Irrelevant speech effect in open plan offices: Comparison of two models explaining the decrease in performance by speech intelligibility and attempt to reduce interindividual differences of the mental workload by task customisation

Abstract

For employees working in open plan offices, speech noise is now recognised as the main source of disturbance. In the literature, a series of studies on open spaces have found that the increased speech intelligibility happens to decrease the performance during a short-term memory task. The theoretical model proposed by Hongisto relates the decreased performance (DP) to the Speech Transmission Index (STI). The model predicts that for STI values from 0.7 to 1, which correspond to a speech signal with an intelligibility of almost 100%, the DP remains constant at 7%, but few studies have examined the high end of this range. Here, we investigated the relationship between the DP and the STI by varying the STI up to 0.9. Fifty-five subjects between 25- and 59-years-of-age participated in the experiment. Subjects performed a short-term memory task in silence and in four different sound conditions (STI from 0.25 to 0.9). The task itself was personalised by prior measurement of mnemonic span. It was thus possible to define two different cognitive loads (low/high) based on the mnemonic span value determined for each subject. Subjects subjectively assessed the mental workload and sound annoyance at the end of each short-term memory task in each sound condition. Significant interactions between STI and DP, mental workload and sound annoyance were found. In addition, the age of subjects correlated significantly with their performance during the short-term memory task.