Context-sensitive rules

SPE showed how many seemingly desperate phenomena could in fact be explained by rules fliat made direct access to the feature set. Nearly all rules were posited in terms of context-sensitive rules (these rules are used to define a third fundamental type of grammar in addition to finite state rules and context free rules that we saw before.). For example, the tensing rule that describes how word patterns such as divine, divinity and profane, profanity arise is given as ... 

A number of other data driven techniques have also been used to G2P conversion. Pagel et al [343] introduced the use of decision trees for this purpose, a choice which has been adopted as a standard technique by many researchers [249], [198]. The decision tree works in the obvious way, by considering each character in turn, asking questions based on the context of the character and then outputing a single phoneme. This can be seen as a way of automatically training the context-sensitive rules described above. 

A more principled approach is to define a general rule format, a mle processor or engine and a clearly defined set of rules. This at least has the advantage of engine/rule separation meaning that the way the rules are processed and the rules themselves are separate, which helps with system maintenance, optimisation, modularity and e q)ansion to other languages. One common type of rule is the context-sensitive rewrite rule, which takes the form... 

Popular classifiers include context-sensitive rewrite rules, decision lists, decision trees, naive Bayes classifiers and HMM taggers. 

Most approaches use context sensitive rewrite rules of the type introduced in Section 5.2.4. Recall that these are of the form... 

The statistieal approach has been somewhat of a latecomer to grapheme-to-phoneme conversion, perhaps because of the success of other data driven techniques such as pronunciation by analogy or the impression that context-sensitive rewrite rules are adequate so long as they can be automatically trained, e.g. by a decision tree. In recent years however a number of approaches have been developed which give a properly statistical approach. 

Numerous SERS studies of adsorbed molecules have appeared in the literature. Obviously, it is a useful method for the identification of species at the interface, and its inherent surface sensitivity is an attractive feature. In this context it should be noted that the adsorption of a molecule can change the selection rules for Raman scattering, and modes that are Raman inactive in the isolated molecule may show up in SERS. 

---