarding the auditory tasks, including schematic depictions of the stimuli, is available in ( Kuppen et al., 2011 ).

In addition to the auditory tasks, experimental measures of phonological processing were administered (please see Task Appendix for full details). A phonological short-term memory task ( Thomson, Richardson, & Goswami, 2005 ) presented via computer four monosyllabic consonant-vowel-consonant (CVC) words through headphones (e.g., type, rib, nook, bud ). Children were required to repeat back the words as spoken. Sixteen trials were presented in total. In addition, an onset oddity task was also administered by computer. Here, children selected the one spoken word from a set of three, which began with a different sound (e.g., laid, make, mate ). Twenty trials were given overall. Finally, a rapid automatized naming task was given. Children were asked to name line drawings of familiar objects (e.g., fire, cup, bird, leaf). It was first ensured that children were able to name each drawing. They were then shown a page with the pictures repeated 40 times in a random sequence. Children were asked to name the drawings as quickly as possible. Individual performances were timed and errors were counted.

Results

Developmental trajectories were plotted for all tasks. In each case, two linear relationships were calculated for each poor reader group, one assessing the relationship between task and chronological age and the second assessing the relationship between task and reading age. A between-groups analysis of covariance was undertaken for the comparison of each poor reader trajectory against the typically developing group. Two outcomes were of primary importance in ascertaining the appropriate label; these were the presence of a significant main effect of group (indicating delay) or a significant interaction effect (between group and age, indicating a difference in rate of change). An overview of the trajectory outcomes in each case is provided in Table 1 .

To illustrate the power of the developmental trajectories method, Figures 5–9 show the trajectories against chronological age for the three phonological measures and for the three auditory measures that have shown the most consistent results in prior studies (Rise time [1 rise], Duration, and Frequency, see Hämäläinen et al., 2013 ). Figures for all remaining trajectories are presented in the Supplementary Materials (Supplementary Figures 1–10). In all cases, a best-fit linear trendline has been provided. As indicated in Table 1 , in some cases this reflects a significant linear relationship between the key variables, and in others the relationship does not reach statistical significance. Trajectory classifications reflecting the key comparison variables are summarized in Table 2 (CA trajectory analyses) and Table 3 (RA trajectory analyses). For the children with dyslexia, the linear function y = a M YD + b is calculated. Depending on the figure, y represents the total number of correct responses, the response time in a phonological task, or a threshold value in an auditory task; a represents the age-related rate of change in y ; M YD represents age in months relative to the chronological or reading age of the youngest child with dyslexia (CA 81 months, RA 58 months); and b is the value at which the respective trajectory begins. For the low IQ poor readers, M YD in the above equation used for the children with dyslexia is replaced by M YL the age in months relative to the youngest low IQ poor reader’s chronological or reading age (CA 72 months, RA 58 months). In the figures presented, the plotted trajectories reflect the original data before rescaling to the youngest disorder age. They do not therefore match up directly to the accompanying function provided in Table 1 . As a note of caution, while our tasks were undertaken repeatedly with the same participant pool, there is nonetheless some inflated risk of a Type I error (i.e., false positive) in our ANCOVA analyses here.

Performance on Phonological short term memory task against CA. x and dotted lines = poor readers; o and continuous line = TD children.

Performance on Rapid automatized naming tasks against CA. x and dotted lines = poor readers; o and continuous line = TD children.

Performance on 1 rise task against CA. x and dotted lines = poor readers; o and continuous line = TD children.

Performance on Duration against CA. x and dotted lines = poor readers; o and continuous line = TD children.

Performance on Frequency against CA. x and dotted lines = poor readers; o and continuous line = TD children.

Following previous work ( Thomas et al., 2009 ), we used the chronological age comparisons to identify developmental delay. When a delayed onset is demonstrated (a significant main effect of group), or a slowed rate of development is present (a significant interaction between age and group), or both are demonstrated, poor readers are classified as delayed compared to the typically developing group. Poor reader trajectories are classified as atypical when task performance and increasing chronological age are not linearly related for the poor reader group, but are linearly related for the typically developing group. The decision tree for CA trajectory classification is shown in Figure 1 with outcomes in Table 2 . We also checked the trajectory classification on the basis of the RA comparisons, as summarized in Table 3 . Trajectories are classified as atypical when task performance and increased reading age are not linearly related for the poor reader group, but are linearly related for the typically developing group. The decision tree for RA trajectory classification is shown in Figure 2 . When the two classification routes (CA, RA) yield conflicting results, a best fit decision was made and is explained in the text.

Decision tree for trajectory classification using CA comparison.

Decision tree for trajectory classification using RA comparison.

Trajectory Outcomes by Task

In all cases, linear functions are presented in Table 1 by task and reader group. These should accompany any reference to the trajectory figures. Supplementary figures are provided in the supplementary materials which accompany this article.

British Ability Scales Single Word Reading

Trajectories for reading performance by reader group across age are presented in Figure 3 (panels A and B). These trajectories are not classified, as there is no RA comparison with which to undertake the classification procedure (as the task itself measures single word reading ability).

Reading Ability Scores (raw value) against CA. x and dotted line = poor readers; o and continuous line = TD children.

Phonological short-term memory

In the assessment of phonological short-term memory, the children with dyslexia showed delayed trajectories while the trajectories for the low IQ poor readers were classified as atypical. Figure 5 (panels A and B) shows the CA trajectories for each group; the RA trajectories are shown in Supplementary Figure 2. The developmental trajectory with CA was significantly linear for the TD children. While there was no significantly linear relationship for children with dyslexia, the relationship did approach significance and delay was clearly visible. Inspection of the RA trajectory (Supplementary Figure 2A) confirmed that the dyslexic trajectory was significantly linear on this task. Additionally, the trajectory lay on top of that of the TD children, as would be expected in a case of delay. For these reasons the children with dyslexia were classified as delayed. In the LIQPRs, there was no linear relationship with CA ( Figure 5B ) nor with RA (Supplementary Figure 2B), resulting in an atypical trajectory classification.

Rapid automatized naming (RAN)

On the RAN task, the children with dyslexia showed delayed trajectories while the low IQ poor readers showed atypical trajectories. Figure 6 (panels A and B) illustrates the CA trajectories for this task. The children with dyslexia demonstrated the same linear relationship between task and increasing age as the TD children, but with a clear delay, which equated to 37 months (see Table 2 for statistics). The low IQ poor readers showed nonlinear functions for both CA ( Figure 6 Panel B) and RA (Supplementary Figure 3A). Hence the LIQPR group was judged to show an atypical developmental trajectory for RAN.

Auditory Processing

Although analyses were run for all five auditory processing measures (1 rise, 2 rise, Duration, Frequency, and Intensity), we present the CA trajectories for auditory thresholds for Rise time (1 rise, Figure 7 ), Duration (see Figure 8 ), and Frequency (see Figure 9 ) only. The other trajectories are supplied as Supplementary Figures 4–10. For ease of comparison, Table 4 presents a summary of the auditory processing data from our prior studies of English-speaking children, studies that used the same or very similar auditory tasks to those analyzed here. The classification outcomes below should be reviewed in conjunction with Tables 1–3 and Figures 1 and 2 .

On the 1 rise task, both poor reader groups were classified as showing atypical developmental trajectories. Contrary to the TD children, the children with dyslexia did not show a linear relationship between sensitivity to rise time and neither CA ( Figure 7A ) nor RA (Supplementary Figure 4A). There were thus atypical on this task. The low IQ poor readers did show a linear relationship for CA ( Figure 7B ) but not for RA (Supp. Figure 4B ). The CA analyses showed a significant main effect of group (see Table 2 for statistics), indicative of developmental delay for the LIQPR children (equating to 34 months). However, due to the lack of a linear relationship between rise time sensitivity and reading age, this group was also classified as showing an atypical developmental trajectory. It should be noted that the TD children did show a significant relationship between rise time sensitivity and reading age; this is expanded upon further in the Discussion section.

Duration

For the Duration task, the children with dyslexia were classified as showing a delayed developmental trajectory ( Figure 8A ) while the low IQ poor readers were classified as showing an atypical trajectory ( Figure 8B ). For the children with dyslexia there was no main effect of group in the CA analyses (see Table 2 ), indicating that their trajectory was not significantly different from the TD children. However, despite this, a developmental delay of 26 months could be calculated. For the low IQ poor readers, a similar pattern was found in the CA analyses with again no significant group difference present. However, again developmental delay was calculated as 21 months. The RA analysis for the children with dyslexia (Supplementary Figure 5A) demonstrated a linear relationship between task performance and increasing reading age, as was the case for the TD children. However, this was not demonstrated for the LIQPRs (Supplementary Figure 5B), resulting in an atypical trajectory classification for the LIQPR group and a delayed classification for the children with dyslexia.