Popular version of paper 2pNSa1
Presented Tuesday Afternoon, December 2, 1997
134th ASA Meeting, San Diego, CA
Embargoed until December 2, 1997
I. Introduction
During the past few years the state of Washington has passed a number of school bond and levy issues authorizing the rehabilitation and/or new construction of K-12 schools. This recent activity ends a nearly 30-year period when very little school building was undertaken. The principal reason for this long-term dearth of educational building has been demographic. After a decade of frenetic school construction during the middle 1950s to the middle 1960s -- to provide adequate learning space for the "baby boom generation" -- most school districts found that the already-completed facilities offered more than enough educational space to meet student needs (then and for subsequent years.) Only recently, as the children of the baby boomers have begun their primary and secondary educations, have new or rehabilitated school spaces become necessary.
This recent commencement of school construction is occurring while school districts, here in Washington State and elsewhere across the United States, are finding that they do not have up-to-date design guidelines for dictating how school buildings should be planned and outfitted. Instead, school districts are, for the most part, relying on design guidelines which were used as bases for creating the 1960s schools. Only when design and/or construction practices have markedly changed since the 1960's and more current guidelines are well known is new information being included in the design requirements furnished by school districts (to the building designers).
However, many topics exist for which changes in design guidelines have not been instituted, largely because the design and construction professions are unaware of the development of new bodies of knowledge. One such topic concerns means for enhancing speech communication conditions in school classrooms. Presently, little information exists in contemporary architectural literature informing building designers about how to create K-12 classrooms, which will specifically foster good communications between teachers and students.
II. Organizing the pilot study
As a first step toward developing better information about how K-12 classrooms can be designed to achieve better speech communications, a pilot study is being conducted in primary school classrooms in Seattle. The properties of a number of unoccupied, conventional classrooms have been determined. Measurements of these properties have included determination of early decay and reverberation times and background noise levels without and with operating mechanical (HVAC) systems.
Following the brief survey of unoccupied classrooms, intelligibility tests were conducted in a single room occupied by four different groups of children, two from Grades 1-3 and the others from Grades 3-5 (with a total N = 50.) The classroom used for this pilot study is in the B.F. Day Elementary School in the Fremont district of Seattle. The room has tables and chairs for approximately twenty-five students. The principal surface materials are wood, acoustic tile, gypboard, and single-pane glazing. The measured reverberation time (RT60 and unfiltered by frequency) is 0.52 second for the unoccupied classroom.
The intelligibility tests employed pre-recorded AUDITEC word lists and Test of Auditory Comprehension (TAC) short-paragraphs and picture images. On two occasions the test conditions used pre-recorded tapes played through a loudspeaker placed at the front of the class (to simulate the speech of a teacher.) Following these two occasions, on two further occasions, the pre-recorded tapes were played into the loudspeaker whose signal was then picked up by a microphone and played out by an FM-sound field system. The specific FM-sound field unit was manufactured by Custom All Hear Systems, Lynnwood, WA. The centralized speaker array was attached to the ceiling at the center of the room and consisted of five speakers, four of which were directed respectively to the four cardinal compass directions and the fifth was pointed downward. This use of a loudspeaker-to-FM-sound field assembly was intended to simulate a teacher speaking to students with the aid of a FM-sound field system.
For each of the four occasions the tapes were played at alternative levels and against alternative background noise levels. The principal noise source in the room was a partially-enclosed, in-the-wall unitary heat pump. The major noise components of this heat pump were the fan and, when operating, the compressor.
The four playback study conditions included the following parameters: first, with the heat pump turned "off" and a background noise level of 47 dB(A), the front loudspeaker operated at 67 dB(A) (measured at 1 m from the speaker); second, with heat pump "on" (but the compressor "off"), the background noise level was 66 dB(A) and the front loudspeaker operated at 67 dB(A) (measured at 1 m from the speaker); third, the background noise level with the heat pump and compressor both in operation, the background noise level was 70 dB(A) and the sound level coming from the FM-sound field system speaker was 67 dB(A) (measured at 1 m under the FM system central speaker); and, fourth, the background noise level with the heat pump "on" and the compressor "off" was 66 dB(A) and the sound level coming from the FM-sound system speaker was 77 dB(A) (measured at 1 m under the FM system central speaker.)
III. Results of the intelligibility testing
The scores from these four test conditions and the associated signal-to-noise
ratios (in dB(A)) are shown in the following table:
| Test # | S/N(A), dB | AUDITEC, av & SD | TAC, av & SD |
| 1 | + 20 | 0.88 (0.11) | 0.90 (0.13) |
| 2 | +1 | 0.90 (0.11) | 0.92 (0.15) |
| 3 | -3 | 0.78 (0.11) | 0.86 (0.13) |
| 4 | +11 | 0.93 (0.11) | 0.93 (0.13) |
Statistical analyses carried out using paired-sample t-tests of the mean difference scores demonstrate that the average AUDITEC test scores for Test Condition #3 were significantly lower than the scores for the other three Test Conditions. These results for the AUDITEC tests are basically consistent with those reported earlier by a number of scientists such as Finitzo-Hieber and Tillman (1978), Nabelek and Robinson (1982), and Bradley (1986). Statistical analyses for the TAC scores have not yet been performed.
One set of results that appears compelling is a comparison of the performances
of native-English-speaking students (non-ESL) and English-as-a-second-language
students (ESL) on these tests. As the following table shows, the ESL students
performed less well on the AUDITEC tests. Indeed, a Pearson Correlation
test for AUDITEC Test #3 performances versus ESL students showed a correlation
significant at the 0.05 level (2-tailed). But, on the TAC tests -- those
involving listening to short-paragraph stories and reporting on the contents
of these stories -- the ESL students performed at least as well as the
native-English-speaking students.
| Test # | AUDITEC, av &SD | TAC, av & SD | |
| 1 | ESL | 0.84 (0.15) | 0.92 (0.10) |
|---|---|---|---|
| non-ESL | 0.89 (0.10) | 0.89 (0.14) | |
| 2 | ESL | 0.85 (0.13) | 0.96 (0.07) |
| non-ESL | 0.91 (0.10) | 0.91(0.16) | |
| 3 | ESL | 0.70 (0.11) | 0.85 (0.10) |
| non-ESL | 0.80 (0.10) | 0.86 (0.14) | |
| 4 | ESL | 0.88 (0.23) | 0.94 (0.08) |
| non-ESL | 0.94 (0.08) | 0.92 (0.14) |
Proximities to the sound source and to the heat pump for the students
have also been examined. The distance ranges between students and the sound
source and between students and the heat pump were, respectively, 6-18
ft and 3-17 ft. Whether these proximities (or distances) affected performances
was investigated by computing Pearson Correlations using the differences
of paired-samples for the AUDITEC tests versus these two distances. The
results of these correlations appear in the following table.
| Differences of paired samples from | Pearson Correlations for Distance to Sound Source | Pearson Correlations for Distance to Heat Pump |
| Test 1 vs. Test 2 | 0.311 | -0.363* |
|---|---|---|
| Test 1 vs. Test 3 | 0.251 | -0.252 |
| Test 1 vs. Test 4 | -0.021 | -0.089 |
| Test 2 vs. Test 3 | -0.013 | 0.056 |
| Test 2 vs. Test 4 | -0.302 | 0.244 |
| Test 3 vs. Test 4 | -0.204 | 0.116 |
The lack of significant correlation results for all but the one case (of the paired- samples for Tests 1 and 2 versus the distance to the heat pump) suggests that these distances do not influence the test performances. One question that arises is whether this classroom exhibits true diffuse-field behavior with the concomitant decrease in sound levels with increasing distances (see Hodgson, 1996)? In the event that a true diffuse-field is indeed present, then perhaps the distances between students and the sound source and the heat pump were too short to cause performance disparities?
Briefly and in summary, the results from the four AUDITEC tests indicate that the signal-to-noise ratio is an important influence on the children's success on these tests. Secondly, the ESL students performed less well than native-English-speakers on the AUDITEC word list tests. And, thirdly, the proximities of students to the sound sources and to the heat pump (as the major noise source) appear generally insignificant. One last qualification concerns the quite benign reverberation conditions found in the classroom used in this study (i.e., an RT60 of 0.52 s.) This RT60 quantity is substantially lower than has been measured in a number of other Seattle classrooms. What effects a higher reverberation time -- present in addition to the S/N(A) ratio conditions employed in this study -- would have on test performance seems to be worthy of future investigation.
IV. Acknowledgments
This study has been supported by the University of Washington Royalty Research Fund. The principal author also wishes to thank Paul Sampson, Todd Alonzo, and Samantha Bates of the Department of Statistics, University of Washington, for their considerable guidance.
V. References