Web-based language listening tests require digitized and compressed audio files which can be stored on computer storage media and then delivered through the Internet, intranet or local area network (LAN). For digital audio, higher sound quality is achieved at the expense of bigger data size and greater bit rate since higher sound quality requires higher sample rate and higher sample depth. This causes a dilemma for testers in choosing the proper sample rate and sample depth for digital audio recording for the purpose of web-based English listening comprehension tests. While higher quality is desirable for better test validity and reliability, lower quality is preferred so that the audio can be transferred smoothly through the Internet, intranet or LAN. This study examined the effects of digital audio quality on students' performance in a LAN delivered English listening comprehension test created with the World Wide Web (WWW) technology. Based on the data collected through the English listening comprehension test participated by 335 EFL students in a LAN environment, the researcher compared the effects of three different sample rates (44 kHz, 22 kHz and 11 kHz) and two different sample depths (16 bit and 8 bit). Using the statistical method of 2 x 3 ANOVA, the researcher has found that students tested with 44 kHz audio and 22 kHz audio performed significantly better than those tested with 11 kHz audio, but students tested with 16 bit audio did not perform significantly better than those tested with 8 bit audio. No interaction effect was found between sample rate and sample depth. The researcher concluded that 22 kHz and 8 bit should be set as the standard for digital audio for the purpose of web-based English listening comprehension tests and all equipment involved in such a test must be qualified for effectively recording, transmitting or rendering digital audio of 22 kHz and 8 bit.
展开▼
