pragmatic-code-switch-blindspot / agreement_results.txt
rshahbaz's picture
Upload folder using huggingface_hub
98739f8 verified
Raw History Blame Contribute Delete
2.55 kB
Agreement, coder 1 vs coder 2, n=48, quadratic-weighted Cohen's kappa
intent_inference kappa 0.41 exact 52% coder 2 used {0: 17, 1: 21, 2: 10}
social_appropriateness kappa 0.14 exact 46% coder 2 used {0: 0, 1: 46, 2: 2}
unsupported_assumptions kappa 0.69 exact 62% coder 2 used {0: 14, 1: 16, 2: 18}
semantic_preservation kappa 0.63 exact 60% coder 2 used {0: 25, 1: 20, 2: 3}
pooled kappa 0.54
totals: correlation 0.77, mean absolute difference 1.48
Coder 1, total out of 8
language, context implicit -2.17 [-3.83, -0.58] n=12
language, context explicit -1.83 [-3.42, -0.17] n=12
context, English +1.17 [+0.33, +2.17] n=12
context, Roman Urdu +1.50 [+0.50, +2.50] n=12
interaction +0.33 [-1.00, +1.58] n=12
moderator: whole situation in RU -4.80 [-6.40, -3.20] n=5
moderator: RU quote in EN narration -0.29 [-1.71, +0.86] n=7
Coder 2, total out of 8
language, context implicit -1.58 [-2.92, -0.25] n=12
language, context explicit -1.17 [-2.75, +0.33] n=12
context, English +0.58 [-0.33, +1.50] n=12
context, Roman Urdu +1.00 [+0.17, +2.08] n=12
interaction +0.42 [-0.83, +1.75] n=12
moderator: whole situation in RU -3.80 [-4.80, -2.80] n=5
moderator: RU quote in EN narration +0.00 [-1.14, +1.14] n=7
Mean of both coders, total out of 8
language, context implicit -1.88 [-3.29, -0.50] n=12
language, context explicit -1.50 [-2.88, -0.12] n=12
context, English +0.88 [+0.21, +1.54] n=12
context, Roman Urdu +1.25 [+0.54, +2.08] n=12
interaction +0.38 [-0.42, +1.12] n=12
moderator: whole situation in RU -4.30 [-5.40, -3.10] n=5
moderator: RU quote in EN narration -0.14 [-1.29, +0.86] n=7
Mean of both coders, excluding social appropriateness, total out of 6
language, context implicit -1.71 [-2.88, -0.54] n=12
language, context explicit -1.17 [-2.38, +0.04] n=12
context, English +0.54 [-0.08, +1.17] n=12
context, Roman Urdu +1.08 [+0.50, +1.79] n=12
interaction +0.54 [-0.21, +1.29] n=12
moderator: whole situation in RU -3.80 [-4.50, -3.10] n=5
moderator: RU quote in EN narration -0.21 [-1.14, +0.64] n=7