147th ASA Meeting, New York, NY

[ Lay Language Paper Index | Press Room ]

Verbal Memory Performance
Not Affected by Speech Interference
in Typical Learning and Working Environments

Dana M. Lodico -
Illingworth & Rodkin, Inc.
505 Petaluma Boulevard, South
Petaluma, CA 94952

Coauthors: Rendell R. Torres, Yasushi Shimizu, Claudia Hunter
Rensselaer Polytech. Inst., Troy, NY 12180-3590

Work supported by the Program in Architectural Acoustics, School of Architecture, Rensselaer Polytechnic Institute

Popular version of paper 1aAA6
Presented Monday Morning, May 24, 2004
147th ASA Meeting, New York, NY

Current psychological theories indicate that background noise, especially extraneous speech, is one of the most common forms of distraction in open-plan office and classroom environments. Although cognitive psychology literature provides an abundance of information on the destructive effects of irrelevant speech on performance, very few of these studies examine topics that relate to complex tasks or the built environment. The majority of experiments have thus far focused on simple task performance studies (i.e., remembering a list of letters displayed on a computer screen, remembering a sequence of dot locations on a screen, etc) with visual material (i.e., text, numbers, or colors), and sound levels that are much higher than those typically found in classroom and office environments. However, most classroom activities (lectures, group projects) and many office tasks (meetings, telephone conversations) depend on verbally presented material for information. Classroom acoustics guidelines have recently emerged with a focus on intelligibility. This study investigates the effects of interference speech and the built acoustical environment on human performance. The effects of room size, geometry, and acoustical parameters on human performance were studied through human subject testing.

Four studies (Experiment 1A, Experiment 1B, Experiment 2, and Experiment 3) were conducted to evaluate human participant performance under constrained laboratory conditions. In each experiment, participants were asked to listen to a short news passage read to them by a recorded female voice through a two-channel loudspeaker setup and produce a verbatim written-recall of the passage to the best of their ability. A secondary distracting noise source was played during both the passage reading and the recall portion of the experiment. The focus readings were news article excerpts ranging from 1-minute and 30-seconds to 2-minutes in length. Articles were chosen to have similar difficulty levels, low acoustical confusion, and low ease of memory. The secondary noise was a radio talk show recording consisting of speech that was unrelated in content to the focus passage. A twenty-one minute segment was selected based on continuity of the sounds (by excluding phone conversations and unusual noises) and was composed entirely of speech. In addition to the prose recall task, the fourth study (Experiment 3) also included a card-sorting task. The participants were instructed to give the card-sorting task preference, thereby changing the memory for prose into a secondary task.

Experiments were conducted within a hemi-anechoic laboratory environment. Human performance was rated within acoustical images of four different spaces and at five different background noise levels, as well as a quiet condition. The impulse responses (IR) of three classrooms on the Rensselaer campus, selected to represent a variety of listening environments, were measured and analyzed. The classrooms were selected for size and Reverberation Time (RT) and based on a standard range for offices and classrooms. The room volume ranged from 2,300 ft3 for a small classroom/office environment to 28,200 ft3, modeling a large open plan office environment or a small lecture space. The Reverberation Times ranged from 0.8 seconds, representing an acoustically dry space, to 2.5 seconds, representing a more reverberant environment.


Acoustical Parameter


Near Field Distance (ft)

Far Field Distance (ft)

Reverb Radius (ft)

RT (s)

D50 [-]

Room 203







Room 206














A computer-generated acoustical-image was created for each classroom by convolving the recorded speech signals with near and far-field impulse responses (IR) from the three classrooms (Gallery, Room 206, Room 203) to create a total of four room conditions (including an anechoic condition) for each recorded file. The convolved near/far field files were then matched up to create an acoustical image model of each of the four spaces and individual sound files were calibrated to maintain the desired signal-to-noise ratio (SNR) for each experiment.

In all of the experiments, the focus passages had a mean sound pressure level (SPL) of 60dBA, a typical level for speech. The secondary distracting noise was calibrated to be at one of the following mean sound pressure levels: 45 dBA, 40 dBA, 35 dBA, 30 dBA, 25 dBA depending on the specifications of the individual experiment. Studies have shown a +15 dB signal-to-noise ratio (SNR) to be the limit below which speech becomes less intelligible for most listeners with bland non-speech background noise. With an average conversational voice level of 60 dBA, the background noise level should not exceed 45 dBA. It is likely that intelligibility would be more affected by background speech than by bland background noise.

The experiments found no statistically significant effect of sound pressure level or acoustical room environment on short-term memory performance with verbally presented material. Distracting background noise sound pressure levels of up to 45 dBA (with a 60 dBA focus voice), with SNR from +35 to +15 dBA, did not affect short term memory for verbal prose even when compared to the no-noise condition. Likewise, memory for verbal prose was not affected by acoustical room environment. Furthermore, the performance under all room environments with 45-dBA background speech was shown to be the same as that of the quiet condition. The slight differences between means for each background condition (1 to 2 units) fall well within the standard deviations (4 to 7 units) and no trends are apparent, signaling that there is very little chance that an effect is present.

Conversely, when the average idea units recalled in Experiment 3 (with the sorting task) are compared to the average recall scores under the same room conditions with memory of verbal prose as the primary task, the secondary scores are significantly lower.

The addition of the sorting task as a distracter produced far more interference on the memory for verbal prose task that the differences between room conditions. In fact, the average mean idea units recalled for Experiment 3 were less than two-thirds of those from the previous experiments, whereas the differences between scores under different room conditions only varied by 1-2 idea units, even at the greatest point of contrast. Sorting time differences were also apparent between background conditions. The average sorting times ranged from 2 minutes and 11 seconds to 2 minutes and 45 seconds. The sorting times for the silent background condition are lower than those for any other room condition (by 23-34 seconds).

Additional testing would be necessary before any conclusions could be drawn from the results. However, this pilot study suggests that further study would reveal significant results.

Possible Future Investigations

In contrast to the memory for prose task, the performance for sorting playing cards showed perceptible differences under varied room conditions. Statistical significance could not be inferred from the results due to the small subject sample size. This line of research should be pursued further.

The experiments in this investigation found no grounds in which to suppose that additional testing using extreme spaces that are not considered acceptable for speech would reveal performance effects. However, it may be possible to define a threshold, if one exists, at which effects become apparent.

[ Lay Language Paper Index | Press Room ]