1. Block Design X Language Picture Completion X Language
Does language demand on fluid reasoning tasks impact the predictive power for predicting performance on pre-literacy tasks?
Hope S. Lancaster  Stephen Camarata
Vanderbilt University School of Medicine
INTRODUCTION
METHODS
discussion
Children with language impairments (LI) are at a higher risk than their typical peers for reading deficits (Bishop & Adams, 1990; Catts, Hu, Larrivee, & Swank, 1994). Research has indicated that children with LI show deficits in pre-literacy (PL) skills, such as phonological awareness (Larrivee & Catts, 1999). One possible reason that children with LI have deficits in PL skills might be related to other cognitive skills (Catts, Fey, Zhang, & Tomblin, 1999), in particular their ability to manipulate information (fluid reason, (Gf).
Two pre-literacy (PL) skills will be examined in this study. When children enter kindergarten they are commonly taught phonological awareness (PA) and also letter names. Both of these skills support reading in later grades (Cabell, Justice, Konold, & McGinty, 2011; Raitano, Pennington, Tunick, Boada, & Shriberg, 2004; Yamada et al., 2011).
One intriguing, and understudied, predictor of PL skills is overall intellectual or cognitive ability. For example, Catts et al. (1999) reported that children who were poor readers had lower full scale IQ than their peers with good reading skill. Yet, the actual nature of these relative weaknesses in intellectual abilities is unclear because there are a number of broad abilities, such as oral vocabulary and pattern recognition that contribute to “full scale IQ,” and more recent models of cognitive abilities may yield important insights into which aspects of cognition relate to reading development.
The Cattell-Horn-Carroll (CHC) theory of intelligence views intelligence as a composite of nine broad abilities, which can be estimated using narrow abilities within each cognitive domain, including fluid reasoning (Gf; Cattell, 1979; Carroll, 1993). Several of these broad abilities are likely predictors of pre-literacy skills . For example, visual-spatial (Gv) has a moderate correlation between measures of Gv and PL (r = 0.35 and r = 0.42; Woodcock, McGrew, & Mather, 2001). Woodcock et al. (2001) also reported a moderate relationship between PL skills and Gf abilities (r = 0.34 and r = 0.44). The size of this correlation is similar to that reported in Gv. This project expands upon this intriguing preliminary finding by examining the relationship between PL skills and Gf in more detail.
Gf ability can be tested using both high and low language demands; this means that children with LI might show a different relationship when predicating PL performance based on whether a test of Gf had high or low language demands. The two tasks in this study that measured Gf had varying levels of language demand. Both tasks required the use of receptive language to some extant, but Picture Completion also required a verbal response for the participant. Given that previous research has indicated that non-verbal or minimal verbal measures of cognition yield better estimates within the LI population (Camarata & Swisher, 1999), there is a possibility that the differences of language demand for the tasks might indicate help to begin to understand how language and cognition work in unison to predict performance on academic skills.
The purpose of this study was to examine how Gf and language interact to predict performance on pre-literacy tasks.
The overall results for the regression analyses indicated that the main effects model was the best predictor model. For all the groups adding Gf measures significantly improved model fit, and increased the amount of variance explained.
The hypothesized results were that Block Design would be a stronger predictor than Picture Completion. For most of the results Block Design was a significant predictor, whereas Picture Completion was only a significant predictor in two cases. The Typical and SLI groups show the predicted pattern for both PL measures. However, the NLI group differs for Letter-Word Identification.
In examining the two LI samples, the patterns for predicting performance suggest that children with SLI and children with NLI have different relationships between PL and Gf. The implication for these findings is that there are differences in the strengths and needs for children with SLI and children with NLI.
The results of the regression analyses provide support for the argument that tasks Tomblin considered nonverbal do in fact utilize language to varying degrees. The implications of these results support the notion that considering how much language is used when testing Gf is important. Additionally, this implication is supported by the fact that adding Gf measures, not only increased the variance explained, but also decreased the beta weight for language. N
Percent male
Age
Language
Block Design
Picture Completion
Word-Sound Deletion
Letter-Word ID
Typical
1424
54.2
6.02 (0.32)
(0.67)
0.19 (0.94)
0.18 (0.93)
9.07 (6.34)
27.23 (9.09)
SLI
277
59.2
5.95 (0.31)
(0.38)
0.10 (0.73)
0.13 (0.76)
4.58 (5.14)
22.8 (10.79)
NLI
228
51.8
5.99 (0.32)
(0.47)
(0.65)
(0.73)
2.3 (3.55)
17.3 (10.8)
Notes. Language is a composite z-score composed of scores from TOLD:P-2 (Newcomer & Hammill, 1988) and narrative task (Culatta, et al., 1983). Block Design and Picture completion are z-scores. Word-Sound Deletion and Letter-Word ID are raw scores. Mean (sd).
This project used data from the EpiSLI database (Tomblin, 2010) to explore the research questions. See above table for descriptive information.
Analysis
Hierarchal regression analysis was used to explore main effects and interactions. A three-step model was designed to account for the added variance explained by each variable.
Step 1: Control (Language)
Step 2: Main Effects (Language, Block Design, and Picture Completion)
Step 3: Interactions (Language, Block Design, Picture Completion, and Interaction terms)
Analyses were examined for all three identified groups (Typical, SLI, & NLI).
Transformations
Block Design and Picture Completion were transformed into z-scores.
Interaction terms were created by multiplying specified variables (e.g., Block Design * Language) and reliability was examined.
Y = Constant + Language + Block Design + Picture Completion + Language * Block Design + Language * Picture Completion
Reliability
SELECTED REFERENCES
Bishop, D. V. M., & Adams, C. (1990). A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation. Journal of Child Psychology and Psychiatry, 31, 1027 – 1050.
Cabell, S. Q., Justice, L. M., Konold, T. R., & McGinty, A. S. (2011). Profiles of emergent literacy skills among preschool children who are at risk for academic difficulties. Early Childhood Research Quarterly, 26, 1 – 14.
Catts, H. W., Fey, M. E., Zhang, X., & Tomblin, J. B. (1999). Language basis of reading and reading disabilities: Evidence from a longitudinal investigation. Scientific Studies of Reading, 3,
Catts, H.W., Hu, C.F., Larrivee, L. & Swank, L. (1994). Early identification of reading disabilities in children with speech-language impairments. In R. Watkins and M. Rice (eds.) Specific language impairments in children. Baltimore: Paul H Brookes,
Larrivee, L. S., & Catts, H. W. (1999). Early reading achievement in children with expressive phonological disorders. American Journal of Speech-Language Pathology, 8,
Raitano, N. A., Pennington, B. F., Tunick, R. A., Boada, R., & Shirberg, L. D. (2004). Pre-literacy skills of subgroups of children with speech sound disorders. Journal of Child Psychology and Psychiatry, 45, 821 – 835.
Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Technical manual: Woodcock-Johnson III. Itasca, IL: Riverside Publishing.
Yamada, Y., Stevens, C., Dow, M., Harn, B. A., Chard, D. J., & Neville, H. J., (2011). Emergence of the neural network for reading in five-year-old beginning readers of different levels of pre-literacy abilities: An fMRI study. NeuroImage, 57, 704 – 713.
Block Design X Language
Picture Completion X Language
Typical
.470
.416
SLI
.215
.251
NLI
.262
.226
Based on the reliability results, model 3 results will not be presented as the interaction terms are unreliable.
RESULTS
Word-Sound Deletion
Typical
SLI
NLI
Variables
Control
Main effects
Main effectsa
Language
.472***
.41***
.274***
.258***
.15*
.119 BD
.17***
.148**
.13 PC
.035
- .017
.083 R2
.222
.252
.075
.022
.018
.031
F change in R2
405.9***
29.45***
22.17***
3.31*
5.176*
2.538
ACKNOWLEDGEMENTS
Notes. *** p < 0.001; ** p < 0.01; * p < 0.05; Reporting standardized beta weights only. aNLI main effects model is significant (F(3, 224) = 3.441, p = .018) and Block Design approaches significant (t = 1.957, p = .052).
This study was supported by a Preparation of Leadership Personnel grant (H325D080075; PI: Schuele) US Department of Education.
Acknowledgement is given to original grant #N01-DC and supplement #3 P50 DC S1 from the National Institute on Deafness and Other Communication Disorders, a division of the National Institutes of Health. The authors also acknowledge support given by the Vanderbilt Kennedy Center.
The content is solely the responsibility of the authors and does not necessarily represent the views of Vanderbilt University.
Poster available at:
Letter-Word ID
Typical
SLI
NLI
Variables
Control
Main effects
Language
.22***
.079
.044
.169**
.132* BD
.194***
.2***
.127 PC
.077***
.139* R2
.089
.137
.003
.041
0.24
.048
F change in R2
140.11***
40.412***
1.735
6.438**
6.664**
3.844*
RESEARCH QUESTIONS
1. Is there a difference in the strength of predicative power when considering the amount of language demand for a measure of fluid reasoning?
2. Is the predictive pattern the same for children with typical language, specific language impairment (SLI), and nonspecific language impairment (NLI)
Notes. *** p < 0.001; ** p < 0.01; * p < 0.05; Reporting standardized beta weights only.