text_transcript/README.md

Text transcript

This folder contains a transcript file for each session (e.g., G1S1F_transcript.txt). 　

The files were exported from ELAN, the software used for transcription, and are UTF-8-encoded, tab-delimited text files consisting of the following four columns:

[Participant ID] [Start time (seconds)] [End time (seconds)] [Japanese transcription]

The transcription followed the rules below.

1. Utterance segments were identified with reference to the audio waveform. When a silent interval of 200 ms or longer was observed between non-silent portions, the portions before and after the silence were treated as separate utterance segments.

2. Non-silent portions shorter than 30 ms were not identified as utterance segments.

3. Spoken words were transcribed as uttered, including speech errors.

4. Exhalations, inhalations, breathing sounds that could not be identified as either, and laughter were represented in the transcripts as 《呼気》(exhalation), 《吸気》(inhalation), 《呼吸》(breathing), and 《笑い》(laughter), respectively.

5. Breathing sounds (《呼気》, 《吸気》, and 《呼吸》) were generally treated as segments separate from spoken words. However, when the portions of speech before and after a breathing sound were less than 200 ms apart and were identified as a single utterance segment, the intervening breathing sound was included in the same segment.

6. When laughter was accompanied by intelligible speech, the speech was transcribed together with 《笑い》 in the same segment. When laughter was not accompanied by intelligible speech, only 《笑い》 was recorded.

7. Unintelligible speech was indicated using curly brackets{ }.

8. Each sentence ended with the Japanese period 。, except that the question mark ？ was used for utterances with rising intonation. Japanese commas 、 were inserted where appropriate, taking the content and pauses into account.

9. Kanji use was generally limited to the jōyō kanji (常用漢字).

10. Kanji, hiragana, and katakana were used in preference to alphabetic spelling. Exceptions were made for proper nouns and cases in which the speaker intentionally used English or another foreign language.

11. Kanji, hiragana, katakana, and numerals were written using full-width, multibyte characters, whereas alphabetic characters were written using half-width, single-byte ASCII characters.