The "Morphology" of English Spelling: A Look at the SRS Text-Modification Rules for English

Any scheme that accurately converts English text to phonemes must be based on morphological analysis--i.e., an analysis of words into prefixes, suffixes, and roots. This morphological analysis is accomplished in the SRS synthesis rule set for English [Hertz 1982] by a set of text modification rules, which divide words into morphological units on which subsequent phoneme-generating rules are based.

Because these units are not always true morphemes (for example, fibe + er for fiber on analogy with bribe + er for briber), they are more accurately called "spelling morphs." By analyzing text into spelling morphs, the text-modification rules not only simplify phoneme and stress prediction, but the handling of exceptions as well.

In addition, by marking particular characteristics of these morphs (for example, syllable structure), the text-modification rules also simplify the prediction of certain low-level effects, such as aspiration.