The following table shows the sources of spontaneous and other speech material contained in CSJ.
Click the underlined source names to listen to the sample sounds

APS (Academic Presentation Speech) is live recordings of academic presentations in nine different academic societies held in 1999-2001. The societies range from engineering (3 societies, 621 files), humanities (4 societies, 187 files), and social and behavioral sciences (2 societies, 169 files). Most of the APS are 12-25 minutes long, but there are some hour-long plenary lectures, too.

SPS (Simulated Public Speaking) is layman's "speech" on everyday topics of about 10-12 minutes in front of a small friendly audience. Speakers of SPS are paid layman subjects balanced both in their sex and age. 50 speakers (5 males and 5 females for each of 5 different age groups ranging from their twenties to sixties or elder) made a group and were given a set of topics. A set consists of three broad topics.

The following table shows the topics used in the recording. Speakers were given a set of topics about 48 hours prior to the recording time, and asked to prepare three independent talks. Prepared texts were not allowed, but the speakers were encouraged to make outlines of the talks. As for the 100 speakers recorded in the first year, no broad topic was specified. These speakers spoke one or two topics that they chose freely.

Most of the SPS recordings were done in the recording studio of the NIJLA using the same equipment as in the APS recordings. But the recording of the first 50 SPS speakers were outsourced and done in a non-soundproof room. Although the same recording equipment was used, the room acoustics were different.

APS and SPS are the two main sources of CSJ. In addition to these two, we also made recordings of speech materials of a different nature.

"SPS by interviewee" means, literally, SPS spoken by those speakers who took part in the dialogue recording.

"Reading text" means reading of two written passages excerpted from two popular science books. 487 SPS speakers read these passages. Time required for reading was 3-4 minutes.

"Interview APS" is the interview with an APS speaker about the content of his/her APS. The interviewees are those speakers of "SPS by APS speakers". "Interview SPS" is the interview with the same speakers of "Interview APS" about the content of their SPS. These interviews were 10-15 minutes long.

The same pair of interviewee and interviewer also participated in "Task oriented dialogue," in which they were asked to guess the rank order of the performance fees of 8 TV personalities. The same pair of speakers participated in "free dialogue", also

"Reproduced APS" is the reading of transcribed APS by the same original speaker. Speakers read faithfully all filled-pauses and other disfluencies involved in the transcribed text.