Full proceedings volume - Australasian Language Technology ...

More documents

Recommendations

Info

Using a large annotated historical corpus to study word-specific effects in sound change Jennifer Hay School of Languages, Cultures and Linguistics University of Canterbury jen.hay@canterbury.ac.nz Abstract The Origins of New Zealand English Corpora (ONZE) at the University of Canterbury contain recordings spanning 150 years of New Zealand English. These have all been force-aligned at the phonemelevel, and are stored with many layers of annotation some which have been automatically generated, and some which have been manually annotated. We interact with the corpus via our custom LaBB-CAT interface (LAnguage, Brain and Behaviour Corpus Analysis Tool). I will begin the talk by describing and demonstrating the corpus, and its associated LaBB-CAT tool. I will then focus on one particular recent study which has used the corpus, which aims to understand processes of sound change. The combination of the time-depth of the ONZE collection, and the degree of careful annotation it contains, makes it an ideal data-set for the study of mechanisms underlying sound change. In particular, we aim to address the question which has been the subject of long-standing debate in the sound-change literature do sound changes proceed uniformly through the lexicon, or are there word-specific changes, with some words more ahead in the change than others? I describe a study which aimed to investigate this question by focusing on the mechanisms underpinning the New Zealand English front short vowel shift, of the vowels in words like bat, bet and bit. We automatically extracted formant values for over 100,000 tokens of words containing these vowels, We show that this data contains good evidence for word-specific effects in sound change, and argue that these are predicted by current models of speech production and perception, in combination with well-established psycholinguistic processes. Jennifer Hay. 2012. Using a large annotated historical corpus to study word-specific effects in sound change. In Proceedings of Australasian Language Technology Association Workshop, pages 2−2.
Diverse Words, Shared Meanings: Statistical Machine Translation for Paraphrase, Grounding, and Intent Chris Brockett Microsoft Research Redmond, WA Chris.Brockett@microsoft.com Abstract Can two different descriptions refer to the same event or action? Recognising that dissimilar strings are equivalent in meaning for some purpose is something that humans do rather well, but it is a task at which machines often fail. In the Natural Language Processing Group at Microsoft Research, we are attempting to address this challenge at sentence scale by generating semantically equivalent rewrites that can be used in applications ranging from authoring assistance to intent mapping for search or command and control. The Microsoft Translator paraphrase engine, developed in the NLP group, is a large-scale phrasal machine translation system that generates short sentential and phrasal paraphrases in English and has a public API that is available to researchers and developers. I will present the data extraction process, architecture, issues in generating diverse outputs, applications and possible future directions, and discuss the strengths and limitations of the statistical machine translation model as it relates to paraphrasing, how paraphrase is like machine translation, and how it differs in important respects. The statistical machine translation approach also has broad applications in capturing user intent in search, conversational understanding, and the grounding of language in objects and actions, all active areas of investigation in Microsoft Research. Chris Brockett. 2012. Diverse Words, Shared Meanings: Statistical Machine Translation for Paraphrase, Grounding, and Intent. In Proceedings of Australasian Language Technology Association Workshop, pages 3−3.
Page 1 and 2: Australasian Language Technology As
Page 3 and 4: ALTA 2012 Workshop Committees Works
Page 5 and 6: ALTA 2012 Programme The proceedings
Page 7 and 8: Contents Invited talks 1 Using a la
Page 9: Invited talks 1
Page 13 and 14: A Citation Centric Annotation Schem
Page 15 and 16: The structure of this paper is as f
Page 17 and 18: understanding of sentences surround
Page 19 and 20: henceforth referred to as Annotator
Page 21 and 22: References Angrosh, M A, Cranefield
Page 23 and 24: Semantic Judgement of Medical Conce
Page 25 and 26: 2012). The TE model provides a form
Page 27 and 28: Correlation coefficient with human
Page 29 and 30: Pair # Concept 1 Doc. Freq. Concept
Page 31 and 32: Active Learning and the Irish Treeb
Page 33 and 34: 2.2 Sources of annotator disagreeme
Page 35 and 36: complement (csubj) 2 . See Figure 4
Page 37 and 38: top Y trees from this ordered set a
Page 39 and 40: References Ron Artstein and Massimo
Page 41 and 42: Unsupervised Estimation of Word Usa
Page 43 and 44: not lend itself to determining unsu
Page 45 and 46: T 0: think, want, thing, look, tell
Page 47 and 48: correlation is much stronger than t
Page 49 and 50: References Eneko Agirre and Philip
Page 51 and 52: eaders with sentences with a negati
Page 53 and 54: Question Which word is closest in m
Page 55 and 56: Acceptability and negativity: conce
Page 57 and 58: MORE NEGATIVE / LESS NEGATIVE MR Δ
Page 59 and 60: Julie Elizabeth Weeds. 2003. Measur
Page 61 and 62:
cation produced by a supervised mac
Page 63 and 64:
understood to carry a lower weight
Page 65 and 66:
Figure 1: Average classification ac
Page 67 and 68:
0.8 0.75 0.7 AuthorA4 AuthorB4 Auth
Page 69 and 70:
Segmentation and Translation of Jap
Page 71 and 72:
tomatosōsu “tomato sauce”, rev
Page 73 and 74:
“markup” and “Mach”, so the
Page 75 and 76:
MWE Segmentation Possible Translati
Page 77 and 78:
English Term Pairs from Search Engi
Page 79 and 80:
techniques used are, of necessity,
Page 81 and 82:
ation metrics robust, if they are v
Page 83 and 84:
Figure 2: Methodologies of human as
Page 85 and 86:
an optimal method? Machine translat
Page 87 and 88:
Towards Two-step Multi-document Sum
Page 89 and 90:
archical and non-hierarchical relat
Page 91 and 92:
We use the MetaMap 7 tool to automa
Page 93 and 94:
T T & C CC z -1.5 -1.27 -1.33 p-val
Page 95 and 96:
References Alan R. Aronson. 2001. E
Page 97 and 98:
techniques and results are shown. F
Page 99 and 100:
Rock Hip-Hop Electronic Punk Metal
Page 101 and 102:
Rank Frequency Frequency (filtered)
Page 103 and 104:
Ex. 1 Ex. 2 Ex. 3 Score Song Title
Page 105 and 106:
Free-text input vs menu selection:
Page 107 and 108:
consistent with the much earlier ed
Page 109 and 110:
3.2 Dialogue system architecture Th
Page 111 and 112:
Control Free-text Menu-based n=119
Page 113 and 114:
tions in analyzing algebra example
Page 115 and 116:
langid.py for better language model
Page 117 and 118:
documents of MIME type text/html wi
Page 119 and 120:
Acknowledgments NICTA is funded by
Page 121 and 122:
LaBB-CAT: an Annotation Store Rober
Page 123 and 124:
elationship between the coincidence
Page 125 and 126:
Communication 33 (special issue on
Page 127 and 128:
matically extracting location infor
Page 129 and 130:
Classifier DBp:1R Geo:1R D+G:1R DBp
Page 131 and 132:
ALTA Shared Task papers 123
Page 133 and 134:
All Struct. Unstruct. Total - Abstr
Page 135 and 136:
System Population Intervention Back
Page 137 and 138:
tence positions (absolute and relat
Page 139 and 140:
sitional attributes, and sequential
Page 141 and 142:
ordering labels, All includes BOW,
Page 143 and 144:
3 Software Used All experimentation
Page 145 and 146:
Combination Output Public Private C
Page 147 and 148:
Experiments with Clustering-based F
Page 149 and 150:
3 Results For the initial experimen
show all

Full proceedings volume - Australasian Language Technology ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?