PHONOLOGICAL ROUTE PROCESSING: EVIDENCE OF INTUITION IN PORTUGUESE SPELLING

Resumo

A new procedure for calculating spelling difficulty in Brazilian Portuguese is presented. It aims at predicting the risk of committing spelling errors. It is based on phone to grapheme prevalence indexes.  For any given phone there are n graphemes, each with its prevalence index.  For instance, 11 graphemes encode phone [s], each with its own prevalence: 54% as “s” [sa'livɐ], 27,4% a “c” [si'ɡahɐ] , 9,2% as “ç” ['pɾasɐ],  5,38% as “ss” ['masɐ], 1,82% as “sc” [fasina'doɾ, 1,51% as “x” [espe'liɾ], 0,40% as “z” ['dɛs], 0,12% as “xc” [ese'deɾ], 0,0079% as “xs” [esu'dʲiɾ], 0,0026% as “sç” [kɾe'sɐ], and 0,0009% as “cç” [se'sɐ̃w̃]. According to it, the lower the prevalence with which a given grapheme encodes a given phone, the greater the spelling error vulnerability. Prevalent graphemes tend to intrude upon non-prevalent ones. A study assessed whether prevalence indexes could account for spelling error distribution. A sample of 154 students (61 college and 93 elementary school ones) was exposed to a spelling under dictation task involving 280 different phone-grapheme prevalence indexes. Each student had to spell 560 rare words, 6.6 phones each on average, totaling 3,676 spelling events. Regression analysis results revealed spelling precision is directly proportional to prevalence index. Thus the procedure has been found empirically valid for calculating spelling difficulty in Brazilian Portuguese.