Confusion matrices for judges 1/2. Judge 1 is represented on the horizontal axis, while Judge 2 is on the vertical. The axes represent the accent judgement, and the values in the matrix are the number of file judgements at each vertex. Arabic 112 utterances 1 2 3 4 1 8 2 9 7 3 3 2 39 39 3 4 2 BPortuguese 459 utterances 1 2 3 4 1 33 9 2 24 38 6 3 14 194 134 2 4 2 3 Cantonese 261 utterances 1 2 3 4 1 12 5 2 12 12 2 3 2 113 94 1 4 1 6 Czech 102 utterances 1 2 3 4 1 7 3 2 2 10 3 3 5 33 38 1 4 Farsi 261 utterances 1 2 3 4 1 6 3 2 7 26 7 3 1 98 101 2 4 1 5 2 French 284 utterances 1 2 3 4 1 26 5 2 11 26 10 3 12 108 78 4 4 4 German 326 utterances 1 2 3 4 1 39 27 2 2 16 30 13 3 10 107 78 2 4 1 Hindi 348 utterances 1 2 3 4 1 3 3 2 2 16 5 3 4 135 174 1 4 3 1 Hungarian 277 utterances 1 2 3 4 1 15 15 1 2 9 14 5 3 5 97 108 3 4 4 1 Japanese 194 utterances 1 2 3 4 1 12 7 1 2 5 18 1 3 2 65 74 3 4 6 Korean 169 utterances 1 2 3 4 1 8 3 1 2 7 2 1 3 1 65 70 1 4 1 8 Malay 56 utterances 1 2 3 4 1 3 2 1 4 1 3 21 23 1 4 2 Mandarin 282 utterances 1 2 3 4 1 9 7 2 11 14 3 3 9 107 117 4 4 1 Italian 214 utterances 1 2 3 4 1 23 14 2 2 3 9 5 3 1 65 78 4 4 8 2 Polish 143 utterances 1 2 3 4 1 9 4 2 5 2 1 3 7 40 63 2 4 8 2 Spanish 308 utterances 1 2 3 4 1 54 26 2 2 12 24 7 3 10 110 54 3 4 5 IPortuguese 66 utterances 1 2 3 4 1 7 1 2 1 4 3 1 23 29 4 Russian 236 utterances 1 2 3 4 1 7 6 1 2 8 14 1 3 13 102 74 3 4 5 2 Swedish 203 utterances 1 2 3 4 1 31 19 2 14 37 4 3 9 65 23 4 1 Swahili 71 utterances 1 2 3 4 1 2 2 5 1 3 30 31 1 4 1 Tamil 326 utterances 1 2 3 4 1 2 4 2 5 15 3 3 2 123 155 10 4 5 1 Vietnamese 134 utterances 1 2 3 4 1 5 4 1 2 4 8 2 3 3 44 50 3 4 1 6 1 ALL Languages 4961 utterances 1 2 3 4 1 348 172 11 2 173 338 86 3 114 1818 1730 50 4 6 92 13