Hierarchical Trees: Relations of Dominance and Dependence
Trees Can Express Recurrent Feature Groupings
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, two or more features can be expected to co-occur in rules or constraints only if they share a common node in the tree. In other words, defining phonological rules over nodes in the tree graph makes certain feature groupings simple to express and others more complex. In (2), for instance, {a} and {b} can be simply isolated by mentioning {f}; on the other hand, {b} and {c} can be grouped only by also including {a}, {d} and {e}, which would be an instance of complete assimilation.
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:
<s>
<np><det>The</det><n>woman</n></np>
<vp><v>saw</v>
<np><det>the</det><n>man</n>
<pp><p>in</p>
<np><det>the</det><n>park</n></np>
</pp>
</np>
</vp>
</s>As a hierarchical tree, the sentence looks like this:
