You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
47 lines
2.3 KiB
47 lines
2.3 KiB
From text to speech: The MITalk system
|
|
|
|
4. its ability to change a preceding y to i or to cause the omission of a
|
|
preceding e.
|
|
Prefixes are given no further specification.
|
|
|
|
Detection of suffixes proceeds in a right-to-left, longest-match-first fashion.
|
|
When no additional suffixes can be detected, or when a possible suffix is judged
|
|
syntactically incompatible with its right-adjacent suffix by a part-of-speech test
|
|
using the first two classifications above, the process is terminated. Finally,
|
|
prefixes are detected left-to-right, also by longest match first. If at any time the
|
|
removal of an affix would leave no letter in the remainder of the word, the affix is
|
|
not removed.
|
|
|
|
An example of affix detection and analysis is furnished in Figure 6-1 below.
|
|
Two possible suffixes, ish and ing, are detected. The suffix ing terminates either a
|
|
noun or a verb, and is constrained to follow either a noun-forming or a verb-
|
|
forming suffix. The suffix ish, however, is adjectival. Therefore, this possible
|
|
analysis is rejected, and the correct analysis is chosen. If the string ish had been
|
|
selected as a suffix, the root to which it attaches would have been assumed to end
|
|
in e, and would have been pronounced fine.
|
|
|
|
finishing
|
|
fin+ish+ing possible suffix analysis
|
|
ing: (a) nominal or verbal suffix
|
|
(b) follows nominal or verbal suffix
|
|
ish: (a) adjectival suffix
|
|
parts of speech not compatible
|
|
(b) follows nominal or adjectival suffix
|
|
finish+ing correct analysis
|
|
|
|
Figure 6-1: Suffix detection in the word finishing
|
|
|
|
6.2.2 Domain of application
|
|
The domain of application of the second stage rules excludes any previously
|
|
recognized affixes and is assumed to be a single-root morph. This stage is in-
|
|
tended primarily for consonant rules and proceeds from the left of the string to the
|
|
right. Extending the domain to the whole letter string once again for the third
|
|
stage, a phonemic representation is given to affixes, vowels, and vowel digraphs.
|
|
Phonemic representations are produced by a set of ordered rules which con-
|
|
vert a letter string to a phonetic segment string in a given context. Both left and
|
|
right contexts are permitted in the expression of a rule. Any one context may be
|
|
composed of either letters or segments. Combination of these possibilities for both
|
|
left and right contexts allows for four possible context types.
|
|
|
|
58
|