Abstract
Thirty generative-style context sensitive rules, critically ordered, were formulated and applied to a computer input consisting of place names of English-speaking countries in English orthography. The output was a Hebrew representation of the place names as pronounced in their several countries. The high-level rules first assigned primary stress, then reduced unstressed, non-high vowels to shwa. The rest of the rules, A-rule, B-rule, etc. to Z-rule generated representations of the many-varied pronunciations of the letters in all possible contexts. To assure double stresses in compounds, e.g. Southend, Bridgeport, a list was provided of compound elements, e.g. Bridge-, Broad-, -hall, -cross, etc., with an instruction to insert a word-boundary in such compounds, so that the stress rule would apply twice and the Vowel-Reduction rule would be blocked on the first vowel of the second element of the compound. The rules for England were modified for the U.S.A. and for other English-speaking countries.
Users
Please
log in to take part in the discussion (add own reviews or comments).