A spectrographically grounded scale for evaluating reading expressiveness

Benjamin, R. G., Schwanenflugel, P.J., Meisinger, E. B.,Groff, C., Kuhn, M.R., & Steiner, L. (2013). A spectrographically grounded scale for evaluating reading expressiveness. Reading Research Quarterly, 48(2), 105-133.


Capturing the complexities of the reading process so that we can observe and measure what readers do has always been a challenge in reading assessment. This has been especially true for the construct of reading fluency. Reading researchers seem to agree that fluent reading consists of several important and interlocking factors. Researchers have sometimes used slightly different names for these factors, but they mostly look like some variation on speed, accuracy, expression, and comprehension. Sometimes expression is referred to by the more general term, prosody, and some researchers combine speed and accuracy into a factor they call “automaticity”, as the authors do here.

Whatever you call these factors, it is clear that currently popular measures of fluency do not account for expression or prosody well, if at all. Most fluency assessments today look at words correctly read per minute, and that is it. Some may attempt some rough measures of reading comprehension like questions or retelling after reading. Counting how many words a child can read accurately in a minute is observable, can be documented, is easy to report and understand, and requires little training or time to do. It is quick and efficient, and seems to be objective. The assessment of Words Correct Per Minute (WCPM) has been widely accepted as a fluency measure in elementary schools. The well-known grade-level fluency norms developed by Hasbrouck and Tindall rely on these WCPM assessments, and are used to make decisions about children’s reading levels and instruction, even though that kind of assessment leaves out an important part of fluency.

Until now there has been no really observable way to capture and present observations of expression/prosody in children’s reading. We have rating scales provided by researchers, (notably Jerry Johns and Tim Rasinski), but these have always seemed subjective, no matter how carefully they were developed and validated. The authors here present an assessment tool that they believe can capture the missing element of prosody (the word they use for expression). This tool is potentially a breakthrough because it uses technological advances to map, capture, and document a part of the reading process that has heretofore been impossible to show graphically. Through spectrographic analysis, the authors created visual representations of children’s readings of texts, and were able to describe some of the factors that distinguished more fluent readers from less fluent readers. These factors can be seen on the spectrographic representations of the children’s reading.

The authors describe several aspects that are observable in representations of fluent reading, including the number of pauses, the location of pauses in grammatically appropriate places in sentences (here, the researchers mainly looked at pauses between sentences), the variation in voice pitch, the dropping of pitch at the end of declarative sentences. Fluent readers pause less, and when they do pause, it is at appropriate places. They have more variety in pitch than do less fluent readers, and they tend to drop their pitch at the end of declarative statements. This can clearly be seen and documented on WAV files with recently developed technology. The software the researchers used (Audacity) is accessible and readily available for use on computers that are available to most educators. With this technology, what was heretofore only indirectly observable now seems to be at least more directly and visually observable. Having data this concrete could potentially allow educators to consider prosody in its true place, as a factor that is just as important as rate and accuracy and just as documentable and reportable.

This article is likely an important one, which this journal’s editors recognized by making it the lead article in this issue of the journal. For that reason, I recommend it. Though parts of the article may be difficult reading for educators without graduate coursework in research methods, the article (like most in this particular journal) is fairly accessible and readable for most educators. The research described here is fascinating, and opens up many possibilities. Knowing how to assess prosody could radically change the way fluency is now measured, which in my view badly needs to happen. It is clear that the authors plan to continue this line of research for some time, and there are plenty of avenues to explore. I was especially intrigued by the technology, and plan to explore software like Audacity so I can see what readers’ spectrographic representations actually look like. I am also interested in the gender factor that the authors mention briefly in the article. Apparently, female children scored higher on the prosody measure than male children did. Why did that occur? Is prosody something that is innately linked to gender, or is it a socially constructed thing (i.e., is it somehow not “cool” for young boys to read with expression?)? Alternatively, was this a factor within the researchers who tested the assessment? They were all female. Was it something in the way females perceive the reading and the spectrographic representations of males versus females? I look forward to following this line of research.