 |
|
WinterFest CoursesTopic areas: [language] [speech] [sonics] Jump to:
Jane Davidson - Musical Performance [sonics]Course DescriptionGenerating a musical performance involves complex motor programming and cognitive and perceptual operations. Studies of musical performance have focused on: the physical skill requirements; the generation of the music including the production of its expressive features; the aspects involved in coordinating ensemble performance; the social and communicative elements involved in inter-performer and performer-audience communication. In this Winter School, the session will delineate the field for students, providing an overview of the different types of research undertaken. Students will be given examples of new research in the area, exploring a range of methodologies and theoretical perspectives.About JaneJane Davidson is the Callaway-Tunley Chair of Music at the University of Western Australia. She has written more than 100 scholarly publications and secured a range of research grants. In the area of musical development, her major output is based on two longitudinal research studies: one focused on children showing exceptional skills, and the second traced how children develop musical lives and identities, whether or not they persist with instrumental learning. Pioneering work on expressive body movement and social interaction in performance include topics ranging from the solo classical pianist and the chamber orchestra to Robbie Williams and Annie Lennox. Music and health research has examined music therapy interventions with multiple sclerosis patients and a series of investigation into the health benefits of singing. Jane is currently investigating singing interventions for older people, especially those facing social isolation. Two edited volumes -The Music Practitioner (Ashgate, 2004) and La pùrpura de la rosa: the staging of an opera, bringing the first Latin-American opera to life (DMLS, 2007) -reveal Jane's interest in the social psychology of operatic rehearsal and production.Download resources (slides etc): 1 2 3 4 5 6 7 Karen Mattock - The Littlest Language Learners: Spotlight on Infant Speech Perception [speech]Course DescriptionAnyone who has tried to learn a second language as an adult (or develop a speech recognition system) can attest to the challenges of speech perception. However, decades of research has revealed that young infants have remarkable speech perception skills. So, how do infants tackle the speech signal in the absence of a fully-developed knowledge of language? This course explores the development of speech perception from the prenatal and newborn periods through to the infants' second birthday, with reference to seminal studies, influential theories, and general perceptual learning mechanisms. The course will cover infants' perception of phonetic and prosodic units of speech, segmentation of continuous speech input, and sensitivity to phonotactic structure. We will focus on infants' initial biases for speech, and how infants learn about speech sounds and discover regularities in the input via exposure and experience with their native language. We will also touch on links between speech perception in the first year of life, and language acquisition milestones in the second year of life. The course will highlight recent directions in infant speech perception research, including bilingualism and hearing impairment, and the contribution of methodological advances in electrophysiology and brain-imaging to our knowledge.About KarenKaren Mattock is a Research Council U.K Fellow at the Department of Psychology and Centre for Human Development & Learning at Lancaster University, U.K. Karen obtained her undergraduate degree in Psychology from the University of Wollongong and her PhD in Psychology from MARCS, University of Western Sydney. Prior to joining Lancaster in 2007 she was a postdoctoral fellow at the Centre for Research in Language Mind and Brain and School of Communication Sciences and Disorders, McGill University, Canada. Karen investigates the development of speech (and auditory) perception, babbling, and word learning in infancy, and how these are influenced by exposure and experience with language(s) in the first years of life. Her primary contribution is in the area of infants' lexical tone perception.Download resources (slides etc): 1 Jon Drummond - Software for Music Processing [sonics]Course DescriptionThis workshop will present an overview of music processing applications with specific focus on creating custom applications in patching environments such as Pure Data (Pd) and Max/MSP. These high-level programming applications support the creation of customised music processing applications and rapid prototyping through the use of intuitive graphic interfaces, object oriented design, extensive code libraries and online communities. The workshop will cover methods for MIDI processing, sound processing, video processing, Open Sound Control (OSC), digital sound synthesis, spatial audio, pitch recognition, processing and analysing data, approaches to system design and creating custom applications for use in experiments. Example patches will be provided and participants in the workshop are encouraged to bring their laptops to work with practical examples.About JonJon Drummond is a researcher at MARCS Auditory Laboratories, freelance programmer and composer. He has a Bachelor of Music degree from the University of Sydney, a Masters of Science (software engineering) from Macquarie University and received his doctorate from the University of Western Sydney. His research interests include interactive systems, haptics, spatial audio, digital musical interfaces and the intersection of improvisation and composition. As a programmer he has collaborated widely with numerous musicians and artists including Lisa Anderson, Ros Bandt, Roger Dean, Ros Dunlop, Ros Gibson, George Gittoes, Nigel Helyer, George Khut, Kate Richards and Sarah Waterson. As an active computer musician and composer his music has been performed widely both in Australia and internationally.Slides coming soon Roger Dean - Time Series Analysis in Cognitive Science [sonics]Course DescriptionAs William James pointed out, as soon as a perception is expressed it is no longer the original perception. This should be taken together with the obvious fact that we face continuously changing temporal streams of information in everyday life as well as in communication and artistic expression. Consequently it is clear that analyses of continuous perceptions and their cognition is an important complement to the aggregated retrospective judgement approach that most experimental psychology involves. Continuous temporal series of events commonly have features (notably autocorrelation) which abnegate most routine statistical techniques, and thus a specialised field of 'time series analysis' has been developed over the last fifty years or so. I will illustrate its basis, its application to psychophysiology (the case of skin conductance responses, contrasted with common approaches to neuroimaging), and its application in assessing real-time perceptual responses to musical and other sonic streams concerned with loudness, structural change, and affective properties. I will point out the potential for much more extensive use of the powerful battery of available time series analysis techniques throughout cognitive science.About RogerRoger Dean has been Research Professor of Sonic Communication at MARCS since mid-2007. He has published around 300 primary journal articles in biochemistry, music and cognition, 16 books, and has an ISI Hirsch Index of 53. After an academic research career in the UK, culminating as a Professor at Brunel, he became the foundation director of the Heart Research Institute in Sydney. Before joing MARCS, and returning to full time research, he was then the Vice-Chancellor of the University of Canberra. He directs the contemporary music and intermedia creative ensemble austraLYSIS and has made many recordings internationally.Slides coming soon Emery Schubert - Emotion and Music [sonics]Course DescriptionOne of the most powerful reasons that people are attracted to music is because of its ability to evoke and express emotions. This workshop examines emotion in music from the introspection of philosophers and musicologists through to modern scientific methods. Questions discussed include: How can emotion be measured? Are the emotions in question subjective, or can they be agreed upon? What is the relationship between musical preference and emotion? Is there a difference between the way that music makes you feel versus what it appears to be trying to convey? Do we have to have another nature/nurture debate? Can a composer or improviser produce music in such a way as to evoke a particular emotion? Why do some harmonies make us cry? Why is it that we can be attracted to music that evokes negative emotions? By the end of the workshop, students will have a basic understanding of the key issues of emotion in music and its research, and be familiar with some important figures in the area.About EmeryEmery Schubert is an Associate Professor at the School of English, Media and Performing Arts, and co-leader of the Empirical Musicology Group, at the University of New South Wales, Sydney, Australia. He is a founding member and past President of the Australian Music and Psychology Society (AMPS), and is on the Editorial Board for Empirical Musicology Review. Research interests include the scientific study of emotion, preference and aesthetics in music. Among over 100 publications, he has contributed to the Oxford University Press volumes 'Emotion in Music: Theory and Research' (2001), and 'Handbook of Emotion in Music' (2010).Download resources (slides etc): 1 Andy Butcher - The Phonetics of Australian Indigenous Languages [speech]Course DescriptionAcoustically the sound systems of Australian Aboriginal languages are strikingly different from the majority of the world's languages. They are lacking both in contrasts which depend on low frequency acoustic cues (high vowels, voiced obstruents) and in contrasts which depend on cues at the high frequency end of the spectrum (fricatives, aspirated stops). However, they typically have five or six places of articulation. Chronic otitis media with effusion (OME) leaves up to 80% of Aboriginal children with a significant conductive hearing loss (i.e. greater than 25 dB). This affects both the low frequency end of the scale (under 500 Hz) and the upper end of the scale (above 4000 Hz). Aboriginal languages typically lack any sounds which are made in these areas, but are rich in 'place-of-articulation' contrasts, which depend on rapid spectral changes in the middle of the frequency range. Furthermore it seems that speakers use a number of strategies to maximise the clarity of these contrasts. Thus it appears that Aboriginal languages favour sounds whose characteristics exploit precisely that area of hearing ability which is most likely to remain intact in OME.About AndyAndy Butcher is Professor of Communication Disorders at Flinders University, Adelaide. He has degrees in linguistics and phonetics from the Universities of Edinburgh and London and a PhD in phonetics from the University of Kiel, Germany. He worked for many years in the Department of Linguistic Science at the University of Reading, UK and spent 3 years as a research fellow in Linguistics at the ANU before moving to Adelaide in 1993. Andy's main areas of research involve the instrumental measurement of articulatory parameters such as tongue-palate contact, oral and nasal airflow and pressure, vocal fold activity and the acoustic analysis of voice and speech. His particular interest is in the phonetics of Australian aboriginal languages, which he has been researching for over 20 years. His current projects (with Janet Fletcher and Marija Tabain) are looking at consonant articulation in adult speakers of these languages and at the relationship between speech production, speech perception and hearing impairment in Aboriginal children.Download resources (slides etc): 1 2 Myfany Turpin - Words and Music in Australian Aboriginal Song Styles [sonics]Course DescriptionSinging is an activity that happens in every society. But the ways that texts and music align with each other and the ways meanings are conveyed, and how the activity fits into larger contexts vary considerably between societies. In this course we explore these features of traditional Aboriginal songs, which usually form part of a larger event such as a ceremony, storytelling or game. We consider how words and syllables align with musical rhythm, and how the resulting rhythmic-text is then set to a pitch sequence in various ways. The independence of rhythm and pitch raises the question of whether these might be two separate areas of music cognition. We then explore how songs are able to convey multiple meanings and the types of inferences hearers must make to receive a message. By the end of the course you will develop skills to analyse unfamiliar music and explore how meaning is conveyed in poetic forms.About MyfMyfany Turpin is a Research Fellow at Griffith University on Sustainable Futures for Music Cultures, an international project that aims to identify ways to maintain music diversity. She has degrees in linguistics and music from Melbourne University and received her doctorate from Sydney University. She held a Hans Rausing (SOAS) fellowship in Linguistics at the University of Queensland and takes up an ARC fellowship there later this year. Her research interests are in the relationship between language and music, Aboriginal songs and the Arandic languages of Central Australia, where she has worked for many years on language and music maintenance projects. She has produced a learner's guide and dictionary of Kaytetye, as well as articles on Aboriginal music, lexicography and semantics. She is currently also involved in an ARC project that aims to produce text from AFL data in the Aboriginal language Arrernte.Download resources (slides etc): 1 2 David Grayden - Introduction to Speech Processing [speech]Course DescriptionThe aim of this course is to provide an understanding of the basic techniques used in the major applications of speech processing. The topics that we'll cover are:
About DavidDavid is a senior lecturer in the Department of Electrical & Electronic Engineering at the University of Melbourne. He is Discipline Coordinator for the Biomedical Engineering teaching programme and is one of the leaders of the Neuro-Engineering research group. The focus of his teaching and research is in understanding how the brain processes information and how best to present information to the brain using medical bionics, such as the bionic ear and bionic eye.Download resources (slides etc): 1 Richard Sproat - Writing Systems and Decipherment [language]Course DescriptionNearly all of the core data that computational linguists deal with is in the form of text, which is to say that it consists of language data written (usually) in the standard writing system for the language in question. Yet surprisingly little is generally understood about how writing systems work and what the criteria are for calling something a writing system. This tutorial will attempt to shed some light on these issues by surveying existing writing systems, methods for deciphering unknown writing systems, and methods that have been applied to try to determine if an unknown system is in fact writing. The tutorial will thus be divided into three parts. In the first part I discuss the history of writing and introduce a wide variety of writing systems, explaining their structure and how they encode language. In the second, I will discuss the problem of decipherment and how computational methods might be brought to bear on the problem of unlocking the mysteries of as yet undeciphered ancient scripts. I start with a review of three famous cases of decipherment. I then discuss how techniques that have been used in speech recognition and machine translation might be applied to the problem of decipherment. I end with a survey of the as-yet undeciphered ancient scripts and give some sense of the prospects of deciphering them given currently available data. In the third part I turn to the question of whether statistical methods can be used to tell if an uninterpretable ancient symbol system is writing or not. I will focus on two recent strands of work, one involving the Indus Valley symbols, and one involving Pictish symbols. In both cases the authors of the work have claimed that statistical methods can be used to establish that the systems in question were probably writing. I will explain the techniques they use, and argue that these are largely an abuse of the statistical methods employed.About RichardRICHARD SPROAT received his Ph.D. in Linguistics from the Massachusetts Institute of Technology in 1985. Since then he has worked at AT&T Bell Labs, at Lucent's Bell Labs and at AT&T Labs -- Research, before joining the faculty of the University of Illinois, and subsequently the Oregon Health & Science University. Sproat has worked in numerous areas relating to language and computational linguistics, including syntax, morphology, computational morphology, articulatory and acoustic phonetics, text processing, text-to-speech synthesis, writing systems, and text-to-scene conversion.Download resources (slides etc): 1 Jane Simpson - The Syntax of Australian Indigenous Languages [language]Course DescriptionThere are around 300 Australian Indigenous languages. While they show evidence of common ancestry, they cover a wide range of syntactic types, including highly polysynthetic head-marking and case-suffixing languages, languages with free word order and verb-initial languages. Work on Australian languages has strongly influenced theories of syntax, from initial attempts to deal with free word order to recent recognition of its role in information structure. A challenge is how to represent the information provided in complex words. Tense/aspect information may be split between verbs, auxiliaries and nominals; agreement allows widespread zero anaphora; modifiers are often separated from heads; definiteness/specificity may be conveyed by interactions between the presence/absence of classifiers, agreement and noun incorporation. We discuss these in terms of test cases for parameters of syntactic difference provided by geographic distribution (some regions have many closely related languages; others have sharp genetic and typological divisions), and test cases provided by the development of creoles and mixed languages.About JaneJane Simpson is Associate Professor in the Department of Linguistics, University of Sydney. She has degrees in linguistics from MIT and ANU. Her work on Australian languages include long-term study of Kaurna, Warlpiri (using Lexical-Functional Grammar, a unification-based theory) (Warlpiri morphosyntax Kluwer 1991), Warumungu (A learner's guide to Warumungu, IAD Press 2002), and recent collaboration with Ilana Mushin on information structure, anaphora and word order in Australian languages. Her two main current projects are the ACLA project investigating children's language in Indigenous communities, including study of creoles and mixed languages (co-editor and co-CI Gillian Wigglesworth Children's language and multilingualism Continuum:2008), and a team project developing a computationally tractable grammar of Indonesian in the Pargram framework (co-CI/PIs Wayan Arka, Avery Andrews, Mary Dalrymple).Slides coming soon David Hawking - Reasonable Responses to Quixotic Queries [language]Course DescriptionThe search query 'Distance Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch to Taumatawhakatangihangakoauauotamateapokaiwhenuakitanatahu' is unambiguous and (possibly) correctly spelled. Unfortunately, a high proportion of queries received by leading search engines such as Baidu, Google and Funnelback are not lyke thus. This short course provides a behind the scenes look at the many tools and techniques which can be used to guide users to useful resources despite their poor attempts to explain what they want. Topics covered will include:
About DaveDavid Hawking (david-hawking.net) is Chief Scientist at the Funnelback search company and also a professor at the Australian National University. He has been a researcher in the search/information retrieval area since 1991 and his research at ANU and CSIRO, with colleagues of course, led to the successful Funnelback spinoff in 2005. Funnelback search technology is now in widespread use in Australia, New Zealand and the UK. David was jointly responsible for the creation and distribution of text retrieval benchmark collections now in use at over 120 research organisations worldwide. In his spare time he programs in PostScript, procrastinates over home maintenance and plays Ultimate frisbee.Slides coming soon Felicity Cox - Australian English and its Variants [speech]Course DescriptionHave you ever wondered what makes the Australian accent distinctive and how it developed and evolved into its current form? With Australian English as the focus, we will explore how and why accents change, how sociopolitical events can affect accents and how variation is an important parameter in the analysis of group membership. This workshop will explore these issues from an acoustic phonetic perspective; however, no previous knowledge of acoustics is necessary.About FelicityFelicity Cox is regarded as one of the leading experts in the acoustic phonetic analysis of the Australian English accent. Her work on variation and change provides insights into the nature of linguistic evolution and the interaction of social and linguistic factors in the construction of accent. Felicity has 25 years experience in teaching and research in phonetics and phonology both at Sydney University's Cumberland campus and at Macquarie University and has recently developed the Australian Voices website, which provide a wide range of information about Australian accent and encourages community involvement in Australian speech science research.Download resources (slides etc): 1 2 Mark Johnson - Probabilistic Models for Computational Linguistics [language]Course DescriptionThis course explains how probabilistic models are used in several areas of Computational Linguistics, including:
About MarkMark Johnson is Professor of Language Sciences in Macquarie University's Department of Computing, which he joined in 2009 after spending two decades at Brown University in the USA. Mark is known for his work on statistical models for syntactic parsing and Bayesian inference for language learning. He was President of the Association for Computational Linguistics in 2003, and has published over 70 articles on computational linguistics.Download resources (slides etc): 1 Katherine Demuth - Children's Acquisition of Early Words [speech]Course DescriptionThis course provides an introduction to children's early acquisition of words. In particular, it discusses how both frequency effects in the lexicon of the ambient language and phonological context both influence the shape of children's early productions. It then extends these findings to show how the same constraints play an important role in determining the course of morphological acquisition as well. This variability in the production of grammatical morphemes is often thought to be a problem of impoverished syntax. However, the approach taken here shows that various prosodic/phonological factors can best capture this variability. Insights are drawn from both corpus and experimental studies from a wide range of languages. Basic concepts in phonology and morphosyntax will be illustrated. No previous knowledge of linguistics is required.About KatherineKatherine Demuth is Professor of Language Sciences in Macquarie University's Department of Linguistics, which she joined in 2009 after spending two decades at Brown University in the USA. Katherine is known for her crosslinguistic work on the factors influencing the course of child language acquisition, especially in the areas of phonetics, phonology, morphology and syntax. This has involved both corpus analysis and experimental studies of child-directed and child speech.Download resources (slides etc): 1 2 3 Michael White - Statistical Natural Language Generation [language]Course DescriptionNatural Language Generation (NLG) is the sub-field of Natural Language Processing (NLP) concerned with enabling computers to convey information to people through spoken or written language. Traditionally, symbolic techniques have predominated in NLG, though recent years have seen increasing use of statistical and machine learning methods. As in other areas of NLP, these empirical methods hold out the promise of more robust and flexible systems that require less knowledge engineering effort to build. This course provides an overview of the use of statistical methods throughout the conventional NLG pipeline, covering their use in content selection and ordering, sentence planning, surface realization and prosody prediction. It also examines the perennially thorny issue of evaluation in NLG, given that there is almost always more than one right answer, but some are certainly better than others.About MikeMichael White is an Assistant Professor in the Department of Linguistics at The Ohio State University. After obtaining his Ph.D. in computer science from the University of Pennsylvania, he worked for many years at CoGenTex, Inc., a small company dedicated to developing commercial natural language generation software, where he also led research collaborations with Cornell University and the University of Pennsylvania. Following a three-year stint as a Research Fellow at the University of Edinburgh working on dialogue systems, he joined the linguistics faculty at Ohio State, where he has continued to pursue research on NLG, dialogue systems, and the connection between NLG and speech synthesis.Download resources (slides etc): 1 2 Nenagh Kemp - Spelling Development and Mastery: Early, Late, or Never? [language]Course DescriptionWhy do "bees" and "buzz" have different endings? Isn't it "weird" that we're taught "i" before "e", except after "c"? Why does the new word "dack" look okay, but "ckad" look wrong? The English spelling system is largely alphabetic, but it also has many other regularities. Beginning spellers need to learn much more about written language than 26 letter-to-sound correspondences, and even as adults, we're all sometimes challenged by its quirks. The often-neglected "little brother" of reading, spelling research has begun to show that in a variety of languages, children often know more, and adults less, than we might expect. I will present some examples of the kinds of experiments you can do to determine the type of strategies that individuals of all ages use to write words. You may even come away with some strategies to improve your own spelling.About NenaghNenagh Kemp is a lecturer in Psychology at the University of Tasmania. Her doctorate, from the University of Oxford, was on children's spelling development, and she held post-doctoral positions at the Universities of Manchester and British Columbia on early language development, especially on the acquisition of words and grammar. Nenagh's current research interests focus on children's and adults' understanding of word structure, and how this is manifested in their spelling - as well as whether improving such understanding can help to improve spelling scores. She also conducts research on the use of text-message-style writing, and its links to more conventional literacy skills.Download resources (slides etc): 1 Karen Croot - Psychological Models of Speech Production [speech]Course DescriptionThis tutorial will provide an introduction to four influential models of word production and speech motor control: the model of lexical access developed by Levelt, Roelofs & Meyer, 1999, the interactive activation models developed by Dell and colleagues, the DIVA (Directions Into Velocities of Articulators) model of Guenther, Perkell and colleagues, and the Coupled Oscillator model developed by Goldstein, Pouplier and others within the articulatory phonology framework. The tutorial will outline the types of data from which these models are derived, and discuss some of their applications and limitations.About KarenComing soon. Slides coming soon Paul McCormack - Psycholinguistic Perspectives on Developmental Speech Impairment [speech]Course DescriptionChildren with developmental speech impairment are the largest group of children with a communication disorder. But what is the nature of this impairment? Linguistic and phonetic frameworks have enabled increasingly accurate descriptions of the surface speech characteristics of these children but have limited capacity on their own to contribute to our understanding of the nature of the impairment and whether there are, in fact, subtypes. The existence or not of subtypes is not a trivial matter. It is not only of theoretical importance but has significant consequences for the efficacy and efficiency of approaches to intervention. Psycholinguistic models of speech development and impairment provide a perspective that draws on other domains of knowledge about human functioning besides linguistics and phonetics to explain speech processing in children and explore differences in impairment that may arise from interference at different levels of processing. Using a psycholinguistic perspective, this workshop will explore a range of experimental and clinical evidence for and against the existence of subtypes of developmental speech impairment and their possible loci of explanation. The presenter will also demonstrate the application of a psycholinguistic perspective to a series of experiments exploring the nature of the developmental prosodic disorder found in adolescents and young adults with Aspergers Syndrome.About PaulPaul McCormack is an associate professor in the speech pathology & audiology group at Flinders University. He teaches in the areas of speech development & impairment and linguistics & phonetics. His current research is in the areas of prosody in Aspergers Syndrome, speech development in Singapore English, and the development of connected speech processes in preschool children.Slides coming soon |