<?xml version="1.0" standalone="yes"?>
<Paper uid="P95-1002">
  <Title>Automatic Induction of Finite State Transducers for Simple Phonological Rules</Title>
  <Section position="9" start_page="14" end_page="14" type="concl">
    <SectionTitle>
8 Conclusion
</SectionTitle>
    <Paragraph position="0"> Inferring finite state transducers seems to hold promise as a method for learning phonological rules. Both of our initial augmentations of OSTIA to bias it toward phonological naturalness improve performance. Using information on the alignment between input and output strings allows the algorithm to learn more compact, more accurate transducers. The addition of decision trees at each state of the resulting transducer further improves accuracy and results in phonologically more natural transducers. We believe that further and more integrated uses of phonological naturalness, such as generalizing across similar phenomena at different states of the transducer, interleaving the merging of states and generalization of transitions, and adding memory to the model of transduction, could help even more.</Paragraph>
    <Paragraph position="1"> Our current algorithm and most previous algorithms are designed for obligatory rules. These algorithms fall completely when faced with optional, probabilistic rules, such as flapping. This is the advantage of probabilistic approaches such as the Riley/Withgott approach. One area we hope to investigate is the generalization of our algorithm to probabilistic rules with probabilistic finite-state transducers, perhaps by augmenting PFST induction techniques such as Stolcke &amp; Omohundro (1994) with insights from phonological naturalness.</Paragraph>
    <Paragraph position="2"> Besides aiding in the development of a practical tool for learning phonological rules, our results point to the use of constraints from universal grammar as a strong factor in the machine and possibly human learning of natural language phonology.</Paragraph>
  </Section>
class="xml-element"></Paper>