EyeTracking data (24 Japanese Native Speakers) on BCCWJ Newspaper Samples

Masayuki Asahara (National Institute for Japanese Language and Linguistics, Japan)

Masayuki Asahara (National Institute for Japanese Language and Linguistics, Japan)

CC BY-NC-SA 3.0 https://creativecommons.org/licenses/by-nc-sa/3.0/deed.ja

National Institute for Japanese Language and Linguistics (2018) `BCCWJ-EyeTrack' (ver. 1.0.0)

the files without the original text for Stan analysis

• fft: First Fixation time
• fpt: First Pass Time
• spt: Secound Pass Time
• rpt: Regression Path Time
• total: Total Time
• self: Self paced reading Time

{fft,fpt,spt,rpt,total,self}.csv files include titles, subtitles, lists, and captions.

{fft,fpt,spt,rpt,total,self}-def.csv files exclude titles, subtitles, lists, and captions.

• logtime: (logarithmic values (log(x)) of fixation/reading time)
• invtime: (reciprocal values (1/x) of fixation/reading time)
• surface: the original text masked by "_" due to the copyright issue
• time: fixation/reading time (msec)
• measure: (fft, fpt, spt, rpt, total or self)
• sample: sample id (A, B, C, D)
• article: article id in BCCWJ
• metadata_orig: (the BCCWJ original metadata categories of titles, subtitles, lists, captions, texts, and so on.)
• metadata: the modified version of the metadata categories of titles, subtitles, lists, captions, texts, and so on.
• sessionN: session order
• length: length of the surface, namely the number of characters in the original text
• space: presented with or without spaced between the phrases
• subj: subject participant id
• setorder: (the presentation order for the subject participants)
• rspan: the subject feature -- the result of reading span test
• voc: the subject feature -- the result of vocab test
• dependent: the number of the dependent for the phrase
• articleN: article presentation order
• screenN: screen presentation order
• lineN: line presentation order -- the vertical offset
• segmentN: segment presentation order -- the horizontal segment order in the line
• sample_screen: (screen id)
• is_first: is_first segment in the line (1: FALSE, 2: TRUE)
• is_last: is_last segment in the line (1: FALSE, 2: TRUE)
• is_second_last: is_last segment in the line (1: FALSE, 2: TRUE)
• infostatus, definiteness, specificity, animacy, sentience, agentivity, commonness: the annotation of the information structure from BCCWJ-InfoStr
• CLRHST, CLRMST, CLRFUT, CLRHRT, CLRHSL, CLRMSL, CLRFUL, CLRHRL, CLMHST, CLMMST, CLMFUT, CLMHRT, CLMHSL, CLMMSL, CLMFUL, CLMHRL: the annotation of the clause boundary infomation from BCCWJ-ToriClause
• WLSPSUWA, WLSPSUWB, WLSPSUWC, WLSPSUWD, WLSPLUWA, WLSPLUWB, WLSPLUWC, WLSPLUWD: the annotation of the thesaurus categories from BCCWJ-WLSP
• cbow_norm, cbow_prev, skip_norm, skip_prev: word embedding features from NWJC2vec -- Japanese Word Embeddings
• count, count_ave, count_log, count_ave_log, count_ave_kika: word frequency features from NINJAL Web Japanese Corpus
• bid, did: word dependency information from BCCWJ-DepPara we used `dependent' column for analysis
• type_pred, dist_invga, type_ga, dist_invo, type_o, dist_invni, type_ni: predicate argument relation from BCCWJ-PAS

these files includes the annotation data

• 0th column the number of column in *.csv
• 1st column the label of annotation data
• 2nd column the id in *.csv

masayu-a -at- ninjal.ac.jp