An Evaluation of the " Basic Orthographic Syllabic Structure " in a Phonologically Shallow Orthography

The notion of a "Basic Orthographic Syllabic Structure" (BOSS) (Taft, 1979a) was examined in the phonologically shallow orthography of Serbo-Croatian, which is a highly inflected language written in two alphabets--Roman and Cyrillic. Some characters are shared by both alphabets and retain the same pronunciation in each, some are unique to one alphabet, and some are ambiguous, i.e., receive different readings in the two alphabets. Thus, a letter string composed of common and ambiguous characters might be pronounced in one way if read in Roman and in a different way if read in Cyrillic. Lexical decisions were made on a set of words that met the following criteria: When written in Cyrillic, the nominative singular form of the word was phonologically ambiguous while the dative singular form of the word was unambiguous; when written in Roman, both grammatical forms of the word had only one possible pronunciation. The relation between the lex ica1 deci sions to the nominative singular and dative singular forms of the same word depended upon the alphabet in which the words were written. Decision times for the Cyrillic nominative singular forms were very slow relative to those for the Roman nominative singular; in contrast, the decision times for the Roman and the Cyrillic dative singular forms were virtually identical. The BOSS perspective anticipates the same relationship between grammatical forms in both alphabets, since inflected forms of the same word must share the same BOSS and their affixes must occur with the 3ame frequency. In addition, the results showed that the number of wlbiguous characters is a significant determinant of the decision latencies when no unique characters are present. The BOSS perspective was dismissed in favor of the view that the lex ical representation of Serbo-Croatian words is phonological and not purely orthographic. How does a reader determine that a string of letters is a word? The words a reader knows are said to be representedl in a special memory conventionally termed the internal lexicon. Roughly, a representation is a structure whose elements (symbols, predicates, or what~ver) putatively corres~ +A1so University of Connecticut and University of Belgrade. ++University of Belgrade. +++Also University of Connecticut. Acknowledgment. This research was supported in part by NICHD Grant HD 08495 to the University of Belgrade, in part by IUCHD Grant HD 01994 to Haskins Laboratories, and in part by NSF dissertation fellowship BNS 7924409 to Laurie Feldman. [HASKINS LABORATORIES: Status Report on Speech Research SR-66 (1981)]