In generative phonology, the hierarchical feature tree provides a natural formalism to express recurrent feature groupings. Mentioning a particular node implies all the information dominated by that node. An essential property of the feature tree and the segments it generates is this relation of dominance or dependence, or both.
This relation provides a simple formalism to explain several
additional phonological processes. Here's why: As (2)
illustrates
By establishing relations of dominance and dependence, the feature tree depicted in (1) explains why [high], [low] and [back] group together -- because they are all daughters of the dorsal node. On the other hand, [high] and [nasal] should not join up in a rule or constraint because there is no single node that dominates only these features.
Hierarchical trees are also frequently used to express relationships in syntax. The following example, marked up in XML, reveals the hierarchical structure of the atomic syntactic elements that make up a sample sentence:
The woman ]]> saw the man in
the park
As a hierarchical tree, the sentence looks like this:
Abstract: Abney's Memory Requirements and Lexical Ambiguities of Parsing Strategies
Books on Computational Linguistics