Reading while listening (RWL) is an instructional technique in which children associate unfamiliar written words with their corresponding familiar sounds. In RWL, children listen to a text (often a “talking book” on audiotape or computer software) while following along in a printed version. Although RWL is a popular technique, particularly in tutorial settings, research about its effectiveness is unclear and contradictory. Van der Leij (1981) noted that the technique supplies children with a model of a reading “end product.” From this model, they can derive both phonetic rules (Carbo, 1978) and correct pronunciation of irregular words (Reitsma, 1988). The model allows children to shift their attention from the laborious effort of reading individual words to the far more interesting job of understanding the narrative. Thus, RWL increases the children's motivation to read (Carbo; Gamby, 1983; Reitsma; Van der Leij), and consequently can help break the vicious cycle of “Matthew effects” in which poor reading reduces motivation, which in turn lessens reading practice so reading does not improve (Stanovich, 1986). To complement these positive views about RWL's effectiveness is some evidence that suggests the technique promotes reading fluency and speed, word recognition, motivation, and reader self-confidence (Chomsky, 1976; Carbo; Klein, 1989; Van Bon, Boksebeld, Font Freid, & Van der Hurk, 1991).

However, a significant body of research contradicts some of these findings. Rasinski (1990) found that repeated RWL did not improved fluency or accuracy any more than did repeated silent reading; Bruland (1970) and Hasselriis (1968) reported similar findings. Gbenedio (1986) found RWL to be less efficient than an eclectic method in English as a second language instruction, and Holmes and Allison (1985) found that reading aloud and silent reading resulted in better comprehension than did RWL.

McMahon (1983) offers an explanation for these contradictions, suggesting that the reason some children appear not to benefit from RWL is that the narration rate used is generally significantly faster than the children's own reading rates. When children cannot follow the narration, they are unable to make the connection between the graphemes they see and the phonemes they hear. Moreover, the seemingly unrelated phonemes distract them from the reading process. McMahon's investigation found tape-recorded story rates that averaged 112 words per minute (wpm) for first-grade cassettes and 141 wpm for third grade, while the first graders in her study read at 18 to 50 wpm and the third graders at 50 to 91 wpm. The children at both grade levels performed a mismatch task quite well when the narration rate equaled their own reading rate or exceeded it by up to 35 percent, but they failed to notice mismatches between the written and voiced texts when the narration was at the average rate of the cassettes intended for their age groups.

Some research has found that slowing the rate of narration can be beneficial (see, e.g., Carbo, 1988; Neville, 1975). However, slowing it down too much seems to result in comprehension problems related to the limited capacity of young children's short-term memories (Breznitz & Share, 1992). Worse still, if the rate is too slow, the children may become bored and lose interest in the story and activity (McMahon, 1983).

It would seem, then, that talking books should be narrated at a rate that matches or slightly exceeds the listener's reading rate. Determining that rate for a particular grade level is, however, an impossible task. Reading rates vary across individuals according to many factors, including age, ability, and familiarity with the text. A simple solution to this problem might be to give the child control of the narration rate. To test this hypothesis, Shany and Biemiller (1995) used a tape recorder that could be played back at different speeds to test the practice effect on poor readers in grades 3 and 4. In testing for reading comprehension, speed, and accuracy, the research showed that this method resulted in almost the same effect as having the poor readers read aloud with a tutor who provided correction -- a technique that is both labor intensive and expensive. Moreover, the tape-assisted group read twice as much as the “tutor” group, which resulted in a significant improvement in listening comprehension.

Since computers are replacing tape recorders in the talking books field, it is now possible to store, in digital form, narration prerecorded at different rates, thereby avoiding problems of pitch and sound distortion caused when traditional audiotapes are played back at different speeds; these different rates of narration can then be accessed on command, either by the teacher or the student (patent no. 08/793,715; pending). Surprisingly, no software that takes advantage of this possibility has been produced commercially, nor has the effect of changing narration rate been checked independent of the RWL effect.

The purpose of the present study was to examine the effect of RWL with reader control over narration rate in comparison to standard RWL, testing for reading comprehension, speed, and accuracy at both text and word levels. The research hypothesized that

• Children will perform better on the test measures when they are given control of the narration rate
• Because poor readers read more slowly and find standard RWL harder to follow, they will profit more from control of narration rate than will their more able peers
• When asked, children will indicate that when they controlled the narration rate, reading was both easier and more enjoyable than in standard RWL

Back to abstract

Children from two first-grade classes at the Afek School in Rosh-Haayin, Israel, participated in the study. The experiment was conducted in Hebrew, the children's first language. A within-subject design was used. Each child engaged in RWL with two stories under two conditions, one with control over the story's narration rate, and the other with narration at the fixed rate determined by the materials' designers and publishers. After each of the conditions, study subjects read aloud a list of words extracted from the story for an accuracy analysis, they responded to a comprehension test, and they read a page of the story aloud to measure both reading accuracy and speed. The children were also asked which of the two conditions was easier for them, and which was more enjoyable.

Sample. Parents of 71 of the 79 first-graders at the school gave consent for their children to participate; these 71 children also agreed themselves. From this sample, two children were excluded: one suffered from attention-deficit hyperactivity disorder and the other from behavioral problems. Of the 69 children who participated, 38 were boys and 31 were girls. A technical problem with the recordings occurred with four of the children; for them, only the written comprehension tests were used in calculation of results and statistical analysis. The teaching method in the school was eclectic, but tended toward whole-word reading.

Assignment to groups. All children received the same treatment, but in order to test the second hypothesis, they were assigned to one of two ability groups. Thirteen children (18.8% of the study sample) were assigned to a “poor readers” group, on the basis of two criteria:

1. Their overall test scores during the experiment were more than a standard deviation (SD) below the average
2. They were considered to be poor readers by their teacher, had been assigned to the lowest reading group in the class, and were receiving additional instruction from a special education teacher

Tools. The reading materials used were taken from the Step by Step series produced by the Israeli Center for Educational Technology. All are recommended for use in first grade by the Israeli Ministry of Education. The series includes books and read-along audiocassettes designed for RWL. The average rate of narration on the cassettes is 90 wpm. The story “For Children Only” was used for a training phase; “Eran's Morning Talk” and “Chen's Luck Chain,” both by Sara Shalom, where used during the experiment itself. The text of “Eran's Morning Talk” can be viewed by clicking here.

At the facilities of the Hamon Recording Studio, the stories were read aloud at rates of 30, 45, 60, 75, and 90 wpm, and were professionally taped. The narrator, Israeli actor Ben Yegendorf, was asked to use the same tone in each of the narration rates. Using Macromedia's Director software for creating interactive multimedia, a projector was created to display the story page by page on the screen while the narration played back, either at a fixed rate of 90 wpm (the rate of the audiocassettes in the series) or at a rate selected by the viewer. When the latter feature was activated, two images displayed on the computer screen allowed children to control the rate: clicking on a rabbit icon increased the narration rate by one step, and clicking on a turtle icon reduced it by one. The icons were felt to be preferable to a scale, which would have opened the possibility for children to make overly ambitious initial selections of narration rate. To see a demonstration of the software (618 Kb; requires Macromedia's free Shockwave plug-in, available from the download center of Shockwave.com), click here.

Measures. The following tests were developed for each of the two stories:

• Reading comprehension
  A multiple-choice test was prepared for “Eran's Morning Talk” and “Chen's Luck Chain.” (The one developed for the former story can be viewed by clicking here.) The tests included various levels of comprehension questions -- literal, inferential, critical, and affective (see Salvia & Ysseldyke, 1995).
  
  
• Reading accuracy, word level
  Each child read a list of words taken from the text into a tape recorder. Tapes were analyzed, and mistakes were weighted as follows: -1 point for a mispronounced or omitted word; -0.5 points for a self-corrected word or a word pronounced as fragmented phonemes (e.g., /ta/ /b/ /le/ instead of /table/); and -0.25 points for a single stop in pronunciation (e.g., /ta/ /ble/) or for a word that was pronounced more than once (taken as an indication that the child was unsure of the pronunciation). Of course, this scaling method checks only the accuracy and fluency of pronunciation and not comprehension. The assumption was that the better the word was learned (internalized and atomized), the better it would be pronounced (fewer mistakes, corrections, or repetitions).
  
  
• Reading accuracy, text level
  Children read aloud a page from each story into a tape recorder. Mistakes were scaled as in the word-list test.
  
  
• Reading speed
  The one-page reading for the text-level accuracy test was timed. Speed was calculated as the average number of words per minute pronounced by the participant.
  
  
• Attitude questionnaire
  The children were asked two questions at the conclusion of the testing procedure: (1) At which narration rate was the reading easier for you? (2) At which narration rate was the reading more fun for you?
  

Analysis of the answer sheets and tape recordings was blind, undertaken without knowledge of the condition (with or without control of narration rate) to which the data related. Raw scores underwent a statistical cleaning process described below.

Procedure. Each child was tested separately. The testing procedure had six phases:

1. Training
2. First condition -- RWL either at the fixed rate or with choice of rates
3. First set of measures
4. Second condition -- RWL in the complementary condition
5. Second set of measures
6. Attitude questionnaire

In the training phase, the children were taught how to use the computer software to control the narration rate. At the bottom of each page of displayed text, two icons -- a rabbit and a turtle -- appeared on the screen, and the children were asked whether they wanted to make the narration rate faster or slower. Each click of the mouse on one of the icons changed the rate to the next discrete level, either faster (e.g., from 60 to 75 wpm) or slower (e.g., from 60 to 45 wpm). The researcher encouraged the children to try each of the different rates and eventually to choose the one they felt was most suitable for them.

For each of the conditions, children were asked to follow along with their finger on a computer screen that displayed one of the stories, while listening to the story being read through the computer's speakers. The order of the conditions (standard RWL or control over narration rate) and their coupling with the two stories were counterbalanced. The rate for the standard RWL condition was set at 90 wpm, the rate of the talking book series from which the stories were selected (the average rate of Israeli talking books for this age group is 105 wpm). In the control-over-narration-rate condition, the story was read at the rate chosen by each child in the training phase, in order to prevent participants from being distracted by the process of actual rate selection during the experimental condition.

After each reading, the comprehension test and the two accuracy tests were administered for the appropriate story (reading speed was calculated from the text-level reading accuracy test). Finally, the attitude questionnaire, delivered orally, asked children to identify the condition that was (1) easier and (2) more enjoyable for them.

Statistical analysis. The parameters of reading comprehension, text-reading accuracy, and accuracy on the word-list reading were measured as a proportion of correct answers or correctly pronounced words; reading speed was measured as the average number of words per minute pronounced by the child (see Measures). Because a within-subject design was used, it was not possible to test the two conditions with the same story. Although the two stories used were written by the same author for the same reading level in the same book series, children did find “Eran's Morning Talk” to be significantly more difficult to read than “Chen's Luck Chain.” To eliminate this variance the measures were cleaned as follows:

• The mean scores for each of the stories were calculated. For each story half of the subjects were tested under standard RWL and the other half under the control-over-rate condition. Therefore, the difference between the means was not caused by the experimental conditions, but by the difference between the stories.
• The difference between the mean story score was calculated and then divided by the story's standard deviation. This resulted in a Z score, defined as the original score cleaned from the story's effect.

To check the first hypothesis -- that reading will improve under the control-over-rate condition -- the Z score of the two conditions was compared in a paired t test. To check the second hypothesis -- that the improvement for the “poor readers” group would be greater than that for the other children -- an added value score was created by subtracting for each participant the Z score of the standard RWL condition from its paired Z score under the control-over-rate condition. Then, a t test was used to compare the added value score of the two groups.

Back to abstract

Four results from the descriptive statistics stand out. First, the average reading rate for the study sample was determined to be 53.4 wpm, about half the average 105 wpm found in tapes and software intended for this age group. Only six children (less than 10% of participants) read at or above 105 wpm. Second, the standard deviation of the reading rate was 33.9. This means that a reading rate 1 SD above the average (i.e., 87.3 wpm) is more then four times faster than the rate 1 SD below the average (i.e., 19.5 wpm). Third, distribution of the children's reading rates was clearly bimodal, as shown in Figure 1.

Finally, there was a significant correlation between the children's actual reading rates and the narration rate that each chose (R (0.66,65) < 0.001). When examining only the participants who claimed that standard RWL was “more fun,” the correlation remained high (R (0.63,24) = 0.01).

The effect of control of narration rate on the reading performance of the participants is shown in Table 1.

The results shown in Table 1 indicate that having a choice of narration rates had a significant effect on the participants' comprehension and their accuracy in text reading, but not on word-list accuracy or speed. The overall results indicate with high significance that giving children a choice of narration rate had a positive effect on their learning with the RWL technique.

The effect specifically for poor readers is shown in Table 2.

* p < 0.05
** p < 0.01

The results shown in Table 2 indicate that choice of narration rates had a significant positive effect on the poor readers' group in the area of reading accuracy (both text and word levels), but not for comprehension or speed. A comparison of the results for the poor readers and the other study participants is shown in Table 3.

* p < 0.05

Table 3 shows that the poor readers' group benefited more from the choice of narration rate than did other readers on the text accuracy test. Notice also that, contrary to the research hypothesis, the other readers gained (insignificantly) more from the choice of narration rate than did the poor readers.

As expected, participants found that reading with a narration rate they chose was “easier” and “more fun” than undertaking RWL with a fixed rate (see Table 4).

Back to abstract

The results of this study confirm the hypothesis that having a choice of narration rate is more effective than standard RWL for students in the initial stages of learning to read. The results indicate that, overall, control over narration rate yields improvements in reading comprehension and text-level reading accuracy. Results regarding word-list accuracy reached significance only for the poor readers. There were no significant results regarding reading speed. The positive effect on the children's performance under the control-over-rate condition was confirmed with high significance.

These results come as no surprise when one examines the descriptive statistics: the average reading rate of first-grade participants in this study (53 wpm) is about 60 percent slower than the fixed-rate narration in the two stories selected from the talking books series (90 wpm) and 50 percent slower than the average narration rate of Israeli tapes and software designed for this age group (105 wpm). These results are not unique to readers in Israel, however: McMahon (1983) found the average rate of narration on audiocassettes designed for first-grade American students to be 112 wpm, while the reading rate of the first graders she tested ranged from 18 to 50 wpm.

It seems reasonable to suggest that if children input written and spoken words at different rates, this reduces the chance for cognitive association between the representations. Moreover, the difference between the narration rate and the child's actual reading rate turns RWL into a task that splits young reader's attention -- that is, the narration becomes a distraction that seems to relate almost to a different text. This is not to say that RWL is necessarily a split-attention task, as was suggested in the 1970s (Daly, Neville, & Pugh, 1975; LaBerge & Samuels, 1974); rather, it becomes such a task when the narration rate is too fast for the children.

The study also found a very high variance (SD of 33.9 wpm) in reading rate among the children. A student who read at 1 SD above the average (i.e., at 87.3 wpm) read more than four times as fast as one who read 1 SD below (i.e., 19.5 wpm). This means that even if software and audiotape producers of talking books included narration at the reading rate of the average student, it would still be either too fast or too slow for most children. In other words, there is no single narration rate suitable even for a homogeneous age group.

This leads to the conclusion that choice of narration rate is required. The next question is whether children can actually determine the appropriate rate for themselves. The results of this study imply that they can. A significant correlation was found between the children's actual reading rates and the narration rate they selected. This correlation was also found for the children who claimed that the standard reading rate was “more fun.” This suggests that children tend to choose realistically according to their abilities, overcoming their desire to hear the story at a faster pace.

Improvement of poor readers. Based on results from the text-level reading accuracy measure, the poor readers' group seemed to have profited more from control over the narration rate than did the other readers. This could be explained by the fact that the poor readers are slower than their classmates, so it is more difficult for them to follow the standard narration. In contrast to this, and contrary to the prediction of the second research hypothesis, the poor reading group did not profit more than the rest of the readers from control over narration rate in terms of reading comprehension. An explanation for this could be found in an anecdotal observation: when these children were presented with narration at the standard rate, their fingers wandered randomly across the computer screen -- apparently because they were not able to follow the story as it was displayed. Since they did not try to read, they could listen attentively and, hence, they had no difficulty understanding the story at the fixed-rate condition.

Psychological and emotional considerations. Most children found the narration rate they selected themselves to be easier and “more fun” than the standard narration rate. They responded to the standard rate with stress signals (saying, for example, “What is this? I can't keep up!”), and were enthusiastic about the actual rate-selection process. This suggests another explanation for the positive results: it is possible to view a student who is unsuccessfully trying to follow narration as engaging in the behavior of learned helplessness (Petterson & Seligman, 1993). By giving them the choice over narration rate, the children gain control of the situation and view themselves as being trusted to make decisions.

Back to abstract

It would be useful for this study to be replicated in a course of longitudinal research in order to test its implementation in an actual school environment. Its effect could also be tested with dyslexic children and students studying a second language.

In addition, the study's results suggest two other directions for future research:

1. The distribution of the participants' reading rates was unexpectedly bimodal. This could be attributed to interpersonal differences (e.g., Matthew effects, as described in Stanovich, 1986), or to a sudden increase in children's reading rates as part of the development of reading skills. During the first stages of reading acquisition, for example, the child might read fragmented phonemes in the context of a word, while in the second stage, the child reads whole words in the context of a sentence. A longitudinal study that monitors and follows children through the stages of reading skills acquisition is required to test these hypotheses.
2. Many software versions of talking books focus the child's attention on the part of the text being narrated by highlighting on the computer screen the text block that is being verbalized. But how much of the text should be highlighted? When selection of narration rate is enabled, the extent of highlighting can be adapted to the narration rate. Thus, at a slow narration rate, highlighting could cover only one word at a time, while for a faster narration rate, the highlighting could cover a few words or a sentence or two. Research is needed to determine if this modification of the control-over-rate technique would beneficial.

Back to abstract

The central question addressed by this research was “Does giving children control of the narration rate help them acquire reading skills in RWL activities?” The study's results indicate that it does. Not only did students show a significant improvement in reading comprehension and accuracy (though not in speed), they also had a more positive attitude toward the control condition, possibly because it allowed them to be in command of the learning activity.

For more than a hundred years, the progressive reform movement in education has asserted that children should be allowed to learn at their own pace. In practical terms, this has been difficult to implement -- teachers generally have a wide range of children and ability levels in their classes and are usually required to “cover” a prescribed curriculum, often with limited or inadequate resources. The advent of computers in the classroom, however, has made it possible to put this educational philosophy into practice (see, e.g., Schank & Cleary, 1995) -- and this study shows its benefits for one facet of reading instruction.

Author's Note: I wish to thank the following for assistance, encouragement, and support in the research and writing of this article: Miriam Faust and Lilly Dimitrovsky, Bar Ilan University; Eyal Gever, zapa.com; Lior Shur, MediaMate; Afek School principal Ariela Segal, teachers Avital and Rinat, and the first-grade children who participated in the study; Haggith Gor-Zeev, Yoad Nevo, Orit Bergman, Zeev Bergman, Oren Zukerman, Uzi Dvir, Michal Shnayder-Beery, Alex Aldor, Etgar Keret, MESSER - Idea Promotion Center, and finally, my wondrous wife, Rutu Modan, without whom nothing would have occurred.

Breznitz, Z., & Share, D.L. (1992). Effects of accelerated reading on memory of text. Journal of Educational Psychology, 84(2), 193-199.
Back

Bruland, A. (1970). Listening and listening-reading at two rates of presentation by fifth grade pupils (Doctoral dissertation, Syracuse University). Dissertation Abstracts International, 32, 833A.
Back

Carbo, M. (1978). Teaching reading with talking books. Reading Teacher, 32, 267-273.
Back

Chomsky, C. (1976). After decoding: What? Language Arts Journal, 53, 288-296.
Back

Daly, B., Neville, M., & Pugh, A. (1975). Reading while listening: An annotated bibliography of materials and research (Paper no. 13). Leeds, UK: University of Leeds, School of Education.
Back

Gamby, G. (1983). Talking books and taped books: Materials for Instruction. The Reading Teacher, 36, 366-369.
Back

Gbenedio, U.B. (1986). Two methods of teaching reading in Nigerian primary classes. ELT Journal, 40(1), 46-51.
Back

Hasselriis, P. (1968). Effects on reading skill and social studies achievement from three models of perception: Simultaneous reading-listening, listening and reading (Doctoral dissertation, Syracuse University). Dissertation Abstracts International, 29, 4376A.
Back

Holmes, B.C., & Allison, R.W. (1985). The effect of four modes of reading on children's comprehension. Reading Research and Instruction, 25, 9-20.
Back

Klein, H.A. (1989). The effect three testing modes -- reading while listening, listening, and silent reading -- have on sixth grade boys and girls. Reading Improvement, 26(4), 298-304.
Back

LaBerge, D., & Samuels, S.J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 292-323.
Back

McMahon, M.L. (1983). Development of reading-while-listening skills in the primary grades. Reading Research Quarterly, 19(1), 38-52.
Back to introduction
Back to discussion

Neville, M.H. (1975). The effect of rate of an aural message on listening and on reading while listening. Educational Research, 18, 37-43.
Back

Petterson, C., & Seligman, M.E.P. (1993). Learned helplessness. San Diego, CA: Academic.
Back

Rasinski, T.V. (1990). The effect of repeated reading and reading while listening on reading fluency. Journal of Educational Research, 83(3), 147-150.
Back

Reitsma, P. (1988). Reading practice for beginners: Effect of guided reading, reading while listening, and independent reading with computer-based speech feedback. Reading Research Quarterly, 23(2), 219-235.
Back

Salvia, J., & Ysseldyke, J.E. (1995). Assessment (6th ed.). Boston, MA: Houghton Mifflin.
Back

Schank R.C., & Cleary, C. (1995). Engines for education. Hillsdale, NJ: Erlbaum.
Back

Shany, T.M., & Biemiller, A. (1995). Assisted reading practice: Effects on performance for poor readers in grades 3 and 4. Reading Research Quarterly, 30(3).
Back

Stanovich, K.E. (1986). Matthew effect in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21(4), 360-406.
Back to introduction
Back to future research directions

Van Bon, W.H.J., Boksebeld, L.M., Font Freid, T.A.M., & Van der Hurk, A.J.M. (1991). A comparison of three methods of reading-while-listening. Journal of Learning Disabilities, 24(8), 471-476.
Back

Van der Leij, A. (1981). Remediation of reading disabled children by presenting text simultaneously to eye and ear. Bulletin of the Orton Society, 31, 229-243.
Back

Back to abstract

Go to story sample

Go to comprehension test sample

Go to software sample

Go to discussion forum