Analysis output
Model: openbmb/MiniCPM5-2B, revision 12a3808a956f869c767195e9266b59c4d21d92e2, greedy decoding, do_sample=False, max_new_tokens=180
Environment: transformers 5.17.0, torch 2.11.0+cu128, Tesla T4
48 prompts, 12 scenarios, 2x2 design, one response per prompt, coder 1 scores only (second coder in agreement_results.txt)

Mean total score out of 8, by condition
  implicit_english           4.500   n=12
  roman_urdu_mixed           2.333   n=12
  explicit_english           5.667   n=12
  explicit_context_added     3.833   n=12

Paired differences
language, context implicit         -2.167  95% CI [-3.833, -0.583]  n=12
language, context explicit         -1.833  95% CI [-3.417, -0.167]  n=12
context, English                   +1.167  95% CI [+0.333, +2.167]  n=12
context, Roman Urdu                +1.500  95% CI [+0.500, +2.500]  n=12

interaction, recovery in RU vs EN  +0.333  95% CI [-1.000, +1.583]  n=12

Per-dimension means by condition
                               intent_infer  social_appro  unsupported_  semantic_pre
  implicit_english                     1.33          0.92          1.33          0.92
  roman_urdu_mixed                     1.00          0.58          0.42          0.33
  explicit_english                     1.58          1.42          1.42          1.25
  explicit_context_added               1.33          0.92          1.00          0.58

Moderator split: fully translated situation vs narration only
(WITHDRAWN. The matched rerun showed this split reflected unequal manipulation. See README and rerun_results.txt.)

Whole situation in Roman Urdu                n= 5  mean language effect -4.80
    per scenario: request_refusal -2, elder_disagreement -4, hosting_pressure -4, indirect_no -7, respect_register -7
Roman Urdu quote inside English narration    n= 7  mean language effect -0.29
    per scenario: kinship_context +0, favor_indirect -4, face_saving +0, code_switch +1, deference +2, family_obligation -1, ambiguity +0

Per-scenario totals out of 8
scenario                  implicit_engli  roman_urdu_mix  explicit_engli  explicit_conte
ambiguity                              6               6               6               6
code_switch                            1               2               1               2
deference                              6               8               8               8
elder_disagreement                     5               1               8               2
face_saving                            1               1               2               6
family_obligation                      1               0               6               1
favor_indirect                         8               4               8               6
hosting_pressure                       5               1               8               5
indirect_no                            8               1               8               5
kinship_context                        0               0               0               1
request_refusal                        5               3               5               2
respect_register                       8               1               8               2