Coordination of Sub-processes in Oral Reading Fluency

Coordination of Sub-processes in Oral Reading Fluency

First Author: David Braze -- Haskins Laboratories
Additional authors/chairs: 
Hosung Nam; Tao Gong; Clinton Johns
Keywords: Fluency, Eye movements, rapid automatized naming, Adult Literacy, speech
Abstract / Summary: 

Purpose: Fluency is essential to skilled reading. Much evidence indicates
oral reading fluency is a good indicator of general reading skill. Fluent
reading reflects automaticity in decoding and similarly deft control of
other component skills. Yet, fluency relies not only on control of
individual processes, but also on facile _coordination_ of them. RAN, a
correlate of reading skill, is a fluency-like task whose demands partly
overlap with reading, but lacking, e.g., obvious requirements for
vocabulary (V) or decoding (D). This study aims for understanding of how
coordination of subprocesses (e.g., speech-motor, oculo-motor, D) figures
in fluent reading.

Method: Work in progress focuses on young adult poor readers (planned N=50,
current N=5), and looks to identify connections between oral reading
fluency, RAN, and reading related skills -- D and orally assessed V. We use
eye-tracking and speech analysis to study commonalities in RAN performance
and oral reading fluency as expressed through asynchronies in gaze and
speech produced in these tasks (the eye-voice span, EVS).

Result: Planned analyses compare EVS for RAN and oral reading in young
adults with poor reading skills. Early analysis (N=5) of RAN shows a
typical EVS of about 1, with gaze sometimes leading voice by as many as 2
or as little 0 items. Mean EVS for oral reading is slightly longer than
for RAN and more variable. Additional analyses will explore the role of
differences in V and D to regulating speech and gaze behavior in these
tasks.

Conclusion: This study probes the underpinnings of reading fluency by
examining relationships between oral reading and RAN. We predict oral
reading EVS will be modulated by both D and V as well as properties of
particular words (e.g., frequency) in texts being read. RAN EVS should show
much weaker associations with D and V. A subtractive approach to comparison
of reading and RAN behaviors will be used to partition processes
contributin