Automatic Conversion of Dialectal Tamil Text to Standard Written Tamil Text Using Fsts Rules, Analogy, and Social Factors Codetermine Past-tense Formation Patterns in English Revisiting Word Neighborhoods for Speech Recognition

ii Preface These proceedings contain the contributions presented at the MORPHFSM workshop held on June 27, 2014 in conjunction with ACL 2014 in Baltimore, Maryland, USA. The workshop was a joint meeting of two special interest groups of the ACL: • SIGMORPHON — the special interest group in computational morphology, phonology and phonetics, and • SIGFSM — the special interest group on finite-state methods. It was the thirteenth meeting of SIGMORPHON and an off-year event for SIGFSM. The full-day workshop consisted of an invited presentation by JASON EISNER, contributed presentations, and a special panel session on open problems. The workshop covered a wide range of topics from theoretical to applied morphology and finite-state technology in natural language processing. This volume contains the 7 regular and 1 panel paper that were presented at the workshop. In total, 12 papers (10 regular and 2 panel papers) were submitted to a doubly blind refereeing process, in which each paper was reviewed by 3 program committee members. The overall acceptance rate was 67%. The program committee was composed of internationally leading researchers and practitioners selected from academia, research labs, and companies. The organizing committee would like to thank the program committee for their hard work and valuable feedback, the invited speaker JASON EISNER for his innovative and inspiring keynote, our panelists for their interesting discussion and expertise, the local organizers for their tireless efforts, the ACL administration for their support, and last but not least the authors for their contributions. Abstract Word neighborhoods have been suggested but not thoroughly explored as an explanatory variable for errors in automatic speech recognition (ASR). We revisit the definition of word neighborhoods, propose new measures using a fine-grained artic-ulatory representation of word pronunciations , and consider new neighbor weight-ing functions. We analyze the significance of our measures as predictors of errors in an isolated-word ASR system and a continuous-word ASR system. We find that our measures are significantly better predictors of ASR errors than previously used neighborhood density measures.