개회사 안녕하십니까? 국립국어원장소강춘입니다. 귀한시간을내어참석해주신모든분들께감사의말씀을올림니다. 풍요로운계절에이자리에참석하신분들을뵙게되어반갑습니다. 국립국어원은 2010년부터 2년에한번씩세계의언어정책을국내에소개하고관련기관들과교류와협력관계를다지는국제학술대회를마련해왔습니

Similar documents
Page 2 of 5 아니다 means to not be, and is therefore the opposite of 이다. While English simply turns words like to be or to exist negative by adding not,

11¹Ú´ö±Ô

Page 2 of 6 Here are the rules for conjugating Whether (or not) and If when using a Descriptive Verb. The only difference here from Action Verbs is wh

본문01

step 1-1


장양수

歯1.PDF

Output file

<BFA9BAD02DB0A1BBF3B1A4B0ED28C0CCBCF6B9FC2920B3BBC1F62E706466>

27 2, * ** 3, 3,. B ,.,,,. 3,.,,,,..,. :,, : 2009/09/03 : 2009/09/21 : 2009/09/30 * ICAD (Institute for Children Ability

歯kjmh2004v13n1.PDF

<30322D28C6AF29C0CCB1E2B4EB35362D312E687770>

pdf 16..

<32382DC3BBB0A2C0E5BED6C0DA2E687770>

Journal of Educational Innovation Research 2018, Vol. 28, No. 4, pp DOI: * A Research Trend

Journal of Educational Innovation Research 2019, Vol. 29, No. 1, pp DOI: (LiD) - - * Way to

大学４年生の正社員内定要因に関する実証分析

- 2 -

Journal of Educational Innovation Research 2019, Vol. 29, No. 1, pp DOI: * Suggestions of Ways

<B1E2C8B9BEC828BFCFBCBAC1F7C0FC29322E687770>

#Ȳ¿ë¼®

06_ÀÌÀçÈÆ¿Ü0926

04-다시_고속철도61~80p


¹Ìµå¹Ì3Â÷Àμâ

Journal of Educational Innovation Research 2017, Vol. 27, No. 2, pp DOI: : Researc


하나님의 선한 손의 도우심 이세상에서 가장 큰 축복은 하나님이 나와 함께 하시는 것입니다. 그 이 유는 하나님이 모든 축복의 근원이시기 때문입니다. 에스라서에 보면 하나님의 선한 손의 도우심이 함께 했던 사람의 이야기 가 나와 있는데 에스라 7장은 거듭해서 그 비결을

Journal of Educational Innovation Research 2018, Vol. 28, No. 1, pp DOI: * A Analysis of

IKC43_06.hwp

저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할

항공우주뉴스레터-제13호-컬러3

Journal of Educational Innovation Research 2016, Vol. 26, No. 2, pp DOI: * Experiences of Af

I&IRC5 TG_08권

歯3이화진

Stage 2 First Phonics

07_À±¿ø±æ3ÀüºÎ¼öÁ¤

<B7CEC4C3B8AEC6BCC0CEB9AEC7D B3E23130BFF9292E687770>

한국성인에서초기황반변성질환과 연관된위험요인연구

<BCF6BDC D31385FB0EDBCD3B5B5B7CEC8DEB0D4C5B8BFEEB5B5C0D4B1B8BBF3BFACB1B85FB1C7BFB5C0CE2E687770>

272 石 堂 論 叢 49집 기꾼이 많이 확인된 결과라 할 수 있다. 그리고 이야기의 유형이 가족 담, 도깨비담, 동물담, 지명유래담 등으로 한정되어 있음도 확인하였 다. 전국적인 광포성을 보이는 이인담이나 저승담, 지혜담 등이 많이 조사되지 않은 점도 특징이다. 아울

<B3EDB9AEC1FD5F3235C1FD2E687770>

Journal of Educational Innovation Research 2018, Vol. 28, No. 1, pp DOI: A study on Characte

Journal of Educational Innovation Research 2018, Vol. 28, No. 3, pp DOI: NCS : * A Study on


182 동북아역사논총 42호 금융정책이 조선에 어떤 영향을 미쳤는지를 살펴보고자 한다. 일제 대외금융 정책의 기본원칙은 각 식민지와 점령지마다 별도의 발권은행을 수립하여 일본 은행권이 아닌 각 지역 통화를 발행케 한 점에 있다. 이들 통화는 일본은행권 과 等 價 로 연

Microsoft PowerPoint - 27.pptx

2011´ëÇпø2µµ 24p_0628

퇴좈저널36호-4차-T.ps, page Preflight (2)

대한한의학원전학회지26권4호-교정본(1125).hwp


중국 상장회사의 경영지배구조에 관한 연구

2009년 국제법평론회 동계학술대회 일정

... 수시연구 국가물류비산정및추이분석 Korean Macroeconomic Logistics Costs in 권혁구ㆍ서상범...

<313020C1A4BFECBAC034332E687770>

216 동북아역사논총 41호 인과 경계공간은 설 자리를 잃고 배제되고 말았다. 본고에서는 근세 대마도에 대한 한국과 일본의 인식을 주로 영토와 경계인 식을 중심으로 고찰하고자 한다. 이 시기 대마도에 대한 한일 양국의 인식을 살펴볼 때는 근대 국민국가적 관점에서 탈피할

<C1DF3320BCF6BEF7B0E8C8B9BCAD2E687770>

2. 박주민.hwp

_KF_Bulletin webcopy

Hi-MO 애프터케어 시스템 편 5. 오비맥주 카스 카스 후레쉬 테이블 맥주는 천연식품이다 편 처음 스타일 그대로, 부탁 케어~ Hi-MO 애프터케어 시스템 지속적인 모발 관리로 끝까지 스타일이 유지되도록 독보적이다! 근데 그거 아세요? 맥주도 인공첨가물이

<B3EDB9AEC1FD5F3235C1FD2E687770>

300 구보학보 12집. 1),,.,,, TV,,.,,,,,,..,...,....,... (recall). 2) 1) 양웅, 김충현, 김태원, 광고표현 수사법에 따른 이해와 선호 효과: 브랜드 인지도와 의미고정의 영향을 중심으로, 광고학연구 18권 2호, 2007 여름

원고스타일 정의

1. 서론 1-1 연구 배경과 목적 1-2 연구 방법과 범위 2. 클라우드 게임 서비스 2-1 클라우드 게임 서비스의 정의 2-2 클라우드 게임 서비스의 특징 2-3 클라우드 게임 서비스의 시장 현황 2-4 클라우드 게임 서비스 사례 연구 2-5 클라우드 게임 서비스에

DBPIA-NURIMEDIA

6자료집최종(6.8))

대한한의학원전학회지24권6호-전체최종.hwp

2 min 응용 말하기 01 I set my alarm for It goes off. 03 It doesn t go off. 04 I sleep in. 05 I make my bed. 06 I brush my teeth. 07 I take a shower.

발간사 반구대 암각화는 고래잡이 배와 어부, 사냥하는 광경, 다양한 수륙동물 등 약 300여점의 그림이 바위면에 새겨져 있는 세계적 암각화입니다. 오랜 기간 새겨진 그림들 가운데 고래를 잡는 배와 어부모습은 전 세계적으로 유례를 찾기 힘들 정도로 그 중요성과 가치가 큽

민속지_이건욱T 최종

<BFACBCBCC0C7BBE7C7D E687770>

<BABBB9AE2E687770>

2014 HSC Korean Continuers

歯kjmh2004v13n1.PDF

2 佛敎學報 第 48 輯 서도 이 목적을 준수하였다. 즉 석문의범 에는 승가의 일상의례 보다는 각종의 재 의식에 역점을 두었다. 재의식은 승가와 재가가 함께 호흡하는 공동의 場이므로 포 교와 대중화에 무엇보다 중요한 역할을 수행할 수 있다는 믿음을 지니고 있었다. 둘째

Output file

에너지최종-수정(색변화)

30이지은.hwp

DBPIA-NURIMEDIA

,,,,,,, ,, 2 3,,,,,,,,,,,,,,,, (2001) 2

#중등독해1-1단원(8~35)학

[ 영어영문학 ] 제 55 권 4 호 (2010) ( ) ( ) ( ) 1) Kyuchul Yoon, Ji-Yeon Oh & Sang-Cheol Ahn. Teaching English prosody through English poems with clon

Journal of Educational Innovation Research 2017, Vol. 27, No. 3, pp DOI: (NCS) Method of Con

OP_Journalism

<30362E20C6EDC1FD2DB0EDBFB5B4EBB4D420BCF6C1A42E687770>

아니라 일본 지리지, 수로지 5, 지도 6 등을 함께 검토해야 하지만 여기서는 근대기 일본이 편찬한 조선 지리지와 부속지도만으로 연구대상을 한정하 기로 한다. Ⅱ. 1876~1905년 울릉도 독도 서술의 추이 1. 울릉도 독도 호칭의 혼란과 지도상의 불일치 일본이 조선

1_2•• pdf(••••).pdf

0125_ 워크샵 발표자료_완성.key

11이정민

38이성식,안상락.hwp

Journal of Educational Innovation Research 2018, Vol. 28, No. 4, pp DOI: 3 * The Effect of H

09김정식.PDF

±èÇö¿í Ãâ·Â

16회말하기

Microsoft Word - Westpac Korean Handouts.doc

WHO 의새로운국제장애분류 (ICF) 에대한이해와기능적장애개념의필요성 ( 황수경 ) ꌙ 127 노동정책연구 제 4 권제 2 호 pp.127~148 c 한국노동연구원 WHO 의새로운국제장애분류 (ICF) 에대한이해와기능적장애개념의필요성황수경 *, (disabi

000표지


Á¶´öÈñ_0304_final.hwp

Transcription:

발표자료집 2018 국립국어원국제학술대회 보이는 언어의기록, 수어사전 Records of Visible Language: Sign Language Dictionary 2018. 11. 8. ( 목 ) 이화여자대학교이화삼성교육문화관 1 층강당 주최. 주관

개회사 안녕하십니까? 국립국어원장소강춘입니다. 귀한시간을내어참석해주신모든분들께감사의말씀을올림니다. 풍요로운계절에이자리에참석하신분들을뵙게되어반갑습니다. 국립국어원은 2010년부터 2년에한번씩세계의언어정책을국내에소개하고관련기관들과교류와협력관계를다지는국제학술대회를마련해왔습니다. 지난 2016년의 수어언어와사회적의사소통 주제에이어, 올해에는 보이는언어의기록, 수어사전 이라는주제로국제학술대회를개최하게되었습니다. 국립국어원장소강춘 이러한뜻깊은자리에소중한경험을나누어주시기위해멀리서오신예테크리스토페르센선생님, 요한나메시선생님, 트레버존스턴선생님, 토마스항케선생님께감사와환영의말씀을올립니다. 애초기조연설을해주시기로하셨던글래디스탕교수님은안타깝게도갑작스러운개인사정으로참석하지못하셨습니다. 2016년 한국수어언어법 의제정과시행으로농인의고유한언어인한국수어는우리사회에서언어로서의자격을법적으로보장받게되었으며, 27만농인과한국수어사용자의언어권과삶의질을실질적으로보장하기위한발판이마련되었습니다. 국립국어원은기존에구축된수어사전을정비하여더욱체계적인수어사전구축을계획하고있는데이번학술대회를계기로한걸음더나아간한국어수어사전의모습을전망하고농사회의발전과공감을이끌어내려고합니다. 이번학술대회에서많은논의가이루어짐으로써우리사회가소통의장벽없는사회로한발더다가설수있는계기가마련될것이라고생각합니다. 오늘이자리에서말뭉치중심의수어사전, 수어사전의구조와표제어표기, 정보기술을활용한수어사전의미래등에대하여논의하면서앞으로의한국수어사전의발전방향을모색하는귀중한자리가되기를바랍니다. 오늘이자리를빛내주신귀빈여러분, 바쁜일정에도우리학술대회를찾아주신여러청중들, 그리고발표를맡아주신세계각국의수어언어전문가여러분께다시한번깊은감사의말씀을올립니다. 모쪼록이번국제학술대회를통해수어사전의발전을이끌어내는풍성한수확을거두기를기원합니다. 2018. 11. 8. 4 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Opening Remarks Greetings. I am So Kang-Chun, Director General of the National Institute of Korean Language (NIKL). I would like to start by expressing my gratitude to all participants for taking the time to attend this conference. It is a pleasure to see everyone here in this season of harvest. The National Institute of Korean Language has hosted the international academic conference biannually since 2010 to introduce language policies from around the world and to strengthen ties for cooperation and exchange with other organizations. Following the theme of Sign Language and Social Communication in 2016, this year s conference is being held under the theme of Records of Visible Language: Sign Language Dictionary. Director General of NIKL So Kang-Chun I want to extend my heartfelt gratitude and welcome to our presenters Dr. Jette Kristoffersen, Dr. Johanna Mesch, Dr. Trevor Johnston, and Dr. Thomas Hanke, who travelled far to share their valuable insights with us. Dr. Gladys Tang was also sched uled to join us to deliver the keynote address, but unfortunately, she could not be here with us due to personal circumstances. With the enactment of the Korean Sign Language Act in 2016, Korean sign language gained legal recognition as a language and laid the foundation for 270,000 deafs and sign language users to ensure their right to language and quality of life. The National Institute of Korean Language is currently working on a more systematic revision of the Korean Sign Language Dictionary. We believe this conference will bring us one step closer to the new and improved dictionary and help us develop and gain a better understanding of the deaf society. Through active discussions, we can lay the groundwork toward a society free of communication barriers. As we touch on topics such as corpus-based dictionaries, dictionary structure and entries, and the future of sign language dictionaries with IT, I hope there will be meaningful discussions about the future direction of the Korean Sign Language Dictionary. I would like to once again express my deepest appreciation to our distinguished guests and members of the audience for attending the conference, as well as sign language experts from around the world who will be presenting today. I hope we can reap a fruitful and rich harvest from this conference for the future growth of sign language dictionaries. November 8, 2018 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 5

축사 안녕하십니까? 문화체육관광부문화예술정책실장이우성입니다. 보이는언어의기록, 수어사전 국립국어원 2018 국제학술대회의개최를축하드리며, 이행사에참석하여자리를빛내주신모든분들께감사와환영의마음을전합니다. 특히, 이번학술대회를위해홍콩, 덴마크, 스웨덴, 호주, 독일등에서찾아와주신발표자여러분께특별한감사의인사를전합니다. 문화체육관광부문화예술정책실장이우성 올해는한글을창제하신세종대왕이즉위하신지 600돌이되는해입니다. 누구나자유롭게소통할수있는세상. 이것이세종대왕이꿈꾸던세상이고, 한글은바로그꿈의산물이아닌가합니다. 600년이흐른지금한글은한국어사용자들을잇는다리로서그소임을다하고있습니다. 2016년에제정된 한국수어언어법 에는한글창제에담긴정신을한국수어사용자와도함께나누기위한뜻이담겨있습니다. 한국수어언어법 은한국수어가한국어와동등한자격을가진농인의고유한언어임을천명하고, 한국수어사용자가자유롭게소통할수있는환경을만들어나가는것을목적으로하고있기때문입니다. 이번학술대회는이러한목표를이루기위한의미있는발걸음이라고생각합니다. 지난해문체부는 제1차한국수어발전기본계획 (2018~2022) 을발표한바있습니다. 이계획에따라한국수어교육자료개발, 한국수어교육기관지원, 한국수어교원자격제도운영, 한국수어문화정보구축, 한국수어사전편찬등한국수어발전을위해다양한정책을추진하고있습니다. 특히, 새롭게편찬하고있는 한국수어사전 은표제어, 뜻풀이, 용례모두한국수어가중심이되는사전을목표로하고있습니다. 농인과농인, 농인과청인을이어주는 한국수어사전편찬 은매우중요한작업입니다. 이번학술대회가우리수어사전의편찬수준을한층높일수있는계기가되길바랍니다. 오늘학술대회를위해멀리서오신발표자선생님을비롯한청중여러분들께다시한번감사드리며, 행사준비를위해애써주신소강춘국립국어원장님을비롯한관계자여러분께도감사의마음을전합니다. 유난했던여름에여봐란듯이쌀쌀한날씨가이어지고있습니다. 모두건강에유의하시면서즐겁게한해를마무리하시기바랍니다. 고맙습니다. 2018. 11. 8. 6 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Congratulatory Remarks Greetings. I am Lee Woo-sung, Head of the Culture and Arts Policy Office at Korea s Ministry of Culture, Sports and Tourism. First of all, I would like to congratulate the opening of the National Institute of Korean Language s 2018 International Academic Conference, held under the theme of Records of Visible Language: Sign Language Dictionary. I would like to extend my warmest appreciation and welcome to everyone who came out to the conference today. In particular, I especially want to thank the presenters who travelled all the way from Hong Kong, Denmark, Sweden, Australia, and Germany. Head of Culture and Arts Policy Office Ministry of Culture, Sports and Tourism Lee Woo-sung This year marks the 600th anniversary of the enthronement of King Sejong, the inventor of the Korean alphabet Hangul. King Sejong dreamt of a world where all people can communicate freely, and Hangul is the product of that dream. Now, 600 years later, Hangul is fulfilling its role as a bridge between Korean language users. The Korean Sign Language Act, enacted in 2016, captures the determination to share the spirit of Hangul with sign language users. This Act declares Korean sign language as the native language of the deaf people with an equal status as Hangul, and aims to create an environment where Korean sign language users can communicate freely. I believe this conference is a meaningful step toward achieving this goal. Last year, Korea s Ministry of Culture, Sports and Tourism announced its first Basic Plan on Korean Sign Language Development (2018-2022). Following this plan, many policies have been carried out for the advancement of Korean sign language, including development of educational resources, support for educational institutions, operation of an instructor qualification system, building culture information, and compilation of a dictionary. In particular, the Korean Sign Language Dictionary will consist of entries, definitions, and examples that are specific for Korean sign language. Compilation of a Korean Sign Language Dictionary is an important task, bringing together deaf people with other deaf people, as well as non-deaf people. I hope this conference will be a valuable opportunity to enhance the quality of the new dictionary. Once again, I would like to thank the presenters and the members of the audience for attending today s conference. I also want to extend my gratitude to Director General So Kang-Chun of the National Institute of Korean Language and his staff for all of their hard work in preparation of this event. With the scorching summer behind us, the chilly weather is finally making an ostentatious appearance. I hope you take care of your health, and I wish you all a good rest of the year. Thank you. November 8, 2018 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 7

일정표 Program 시간 내용 09:30~10:00 ( 30) 등록 / Registration 10:00~10:30 ( 30) 개회식(개회사, 축사) / Opening Ceremony 1부 / Session 1 10:30~11:10 ( 40) 주제 1: 수어 말뭉치 중심의 수어사전 발표자: 트레버 존스턴(Trevor Johnston, 호주, Macquarie University) 11:10~11:50 ( 40) 주제 2: 수어사전의 구조 발표자: 예테 크리스토페르센(Jette H. Kristoffersen, 덴마크, University College Capital) 11:50~13:50 ( 120) 점심 식사 / Luncheon 2부 / Session 2 13:50~14:30 ( 40) 주제 3: 수어사전 편찬의 과제 발표자: 요한나 메시(Johanna Mesch, 스웨덴, Stockholm University) 14:30~15:10 ( 40) 주제 4: 대한민국 수어사전의 현황과 발전 방향 발표자: 최혜원, 이현화(한국, 국립국어원) 15:10~15:30 ( 20) 휴식 / Coffee Break 3부 / Session 3 15:30~16:10 ( 40) 주제 5: 미래의 수어사전 발표자: 토마스 항케(Thomas Hanke, 독일, Hamburg University) 16:10~16:30 ( 20) 폐회 / Closing 8 2018 국립국어원 국제 학술 대회 보이는 언어의 기록, 수어사전

2018 국립국어원 국제 학술 대회 보이는 언어의 기록, 수어사전 Records of Visible Language: Sign Language Dictionary 목차 / Contents 1부 발표 Session 1 주제 1 수어 말뭉치 중심의 수어사전 11 트레버 존스턴_ 호주, 맥쿼리 대학교 A Corpus-oriented Sign Language Dictionary 28 Trevor Johnston_ Macquarie University, Australia 주제 2 수어사전의 구조 43 예테 크리스토페르센_ 덴마크, 유시시(UCC) 대학교 Structure of a Sign Language Dictionary 52 Jette H. Kristoffersen_ University College Capital, Denmark 2부 발표 Session 2 주제 3 수어사전 편찬의 과제 61 요한나 메시_ 스웨덴, 스톡홀름 대학교 Challenges of creating a Sign Dictionary 70 Johanna Mesch_ Stockholm University, Sweden 주제 4 대한민국 수어사전의 현황과 발전 방향 79 최혜원, 이현화_ 한국, 국립국어원 Current Status and Development Direction of Korean Sign Language Dictionary 95 Hyewon Choi, Hyunhwa Lee_National Institute of Korean Language, Korea 3부 발표 Session 3 주제 5 미래의 수어사전 113 토마스 항케_ 독일, 함부르크 대학교 Sign Language Dictionaries of Tomorrow 125 Thomas Hanke_ Hamburg University, Germany 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 9

주제 1 수어말뭉치중심의수어사전 트레버존스턴호주, 맥쿼리대학교 Trevor Johnston Macquarie University, Australia

1 부발표 수어말뭉치중심의수어사전 트레버존스턴호주, 맥쿼리대학교 개요 [ 슬라이드 2] 본강의에서는말뭉치중심수어사전, 특히 오슬란사인뱅크 (Auslan Signbank) 로알려진호주수어 ( 이하 오슬란 ) 온라인사전을다룬다. 본강의는두부분으로구성된다. 먼저파트 1에서는 1985년이후로출판되어온다양한오슬란사전의발전과역사를살펴본다. 특히 1984년에만들어졌으며 1985년이후책, 시디롬, 온라인형식으로출판되어온모든오슬란사전의기초가되는수어어휘데이터베이스를논의한다. 이들모두오슬란말뭉치 (Auslan Corpus) 가만들어지기전에출판되었다. 해당데이터베이스는 오슬란어휘데이터베이스 라불린다. 또다양한사전형식의항목구조에대해서도설명한다. 파트 2에서는오슬란사인뱅크를사례로들어말뭉치중심수어사전을설명한다. 오슬란사인뱅크는 2004~2007년사이에오슬란기록보관소가만들어지고 (Johnston, 2008) 해당기록보관소가오슬란말뭉치라는이름의언어말뭉치에주석이달리기시작하면서말뭉치중심수어사전이되었다. 본사전에기록된수어는현재크게두가지측면에서말뭉치에서확인된수어와연관이있다. 첫째는식별용주석을사용한것이고, 둘째는수어사용자와그들이사용하는수어관련말뭉치로부터정보를얻은사전의항목으로수어를통합한것이다. 파트 1: 말뭉치이전수어사전 [ 슬라이드 3-39] 오슬란어휘데이터베이스 [ 슬라이드 4-5] 오슬란어휘데이터베이스는엠에스워드 (MS word) 문서형태로된일련의표항목으로 1984 년에시작 되었다. 1984 년부터 2018 년까지 34 년동안해당데이터베이스는지속적인조사에따른수정및추가를통 12 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어말뭉치중심의수어사전 트레버존스턴 해끊임없이개선되어왔다. 1984년이후로반복을거듭해온오슬란어휘데이터베이스는표제어로입력된기존오슬란어휘의저장소로간주되어왔다. 수어사용자들이서로의사소통할때만들어내곤하는다른유형의수어항목은체계적으로본데이터베이스에저장되지않는다. 비어휘적인이들수어는묘사 ( 또는 식별자 기호 ), 지시기호, 행위및몸짓으로불려왔다. 굴절형태소에해당하는일부표제어가변형되거나바뀐경우가있다. 동작주 피동작주역할이나상정보의동사를표시하는변화또는명사의복수형을표시하는변화가그예이다. 그처럼굴절된형태는사전에별도로들어가지않는다. 대부분의언어사전들과마찬가지로변형된형태는어휘보다는해당언어의문법에서논의되도록한다. 하지만경우에따라서는기호의이전굴절변경은사실상, 그리고불규칙적으로새롭고예측불가능한의미, 다시말해새로운어휘와관련이있다. 그와같은경우는사전에추가된다. 이조사는다음의몇가지기법을이용하여수행되었다. (1) 원어민수어사용자의직관과자기성찰, (2) 농인공동체에서의참여자관찰세션동안녹화한청각장애원주민수어사용자간의대화영상시청및분석, (3) 원주민수어사용자를대상으로실시한그룹 개인면담상에서개별기호의용법과의미를논의, 검사또는유도 2004년오슬란어휘데이터베이스는오슬란사인뱅크사전의온라인출판을목적으로대대적인정비를거쳤다. 이사전에는피드백메커니즘이적용되어있어서등록사용자들이누락된수어나부정확한정의를지적하거나보고할수있었다. 머지않아오슬란말뭉치가탄생했다. 그이후로오슬란사인뱅크상에이미기록되어있는수어와관련하여식별용주석의형태로말뭉치의주석을표시했다. 시간의경과에따른오슬란어휘데이터베이스 [ 슬라이드 6] 오슬란어휘데이터베이스상의데이터는 1984년이후로다양한유형의플랫폼사이를이동했다. 다양한버전의오슬란사전이기초를두고있는데이터셋인것이다. 오슬란어휘데이터베이스는 1985-1989년부터엠에스워드 (MS Word) 의초창기버전에저장된기록약 2,000건의목록으로존재했다. 1990년대초엠에스워드 (MS Word) 표는애플의하이퍼카드 (HyperCard) 스택으로이동했다 (1990-1992). 하이퍼카드스택덕분에컴퓨터상의각기호기록내에시각정보를쉽게포함시킬수있게되었다. 각기호항목 / 기록은라인드로잉 ( 라인드로잉이오슬란사전에등장한것은 1989 년이다 ) 을이용하여표시하였다. 1990년대초데이터는또다시상업적으로이용가능한데이터베이스로이동했는데처음은폭스프로 (FoxPro) 였고그다음이파일메이커프로 (FileMaker Pro) 였다. 이들새로운데이터베이스상의각기록은기호별로음운론, 의미론, 방언, 도상성등다양한정보가기호화된 100여개의필드를지니고있었다 ( 본강의파트 2에서이를다시다룰예정이다 ). 이들프로그램은일부필드에영상및그래픽정보를입력할수있었으며이를통해스프레드시트내에서각각의기호표제어를시각적으로보여주는것이가능했다. 이와비슷한신규오슬란어휘데이터베이스가 1990년대말에생성되었다. 파일메이커프로데이터베이 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 13

1 부발표 스의데이터를내보내시디롬을지원하는전용컴퓨터프로그램으로통합시키는방식이었다. 해당프로그램은멀티미디어시디롬오슬란사전을만들기위해사용되었다. 시디롬을지원하는프로그래밍언어는각기호와관련있는복수의필드, 그리고데이터베이스상의기호사이에, 다시말해동의성관계, 방언변이, 제한적용법등다양한기록및필드사이에수립된수많은교차참조를이용할수있었다. 2000년대초비디오필드를포함한파일메이커프로의전체데이터셋은시디롬과기본적으로기능이동일한온라인웹사이트로이동했다. 이웹사이트에서는파이선 (Python) 프로그래밍언어로구현된장고 (Django) 웹애플리케이션프레임워크를사용했다. 2004년에출범한이웹사이트는오슬란사인뱅크 (Auslan Signbank) 라는이름으로불렸다 (2004). 오슬란사인뱅크를지원하는스크립팅및프로그래밍은 2014년부터오픈소스가되었고깃허브 (Github) 웹사이트 (https://github.com/signbank) 상에서제공된다. 2018년현재오슬란의항목수는 7,298개에이른다 ( 이가운데대략절반은 stand-alone 핵심표제어이고나머지절반은핵심표제어가변형된형태이다 ). 사전형식 : 인쇄, 시디롬, 온라인 [ 슬라이드 7-14] 오슬란어휘데이터베이스를이용하여출판된사전은세가지형태로나뉜다. 첫째, 인쇄형식 : 1987 호주수어 ( 오슬란 ) 예비수어사전, 1989 오슬란사전 : 호주농인공동체수어사전, 1997 호주수어 : 새로운오슬란사전, 2003 오슬란생존가이드 : 초보자를위한호주수어포켓사전. 둘째, 시디롬멀티미디어형식 : 1997 호주수어 : 새로운오슬란사전, 2002 오슬란의발견 : 초보자를위한호주수어사전. 셋째, 웹기반온라인형식 : 2004, 2008, 2018 오슬란사인뱅크 (http://www.auslan.org.au). 사전의다양한버전과서로다른형식에도불구하고위에서언급한모든사전의구성원칙이바뀌지않았다는사실은주목할필요가있다. 지난수년동안있었던변화는 (1) 형식상의특징에따른기호의순차적처리와관련한사소한변경, (2) 용어검색및분류항목의데이터베이스기능개선, (3) 기존매체에서새로운매체로의전환, 즉인쇄형식에서시디롬형식, 나아가온라인형식으로의전환에따른적응및발전정도이다. 어떤형식으로든오슬란어휘데이터베이스는호주농인공동체의수어어휘를정확하게문서화하고자한다. 사전구조 [ 슬라이드 15-17] 오슬란사전은모두제한적인양방향, 2 개국어사전이었다. 단, 한가지예외가있었는데그것은바로 오슬란생존가이드 : 초보자를위한호주수어포켓사전 (Johnston & Schembri, 2003) ( 아래내용참 14 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 조 ) 이다. 진정한사전이라면한언어의단어나기호는단순히나열되는것이아니라정의가내려진다. 2개국어사전에서는 A 언어의단어나기호가 B 언어로정의내려진다. 양방향사전의경우에는 B 언어로정의내려진 A 언어단어또는기호의목록과 A 언어로정의내려진 B 언어단어또는기호의목록으로구성되어있다. 그러한이유에서수어의양방향, 2개국어사전은어휘정보에접근하는두가지방법이존재한다. 하나는수어를통한방법, 다른하나는음성언어를통한방법이다. 한언어의각단어또는기호는상대언어의단어또는기호로정의내려진다. 오슬란사전의본문은영어 (B 언어 ) 로정의내려진오슬란기호 (A 언어 ) 를열거한다. 오슬란의표제어는손모양과위치등언어내적인구성원칙에따라순서대로나열된다. 인쇄사전의표제어는라인일러스트로나타냈다. 1990년대말까지는데이터베이스의표제어또한함노시스 (HamNoSys) 를이용한전용수어전사로나타냈으나일반인이볼수있는것은아니었고인쇄사전에실리지도않았다. 함노시스를이용한전사는기호의형태를알고자할때더이상필수적인것으로간주되지않는다. 영상클립 ( 또는이전버전의경우에는라인드로잉 ) 이포함된필드를확인하는것만으로도기호의형태를쉽게알수있기때문이다. 또음운론적특성과관련하여기호항목을분류또는설명하려고할때전사가필요하지않다 ( 과거에는손모양, 위치등에사용된유니코드심볼에순서를지정하는방식으로함노시스전사를분류하는것이가능했다.). 게다가새로운데이터베이스는특별한전사스크립트에의존하지않고도기호의음운론적특징또는그밖의특징을태그로달수있는상당수의필드를수월하게수용할수있다. 함노시스전사는 1990년대말까지생성되었고이후에는디지털영상기술과데스크톱성능이전용영상클립으로각각의표제어를나타내는것이가능할만큼충분히강력해졌고비용도저렴해졌다. 결국함노시스전사는중단되었다. 그에따라오슬란어휘데이터베이스의모든항목은영상클립으로만족해야하며음운론적형태를기호화하는관련필드를지니고있다. 인쇄, 시디롬, 웹사이트형식의사전에는오슬란-영어란또는데이터를볼수있는뷰가있다. 하지만세가지형식모두영어-오슬란뷰 (B 언어-A 언어 ) 는제한적이다. 인쇄형식의경우그이유는단순하다. 라인드로잉또는함노시스를이용하여오슬란으로영어표제어를정의하기란사실상불가능했기때문이다. 따라서비교방향은단순히각단어에해당하는오슬란번역어의항목번호를사용자에게알려주는영어단어색인을통해드러냈다. 시디롬과사인뱅크웹사이트형식의경우에는음운론적특성 ( 손모양, 위치, 움직임등 ) 및관련영어키워드에대한각기호항목 ( 기호형태 ) 을태그할수있다. 사용자는시디롬또는사인뱅크웹사이트의검색창에영어단어를입력하여해당키워드와관련있는기호를찾는다. 따라서여전히양방향사전의특징을지니고있지만제한적인의미에서만그러하다. 오슬란사전은형식상의특징 ( 아래참조 ) 에기반한원칙에따라표제어를분류하고배열했다. 본사전은수어초반우세손의모양을바탕으로하여수어를손모양섹션으로분류한다. 각섹션내에는그외에도손의위치와방향, 움직임및손의개수, 그리고경우에따라서는얼굴표정등형식상의특징이담겨있다. 비 ( 非 ) 우세손의모양과기호의대칭은우세손모양이동일한기호의분류및배열을결정한다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 15

1 부발표 인쇄사전의정렬원칙 ( 알파벳순서 ) [ 슬라이드 18-25] 여기서는기호항목의 알파벳순서 근간을이루는주요원칙만을다룬다. 더자세한내용은존스턴 (Johnston, 2003 년 ) 을참조하기바란다. 손모양기준 [ 슬라이드 18-20] 손모양은하나의기호를다른기호와구분하고여러기호를섹션으로분류하기위한원칙으로삼기에이상적이다. 오슬란의손모양은어느정도자의적인순서에따라정렬된다. 초창기오슬란사전 (1987년, 1989년 ) 의경우에는어휘에서차지하는손모양의빈도를바탕으로손모양의순서를정하되가장빈도가높은손모양 ( 평평한손모양 ) 이가장먼저나온다. 호주의수어 (1997) 에서는수를셀때사용하는오슬란손모양을바탕으로더자연스럽고직관적인순서를사용했다. 순서는 0부터시작해서수를셀때사용하는손모양을따른다. 이러한손모양의미세한변형은수를세는주요손모양들가운데에서순서가정해진다. 예를들어둥글게오므린손 (Round Hand) 은오슬란손모양으로숫자 0을의미하기때문에사전상에서가장먼저등장한다. 오케이모양의손 (Okay Hand) 은숫자 0 을의미하는대체가능한손모양이기때문에두번째로온다. 그다음은숫자 1을의미하는검지를세운손 (Point Hand) 이다. 검지를세운손과비슷한갈고리모양의손 (Hook Hand) 이그뒤에나오며이와같은식으로숫자 2부터 12까지의손모양을통해순서가정해진다 ( 오슬란에서는 12까지의숫자가운데그어느것도낮은숫자에서이미나온손모양을사용하지않는다 ). 오슬란에서사용하는손모양중몇가지는숫자와아무런관련이없으며임의적으로목록과사전의마지막에온다. 예를들어 SUN 은사전속둥글게오므린손 (Round Hand) 섹션, IGNORE 는검지를세운손 (Point Hand) 섹션, FRONT 는평평한손 (Flat Hand) 섹션에나온다 [ 슬라이드 20]. 기본위치기준 [ 슬라이드 21-22] 그다음으로오슬란사전은위치에따라기호의순서를나열한다. 수어공간또는신체상의높은위치에서표시한기호 ( 예 : 머리또는얼굴위에서의기호 ) 는낮은위치에서표시한기호 ( 예 : 허리또는팔위에서의기호 ) 보다먼저온다. 그러한이유에서 THINK 는 STOMACH 보다앞서는 WHO 의앞에온다. 손의개수기준 [ 슬라이드 23] 각각의손모양섹션내에서한손만을사용하는기호가두손모두를사용하는기호보다먼저나온다. 예를들어한손만사용하는기호 CONSIDER 는두손을모두사용하는변형기호보다앞선다. 동일한또는다른손모양기준 [ 슬라이드 24-25] 각각의손모양섹션내에서두손을사용해야하는기호는한손만사용해야하는기호보다뒤에나온다. 두손을사용하는기호내에서는양손의모양이같은기호가비 ( 非 ) 우세손의모양이다른기호보다먼저나온다. 그러한이유에서 CONSIDER 는 CROSS-OUT 보다앞선다. 두손을사용하지만비 ( 非 ) 우세손의모양이다른기호의순서는단순히사전전반의손모양순서를따른다. 결국우세손의모양이해당기호의손모양섹션을결정하고, 비 ( 非 ) 우세손의모양이우세손모양이동일한두손기호들사이의순서를결정한다. 그러한이유에서 QUESTION, POTATO, DEBT 는상대적으로이같은순서를따른다. 오케이모양의손은검지를세운손보다앞서고, 16 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 검지를세운손은평평한모양의손보다앞서기때문이다. 생존가이드구조 [ 슬라이드 26] 오슬란생존가이드 : 초보자를위한호주수어포켓사전 (Johnston & Schembri, 2003) 은정확한의미에서사전이라기보다는단어목록이라할수있다. 그러한이유에서과거에비언어학자들이만든대부분의이른바수어사전들과다를바없었다. 이들은본질적으로단일어단어목록이었으며표제어는사실해당언어의알파벳문법에따라배열된공동체음성언어의단어였다. 표제어와동일한기호의일러스트는바로옆에위치해있었다 ( 생존가이드 는한정된특정독자를염두에두고출판되었다. 농인과기본적인의사소통을할수있을정도의매우기본적인어휘력을쌓고싶지만시간적여유가많지않은성인이그대상이다 ). 이같은유형의단순한어휘목록에대한특정한틈새분야가있지만오늘날언어학자들이만들고있고본강의에서다루고있는진정한수어사전과혼동해서는안된다. 시디롬구조 [ 슬라이드 27-33] 시디롬버전의 < 호주수어 >(1997) 는오슬란최초의멀티미디어사전이다. 오슬란어휘데이터베이스는시디롬제작에필요한데이터를제공했다. 시디롬의가장중요한혁신을꼽자면시디롬형식상의특징을기반으로기호를검색할수있다는점이다. 친숙한 영어단어로검색 하는방법 ( 인쇄형식사전의영어색인을참고하는방법 ) 대신 오슬란으로검색하기 옵션버튼을누르는방법으로기호검색을요청할수있다. 사용자는시디롬상에서 오슬란으로검색하기 옵션을선택한후다음의구체적인내용을묻는일련의추가질문에답해야한다. (1) 한손을사용하는기호인지, 두손을사용하는기호인지의여부, (2) 수어시작시우세손의모양, (3) 수어의위치, (4) 두손을사용하는기호의경우비 ( 非 ) 우세손의모양, (5) 2차적인위치 ( 즉, 두손이만나거나서로상호작용하는경우의비 ( 非 ) 우세손의위치 ) 그다음으로중요한시디롬의혁신은시멘틱필드 ( 음식, 신체부위, 감정, 건강등 ) 또는방언 ( 호주의주요방언지역또는주기준 ) 과같이다양한기준에따라즉시기호를선택하고분류할수있다는점이다. 세번째로중요한시디롬의혁신은미리설정된여러링크 ( 예 : 동의어, 반의어, 동형이의어 ) 를기반으로사전내의기호사이를즉시이동할수있다는점이다. 온라인사인뱅크구조 [ 슬라이드 34-39] 오슬란어휘데이터베이스는또한일반적으로는사인뱅크 (Signbank), 구체적으로는오슬란사인뱅크 (Auslan Signbank) 라고하는온라인버전오슬란사전의근간을이룬다. 사실데이터베이스자체는현재 웹사이트내에있다. 다시말해오슬란사인뱅크는오슬란어휘데이터베이스의웹사이트명에불과하다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 17

1 부발표 오슬란사인뱅크는단순히데이터베이스상의웹포털인것이다. 하지만연구원들은.csv 파일형태로사인뱅크웹사이트상의내용을다운로드받아추가데이터처리나조사를목적으로다양한데이터베이스로통합할수있다 ( 따라서데이터베이스버전은여전히파일메이커프로및기타스프레드시트형식으로남아있다.). 모든이전버전의오슬란사전과마찬가지로오슬란사인뱅크는영어정의와함께표제어로표시된언어의전통적인어휘적기호를포함하고있다. 오슬란사전에는묘사하는기호또는식별자기반의기호가다수포함되어있는데이들기호의어휘상태에는의심의여지가있다. 오슬란사인뱅크상에서사전을보여주는방식에몇가지혁신이적용되어있다. 첫째, 인터넷접속이가능한사람이라면누구든퍼블릭뷰를통해이용가능하다. 둘째, 등록사용자만이용가능한제한적뷰또는프라이빗뷰가있다. 등록뷰의경우데이터베이스에대한접근이여러단계로나뉘어져있다. 일반사용자는웹사이트에대한피드백 ( 사전상의기호오류또는누락보고등 ) 을작성하기위해등록을해야한다. 그외에도애플리케이션, 통역사, 연구자, 편집자, 출판사를위한네단계의조사용접근이가능하다. 데이터베이스내에는이와관련하여다양한접근및편집권한이존재한다. 데이터베이스열람및변경접근은다양한등급의사용자들에게웹사이트에대한서로다른단계의권한을부여하는방식으로통제한다. 추가적인페이지뷰에서는이미시디롬에서이용중인유형의정보를보여준다. 하지만시디롬과는달리오슬란사인뱅크는어느때라도업데이트가가능하며 ( 단, 접근권한을갖고있다는전제하에 ) 이때의변경사항은즉시효력을발휘하여뷰에반영된다. 또한새로운정보가이용가능해지면기록은삭제되거나추가될수있다. 이는오슬란사전이공동체의피드백또는오슬란말뭉치의점검을통해확보한새로운정보에신속하게대응할수있음을의미한다 ( 다음섹션참조 ). 오슬란사인뱅크의 2004년초기웹버전은당시의스크립트언어에따른제약때문에시디롬의기능몇가지만을지니고있었다. 2004년과 2018년사이에오슬란사인뱅크에사용된스크립트언어는전면적으로개선되었고이전까지시디롬에서만제공하던기능이웹사이트에서도이용할수있게되었다. 하지만현재이기능의대부분은특별접근권한을부여받은등록사용자만이용가능하다. 수어는인쇄버전의사전에서먼저부여받은내재적순서를지니고있지만웹사이트상에서의기호순서는인쇄사전에서만큼명확하지않다. 각각의페이지는단하나의기호항목을보여주므로사용자가사전한페이지에서다음페이지로이동하는경우에만순서가분명히드러난다. 가장최근인 2018년에이행된오슬란사인뱅크의주요혁신은 오슬란상의정의 필드가제공되는것이다. 이필드에는영상클립이삽입될수있으며이때해당표제어에대한오슬란상의정의가제공된다. 이러한유형의항목이생겨난것이불과최근의일이지만향후단일어사전에서처럼오슬란사인뱅크항목을보여주는것이가능할수있음을시사한다. 이는획기적인일이다. 어떤수어라도단일어사전은아직까지출판된적이없다. 수어용문자체계가존재하거나개발되지않고서는인쇄형태의수어단일어사전이란사실상불가능했다. 비록영상의도래에힘입어단일어수어사전이이론상가능해졌지만실행은여전히불가능했다. 그래도디지털영상기술이나오면서단일어사전제작이기술적으로가능해졌지만아직까지제작되거 18 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 나출판된사전은하나도없다. 최신 2018 년버전오슬란사인뱅크는현재단일어항목생성을가능하게해 준다. 오슬란을이용하여오슬란표제어가정의내려지는항목말이다. 이기술을이용한최초의항목이생 성되기시작했다. 파트 2: 말뭉치중심수어사전 [ 슬라이드 40-70] 말뭉치중심수어사전에대한근거 [ 슬라이드 40-44] 앞에서설명했다시피오슬란사전은참여자관찰, 면담, 도출등을수반한 전통적 언어기록과서술연구프로그램의산물이다. 이와같은모델은아직문서화되지않고서술되지않은언어에대한연구를목적으로언어학과인류학에서폭넓게사용되고있다. 일반적으로이러한유형의연구를진행하는언어공동체는대체로서술의역사가없고지리적으로동떨어져있으며소규모인전통사회에서찾아볼수있다. 하지만수어공동체는이와는다른경향을보인다. 거의모든곳에서공동체수어는음성언어 ( 및때로는문자언어 ) 를사용하고구성원대부분이청인인공동체내의소수농인들이사용하고있다. 일부국가또는지역의몇몇농인들은수세대에걸쳐청각장애를앓아온가정에서태어나지만이는비교적드문경우이다. 농인대부분의부모와형제자매는청각장애를갖고있지않다. 그러므로대부분의수어를습득하여사용하는것은음성언어를습득하여사용하는것에비해매우이례적인것으로보인다. 농인공동체의잘알려진사회언어학적특성에따르면상당수연구자들은수어공동체가다른예상보다어휘와문법양측면에서훨씬높은수준의변화를보이는듯하다고말한다. 1990년대말과 2000년대초오슬란연구자들을포함한상당수수어연구자들에게분명해진사실이있다. 그것은바로수어의어휘와문법에대하여경험적으로건전한서술을하려면원어민또는그에가까운수어사용자의자연스러운수어를바탕으로언어학적말뭉치를만들어야한다는점이다 (Johnston, 2010, 2014). 언어학적말뭉치를만드는일은기계가읽을수있도록전사, 주석, 번역을통해언어사용기록을변환하는작업이수반된다. 수어의경우, 물론이는디지털영상기록물을제작하고멀티미디어주석소프트웨어를사용하는것을의미한다. 일단수어말뭉치를만드는준비과정에서기록보관소가마련되면해당수어의기존수어사전은여기에적응해야할필요가있다. 예를들어식별되어말뭉치에주석이달린기존의모든어휘기호들또한사전에실려야한다. 그렇지않을경우해당사전은불완전한것으로여겨질것이다. 사실상사전과말뭉치간의관계는두가지방향으로존재한다. 한편으로는사전이기호의식별과주석을알려주고, 다른한편으로는말뭉치가지금까지문서화되지않은기존기호와관련하여잠재적으로새로운데이터를제공하거나이미문서화된형태의가능한변이를식별해준다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 19

1 부발표 말뭉치주석과사전내기호항목간의결부 [ 슬라이드 45-49] 식별용주석 사전과말뭉치가조화를이루기위해서는사전에입력된기호유형의형태가말뭉치내의토큰 (token) 과부합할수있도록해야한다. 이말뭉치중심데이터베이스의기호항목은영어로된키워드와고유의식별용주석 (Identifying-gloss, ID-gloss) 과관련이있다. 고유의식별용주석은사전내각각의표제어마다배정되어있다. 사전내기호유형과관련한식별용주석은말뭉치내기호유형의모든토큰에주석을다는목적으로사용된다. 하지만식별용주석을효과적, 효율적으로사용하려면어휘유형이문서화되어있는참조데이터베이스가필요하다. 이것이바로말뭉치중심수어사전의정수이다. 해당언어의모든기존어휘기호는사전상에등재되며말뭉치주석으로분할되어표시를할때고유의식별자를갖게된다. 따라서말뭉치에주석을다는동안수어사전에끊임없이참조표시를해야한다. 이를위해오슬란주석자는오슬란사인뱅크를참조한다. 주석자는특정기호의식별용주석에확인이필요할때면언제든지해당기호와관련있는영어키워드를오슬란사인뱅크검색창에입력하고사인뱅크에서제시하는일치여부를확인한다. 그후주석자는말뭉치에서주시하고있는기호의형태와가장유사한기호의식별용주석을선택한다. 현재오슬란사인뱅크는말뭉치구축에사용되는 ELAN 주석프로그램의외부에있다. 주석자는 ELAN 상에서주석을만들때데스크톱상에온라인사전이열려있도록할뿐이다. 하지만사인뱅크 NGT( 오슬란사인뱅크기반의네덜란드수어용사인뱅크. 뒤에서이를다룰계획이다 ) 에서둘은서로연결되어있다 (Crasborn & Meijer, 2012; Crasborn 등, 2012). 예를들어사인뱅크 NGT는 NGT 말뭉치의주석을다는데사용되는 ELAN 소프트웨어템플릿과연결되어있다. 주석자는 ELAN상에기호의식별자주석을입력할수있으며사인뱅크 NGT상의모든관련식별자주석을포함하는통제어휘가생겨나게되면최적의조합을선택할수있을것이다. 만약의구심이들경우주석자는통제어휘내의어떠한주석이라도클릭할수있고사인뱅크로부터의식별용주석을위한섬네일영상클립이팝업창으로나타나게된다. 음운변이 [ 슬라이드 50-51] 가장흔한형태의기호 ( 말뭉치의토큰카운트에서가장빈번하게발생하는변이의횟수로평가 ) 는어휘기호의어간또는기본형태로여겨진다. 바로중심표제어인것이다. 표제어의변이는해당항목에서드러난다. 근본적으로변이는기본기호와는다른형식상의특징이하나또는많아야둘인기본기호의한형태이다 ( 변이는보통다른손모양, 위치또는수어에서사용되는손의개수를수반한다.). 또한변이는형태기반사전의적절한위치에고유의항목기록을갖고있다. 만약말뭉치에기록된것처럼사용데이터에따라기본및변이형태의지위가뒤바뀌어야된다고한다면그대로될것이다. 예를들어사용데이터는두손을모두사용한형태의기호가한손만을사용한형태보다훨씬더빈번하다는점을보여줄수있다. 만약 20 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 한손만을사용한형태가사전상에중심표제어로올라와있을경우관계는뒤바뀌어야한다. 입증된사용 [ 슬라이드 52] 시간이지나면서, 그리고말뭉치가모든사용자와모든텍스트유형을대표하고또충분히크다면수어사전항목의대부분이말뭉치를통해입증된다는점을확인할수있다. 일반적인말뭉치에서찾아보기힘든기호는너무오래전에사용되어쓸모없어졌거나중요도가떨어지는기호정도일것이다 ( 예 : 특정분야의소규모사람들만이사용하는어휘 ). 말뭉치에서입증된사용은사전이경험적으로볼때견고하다는점을보여준다. 2012년, 4년간의주석사용이후 (Johnston 2012 참조 ) 오슬란사인뱅크의 3,500개핵심항목가운데약 75% 가오슬란말뭉치에서증명되었다. 당시오슬란말뭉치의규모 ( 약 70,000 토큰 ) 가음성언어의말뭉치에비해작다는점을고려하면이는매우높은비율이다 ( 오슬란사인뱅크의항목은 7,000여개로두배가넘지만이는모두 3,500개핵심어휘기호가변이된형태이다 ). 누락데이터 [ 슬라이드 53] 물론어떠한사전도완전하지는않기때문에지금까지기록되지않은기호가말뭉치에서식별될가능성도높다. 만약해당언어의진정한기존어휘기호로확인된다면일정기간동안공동체구성원들과의협의및말뭉치에서발견한추가토큰의확인을통해사전에추가될수있다. 이런식으로말뭉치는지금까지기록되지않은기호로어휘데이터베이스를확대할수있다. 그것이기존어휘항목의변이형태이든, 언어에새롭게추가된기호이거나 ( 신조어또는차용어 ) 단순히이전까지간과되었던기호이든말이다. 약 50개의새롭거나이전에기록되지않았던기존어휘기호가오슬란말뭉치에서식별되었고 4년간의주석사용이후오슬란사인뱅크에추가되었다. 사용데이터에근거한정의개선 [ 슬라이드 54] 애초에말뭉치에서의입증된사용을고려했을때지나치게좁은의미가기호의형태에부여되었을수있 다. 그러한경우해당기호의실제자연스러운사용을반영하여사인뱅크내의정의를개선하게된다. 이는 정의에대한변경및개선, 또는관련키워드의추가, 제거또는변경이될수있다. 일부기호의어휘지위조정 [ 슬라이드 55] 초기현장조사에근거하여일부매우상징적이거나묘사적인기호가기존어휘기호로사전에포함되었 을수있다 ( 직관, 자기성찰, 도출또는공동체피드백결과관습적이고어휘화된것으로보고된기호이다 ). 하지만말뭉치에서입증된것처럼기호를사용한다는것은기호가애초에생각했던것만큼구체적이거나 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 21

1 부발표 관습적 ( 또는 어휘적 ) 인의미를지니고있지않음을시사할수있다. 그와같은기호는어휘화측면에서경계선상의기호이다. 즉의미측면에서해당언어의특징을지니지못하고그러한이유로오슬란사전에적절하게포함되지못했을수있다 ( 물론해당언어의문법측면에서논의가필요할것이다 ). 따라서그러한항목은그지위를제대로반영할수있도록업데이트되거나아예제거되어야할수있다. 말뭉치에서사전까지? [ 슬라이드 56-57] 이상적인말뭉치구축상황에서는그누구도먼저언어에대하여기본적인어휘론적연구를수행한다음그결과로나온초기어휘를데이터베이스나사전에기록및반영하는일없이수어의기록을분할하고주석을달려고하지않을것이다. 하지만최근몇년동안대규모수어말뭉치프로젝트의일환으로몇가지수어사전프로젝트가시작되었다. 그목적은말뭉치를사용하여사전을만드는것에있다. 거의아무것도없는상황에서주석자체를바탕으로참조어휘를구축하는것은가능한일이다. 즉, 어휘의기초로말뭉치자체의입증된사례만을사용하는방법이다. 이와비슷한접근방식이 BSL 말뭉치와관련하여설명되었다 (Cormier 등, 2012). 비록이런식으로진행할수도있겠지만이접근방식은어렵고많은시간이소요되며일관성이떨어지기쉽다. 특히주석작업초기에주석자가이용할공통의참조점이없기때문이다. 연구자들은초기의데이터베이스와사전을끊임없이보완하고수정해야할상당수의다양한사용및변이형태와맞닥뜨릴가능성이높다. 이는커져가는어휘데이터베이스에도연쇄효과를일으킬수있다 (Crasborn, Meijer, 2012). 실제로일부연구자들은실증적인수어사전이참조말뭉치상에서입증된기호에만의존해야한다고생각한다. 어쩌면혹자는이를두고본강의에서언급한말뭉치중심수어사전이라기보다는말뭉치기반또는말뭉치주도수어사전이라할것이다. 하지만수어사전에기호를포함시키기위해그처럼엄격한기준을적용하기에는위험이따른다. 한언어의어떠한말뭉치라도 ( 무수히많은단어나기호와접하지않는한 ) 해당언어의알려진모든단어나기호의최소하나의토큰을포함할가능성은매우낮다. 만약매우기본적인일상어휘를말뭉치에서찾아볼수없게된다면특히문제가되며, 규모가그리크지않다는점에서이미구축되었거나구축중인수어말뭉치의경우에는그렇게될가능성이매우높다. 수어사전의사용자라면누구라도기본개념을위한기호가기록되어있지않다는사실을깨닫게될것이다. 말뭉치상에서과거에단한번도없었던일이기때문이다. 이는만족스럽지못한일이며사전의사용자는사전이불완전하다는타당한결론을내릴것이다. 만약입증된데이터를제한하고싶다면이문제를피할수있는방법한가지는관습적인또는서술적인텍스트뿐만아니라도출된텍스트와어휘까지아우르는말뭉치를자체적으로구축하는것이다. 그렇게되면기본적인어휘를도출하는기록과정에서작업을통합할수있다. 말뭉치결과를이용한사전항목강화 [ 슬라이드 58-68] 오슬란어휘데이터베이스와더불어오슬란사인뱅크는각항목의각기호에대한매우다양한정보를수 22 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 용할수있도록만들어졌다. 해당정보는수많은전용필드나셀에서찾을수있다. 여기에는다음에관한정보가담겨있다. (1) 표제어의형태, (2) 표제어의의미, (3) 표제어의사용자, (4) 표제어의검색이나분류또는온라인상표시를위한기준. 이들정보상당부분은말뭉치자체를조사해보면찾을수있다 ( 예 : 방언선호도, 연령집단, 학교기반기호, 소속종교, 빈도등 ). 이정보는끊임없이검토되며사인뱅크연구자용인터페이스상에서쉽게편집가능하다. 일부정보는일반인이즉시열람할수있고또다른일부정보는데이터베이스의프라이빗뷰에서만연구자가참고목적으로열람가능하다. 동형이의어처리 [ 슬라이드 69-70] 오슬란사인뱅크는형태기반의기호데이터베이스이기때문에만약두기호가동형이의어인경우, 다시말해오슬란상에서는서로다른의미를갖고있지만기호형태가동일한경우 ( 예 : 금속용기 와 할수있는 을의미하는영어단어 can ), 두기호모두동일한형태를참조하게된다. 오슬란사인뱅크에서의해결책은영어사전에서와마찬가지이다. 즉표제어가되풀이되고생성된두개 ( 또는그이상 ) 의항목에서동일한음운론적형태에대하여두가지 ( 또는그이상 ) 의의미로각각규정한다. 오슬란사인뱅크에서는의미 1, 의미 2 등과같은식으로번호를부여한다. 다수의수어 [ 슬라이드 71-78] 오슬란사인뱅크 & 기타사인뱅크 [ 슬라이드 72] 오슬란어휘데이터베이스구조의기본원칙은처음만들어져데이터베이스에적용된이후로어떠한수어에도적용가능했다 (Johnston, 2001). 오슬란사인뱅크 ( 및인쇄, 시디롬초기형식 ) 의표제어는형태에기반한것이었다. 이들표제어는그형식상의특징이전용전사시스템 ( 함노시스등 ) 또는각기호가지닌관련형식상의특징 ( 손모양, 위치등 ) 의존재여부를명시한일련의필드에기호화되어있다. 그결과어떠한수어의기호라도데이터베이스내에수용될수있다. 단순히형태별로서술된항목이생성되는것이다 ( 만약데이터베이스가형태별로정렬된다면새로운기호는알파벳순서에따라적절한자리에들어가게된다.). 심지어기호들사이의상관관계유형 ( 가령동의어, 반의어, 동형이의어, 의미필드항목, 방언분포등 ) 이프로그램속으로구축되어있다면이또한어떠한수어의경우에도적절하게이용가능하다. 이러한잠재력이실현되자다른수어를연구하는타언어학자들도오슬란사인뱅크의기본아키텍처를본인의수어연구에도입하게되었다. 그결과네덜란드수어 (NGT) 사인뱅크, 영국수어 (BSL) 사인뱅크, 핀란드수어 (FinSL) 사인뱅크의온라인사전이제작되었고미국수어 (ASL) 사인뱅크도작업에착수했다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 23

1 부발표 동음어에서다수의수어까지 [ 슬라이드 73-77] 오슬란사인뱅크구조의한가지중요한점에힘입어단일사인뱅크, 다시말해범용어휘데이터베이스 (Universal Lexical Database) 에다수의수어를수용하는어휘데이터베이스구축의가능성이열렸다. 그러한데이터베이스는기존오슬란사인뱅크를타사인뱅크의템플릿으로사용하는데그치지않고범용사인뱅크 (Universal Signbank) 를구축해준다. 앞에서언급한것처럼오슬란사인뱅크는형성과정의특징에따라수어형태를기록해낸다. 그결과인쇄형식으로제공된경우에는라인드로잉또는사진이시각적인순서로정렬되고, 스프레드시트의기록과셀의형태로제공된경우에는유사한영상클립들이늘어서게된다. 비록사인뱅크같은온라인사전에서는페이지당하나의항목만눈에보이지만동일한원칙에따라각항목에기호화된관련형식상의특징별로데이터베이스내에서기록을즉시분류할수있다. 그러한이유로기호는손모양 ( 인쇄형식에서처럼 ) 이나위치, 또는요청받은다른특징별로정렬할수있다. 더욱중요한점은이러한형태기반항목들이쉽게기호화될수있고음운론적변이, 방언변이, 의미관계 ( 반의어, 동의어, 동형이의어 ) 및타수어 ( 영국수어, 미국수어등 ) 와의관계를대변한다는사실이다. 오슬란사인뱅크의기호형태는이제방언뿐만아니라수어에도태그로달릴수있다. 태그는기호형태가영국수어나미국수어에도존재한다는것을의미한다. 언어에태그를달수있고더불어다른기호와동형이의어관계에있는기호에태그를달수있게되면서확장가능한데이터베이스의가능성이열렸다. 어떠한수어로부터의기호라도동일한데이터베이스에들어갈수있다. 오슬란어휘데이터베이스가처음구상되었을당시에만해도컴퓨터의데이터처리속도 ( 및비용 ), 디지털영상기록물의품질, 서버의저장공간 ( 및비용 ) 에실질적인제약이있었다. 오늘날에는사인뱅크의항목수, 각항목과관련한코드의수와유형, 표제어나정의에사용되는영상클립의길이등은더이상실행불가능하지도, 많은비용이들지도않는다. 그결과이러한유형의범용사인뱅크의가능성과사용을시험해보려는노력이시작되었다. 이는글로벌사인뱅크 (Global Signbank) 라한다. 글로벌사인뱅크? [ 슬라이드 78] 글로벌사인뱅크는실현가능한가? 지금까지심층적으로연구되어온수어어휘의두가지관찰에따르면그가능성은충분하다. 첫째, 규모이다. 지난 50여년동안출판되어온수어사전대부분은수어표제어의수가수천개를넘지못한다 (3,000~8,000개사이 ). 이는대다수음성언어의사전, 심지어는문자가존재하지않는음성언어와비교해보더라도낮은수치이다. 둘째, 상당수연구자들은수어의도상성이높다보니아주똑같거나매우유사한형태의기호가서로다른수어에서거듭나타나는상황이발생하곤한다고밝혔다. 비록도상적으로자극을받은이들기호형태가 24 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 종종서로관계없는수어사이에서똑같거나유사한의미를갖기도하지만그의미는전혀다를수도있다. 그럼에도불구하고이들수어는동일한기호형태를공유하고있다. 한가지수어를사용하는사람들이외국수어를본후에형태측면에서본인이사용하는수어의기호화매우유사해보인다고말하는경우가종종있다. 때로그들은그의미를정확하게추측할수있을것이고그러지못할때도있을것이다. 반면서로관계없는음성언어사이에서동일한단어형태가거듭나타나는경우는매우드물다. 그이유는다음과같다. 대부분의음성언어에서음운론적으로형태가올바른단어의잠재적수는음소와가능한음절수의조합을고려했을경우사실상무제한인것으로보이기때문이다. 하지만수어에서의상황은일부분도상성의영향때문에다를수있다. 가령글로벌사인뱅크가실현가능하다면사인뱅크의다국어버전이라할수있는글로벌사인뱅크는현재재구성되고있는구성상의순서로부터가장큰수혜를입을가능성이높다. 글로벌사인뱅크는심지어손모양이앞으로도가장 중요한 수어의특징으로남을가능성이높다하더라도대다수수어연구자들이받아들이는 중립적인 ( 다시말해, 특정언어의특징을지니지않는 ) 구성상의순서를갖게될것이다. 예를들어특정언어와관계없는순서는수를세는별도의체계보다는각손모양에서펼치고있는특정손가락의수에근거할수있다. 두가지이상의수어가동일하거나거의동일한기호형태를지니고있다면이들은마치오슬란사인뱅크상의동형이의어와마찬가지로글로벌사인뱅크상의동일한영상에태그가달릴것이다. 글로벌사인뱅크는수어음운목록, 도상성의정도와유형, 어휘화 문법화패턴등에대한교차언어 비교연구목적으로이용가능하다. 참고문헌 Cassidy, S., Crasborn, O., Nieminen, H., Stoop, W., Hulsbosh, M., Even, S., Komen, E.and Johnston, T. 2018. Signbank: Softrware to Support Web Based Dictionaries of Sign Language. In N. Calzolari, K. Choukri, C. Cieri, T. Declerck, S. Goggi, K. Hasida, H. Isahara, B. Maegaard, J. Mariani, H. Mazo, A. Moreno, J. Odijk, S. Piperidis and T. Tokunaga (eds.). Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), May 7-12, 2018, Miyazaki, Japan. Cormier, K., Fenlon, J., Johnston, T., Rentelis, R., Schembri, A., Rowley, K., Adam, R. and Woll, B.. 2012. From Corpus to Lexical Database to Online Dictionary: Issues in Annotation of the BSL Corpus and the Devel-opment of BSL SignBank. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen and J. Mesch (eds.). Proceedings of 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 25

1 부발표 the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Workshop as part of 8th International Conference on Language Resources and Evaluation, LREC] (pp. 7-12). Istanbul, Turkey., Istanbul, Turkey 21-27 May 2012. Crasborn, O & Meijer, A de. 2012. From corpus to lexicon: the creation of ID-glosses for the Corpus NGT. In Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon (pp. 13-18). Crasborn, O., Hulsbosch, M. and Sloetjes, H. 2012. Linking Corpus NGT annotatons to a lexical database using open source tools ELAN and LEXUS. In Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon (pp. 19-22). Istanbul, Turkey., Istanbul, Turkey 21-27 May 2012. Johnston, T. 1987. A preliminary signing dictionary of Australian Sign Language (Auslan). Adelaide: TAFE Na-tional Centre for Research and Development. Johnston, T. 1989. Auslan Dictionary: A dictionary of the sign language of the Australian deaf community. Sydney: Deafness Resources Australia. Johnston, T (ed.) 1997. Signs of Australia: A new dictionary of Auslan. Sydney: North Rocks Press. Johnston, T. 2001. The lexical database of Auslan (Australian Sign Language). Sign Language & Linguistics 4(1/2). 145-69. DOI: 10.1075/sll.4.12.11joh. Johnston, T., Thornton, D. and Napier, J. 2002. Discovering Auslan: A beginner's dictionary of Australian Sign Language. Sydney: North Rocks Press. Johnston, T & Schembri, A (eds.) 2003. The Survival Guide to Auslan: a beginner's pocket dictionary of Australi-an Sign Language. Sydney: North Rocks Press. Johnston, T. 2003. Language standardization and signed language dictionaries. Sign Language Studies 3(4). 431-68. 10.1353/sls.2003.0012. Johnston, T. 2004. Auslan Signbank (http://www.auslan.org.au). Sydney: Royal Institute for Deaf and Blind Children & Catalyst Training Systems. Johnston, T. 2008. The Auslan Archive and Corpus. In D. Nathan (ed.). The Endangered Languages Archive http://elar.soas.ac.uk/languages, London: Hans Rausing Endangered Languages Documentation Project, School of Oriental and African Studies, University of London. 26 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

말뭉치중심수어사전 트레버존스턴 Johnston, T. 2010. From archive to corpus: transcription and annotation in the creation of signed language corpora. International Journal of Corpus Linguistics 15(1). 104-29. DOI: 10.1075/ijcl.15.1.05joh. Johnston, T. 2012. Lexical Frequency in Sign Languages. Journal of Deaf Studies and Deaf Education 17(2). 163-93. Johnston, T. 2014. The reluctant oracle: using strategic annotations to add value to, and extract value from, a signed language corpus. Corpora 9(2). 155 89. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 27

Session 1 A Corpus -oriented Sign Language Dictionary Trevor Johnston, Macquarie University, Australia Outline [Slide 2] In this lecture I describe a corpus-oriented sign language dictionary, specifically the on-line internet dic-tionary of Australian Sign Language (henceforth Auslan) known as Auslan Signbank. The lecture consists of two parts. In Part 1, I give the history and evolution of the various dictionaries of Auslan that have been published since 1985. In particular I discuss the lexical database of sign forms that was first created in 1984 and which underlies all the dictionaries of Auslan published since 1985 in book, CD-ROM, and internet formats. These were all published before the Auslan Corpus was created. The database is referred to as the Auslan Lexical Database. I also explain the structure of entries in the various dictionary formats. In Part 2, I describe a corpus-oriented sign language dictionary, using Auslan Signbank as an example. Auslan Signbank became a corpus-oriented dictionary after the creation of the Auslan Archive between 2004-2007 (Johnston, 2008), and the beginning of annotation of that archive into a linguistic corpus, called the Auslan Corpus. The signs recorded in the dictionary are now related to the signs identified in the corpus in two main ways: by the use of ID-glosses and by incorporating into the entries for these signs in the dictionary of information gleaned from the corpus about signers and the signs they use. Part 1: The pre-corpus sign language dictionary [Slides 3-39] The Auslan Lexical Database [Slide 4-5] The lexical database of Auslan (or Auslan Lexical Database) began in 1984 as a series of tabular entries in a Microsoft Word document. Over the 34 years from 1984 to 2018 the database has been constantly improved with 28 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston corrections and additions as a result of on-going research. In all its iterations since 1984, the Auslan Lexical Database has been conceived as a repository of con-ventional lexical signs of Auslan, which are entered as headsigns. The database does not systematically include entries for the other possible types of signs that signers often produce when communicating with each other. These nonlexical signs have been called depicting (or classifier signs), pointing signs, en-actments and gestures. There are variations or modifications of some headsigns that represent inflectional morphology, e.g., changes that mark verbs for agent and patient roles or for aspectual information, or that mark nouns for plurality. Such inflected forms are not entered separately in the dictionary. As with most language dictionaries, these modified forms are left to be discussed in the grammar of the language, rather than in a lexicon. However, in some cases an erstwhile inflectional modification of a sign is actually and irregularly associated with a new and unpredictable meaning, i.e., a new lexical item. Such cases are added to the dictionary. The research was conducted using several techniques: (i) native signer intuitions and introspections; (ii) videotape recordings of natural signed exchanges between deaf native signers recorded during participant observation sessions in the deaf community were repeatedly viewed and analysed; and (iii) group and individual interviews of native signer informants in which the usage and meaning of individual signs were discussed, tested or elicited. The Auslan Lexical Database underwent a major overhaul for the on-line publication of Auslan Sign-bank dictionary in 2004. This dictionary included a feedback mechanism that allowed registered users to and comment and report on missing signs or inaccurate definitions. Soon after, the Auslan Corpus was created. From that point on, the annotation of the corpus was made with reference to the signs already recorded in Auslan Signbank, by way of ID-glosses. The Auslan Lexical Database over time [Slide 6] The data in the Auslan Lexical Database has migrated across several different types of platforms since 1984. It has been the dataset upon which the various editions of the Auslan dictionaries were based. From 1985-1989 the Auslan Lexical Database existed as a list of approximately 2,000 records stored in a Microsoft Word processing program table (the earliest versions of Microsoft Word). In the early 1990s the Microsoft Word tables was migrated to an Apple HyperCard stack (1990-1992). The HyperCard stack made it possible to include visual information easily within each sign record on a computer. Each sign entry/record was illustrated with a line-drawing (the line drawings appeared in the 1989 Auslan Dictionary). In the early 1990s the data was again migrated to commercially available databases: first to FoxPro, and then to FileMaker Pro. Each record in these new databases had more than 100 fields that encoded a variety of information about each sign, e.g., phonology, semantics, dialect, iconicity, etc. (I will mention some of these again in the second part of this lecture.) These programs allowed for video and graphic information to be entered in some fields, making it possible to have a visual representation of each sign headword within a spreadsheet. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 29

Session 1 A new parallel Auslan Lexical Database was created in the late 1990s by exporting the data from the FileMaker Pro database and incorporating it within into a dedicated computer program that supported the CD-ROM. The program was used to create a multi-media CD-ROM dictionary of Auslan. The programming language that supported the CD-ROM was able to exploit the multiple fields associated with each sign and the numerous cross references that had been established between signs in the database, i.e., between numerous records and fields, such as relationships of synonymy, dialectal variations, restricted usage, and so forth. In the early 2000s the entire FileMaker Pro dataset, including video fields, was migrated into an internet website which had the same basic functionality as the CD-ROM. The site used Django web applica-tion framework written in Python programming language. The site was called Auslan Signbank (2004). From 2014 the scripting and programming that supports Auslan Signbank and has become open source and house on a Github site (https://github.com/signbank). Today (2018), Auslan had at last count 7,298 entries (approximately half of these are core stand-alone headsigns, and half are common variant forms of those core headsigns). Dictionary formats: print, CD-ROM and Internet [Slides 7-14] The dictionaries published using the Auslan Lexical Database have appeared in three formats: One, print format: 1987 A preliminary signing dictionary of Australian Sign Language (Auslan); 1989 Auslan Dictionary: A dictionary of the sign language of the Australian deaf community; 1997 Signs of Australia: A new dictionary of Auslan; and 2003 The Survival Guide to Auslan: a beginner's pocket dictionary of Australian Sign Language. Two, CD-ROM multi-media format: 1997 Signs of Australia: A new dictionary of Auslan; and 2002 Discovering Auslan: A beginner's dictionary of Australian Sign Language. Three, in a web-based internet format; 2004, 2008, 2018 Auslan Signbank (http://www.auslan.org.au). It is important to note that despite the various editions and different formats, the organizing principles of all these dictionaries have remained unchanged. The changes that have happened over the years relate to (i) minor alterations in sign sequencing according to formational features, (ii) improvements to the functionality of the database in terms searching and sorting entries, and (iii) adaptations and developments occasioned by the transfer from old to new media, i.e., from printed book format to CD-ROMs through to the web-based application. In all its manifestations, the Auslan Lexical Database attempts accurately document the lexicon of the sign language of the deaf community in Australia. Dictionary Structure [Slides 15-17] The Auslan dictionaries have all been limited bidirectional, bilingual dictionaries. There has been one exception The Survival Guide to Auslan: a beginner's pocket dictionary of Australian Sign Language (Johnston 30 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston & Schembri 2003) (see below). In a true dictionary the words or signs of a language are defined, not just listed. In a bilingual diction-ary, the words or signs of Language A are defined in Language B. In a bidirectional dictionary one has a list of Language A words or signs defined in Language B, and a list of Language B words or signs defined in Language A. Thus in a bidirectional, bilingual dictionary of a sign language there should be two ways to access lexical information: one through the sign language and one through the spoken language. Each word or sign of each language is defined in the words or signs of the other language. The main part of the Auslan dictionaries lists Auslan signs (Language A) which are defined in English (Language B). The headwords of Auslan, i.e., the headsigns, are ordered according to language-internal formational principles such as handshape and location. In the print dictionaries, the headsign was repre-sented by a line illustration. In the database until the late 1990s, the headsigns were also are represented by a dedicated sign language transcription in HamNoSys, but this was not viewable by the public, nor included in the printed dictionary. Transcriptions in HamNoSys are no longer regarded as essential in order to know what the form of a sign is because one can easily see the form of a sign by looking at the field that contains the video clip (or, in earlier editions, line drawing). Nor are transcriptions required in order to sort or describe sign entries in terms of phonological features. (Previously, it was possible to sort the HamNoSys transcriptions them-selves by assigning an order to the Unicode symbols used for handshape, location, and so on.) Moreover, the new databases can easily accommodate large numbers of fields in which phonological and other features of signs can be tagged, without recourse to a special transcription script. The HamNoSys transcriptions were created up to the late 1990s, before digital video technology and desktop computing power became powerful and cheap enough to make it feasible to represent each headsign with its own dedicated videoclip. They have been discontinued. All entries in the Auslan Lexical Database, therefore, make do with a video clip, and have associated fields that code for phonological form. In the print, CD-ROM and web-site formats of the dictionary there is an Auslan to English section or view of the data. However, in all three the English to Auslan view (Language B to Language A) is limited. In the print format, the reason is simple: a definition of an English headword in Auslan using line draw-ings or HamNoSys was virtually impossible. Thus, this direction of comparison was represent simply by an English word index which pointed users to the entry numbers of relevant Auslan translation equiva-lents of each word. In the CD-ROM and Signbank web-site formats, one can tag of each sign entry (sign form) for its phonological characteristics (handshape, location, movement, etc.) and for the English key-words associated with each sign. The user typed an English word in the search field of the CD-ROM or the Signbank web-site to find signs associated with those keywords. Thus it is still bidirectional, but only in this limited sense. The dictionaries sorted and arranged the headsigns according to principles based on formational fea-tures (see below). The dictionary groups the signs into handshape sections on the basis of the handshape of the strong hand at the beginning of the sign. Within each section, other formational features, such as location, orientation, 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 31

Session 1 movement, the number of hands, and, in some cases, facial expression. The hand-shape on the weak hand, and the symmetry of the sign determine the grouping and sequencing of signs with the same dominant handshape. Ordering principles for print dictionary ( alphabetical order) [Slides 18-25] Only the main principles that underlie the alphabetic order of sign entries are discussed here. For full details, see Johnston (2003). By handshape [Slides 18-20] Handshape is ideally suited to act as the principle for distinguishing one sign from another and for grouping signs into sections. The handshapes of Auslan are ordered according to a semi-arbitrary sequence. The earlier editions of the Auslan dictionaries (1987, 1989) based the se-quence of handshapes on the frequency of occurrence of each in the lexicon, with the most frequent handshape the flat handshape appearing first. Signs of Australia (1997) used a more natural and intuitive sequence based on Auslan handshapes used in counting. The order follows the handshape used in counting starting from zero to then. Slight variations on these handshapes are ordered between the major counting handshapes. For example, the Round hand is first in the dictionary because it is the Auslan handshape for the number 0. The Okay hand is second because it is an alternative handshape for 0. The Point hand follows because it is the handshape for the number 1. It is followed by the Hook hand because the latter is similar to the Point hand (it resembles a hooked 1) and so on through handshapes for the numbers 2 through 12 (In Auslan, none of the numbers above twelve involves the use of a handshape not already found in lower numbers.) A few handshapes found in Auslan cannot be associated with numbers in any way and are arbitrarily placed at the end of the list and at the end of the dictionary. For example, SUN is found in the Round hand section of the dictionary, IGNORE is found in the Point hand section, and FRONT is found in the Flat hand section [Slide 20]. By primary location [Slides 21-22] The dictionary then sequences the signs according to location (tabulation). It lists signs made high in the signing space or high on the body (e.g., on the head or face) before those that are made low in the signing space or low on the body (e.g., on the waist or arm). Thus THINK comes before WHO which comes before STOMACH. By number of hands [Slide 23] Within each handshape section, the dictionary lists as a group the signs that use only one hand before those that use two hands. For example, the one-handed sign CONSIDER precedes its double-handed variant. By same or different handshapes [Slides 24-25] Within each handshape section, the dictionary arranges signs that require the use of two hands after those that require only one hand. Within these two-handed signs those that have the same handshape on both hands are listed before those that have a different handshape on the subordinate or weak hand. Thus CONSIDER precedes CROSS-OUT. The order of two-handed signs with a different handshapes on the weak hand simply repeats the order of handshapes in the dictionary overall. Thus the dominant or strong hand determines the handshape section of the dictionary the sign appears in, and the subordinate or weak handshape determines its ordering with respect to other two handed signs with the same 32 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston dominant handshape. Thus QUESTION, POTATO and DEBT appear in this order relative to each other because the Okay hand comes before the Point hand and the Point hand comes before the Flat hand. The structure of the survival guide [Slide 26] The Survival Guide to Auslan: a beginner's pocket dictionary of Australian Sign Language (Johnston & Schembri 2003) is a wordlist rather than a true dictionary. As such, it was not unlike most so-called sign language dictionaries that have been created in the past by non-linguists.. These were essentially mono-lingual wordlists, in which the headwords were actually the words of the community spoken language arranged according to the alphabetic writing conventions of that language. An illustration of a sign equated with the headword was place next to it. (The Survival Guide was published with a specific narrow audience in mind: adults with limited time, who only wanted to know very simple vocabulary items to facilitate basic communication with deaf people.) There is a specific niche for these types of simple wordlists but they should not be confused with true dictionaries of sign languages which present day linguists are creating and which are the topic of discussion here. Structure of CD-ROM [Slides 27-33] The CD-ROM edition of Signs of Australia (1997) was the first multi-media dictionary of Auslan. The Auslan Lexical database was again the source of data used in creating the CD-ROM. The most important innovation in the CD-ROM was the ability to search for signs based on their for-mational features. A sign search request could be launched selecting the search by Auslan button option rather than the familiar search by English word (by consulting the English index in the book format). After selecting the search by Auslan option on the CD-ROM the user was guided through a series of filtering questions which asked them to specify: (i) if the sign was one-, double- or two-handed; (ii) the handshape on the strong hand at the beginning of the sign; (iii) the location of the sign; and, if the sign was used two hands, (iv) the handshape on the weak hand, and (v) the secondary location (i.e., the place on the weak hand where the two hands contacted or interacted with each other). The second most important innovation was the ability to instantly select and sort signs according to various criteria, such as semantic field (food, body parts, emotions, health, etc.) or dialect (according to major dialect zones or states of Australia). The third most important innovation was the ability to jump instantly between signs within the diction-ary based on a number of pre-established links between them, such as synonyms, antonyms and homo-nyms. Structure of Internet Signbank [Slides 34-39] The Auslan Lexical Database also lies at the basis of the internet version of the Auslan Dictionary, called 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 33

Session 1 Signbank generally, or Auslan Signbank specifically. Indeed, the database itself now resides in the web-site. In other words, Auslan Signbank is just the website name of the Auslan Lexical Database in its cur-rent form: Auslan Signbank is just web portal on the database. It is however possible for researchers to download the contents of the Signbank site as a.csv file and then import the records into a variety of databases for further processing or investigation. (Versions of the database thus still exist in FileMaker Pro and other spreadsheet formats.) As with all previous versions of the dictionary Auslan Signbank contains conventional lexical signs of the language displayed as the headsign with a definition in English. The dictionary also contains a num-ber of depicting or classifier-based signs the lexical status of which is open to question. There are some innovations in the presentation of the dictionary in Auslan Signbank. First, there is now a public view available to anyone with an internet connection. Second, there is a restricted or private view available only to registered users. The registered view has several levels of access to the database. Ordi-nary users register in order to be able to post feedback on the site (including reporting errors or missing signs from the dictionary). There are four other levels of research access available on application, inter-preter, researcher, editor, publisher. These have various access and editing rights within the database. Access to viewing and/or changing the database is controlled allocating different classes of users with different levels of rights and privileges to the site. These additional page views display the type of infor-mation already exploited in the CD-ROM. However, unlike the CD-ROM, Auslan Signbank can be up-dated at any time (provided one has the access privileges) and these changes can be effected and dis-played immediately. Records may also be deleted or added as new information becomes available. This means that the dictionary is able to respond quickly to new information that comes available either through community feedback or through examination of the Auslan Corpus (see next section). The initial web version of Auslan Signbank (2004) had only some of the functionality found in the CD-ROM due to constraints imposed by the scripting language used at that time. Between 2004 and 2018, the scripting language behind Auslan Signbank was completely overhauled and functionality previously found only on the CD-ROM became available on the site. However, to date, most of this functionality is only available to registered users who have special access rights. The signs still have the inherent order first assigned in the print versions of the dictionary, but the or-dering of signs is not transparent on a web-site as it would be in a printed book. The order is only evident if one scrolls through the dictionary from one page to the next, as each page only presents one sign entry. The latest major innovation in Auslan Signbank, implemented in 2018, is the provision of an Defini-tion in Auslan field. In this field, a video clip can be inserted in which a definition in Auslan is given for the headsign. The creation of entries of this type has only recently begun, but it does mean that it will be possible, in the future, to present the Auslan Signbank entries as in a monolingual dictionary. This is ground-breaking. No monolingual dictionary of any signed language has yet been published. Without the existence or development of writing systems for signed languages, monolingual dictionaries of signed languages in printed form were virtually 34 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston impossible. Although monolingual sign language dictionaries became possible in principle with the advent of film and video, they were still totally impracticable. How-ever, with the availability of digitized video, it has become technically possible to make a monolingual dictionary, but none have yet to be made or published. This latest version of Auslan Signbank (2018) now makes it possible to create monolingual entries entries in which an Auslan headsign is defined using Auslan. The first entries using this technique have begun to be created. Part 2: corpus-oriented sign language dictionary [Slides 40-70] Rational for a corpus-based sign language dictionary [Slides 40-44] Recall that the Auslan dictionaries described above were the product of a traditional language documentation and description research program which involved participant observation, interviews, and elicitation. Such a model is widely used in linguistics and anthropology for work on undocumented and undescribed languages. Generally speaking, the language communities in which this type of research is undertake tend to be found in geographically discrete, small-scale traditional societies without a history of writing. Sign language communities, however, tend not to be like this. Almost everywhere, community sign languages are used by small numbers of deaf individuals who are embedded in much larger hearing communities which use a spoken (and sometimes written) language. In some countries or regions, some deaf people are born into families that have several generations of deafness, but this is relatively uncom-mon. Most deaf people have hearing parents and siblings. Thus the acquisition and use of most sign lan-guages appears quite a-typical when compared to that of most spoken languages. These and other well-known sociolinguistic characteristics of deaf communities mean that many researchers have remarked that sign language communities seem to display a much higher degree of variation in both lexis and grammar than might otherwise be expected. In the late 1990s and early 2000s, it became obvious to many sign language researchers, including Auslan researchers, that an empirically sound description of the lexicon and grammar of a sign language requires the creation of linguistic corpus based on naturalistic signing from native or near-native signers (Johnston 2010, 2014). The creation of a linguistic corpus entails the transformation of recordings of language use through transcription, annotation, and translation so that it becomes a machine readable resource. In the case of sign languages this, of course, means creating digital video recordings, and using multimedia annotation software. Once an archive of recordings has been created in preparation for creating a corpus of a sign language, any preexisting sign language dictionary of this sign language needs to orient itself towards it. For ex-ample, one needs to ensure that every conventional lexical sign that is identified and thus glossed in the corpus is also listed in the dictionary, otherwise the dictionary would clearly be judged as incomplete. Actually, the relationship between the dictionary and the corpus is two way: on the one hand, the diction-ary informs sign identification and annotation; 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 35

Session 1 on the other hand, the corpus provides potentially new data on hitherto undocumented conventional signs, or identifies possible variants of already documented forms. Relating corpus annotations to sign entries in the dictionary [Slides 45-49] ID-glosses In order to harmonize the dictionary and the corpus, it is necessary to be able to relate the form of sign types entered in the dictionary with tokens found in the corpus. In this corpus-oriented database, sign entries are associated with keywords and a unique identifying gloss in English. The unique identifying gloss (which I have called the ID-gloss) is assigned to each headsign in the dictionary. The ID-gloss associated with a sign type in the dictionary is used to annotate any token of this sign type in the Corpus. However, in order to use ID-glosses effectively and efficiently, one needs a reference database in which lexical types are documented. This is the essence of a corpus-oriented sign language dictionary: all conventional lexical signs of the language are listed in the sign dictionary and have a unique identifier when segmented and labelled in the corpus annotations. Thus during the annotation and glossing of the corpus, constant reference needs to be made to the sign language dictionary. To do this the Auslan annotator refers to Auslan Signbank. Whenever the ID-gloss for a particular sign needs to be checked, the annotator types in a English keyword associated with the sign into Auslan Signbank search window and views the matches suggested in Signbank. The annotator then select the ID-gloss for the sign that most closely resembles the form of the sign they observe in the corpus. Auslan Signbank is currently external to the ELAN annotation software used to create the corpus. The annotator simply has the on-line dictionary open on their desktop as they create annotations in ELAN. However, in Signbank NGT (a Signbank for Sign Language of the Netherlands based on Auslan Signbak we will discuss this later) the two are linked (Crasborn & Meijer 2012, Crasborn et al 2012). For exam-ple, Signbank NGT is linked to the ELAN software templates that are used to annotate the NGT corpus. Annotators can begin to type the ID-gloss of a sign annotation in ELAN and a controlled vocabulary listing all related ID-glosses in Signbank NGT will appear so they may choose the best fit. If in doubt, the annotator may click on any gloss within the controlled vocabulary and a thumbnail video clip for that ID-gloss from Signbank will appear in a pop-up window. Phonological variation [Slides 50-51] The most common form of a sign (judged by the most frequent occurrence of a number of variant in a token count from the corpus) is treated as the stem or base form of a lexical sign. It is the main headsign. Variants of the headsign are shown at that entry. A variant is basically a form of base sign that has one, or at most two, formational features which are different to the base sign (the variant usually involves a different handshape, 36 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston location, or the number of hands used in the sign). Variants also have their own entry record at the appropriate place in the form-based dictionary If usage data, as recorded in the corpus, suggests that the status of base and variant forms need to be reversed, they will be. For example, usage data may show that a double-handed form of a sign is far more frequent than a one-handed form. If the one-handed form has been listed as the main headsign in the dictionary, the relations should be reversed. Attested usage [Slide 52] Overtime, and providing that the corpus is representative of all users and all text-types, and also large enough, one should expect to find that the majority of one s sign language dictionary entries are attested in the corpus. The only signs expected not to appear in a general corpus are those already known to be archaic, obsolete or marginal (e.g., vocabulary only used by small groups of people in specialist fields). Attested usage in the corpus indicates that the dictionary is empirically sound. In 2012 after 4 years of annotation (see Johnston 2012) approximately 75% of the 3,500 core entries in Auslan Signbank had been attested in the Auslan Corpus. This is a very high percentage of the known lexicon given the relatively modest size of the Auslan Corpus at that time (approximately 70,000 tokens), when compared to spoken language corpora. (There is double this number of entries in Auslan Signbank, i.e., over 7,000 entries, but these are all variant forms of the core 3,500 lexical signs.) Missing data [Slide 53] Of course, no dictionary is ever complete so it is more than likely that hitherto unrecorded signs will be identified in the corpus. If these are confirmed as genuine conventional lexical signs of the language, then, after a period of consultation with community members and/or confirmation by additional tokens being found in the corpus, they should be added to the dictionary. In this way the corpus is able to augment the lexical database with hitherto unrecorded signs, either variants of existing lexical entries, or lexical signs which may be recent additions to the language (neologisms or borrowings) or which were simply previ-ously overlooked signs. Approximately 50 such new or previously unrecorded conventional lexical signs had been identified in the Auslan Corpus and so they were, in turn, added to Auslan Signbank, after 4 years of annotation. Improved definitions based on usage data [Slide 54] A sign form may be initially assigned a meaning which is too narrow given its attested usage in the corpus. In such cases, the definition in Signbank is improved to reflect the way the sign is actually used in naturalistic signing. This could mean changing and improving the definitions, or adding, removing or changing associated keywords. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 37

Session 1 Adjust the lexical status of some signs [Slide 55] Some highly iconic or depicting signs may have been entered in the dictionary as conventional lexical signs on the basis of initial fieldwork (intuitions, introspection, elicitation or community feedback had reported the sign as conventional and lexcialized). However, the use of the sign as attested in the corpus may indicate that the sign does not have a meaning which is as specific or conventional (or lexical ) as first thought. Such signs are borderline signs in terms of lexicalizaiton: they may not be language-specific in their meanings and thus not really appropriately listd in a dictionary of Auslan. (Of course, they would still need to be discussed in a grammar of the language.) Such entries may therefore need to be updated to better reflect their status, or even removed. From corpus to dictionary? [Slides 56-57] In the ideal corpus-building situation, it is not expected that one would begin to segment and gloss a re-cording in a sign language without first having conducted basic lexicological research into the language, and then having recorded and described the resulting initial lexicon in a database or dictionary. However, in recent years some sign language dictionary projects have begun as part of large sign language corpus projects. The aim is to use the corpus to build the dictionary. It is possible to construct a reference lexicon more or less from scratch based the annotations them-selves as they are being made, i.e., using only attested instances in the corpus itself as the basis for one s lexicon. Something akin to this approach has been described for the BSL corpus (Cormier et al., 2012). Though this may provide a way to proceed, this approach is difficult, time consuming and prone to inconsistency because annotators have no common reference point, especially in the early stages of annotation and glossing. Researchers are likely to encounter many different uses and variant forms that require constant revision and correction of the incipient database and dictionary. This can also have knock on effects in the growing lexical database (Crasborn and Meijer, 2012). Indeed, some researchers believe that an empirical dictionary of a sign language should only draw on signs attested in the reference corpus. One could perhaps call these acorpus-based or corpus-driven sign language dictionaries rather corpus-oriented sign language dictionaries, like the one described here. However, there are dangers in adopting such a strict criteria for inclusion of signs in a sign language dic-tionary it is highly unlikely that any corpus of a language (unless it runs into hundreds of millions of words or signs) will include at least one token of every known word or sign in the language. This is par-ticularly problematic if very basic everyday vocabulary items just happen not to be found in a corpus, something that is highly likely to be the case in the sign language corpora that have been or are in the process of being created because they are really not that large. Any user of such a dictionary may find that signs for basic concepts is not recorded because it simply never occurred in the corpus. This is not satisfactory and users of such a dictionary would rightly conclude that it is incomplete. One way to avoid this problem, if one wants to restrict oneself to attested data, is to construct one s 38 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston corpus so that it includes elicited text and vocabulary items, not just conversational or narrative text. One could then incorporate tasks during recording sessions that elicit basic vocabulary items. Enriching dictionary entries from corpus findings [Slides 58-68] The Auslan Lexical Database, and hence Auslan Signbank, is designed to accommodate a large variety of information about each sign at each entry. This information is found in scores of dedicated fields or cells. These contain information on: (i) the form of the headsign; (ii) its meanings; (iii) who uses it; and (iv) criteria by which the entry can be searched, sorted or displayed on-line. Much of this information emerges from the study of the corpus itself (e.g., dialect preferences, age cohorts, school-based signs, religious affiliation, frequency, etc.) This information is constantly under review and is easily editable in the Signbank researcher interface. Some information is visible to the public immediately, other information is only visible on the private view of the database, where it is available for researchers to consult. Dealing with homophones [Slide 69-70] Because Auslan Signbank is a form-based database of signs, it follows that if two signs are homonyms, i.e., two distinct meanings in Auslan have the same sign form (cf. English can meaning metal container and able to ), then they would both reference the same form. The solution in Auslan Signbank is the same as in an English dictionary the headword or headsign is repeated, and at the two (or more) entries created the same phonological form is defined with each of its two (or more) distinct meanings. In Auslan Signbank these are numbered: Sense 1, Sense 2, and so on. Multiple sign languages [Slides 71-78] Auslan Signbank & other Signbanks [Slide 72] From their first conception and application in the Auslan Lexical database, the principles underlying its structure were claimed to be applicable to any sign language (Johnston 2001). Auslan Signbank (and its earlier instantiations in book and CD-ROM formats) has form-based entries. These entries have their formational features encoded either in a dedicated transcription system (such as HamNoSys) or in a series of fields that specify the presence or absence of relevant formational features of each sign (such as handshape, location, etc.). The result is that signs from any sign language can be accommodated within the database entries are simply created which are described by form. (If the database is ordered by form, then the new sign is simply entered at its appropriate slot according to alphabetical order.) Even the types of interrelationships between signs (such as synonymy, antonymy, homonymy, semantic field membership, dialect distribution, etc.) that are built into the 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 39

Session 1 program can also be exploited, as appropriate, for any language. Realization of this potential has lead other linguists working on other sign languages to adapt the basic architecture of Auslan Signbank for their own sign languages. As a result, on-line dictionaries have been created for Sign Language of the Netherlands (NGT Signbank), British Sign Language (BSL Signbank), and Finnish Sign Language (FinSL Signbank), and work on an American Sign Language (ASL Signbank) has commenced. From homophones to multiple sign languages [Slide 73-77] One important aspect of the structure of Auslan Signbank opens up the further possibility of creating a lexical database that accommodates multiple sign languages in the one Signbank, i.e., in a Universal Lex-ical Database as it was described in Johnston (2001). Such a database does not simply use the existing Auslan Signbank as a template for other Signbanks, it creates a Universal Signbank. Recall, that Auslan Signbank creates records of sign forms according to formational features. This re-sults in a visual ordered sequence of line drawings or photographs when presented in print format, or in columns of similar video clips when presented as records and cells in a spreadsheet. Though only one entry per page is visible in an on-line dictionary such as Signbank, the same principle allows instant sort-ing of records within the database by any one of the relevant formational features encoded with each entry. Thus signs may be ordered by handshape (as in the print format), or by location, or by any other feature requested. More importantly, as we have seen, these form-based entries also easily encoded and represent phono-logical variation, dialect variation, semantic relations (antonyms, synonyms, homonyms), and relations to other sign languages, such as BSL and ASL. In Auslan Signbank sign forms can now be tagged not only dialect, but also sign language. The tag means the sign form exists in BSL or ASL. The ability to tag for language together with the ability to tag a sign as being in a homonym relation with another creates the potential for an open-ended database. Signs from any sign language can be en-tered into the same database. When the Auslan Lexical Database was first conceive there were practical limitations on personal computer processing speeds (and cost), the quality of digital video recordings, and the storage capacity (and cost) of servers. Today, the number of entries in Signbank, the number and type of codes associated with each entry, or the length of the video clips used as the headsign or as definitions are no longer impracticable nor too costly. As a result, work to test the possibilities and uses of this type of universal Signbank, has begun. It is called Global Signbank. Global Signbank? [Slides 78] Is a Global Signbank practical? Two observations about the lexicons of sign languages that have been studied in depth to date suggests it may well be. First, size. Most dictionaries of sign languages that have been published over the past 50 years do not seem to contain more than several thousand main headsigns (somewhere between 3,000 to perhaps 8,000). This appears relatively modest compared to most spoken language dictionaries, even spoken languages without a written form. 40 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

A Corpus-oriented Sign Language Dictionary Trevor Johnston Second, many researchers have remarked that the high degree of iconicity in sign languages has tended to create a situation in which the same or very similar sign forms reappear repeatedly in different sign languages. Though these iconically motivated sign forms often have the same or similar conventional meaning in unrelated sign languages, their meanings may also be completely different. Nonetheless they share the same sign form. It is not uncommon for signers of one sign language reporting that when they view a foreign sign language they often see signs that appear to be very similar in form to ones in their own sign language. Sometimes they may be able to correctly guess the meaning, often they will not be able to. By contrast, it is quite rare to see the same word form appear repeatedly in unrelated spoken languages. The reason for this is that the potential number of phonologically well-formed words in most spoken languages appears to all practical purposes to be without limit when the combination of phonemes and the possible number of syllables are taken into account. However, the situation in signed languages may be different partly due to the impact of iconicity. Assuming it is practicable, a multiple language version of Signbank Global Signbank would most likely benefit from the current formational order being re-designed. Global Signbank would preferably have neutral (i.e., non-language specific) formational order accepted by a majority sign language re-searchers, even if the primary feature is most likely to remain handshape. A non-language specific order could, for example, be one based on the number of selected fingers extended in each handshape, rather than any given counting system as such. Whenever two or more sign languages have an identical or near identical sign form they would be tagged to the same video in Global Signbank, just as are homophones in Auslan Signbank. Global Signbank could be used for the cross-linguistic and comparative study of the sign language phonological inventories, degrees and types of iconicity, and lexicalization and grammati-cization patterns. References Cassidy, S., Crasborn, O., Nieminen, H., Stoop, W., Hulsbosh, M., Even, S., Komen, E.and Johnston, T. 2018. Signbank: Softrware to Support Web Based Dictionaries of Sign Language. In N. Calzolari, K. Choukri, C. Cieri, T. Declerck, S. Goggi, K. Hasida, H. Isahara, B. Maegaard, J. Mariani, H. Mazo, A. Moreno, J. Odijk, S. Piperidis and T. Tokunaga (eds.). Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), May 7-12, 2018, Miyazaki, Japan. Cormier, K., Fenlon, J., Johnston, T., Rentelis, R., Schembri, A., Rowley, K., Adam, R. and Woll, B.. 2012. From Corpus to Lexical Database to Online Dictionary: Issues in Annotation of the BSL Corpus and the Development of BSL SignBank. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen and J. Mesch (eds.). Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Workshop as part of 8th International Conference on Language Resources and Evaluation, LREC] (pp. 7-12). Istanbul, Turkey., Istanbul, Turkey 21-27 May 2012. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 41

Session 1 Crasborn, O & Meijer, A de. 2012. From corpus to lexicon: the creation of ID-glosses for the Corpus NGT. In Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon (pp. 13-18). Crasborn, O., Hulsbosch, M. and Sloetjes, H. 2012. Linking Corpus NGT annotatons to a lexical database using open source tools ELAN and LEXUS. In Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon (pp. 19-22). Istanbul, Turkey., Istanbul, Turkey 21-27 May 2012. Johnston, T. 1987. A preliminary signing dictionary of Australian Sign Language (Auslan). Adelaide: TAFE National Centre for Research and Development. Johnston, T. 1989. Auslan Dictionary: A dictionary of the sign language of the Australian deaf community. Sydney: Deafness Resources Australia. Johnston, T (ed.) 1997. Signs of Australia: A new dictionary of Auslan. Sydney: North Rocks Press. Johnston, T. 2001. The lexical database of Auslan (Australian Sign Language). Sign Language & Linguistics 4(1/2). 145-69. DOI: 10.1075/sll.4.12.11joh. Johnston, T., Thornton, D. and Napier, J. 2002. Discovering Auslan: A beginner's dictionary of Australian Sign Language. Sydney: North Rocks Press. Johnston, T & Schembri, A (eds.) 2003. The Survival Guide to Auslan: a beginner's pocket dictionary of Australian Sign Language. Sydney: North Rocks Press. Johnston, T. 2003. Language standardization and signed language dictionaries. Sign Language Studies 3(4). 43168. 10.1353/sls.2003.0012. Johnston, T. 2004. Auslan Signbank (http://www.auslan.org.au). Sydney: Royal Institute for Deaf and Blind Children & Catalyst Training Systems. Johnston, T. 2008. The Auslan Archive and Corpus. In D. Nathan (ed.). The Endangered Languages Archive http://elar.soas.ac.uk/languages, London: Hans Rausing Endangered Languages Documentation Project, School of Oriental and African Studies, University of London. Johnston, T. 2010. From archive to corpus: transcription and annotation in the creation of signed language corpora. International Journal of Corpus Linguistics 15(1). 104-29. DOI: 10.1075/ijcl.15.1.05joh. Johnston, T. 2012. Lexical Frequency in Sign Languages. Journal of Deaf Studies and Deaf Education 17(2). 163-93. Johnston, T. 2014. The reluctant oracle: using strategic annotations to add value to, and extract value from, a signed language corpus. Corpora 9(2). 155 89. 42 2018 국립국어원 국제 학술 대회 보이는 언어의 기록, 수어사전

Ҷ Ҷ ਙ Ҷ ೡ ധ ࠃ י ӟ 3FDPSET PG 7JTJCMF -BOHVBHF 4JHO -BOHVBHF %JDUJPOBSZ 주제 2 수어사전의 구조 예테 크리스토페르센 덴마크, 유시시(UCC) 대학교 Jette H. Kristoffersen University College Capital, Denmark

1 부발표 수어사전의구조 - www.tegnsprog.dk 예테크리스토페르센 덴마크, 유시시 (UCC) 대학교 초록 수어사전학은수어연구계에서신생학문에해당된다. 수어사전학자가직면하는가장큰어려움은수어가시각양식 (visual modality) 을사용하는언어로수어사용자간널리사용되는문어체계가없다는점이다. 지난수십년간수어사전학은전자매체의혜택을크게누렸다. 수어사전은더이상과거와같이어휘만을정리한목록이아닌구어사전과유사한정보를담은과학에기반한사전이다. 하지만수어의시각적요소는어휘검색과수어 (sign) 에관한정보수록측면에서여전히과제로남아있다. 본논문에서는덴마크수어사전 (Danish Sign Language Dictionary, Kristoffersen & Troelsgård, 2016) 의구조를소개하고수어사전의구조를제안하고자한다. 1. 서론 가장최근까지도대부분의수어사전은구어에서해당되는어휘와수어를나타내는그림, 사진또는설명 을수록한어휘목록에불과했다. 이때원천언어는주로구어에해당되며수어가목표언어이다. 지난수십년간수어사전학은빠르게발전하였으며일련의과학적수어사전이편찬되었다. 그첫시도중하나로언어학적원칙에기반한미국수어사전 (A Dictionary of American Sign Language on Linguistic Principles, Stokoe et. al. 1965) 이있다. 오늘날수어사전은더이상어휘의나열이아닌구어사전과유사한정보를담은진정한의미의사전이다. 또한수어가원천언어에해당된다. 수어사전학자는기술발전의혜택을크게누렸다. 과거에는수어를사전에표기하는방식을결정하는데어 려움을겪었다면, 오늘날에는수어를디지털동영상형태로기록하여쉽게전자사전에통합시킬수있다 44 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전의구조 예테크리스토페르센 (Zwitserlood et al, 2013 참조 ). 하지만여전히수어사전학자가겪는몇가지어려움이있다. 예를들어, 수어를지칭하거나기호검색을수행하고검색한결과, 기호의순서정렬등을위해서는수어에대한어떠한공식표기법이필요하다. 또한수지적요소 (manual features) 검색이가능하도록하기위해서수어를원천언어로하는수어사전은구어사전에비해고도화된검색기능이필요하다. 덴마크수어사전 (Kristoffersen & Troelsgård, 2016) 은사실상덴마크수어 (DTS)-덴마크어이중언어양방향사전이지만, 그구성은이중언어사전과단일어사전의중간정도에해당된다 (Kristoffersen & Troelsgård, 2012). 표제어는 DTS 기호만을기준으로하며수어의의미에따라정렬되어있고, 해당되는덴마크어의동형이의어또는다의어를구별하기위한설명을제외한, 용례를포함한거의모든정보는 DTS를기준으로작성되어있다. 덴마크어휘에대한정보는기존의 덴마크단일어범용사전 으로쉽게확인할수있기에본사전에서는 DTS를중심으로기술하도록이접근법을채택했다. 따라서덴마크수어사전은표제어의정의대신해당사전의또다른메타언어인, 검색가능한대응덴마크어를수록한단일어사전이라할수있다. 2. 덴마크수어사전표제어배열 덴마크수어사전의표제어는 4 개부분으로구성되어있다 ( 그림 1 참조 ). 섹션 1: 표제어항이섹션은수어표제어의사진과용어해설 (gloss) 을담고있다. 첫번째수위 (location) 와제일두드러지는수형 (handshape) 을아이콘으로설명한다. 마지막으로버튼을클릭하여해당표제어해설을출력할수있도록별도창을열거나전체사전에수록된기호를포함한모든용례를담은용어색인을확인할수있다. 섹션 2: 동영상창수어표제어의기본형을볼수있다. 단어의음운적변이형 (variant) 이있는경우, 동영상상단에각변이형별재생버튼으로확인할수있다. 컨트롤버튼으로재생속도를낮추거나창을확대할수있다. 용례등의동영상은다른섹션에있는재생버튼을클릭하여확인할수있다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 45

1 부발표 섹션 3: 의미창섹션3에서는수어단어의뜻을설명한다. 대응덴마크어혹은단어의용법에관한설명 ( 기능어의경우 ), 입모양에대한정보, 동의어에관한상호참조, 사용제한설명및예문등이기술되어있다. 의미상이해하기어려운합성어도일반적인뜻아래기술하고있다. 대응덴마크어는덴마크어사전 (Danish Dictionary) 1) 에서검색할수있으며연결링크가제공된다. 따라서사용자들은해당덴마크어의뜻과용법에대해더자세히알아볼수있다. 섹션 4: 추가정보 동일한것으로보이는수어 ( 동형이의어 ) 와분류사의경우에는자주사용되는고정형태 (frozen form) 에 관한상호참조를제시한다. 또한표제어와흔히함께사용되는수어도기록하고있다. [ 그림 1. 덴마크수어사전표제어의 4 개섹션구성예시 ] 1) 덴마크어사전 (Danish Dictionary) 은현대덴마크어를기록한단일어, 범용사전이다. 사전은 www.ordnet.dk/ddo 에서무료로 사용할수있다. 46 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전의구조 예테크리스토페르센 3. 덴마크수어사전검색기능 3.1. 검색기준선택 덴마크수어사전에등재된수어는여러카테고리에따라검색할수있다. 수형, 수위, 주제및텍스트로 검색이가능하다. 수형검색검색이가능한수형은총 65가지이다. 검색기준으로사용될수형은연관수형그룹별로선택되지만원치않는수형은개별적으로제외할수있다. 이방식으로사용자는기본적으로묶이는수형그룹과별개로단일수형또는원하는유형의수형을선택할수있다. 그림 2는집게손가락을편유형을검색한수형선택창의예시를보여주고있다. [ 그림 2. 수형선택창부분 ] 움직이는손 (active hand) 과수동적인손 (passive hand) 별특정수형검색이가능하다. 수위검색수위는 15개의신체부위별혹은신체와가까운위치의수위를나타내는수위아이콘페이지에서선택할수있다. 또한 4개의그룹 ( 선택창상단 ) 중에서수위를선택하는방법도있다. 그룹을지정하면해당그룹에속한모든수위가검색기준에포함된다. 추가로 ( 우측열에서 ) 두개의수위를선택할수있다. 하나는전체몸의앞부분이며다른하나는얼굴앞수위이다. 그림 3은수위선택창을나타낸다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 47

1 부발표 [ 그림 3. 수위선택창 ] 텍스트검색대응덴마크어, 수어용어해설 ( 글로스 ) 및예문 ( 표기와번역문 ) 에대한텍스트검색이가능하다. 이방법을통해기본형은아니지만예문에나타나는수어단어를찾을수있다. 구단위검색및불특정검색도가능하다. 주제검색 70가지주제중에서검색기준을선택할수있다. 70가지주제는 9개그룹과 2개의하위그룹으로구분된다. 상위그룹에해당되는주제를지정하면자동적으로모든하위주제가검색기준으로선택된다. 그림 4 는주제구성체계를묘사하고있다. 복수의검색기준을결합하여검색도가능하다. 카테고리별로는 또는 (or) 조건이적용되며검색카테고 리의조합으로검색하는경우 그리고 (and) 조건을사용한다. 현재선택한기준은화면의좌측에표시된다. 그림 4 는 집게손가락을편 수형과 종교 주제로검색한수 어목록을보여주고있다. 48 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전의구조 예테크리스토페르센 [ 그림 4. 주제검색에사용되는 3 단계체계 ] 3.2. 검색기능조절 검색을조율할수있도록검색기준과검색결과는좌우나열되어표시된다 ( 그림 5. 우측에검색결과확 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 49

1 부발표 인가능 ). 원치않는검색기준을개별적으로클릭하여제외하거나리셋 (reset) 버튼을클릭하여모두삭제 할수있다. [ 그림 5. 수형및주제결합검색에따른기준목록과검색결과일부 ] 모든연관결과가검색결과에포함되도록하기위해덴마크수어사전의검색기능은상당히포괄적이다. 수지적요소의경우, 수어의수위및수형은그중요도와무관하게모두검색기능에고려된다는것을의미한다. 같은이유로텍스트기준으로검색할때글로스, 대응어및용례번역모두가검색결과에반영되어표시된다. 검색결과는연관성순서대로나타난다. 수형및수위로검색하는경우, 수어에나타나는순서에따라점수가부여돼먼저나타나는수형또는수위가뒤따르는수형또는수위보다점수가높다. 텍스트검색결과는글로스, 대응덴마크어, 용례글로스, 용례의번역문에있는단어순으로가중치가부여된다. 즉, 글로스또는대응어가용례에사용된단어보다점수가높다. 이렇게계산된 연관성점수 에따라각검색결과 50 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전의구조 예테크리스토페르센 는 0-3 연관성별 (relevance star) 로평가되어높은순서대로표시된다 ( 그림 5 우측검색결과참조 ). 연관성점수가동일한결과는수위-수형순으로표시된다. 또한사용자는연관성점수와무관하게수위및수형기준으로검색결과를정렬하여확인할수있다. 이방법은검색결과목록상단에있는탭을클릭하여정렬할수있다. 세번째탭은다시연관성기준의정렬방법으로검색한결과를표시한다. 4. 향후과제 덴마크수어사전은몇가지분야에서개선의여지가있다. 단일어구어사전은흔히원천언어를메타언어로사용한다. 이를수어사전에적용하기란거의불가능에가깝지만 DTS에관한추가정보를더함으로써사전이더욱풍부해질수있다 ( 예. 의미의정의 ). 또한굴절형에관한설명또는어원및품사와같이아직연구가많이진행되지않은분야등새로운유형의정보도추가할수있을것이다. 참고문헌 Det Danske Sprog- og Litteraturselskab (2009-2011). Den Danske Ordbog [The Danish Dictionary]. Available at http//:ordnet.dk/ddo (last accessed 18th September 2018) Kristoffersen, J. H. & Troelsgård, T. (2012). The electronic lexicographical treatment of sign languages: The Danish Sign Language dictionary. In S. Granger & M. Paquot (eds.), Electronic Lexicography. Oxford: Oxford University Press, pp. 293-318. Kristoffersen, J. H. & Troelsgård, T. (eds) (2016). Ordbog over Dansk Tegnsprog 3. udgave [The Danish Sign Language Dictionary 3rd edition]. Center for tegnsprog, Copenhagen. Available at http://www.tegnsprog. dk (last accessed 18th September 2018)Stokoe, W.C.; Casterline, D.C.; Croneberg, C.G. (1965). A Dictionary of American Sign Language on Linguistic Principles, Washington D.C.: Gallaudet College Press. Zwitserlood, I, Kristoffersen, J.H. & Thomas Troelsgård (2013). Issues in Sign Language Lexicography. In: Jackson, H. (ed.): The Bloomsbury Companion to Lexicography. London: Bloomsbury Publishing 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 51

Session 1 Structure of a Sign Language Dictionary www.tegnsprog.dk Jette H. Kristoffersen, University College Capital, Denmark 0. Abstract Sign language lexicography is a young discipline in the world of sign language research. The main problem faced by the sign language lexicographer is that sign languages use the visual modality and have no written representation commonly used among the native signers. Sign language lexicography has over the last decades benefited tremendously from the electronic medium, and sign language dictionaries are no longer mere wordlists as in the past, they are real scientifically based dictionaries, with information types similar to those found in dictionaries of spoken languages. The visual modality is still a challenge on both how to perform searches and how to present the information on the signs. This paper describes the structure of the Danish Sign Language Dictionary (Kristoffersen & Troelsgård, 2016) and demonstrate a possible structure of a Sign Language Dictionary. 1. Introduction Until recent times most sign language, dictionaries were mere wordlists, with words from a spoken language, each accompanied by a picture or a verbal description of a sign. The source language was typically the surrounding spoken language and the sign language was the target language. Over the last decades, sign language lexicography has developed rapidly and a series of scientifically based sign language dictionaries has emerged, the first being A Dictionary of American Sign Language on Linguistic Principles, Stokoe et. al. (1965). Today Sign language dictionaries are no longer wordlist they are real dictionaries with the types of information similar to those found in dictionaries of spoken languages. And they typically have sign language as the source language. The sign language lexicography has benefited tremendously from technological advances. Thus, the decision of 52 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Structure of a Sign Language Dictionary Jette H. Kristoffersen how to represent signs in a dictionary used to be a major problem, but now digital video recordings of signs can easily be integrated into electronic dictionaries (cf. Zwitserlood et al, 2013). Still, certain obstacles remain for the sign language lexicographer. For example, some form of formal notation of the signs is needed in order to refer uniquely to the signs, to perform sign searches and to order the signs in search results etc. Also, a sign language dictionary with sign language as the source language requires more advanced search functionalities compared to a dictionary of a written language, for example in order to enable searches on manual features. The Danish Sign Language Dictionary (Kristoffersen & Troelsgård, 2016) is practically a bidirectional bilingual Danish Sign Language (DTS) - Danish dictionary, but its structure places it somewhere between bilingual and monolingual (Kristoffersen & Troelsgård, 2012). There is only one set of entries, with DTS signs as headwords. The entries are structured according to the semantics of the signs, and almost all information, including usage examples, concerns DTS, except for disambiguating explanations of homographic or strongly polysemous Danish equivalents. This approach was chosen in order to keep focus on DTS, as the need for information about Danish words can easily be met by existing Danish all-round monolingual dictionaries. As a result of this decision, the Danish Sign Language Dictionary could be described as monolingual dictionary, which instead of definitions has (searchable) equivalents in another language, Danish, which is also the general metalanguage of the dictionary. 2. Entry layout in the Danish Sign Language Dictionary The entries of the Danish Sign Language Dictionary have four sections (cf. figure 1): Section 1: Entry header In this section the sign headword is shown as a photo and a gloss. The first occurring location and the most prominent handshape of the sign are shown as icons. Lastly, there are buttons for opening separate windows with a printable sign overview and with a concordance view of all usage examples in the entire dictionary including the sign. Section 2: Video window By default the base form of the sign headword is shown. If there are phonological variants of the sign, a play button for each variant is shown in top of the video window. There are also control buttons for slow-motion and enlarged view. Other videos, e.g. of usage examples, are activated by clicking play buttons in different sections of the entry. Section 3: Meanings window In this section the meanings of the sign are described. The meaning description includes: Danish equivalents and/or a description of the sign's usage (for function signs), information about possible mouth movements, cross-references to synonyms etc., information about restricted use, and example 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 53

Session 1 sentences. Semantically opaque compounds with the sign are shown below the regular meanings. Danish equivalents that can be looked up in the Danish Dictionary 1), are supplied with a clickable link to this dictionary. Thus, users can learn more about the meaning and use of the Danish words. Section 4: Additional information This section holds cross-references to homonyms and to common frozen forms of the sign (for classifier entries). In addition to this, frequent co-occurrences with the sign are shown in this section. [Figure 1. The four sections of an entry in the Danish Sign Language Dictionary.] 1) The Danish Dictionary is an all-round, monolingual dictionary of contemporary Danish. The dictionary is freely accessible at: www.ordnet.dk/ddo 54 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Structure of a Sign Language Dictionary Jette H. Kristoffersen 3. Search facilities in the Danish Sign Language Dictionary 3.1. Choosing search criteria The signs in the Danish Sign Language Dictionary can be looked up through search criteria from different categories: handshape, location, topic and text. Handshape There are 65 searchable handshapes. The handshapes to be used in the search criteria are selected in groups of related handshapes, but unwanted handshapes can be deselected individually. This enables the user to select single handshapes or custom groups of handshapes, independently of the default handshape grouping. Figure 2 shows a part of the handshape selection window, with the extended index finger group currently activated. [Figure 2. Partial view of the handshape selection window.] Particular handshapes for the active and the passive hand can be specified. Location Location is chosen from a page with 15 location icons, representing locations on or near the body. In addition to these it is possible to choose location through four group headers (= top row of the selection window). If a group header is chosen, all locations in the group are included in the search criteria. Additionally (in the rightmost column) two options are available for choosing locations in the space in front of the body: one for the neutral space in general, and one for locations in front of the face. The location selection window is shown in figure 3. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 55

Session 1 [Figure 3. The location selection window.] Text Text searches are performed on Danish equivalents as well as on sign glosses and example sentences (both transcriptions and translations). This enables users to find signs that are not themselves lemmas in the dictionary, but appear in example sentences. Phrase searches and wildcard searches are possible. Topic Topics can be chosen as search criteria from a list of 70 topics. The topics are ordered into nine groups, with subgroups in two levels. Topics which are group headers automatically include all subordinate topics in the search criteria. Figure 4 shows an overview of the topic hierarchy. Several criteria categories can be combined. Inside each category an or condition is used, while and conditions are used for category combinations. The currently chosen criteria are shown in the left side of the screen. Figure 4 shows the criteria list for a search for signs that are produced with the handshape extended index finger and concern the topic religion. 56 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Structure of a Sign Language Dictionary Jette H. Kristoffersen [Figure 4. The three-level hierarchy used for topic search.] 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 57

Session 1 3.2. Adjusting the search The search criteria and the result list are shown side by side in order to facilitate adjustments of the search (cf. figure 5, where the result list can be seen to the right). For the same purpose the search criteria can be deselected individually, by clicking them, or cleared altogether, by clicking a reset button. [Figure 5. Criteria list (and part of the result list) for a combined search on handshape and topic.] In order to secure that all possibly relevant matches are included in a search result, the search function in the Danish Sign Language Dictionary is quite broad. For manual features, that means that all occurring locations and handshapes in a sign, both prominent and non-prominent, are taken into account by the search function. Similarly, all text string matches, both among glosses, equivalents, and translations of usage examples are included in the search result. The matches of a search are ordered by relevance. For handshape and location matches this means that they are weighted according to their appearance in the sign, so that the first occurring handshape or location scores higher than the following. Text search matches are weighted in the following order: glosses, Danish equivalents, glosses 58 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Structure of a Sign Language Dictionary Jette H. Kristoffersen in usage examples, words in the translations of usage examples. Thus, a matching gloss or equivalent scores higher, than a match in a usage example. Based on the calculated relevance scores each match receives from none to three relevance stars, and the matches are ordered accordingly (see the result list shown to the right in figure 5). Matches with equal relevance score are ordered firstly by location, secondly by handshape. In addition to this, the user can choose between two sort orders: firstly by location, secondly by handshape, with no regard to the relevance score. These alternative sort orders are activated by clicking their according tabs in top of the result list. A third tab re-establishes the default, relevance-based sort order. 4. Future challenges The Danish Sign Language Dictionary could be developed in several areas. Spoken monolingual dictionaries often use the source language also as meta-language, for sign language dictionaries that would be an almost impossible challenge, but the dictionary would benefit from having more information in DTS, e.g. sense definitions. Also some additional information types could be added, e.g. descriptions of inflected forms, or lesser studied fields like etymology and part-of-speech. References Det Danske Sprog- og Litteraturselskab (2009-2011). Den Danske Ordbog [The Danish Dictionary]. Available at http//:ordnet.dk/ddo (last accessed 18th September 2018) Kristoffersen, J. H. & Troelsgård, T. (2012). The electronic lexicographical treatment of sign languages: The Danish Sign Language dictionary. In S. Granger & M. Paquot (eds.), Electronic Lexicography. Oxford: Oxford University Press, pp. 293-318. Kristoffersen, J. H. & Troelsgård, T. (eds) (2016). Ordbog over Dansk Tegnsprog 3. udgave [The Danish Sign Language Dictionary 3rd edition]. Center for tegnsprog, Copenhagen. Available at http://www.tegnsprog. dk (last accessed 18th September 2018)Stokoe, W.C.; Casterline, D.C.; Croneberg, C.G. (1965). A Dictionary of American Sign Language on Linguistic Principles, Washington D.C.: Gallaudet College Press. Zwitserlood, I, Kristoffersen, J.H. & Thomas Troelsgård (2013). Issues in Sign Language Lexicography. In: Jackson, H. (ed.): The Bloomsbury Companion to Lexicography. London: Bloomsbury Publishing 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 59

주제 3 수어사전편찬의과제 요한나메시스웨덴, 스톡홀름대학교 Johanna Mesch Stockholm University, Sweden

2 부발표 수어사전편찬의과제 요한나메시스웨덴, 스톡홀름대학교 본글은스웨덴수어사전 (Swedish Sign Language Dictionary) 이어떻게개발되었는지설명하고수어사전을편찬하는데오랜시간이걸리는이유를살펴볼것이다. 스웨덴의수어사전화작업은 1998년스톡홀름대학에서시작되었으며, 그결과 2001년첫온라인사전이나오게되었다. 스웨덴수어사전 (SSLD) 은 2008년에처음만들어진후개발이계속되고있다. 2003년말뭉치 (corpus) 구축이시작되고 스웨덴수어말뭉치 2009~2011 3개년프로젝트덕분에어휘주석으로말뭉치데이터가확대되자, 이를온라인스웨덴수어사전의지속적인개발과스웨덴수어사전의새로운수어및어휘변형을위한자원으로활용할수있는방안에대한논의가시작되었다. 키워드 : 수어사전, 사전학, 스웨덴수어 서론 수어사전은여러국가내수어의지위를높인다. 또한학습자와교수자, 가족및수어연구자들의요구에부응하기위해필요한도구이다 (McKee & Vale, 2017). 책형태의스웨덴수어사전은구화주의시대의영향아래 1916년처음출판되었다 (Österberg, 1916). 가장최근책으로출판된사전은스웨덴전국농인협회가 1997년 10년간의작업끝에편찬한것이다 (Svenskt teckenspråkslexikon, 1997). 스웨덴수어사전화작업은 1988년스톡홀름대학교언어학과수어부가처음시작했다. 이프로젝트의결과로 3,132개의수어표제어가담긴첫사전이 2001년스웨덴수어사전디지털버전이라는제목으로온라인에공개되었다. 이사전에는스웨덴전국농인협회가출판한책자사전에서따온수어동작영상파일이포함되었다. 하지만책자사전의기준을따르는것이수어사전을제한적인자원으로만든다는사실이곧명확해졌다. 목적이다른유연한사전을만들어낼필요가있었다. 그이후주요어휘데이터베이스인스웨덴수어사전 (Swedish Sign Language Dictionary) 이 2008 년에 62 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전편찬의과제 요한나메시 만들어졌으며지금까지개발이계속되고있다 (Mesch, Wallin, & Björkstrand, 2012). 현재스웨덴수어사전에는 17,800개의수어표제어와 4,070개의예시문장이포함되어있다 (2018년 9월기준 ). 이프로젝트와관련된사전화실무단은스웨덴수어의단어를기록하고온라인사전을지속적으로업그레이드한다. 사전편찬자들은자신들이속한언어관련기관뿐아니라다른자원을기반으로작업하는경우가많다. 어떤언어라도언어자원은해당언어나그언어에포함된사용역 (register) 을대표적으로보여주는것이중요하다. 그렇기때문에언어자원또한언어의기록문서로사용될수있는것이다. 더중요한것이있다. 수어사전은다양한목적에적합한자료여야하고수어학습자, 통역사, 교수자, 농인커뮤니티및다운증후군이나기타언어장애를가진사람들등여러집단에게좋은자원이될수있어야한다. 사전화작업과작업자 사전팀은농인인 3명의상근직원으로구성되어있으며스웨덴수어단어를기록하고온라인사전을계속해서업데이트하는일을한다. 사전화작업은장애가있는학생들을위해스톡홀름대학교가대학차원에서국가예산을통해자금을지원한다. ( 내부 ) 자문단도함께작업해왔다. 사전의규모가계속커지고있기때문에외부자문단이필요한상황이다. 따라서언어협의회와함께자문단이구성될예정이다. 데이터베이스는파일메이커프로 (FileMaker Pro) 데이터베이스파일로만들어졌고구축이계속되고있다. 스웨덴에서는사인뱅크 (Signbank) 가사용되지않는데, 가까운시일내에사인뱅크를연구도구로사용할계획이있기는하다. 현재사전화실무단은파일메이커와 ELAN을기술적으로연결할수있는방법을찾는중이다. 문장구성방법과관련해서, 작업자 ( 외부자문단위원 ) 는스웨덴어문자없이사람이수어동작을보여주는정지화면만을보게된다. 이는서로의미가다른여러수어동작간연관성을구축한다. 그후수어작업자는자신만의사용예시를문장의형태로고안해야한다. 이러한유도도좋기는하지만, 앞으로는이방식에사전편찬자가말뭉치에서관련문장을검색하고검색된문장을사용에적절한형태로변경하여사전에추가하는방식이더해질것이다. 사전의설계 주요어휘데이터베이스인스웨덴수어사전은 2008년만들어졌으며그이후인터페이스개발이계속되고있다. 스웨덴수어사전웹사이트는 https://teckensprakslexikon.su.se/ 이다. 기술적지원의목적은스웨덴수어사전이좀더사용자친화적이고아이폰이나안드로이드앱등기타플랫폼에서도사용가능하도록개발하는것이었다. 스웨덴수어사전에서수어동작은수형, 수향, 수위, 수동, 스웨덴어번역 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 63

2 부발표 과수어표기로설명된다. 사전의각항목은단어 ( 또는구절 ) 영상, 스웨덴어번역, 음운정보및음운또 는의미상동등한수어동작과의사전내교차연결 ( 동음어와동의어등, 사진 1-3) 로대표된다 (Svenskt teckenspråkslexikon, 2018). 스웨덴수어말뭉치주석자를지원하기위해 ID- 글로스가추가된다. 수어동작은세가지방식 ( 사진 4) 이나주제별로 (46개의하위분류 ) 검색가능하다 ( 가족, 스웨덴도시와지역, 의료서비스나지화등 ). 또한각수어 ( 일반및특수용어 ) 동작에 ID 번호가있다. 하지만 ID 글로스만으로는수어를완벽하게설명할수없다. 각수어동작에대한설명과번역도있다. 스웨덴수어사전에는또한첫스웨덴수어책자사전 (Österberg, 1916) 에포함된 415개의표제어와함께구수어동작이추가되었다. https://teckensprakslexikon.su.se/kategori/osterberg-1916 빈도는특정기한내에방문자가얼마나자주사전사이트를재방문했는지를알려준다. [ 사진 1.] 느린재생, 다시재생, 같은의미를가진다른동작, 이동작의다른의미옵션과함께동작을보여준다 [ 그림 2.] 문장의예를보여준다 64 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전편찬의과제 요한나메시 [ 그림 3.] 정지된화면사진 ( 프린트용 ) 을스웨덴수어사전용표기, ID 글로스, 동작설명, 말뭉치주석용글로스, 사전용글로스와함께보여준다 [ 그림 4.] 수형, 수향, 수위및수동으로검색한다 스웨덴수어사전은 L2 교육과자기학습의표준화된자료로사용된다. 학습목적으로사전을사용하는것은학습자, 가족및교육적환경에있어중요하다. 농인커뮤니티역시페이스북그룹덕분에사전을유용하게이용하고있다. 학습자를위한퀴즈또한만들어졌다. https://tspquiz.se/app/#/start 트위터봇인 Ett techken varje dag ( 하루에한동작 ) 이 2018년 1월에만들어졌으며학습자들사이에서반응이좋다.[https://twitter.com/allatecken]. 이트위터봇은스웨덴수어사전에서무작위로선택한수어동작을하나씩올린다 ( 그림 5). 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 65

2 부발표 [ 그림 5.] 하루에한동작트위터봇 수어사전을지원하는언어자원 스웨덴어수어말뭉치 현재일반대중에게공개된스웨덴수어의주요언어자원은두가지로사전과말뭉치이다 (Mesch, 2012; Mesch & Wallin, 2012, 2015; Mesch, Wallin, & Björkstrand, 2012; Mesch, Wallin, Nilsson, & Bergman, 2012). 두언어자원모두스웨덴수어를위한기록문서역할을한다. 스웨덴수어의첫말뭉치는 2004년 ECHO 프로젝트와연계되어만들어졌다 (O. Crasborn et al., 2007). 이말뭉치는스웨덴수어말뭉치 (SSLC) 로 24시간분량의영상데이터대화, 서술, 프레젠테이션으로구성되어있으며 2009~2011년수어사용자 42명으로부터수집된자료로릭스방켄스유빌레움스방크 (Riksbankens Jubileumsbank) 가자금을지원했다 (Mesch, Wallin, Nilsson, et al., 2012; Wallin & Mesch, 2018). 말뭉치구축, 특히주석작업은매우오랜시간이걸린다. 수집된 24시간분량의데이터중약절반정도에주석이달렸다 (2018년 9월기준 ). 그외스웨덴수어의말뭉치에는촉각수어말뭉치 (Mesch, 2016) 와스웨덴수어학습자말뭉치 (Schönström & Mesch, 2017) 가있다. 스웨덴촉각수어말뭉치는 8명의수어사용자로부터 2007년과 2011년수집된대화와유도로구성되어있다. 수집된 3시간 30 분분량의자료중약 40%( 수어토큰 6,000개 ) 에주석이달렸다. 스웨덴수어말뭉치와다른말뭉치의주석작업은주석자 10명 ( 농인 8명비농인 2명 ) 이비상근으로참여하고있으며여전히진행중이다. 스톡홀름대학언어학과수어부가수어사전과스웨덴수어말뭉치를고안및유지하고있지만, 처음에는서로독립적으로만들어졌다. 이로인해서두자료간기능적불일치가생겨났다. 하지만두프로젝트는작년에통합되기시작했고현재는이미서로를보완하는수준이다. 예를들어, 스웨덴수어말뭉치의수어토큰에는스웨덴수어사전 ID 번호가주석으로달리고 (Riemer Kankkonen, Björkstrand, Mesch, & Börstell, 2018) 수어동작이스웨덴수어말뭉치에포함되면스웨덴수어사전에도추가된다. 또한스웨덴 66 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전편찬의과제 요한나메시 수어사전온라인인터페이스는동의어검색결과를스웨덴수어말뭉치어휘빈도수인터페이스를통해계산된스웨덴수어말뭉치빈도수에따라분류한다 (Börstell & Östling, 2016). 스웨덴수어말뭉치데이터빈도수도구는경쟁적인이형수어간상대적빈도수를보여준다 ( 예 : 수어동작 TIO와 TIO(Y) 10 ). 스웨덴수어말뭉치가확대되면좋은언어자원이될것이다. https://www.ling.su.se/teckenspr%c3%a5ksresurser/verktyg/frekvens 스웨덴수어말뭉치는현재약 113,000개의수어토큰을포함하는데, 말뭉치데이터수집단계에서구체적인어휘변형유도작업이없었기때문에 (Stamp et al., 2014), 많은경우동의어나수어동작의변형이말뭉치데이터사용만으로는조사가불가능하다. 113,000개의수어토큰은언어의변형을체계적으로조사하기위해사용되는구화말뭉치보다는훨씬작지만수어말뭉치로는꽤큰규모다. 크라우드소싱 언어자원개선 사전, 즉스웨덴수어사전을개발하고개선하기위해우리는농인커뮤니티를참여시켜새로운수어동작을수집하고동의어와의유사성및동의어들간차이에대한데이터를수집한다. 바로우리의언어자원, 즉스웨덴수어사전을개선하기위한크라우드소싱이다. 두가지방식으로작업이진행된다 : a) 스웨덴수어사용자와수어동작사용, 어휘의차이, 그리고수어동작구성에대해에스엔에스 (SNS) 상에서토론하고 b) 스웨덴수어사용자와여러장소에서직접적으로교류하며 ( 전국농인기관및농인커뮤니티사업및활동연례참석 ) 수어와수어에대한판단을수집한다 (Riemer Kankkonen et al., 2018). Teckenspråkslexikon ( 수어사전 ) 이라는이름의페이스북그룹은 2014년 10월에시작되었으며사전팀이관리한다. 2,642명의회원이있으며 (2018년 5월기준 ) 회원들은다양한방식으로교류하면서어떤수어동작을사용해야하고수어동작변형간의차이가무엇인지등에대해토론한다. 회원들은스웨덴어나스웨덴수어영상을이용해서질문을올리거나답글을단다. 지난 2년간 (2016년 1월부터 2018년 1월까지 ) 593개의글이올라왔고 5,817건의교류가있었다. 여기에는 X를수어로어떻게하나요?(53%), Y사람에대한수어동작이있나요?(20%) 및사용방법에대한일반적인질문 (10%) 과특정수어동작의어원 (-0.2%) 등의질문이포함된다. 결론 수어사전을개발하고유지하는것은많은시간을요하는자료집약적인작업이다. 비디오파일을업데이 트하거나고장및해킹을방지하면서기술적인발전을지속하는것이중요하다. 사전은또한양질의내용과 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 67

2 부발표 사용자친화적인자원을유지하기위해연구및기술적지원이필요하다. 수어사전을업데이트하고개발하는방법에는언어지식, 이전영상내용등의기타자원, TV 프로그램과영상메시지, 그리고국가별수어에대한데이터를에스엔에스 ( 페이스북그룹등 ) 상에서농인커뮤니티로부터직접크라우드소싱하기등이포함된다. 수어의말뭉치규모는아직작은수준이라말뭉치데이터만으로는수어형태의변형을조사할수없다. 사전을위한크라우드소싱은언어자원을개선하는유용하고빠른방법이다. 오프라인데이터수집은사전팀에게개인적으로농인커뮤니티를만나교류할수있는기회를제공하는데이는언어기록화작업의중요한요소다. 그뿐만아니라개인과의직접적교류로수어동작이형에대해좀더질적인접근이가능할수도있다. 각수어사용자와관련된메타데이터는말뭉치등기록화작업에추가될필요가있는특정개인이나집단을대상으로한다. 말뭉치를기반으로한사인뱅크는수어동작의분류정리화를고려하는연구자들에게필요한도구다. 예를들어사인뱅크에수어동작글로스가변경되거나새롭게추가될경우, 연결되어있는 ELAN 주석파일을통해자동으로업데이트되어주석자와연구자들의작업이쉬워진다. 초기사전이나현재어휘데이터베이스 ( 스웨덴수어어휘데이터베이스 ) 는말뭉치의분류정리화를위한말뭉치프로젝트와함께사용되면충분히좋은자원이될수있다. 구어 / 문어복합의경우 ( 의학, 기술또는직업관련주제에서매우흔함.) 수어사전에표제어를어떻게넣을지는아직도명확하지않으며, 빈도수가높은생산적수어 ( 묘사어나분류사 ) 도사전에추가되어야한다. 참고문헌 Börstell, C., & Östling, R. (2016). Visualizing lects in a sign language corpus : Mining lexical variation data in lects of Swedish Sign Language. In E. Efthimiou, S. E. Fotinea, T. Hanke, J. Hochgesang, J. Kristoffersen, & J. Mesch (Eds.), Workshop Proceedings: 7th Workshop on the Representation and Processing of Sign Languages: Corpus Mining (pp. 13 18). Paris: ELRA. Crasborn, O., Mesch, J., Waters, D., Nonhebel, A., Van Der Kooij, E., Woll, B., & Bergman, B. (2007). Sharing sign language data online: Experiences from the ECHO project. International Journal of Corpus Linguistics, 12(4). McKee, R., & Vale, M. (2017). Sign language lexicography. In P. Hanks & G.-M. de Schryver (Eds.), International Handbook of Modern Lexis and Lexicography (pp. 1 22). Berlin, Heidelberg: Springer Berlin Heidelberg. http://doi.org/10.1007/978-3-642-45369-4_34-1 Mesch, J. (2012). Swedish Sign Language Corpus. Deaf Studies Digital Journal, 3. Retrieved from http://dsdj. gallaudet.edu/index.php?issue=4&section_id=2&entry_id=128 68 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

수어사전편찬의과제 요한나메시 Mesch, J. (2016). Dataset. Tactile Swedish Sign Language Corpus. Department of Linguistics, Stockholm University. Mesch, J., & Wallin, L. (2012). From meaning to signs and back: Lexicography and the Swedish Sign Language Corpus. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen, & J. Mesch (Eds.), Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Language Resources and Evaluation Conference (LREC)] (pp. 123 126). Paris: European Language Resources Association (ELRA). Retrieved from http://www.lrec-conf.org/ proceedings/lrec2012/workshops/24.proceedings_signlanguage.pdf Mesch, J., & Wallin, L. (2015). Gloss annotations in the Swedish Sign Language Corpus. International Journal of Corpus Linguistics, 20(1), 103 121. http://doi.org/10.1075/ijcl.20.1.05mes Mesch, J., Wallin, L., & Björkstrand, T. (2012). Sign language resources in Sweden: Dictionary and corpus. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen, & J. Mesch (Eds.), Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Language Resources and Evaluation Conference (LREC)] (pp. 127 130). Paris: European Language Resources Association (ELRA). Retrieved from http://www.lrec-conf.org/proceedings/lrec2012/ workshops/24.proceedings_signlanguage.pdf Mesch, J., Wallin, L., Nilsson, A.-L., & Bergman, B. (2012). Dataset. Swedish Sign Language Corpus project 2009--2011 (version 1). Sign Language Section, Department of Linguistics, Stockholm University. Retrieved from http://www.ling.su.se/teckensprakskorpus Österberg, O. (1916). Teckenspråket. Uppsala: P. Alfr. Persons förlag. Riemer Kankkonen, N., Björkstrand, T., Mesch, J., & Börstell, C. (2018). Crowdsourcing for the Swedish Sign Language Dictionary. In M. Bono, E. Efthimiou, S.-E. Fotinea, T. Hanke, J. Hochgesang, J. Kristoffersen, Y. Osugi (Eds.), Proceedings of the 8th Workshop on the Representation and Processing of Sign Languages: Involving the Language Community [Language Resources and Evaluation Conference (LREC)] (pp. 171 174). Paris: European Language Resources Association (ELRA). Schönström, K., & Mesch, J. (2017). Dataset. The project From speech to sign learning Swedish Sign Language as a second language. Department of Linguistics, Stockholm University. Svenskt teckenspråkslexikon. (1997). Leksand: Sveriges Dövas Riksförbund. Wallin, L., & Mesch, J. (2018). Annoteringskonventioner för teckenspråkstexter. Version 6, januari 2018. [Annotation guidelines for sign language texts]. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 69

Session 2 Challenges of creating a Sign Dictionary Johanna Mesch, Stockholm University, Sweden The article will describe how the Swedish Sign Language Dictionary has developed, and why it takes a long time to establish such a dictionary. A lexicographic work of Swedish Sign Language was initiated in 1988 at Stockholm University, and it resulted, in 2001, the first dictionary online. The Swedish Sign Language Dictionary was created in 2008 and has been in development since. When the direction of the corpus construction started in 2003, and when the corpus data, thanks to the threeyears project of the Swedish Sign Language Corpus 2009-2011, expanded with gloss annotations, a discussion has arisen about how the Online Swedish Sign Language Dictionary should continue to be in its development and in which direction, and how to use the SSL Corpus as a source of input for new signs and lexical variation in the SSL Dictionary. Keywords: Sign language dictionary, lexicography, Swedish Sign Language Introduction A sign language dictionary elevates the status for existence of sign language(s) in many countries. It is also a necessary tool for serving the needs of learners, teachers, families and researchers of signed languages (McKee & Vale, 2017). The first Swedish Sign Language Dictionary in a book format was published in 1916 in the shadow of the oralism period (Österberg, 1916). The latest dictionary in printed book format was published by the Swedish National Deaf Association in 1997 after ten years work (Svenskt teckenspråkslexikon, 1997). The lexicographic work of Swedish Sign Language was initiated in 1988 by the Sign Language Section of the Department of Linguistics at Stockholm University. In 2001, the first dictionary resulting from the project went online, entitled the Digital version of Swedish Sign Language Dictionary and included 3,132 sign entries. This dictionary includes video files for signs taken from a printed dictionary of the Swedish National Deaf Association. It soon became clear that following criteria of the printed dictionary was making it a limited resource. It needed to be a flexible dictionary for different purposes. 70 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Challenges of creating a Sign Dictionary Johanna Mesch A follow-up dictionary, the main lexical database, the Swedish Sign Language Dictionary was created in 2008 and has been in development since then (Mesch, Wallin, & Björkstrand, 2012). Today, it has approximately 17,800 sign entries and 4,070 sentence examples (September 2018). The lexicographic work group associated with this project documents the vocabulary of Swedish Sign Language and continuously updates the web-based dictionary. The work of the lexicographers is often based on their intuitions concerning language, but also on different sources. An important aspect of any language resource is that it is representative of the language or register that it covers. Because of this, the resources also serve as language documentation. This is perhaps even more important. A sign language dictionary should be a good resource for different purposes and different groups such as sign language learners, interpreters, sign language teachers, the deaf community, and other groups such as people with Downs syndrome and other language disorders. The lexicographic work and actors The dictionary team consists of three full-time employees, all Deaf, who document SSL vocabulary and continuously update the web-based dictionary. The work is financed by Stockholm University through national financing to support for disabled students at university level. There has been the (internal) advisory group. Because the dictionary has been growing bigger, an external advisory group is needed. With The Language Council, an advisory group will be established. The database has been created as FileMaker Pro database files and its construction is ongoing. Signbank is not used in Sweden, although it is planned to have a Signbank as research tool in near future, and the group is trying to find a possible technical connection between FileMaker and ELAN. With respect to the method for creating sentences, the actors (members of the external advisory group) are given still photos only, in which a person is presenting a sign, without any written Swedish words. This creates associations with many signs of different meanings. Then the sign actors have to devise their own usage examples in the form of sentences. This elicitation may be good, but in the future it will be combined with the possibility for lexicographers to look up related sentences in the corpus and change them to a suitable form for use in the dictionary. Dictionary design The main lexical database, the SSL Dictionary was created in 2008 and its interface has been in development 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 71

Session 2 since. The website for Swedish Sign Language Dictionary is https://teckensprakslexikon.su.se/ Technical support aimed to develop the dictionary to become more user-friendly and usable on different platforms such in as apps for iphone and Android. In the SSL Dictionary, a sign is described in terms of its handshape, orientation, location, movement, Swedish translations and sign transcription. Each dictionary entry is represented by a video of the sign (or phrase), a Swedish translation, phonological information, and internal cross-links to phonologically or semantically equivalent signs i.e. homophones and synonyms, see Figures 1-3 (Svenskt teckenspråkslexikon, 2018). ID-glosses are added for supporting SSL corpus annotators. A sign can be searched for using these ways (see Figure 4) or by subject (46 with subcategories), e.g. family, Swedish towns and provinces, healthcare or fingerspelled signs. There are also ID numbers for signs (in general and special terms). However, ID glosses alone cannot describe sign language perfectly. There are also descriptions and translations for each sign. The old signs with 415 sign entries from the first printed dictionary in SSL (Österberg, 1916) are also added to this lexical database. https://teckensprakslexikon.su.se/kategori/ osterberg-1916 Frequency refers to how often visitors return to the dictionary site within a time frame. [Figure 1.] Showing a sign with options for slow motion, play again, other signs with same meaning, this sign has also other meaning [Figure 2.] Showing some sentence examples 72 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Challenges of creating a Sign Dictionary Johanna Mesch [Figure 3.] Showing still photos (for printing) with sign transcription for SSL, ID gloss, sign description, gloss for corpus annotation, gloss for dictionary [Figure 4.] Searching via handshape, orientation, location and/or movement The SSL Dictionary is used as standardized learning resource for L2 teaching and self-learning. The use of the dictionary for learning purposes is important for learners, families and educational settings. The Deaf community has also found it valuable thanks to a Facebook group. A Quiz has also been created for learners, https://tspquiz. se/app/#/start. A twitter bot Ett tecken varje dag (One sign every day) was created in January 2018, and it has been well-liked by learners [https://twitter.com/allatecken]. It publishes a randomly selected sign from the SSL Dictionary, see Figure 5. [Figure 5.] Twitter bot One sign every day 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 73

Session 2 Language resources supporting for the sign language dictionary Swedish Sign Language Corpus There are currently two main language resources of SSL publicly available, a dictionary and a corpus (Mesch, 2012; Mesch & Wallin, 2012, 2015; Mesch, Wallin, & Björkstrand, 2012; Mesch, Wallin, Nilsson, & Bergman, 2012), both of which also serve as a form of language documentation for SSL. The first corpus of Swedish Sign Language was created in 2004 in connection to the ECHO project (O. Crasborn et al., 2007). The resource is the Swedish Sign Language Corpus (SSLC), which consists of 24 hours of video data conversations, narratives, and presentations) from 42 different signers collected in the years 2009 2011, financed by Riksbankens Jubileumsbank (Mesch, Wallin, Nilsson, et al., 2012; Wallin & Mesch, 2018). Corpus work, especially annotation work, is very time-consuming. Approximately half of the collected 24 hours of data have been annotated (September 2018). Other corpora in SSL are the Tactile Sign Language Corpus (Mesch, 2016) and the Learner Corpus in SSL (Schönström & Mesch, 2017). The Tactile SSL Corpus consists of dialogues and elicitation from 8 signers collected in 2007 and 201. Approximately 40 % (6,000 sign tokens) of the collected four hours and thirty minutes hours have been annotated. Annotation work of the SSL corpus and other corpora is still ongoing with ten (eight deaf and two hearing) annotators working on it part-time. Although both the SSL Dictionary and the SSL Corpus are designed and maintained by the Sign Language Section at the Department of Linguistics, Stockholm University, they were initially set up independently from each other. This has led to some discrepancies between the functions of the two resources. However, the two projects have started to converge in the last year, and now they are already supporting each other. For example, sign tokens in the SSLC are annotated with the SSLD ID number (Riemer Kankkonen, Björkstrand, Mesch, & Börstell, 2018), signs are added to the SSLD as they appear in the SSLC, and the SSLD online interface sorts synonyms search hits according to SSLC frequencies calculated through the SSLC lexical frequency interface (Börstell & Östling, 2016). This frequency tool for the SSLC data shows the relative frequencies of two competing sign variants, e.g. the signs TIO and TIO(Y) ten. It will be a good resource when the SSLC has been expanding. https://www.ling.su.se/teckenspr%c3%a5ksresurser/verktyg/frekvens The SSLC contains currently around 113,000 sign tokens, and, as there was no explicit lexical variation elicitation task during the collection of the corpus data (Stamp et al., 2014), many sign synonyms or form variations cannot yet be investigated solely with the use of corpus data. A set of 113,000 sign tokens is fairly large for a sign language corpus, although it remains much small when compared to spoken language corpora used to investigate variation systematically. 74 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Challenges of creating a Sign Dictionary Johanna Mesch Crowdsourcing - Improving language resources To develop and improve the dictionary, specifically the SSLD, we involve the Deaf community to collect new signs and to gather data on familiarity with and variation within sign synonyms that is, crowdsourcing to improve our language resources, specifically the SSLD. This works in two ways: a) discussing sign usage, lexical variation, and sign formation with SSL signers on social media; and b) interacting with SSL signers directly at various venues (the annual participation in Deaf events of national Deaf organizations and Deaf community business and activities) collecting signs and judgments about signs (Riemer Kankkonen et al., 2018). A Facebook group entitled Teckenspråkslexikon ( sign language dictionary ) was started in October 2014 and is administrated by the dictionary team. There are 2,642 members (in May 2018), who interact in different ways, discussing which signs are to be used, or the difference between sign variants. They can write questions or comments in Swedish or in SSL as video comments. In the last two years (i.e., January 2016 to January 2018), there have been 593 posts and 5,817 interactions. This includes questions about signs, such as How do you sign X? (53%), Is there a sign for person Y? (20%), general questions about the usage (10%) and etymology (<0,2%) of specific signs. Conclusions Developing and maintaining the sign language dictionary is time-consuming and resource-intensive. It is necessary to keep up with technical development, for example by updating video files, and protecting against crashes and hackers. The dictionary also needs research and technical support to maintain a good quality and user-friendly resource. The methods for updating and developing the sign language dictionary include language knowledge, other sources such as old video stories, TV programmes and video messages, and crowdsourcing data about the national sign language(s) directly from the Deaf community using social media, e.g. Facebook group. The corpus size for signed languages is still small, so form variations cannot yet be investigated solely with the use of corpus data. Crowdsourcing for the dictionary is a useful and rapid method for enhancing the language resource. Data collected through offline methods may provide a more qualitative approach to sign variants, interacting with individuals directly aside from giving the dictionary team an opportunity to personally meet and interact with the Deaf community, which is an important aspect of any language documentation work. Metadata about each signer target specific individuals or groups that are needed to add to the documentation work, for example, from the corpus. The corpus-based Signbank is a necessary tool for researchers considering sign lemmatization. When changes or additions of new sign glosses, for example, have been made to the Signbank, they are automatically updated through linked ELAN annotation files, which makes annotators and researchers work easier. Early dictionaries or current lexical databases, e.g. the SSL lexical database, are good enough sources, if they are used together with the corpus projects for lemmatization of a corpus. How to put a 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 75

Session 2 headword on a sign dictionary is still not clear when thinking about compounds from spoken/written language (very common in subject areas such as medicine, technical or vocational words), and the high frequency part of productive signs (depicting or classifier signs) should be added to the dictionary. References Börstell, C., & Östling, R. (2016). Visualizing lects in a sign language corpus : Mining lexical variation data in lects of Swedish Sign Language. In E. Efthimiou, S. E. Fotinea, T. Hanke, J. Hochgesang, J. Kristoffersen, & J. Mesch (Eds.), Workshop Proceedings: 7th Workshop on the Representation and Processing of Sign Languages: Corpus Mining (pp. 13 18). Paris: ELRA. Crasborn, O., Mesch, J., Waters, D., Nonhebel, A., Van Der Kooij, E., Woll, B., & Bergman, B. (2007). Sharing sign language data online: Experiences from the ECHO project. International Journal of Corpus Linguistics, 12(4). McKee, R., & Vale, M. (2017). Sign language lexicography. In P. Hanks & G.-M. de Schryver (Eds.), International Handbook of Modern Lexis and Lexicography (pp. 1 22). Berlin, Heidelberg: Springer Berlin Heidelberg. http://doi.org/10.1007/978-3-642-45369-4_34-1 Mesch, J. (2012). Swedish Sign Language Corpus. Deaf Studies Digital Journal, 3. Retrieved from http://dsdj. gallaudet.edu/index.php?issue=4&section_id=2&entry_id=128 Mesch, J. (2016). Dataset. Tactile Swedish Sign Language Corpus. Department of Linguistics, Stockholm University. Mesch, J., & Wallin, L. (2012). From meaning to signs and back: Lexicography and the Swedish Sign Language Corpus. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen, & J. Mesch (Eds.), Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Language Resources and Evaluation Conference (LREC)] (pp. 123 126). Paris: European Language Resources Association (ELRA). Retrieved from http://www.lrec-conf.org/ proceedings/lrec2012/workshops/24.proceedings_signlanguage.pdf Mesch, J., & Wallin, L. (2015). Gloss annotations in the Swedish Sign Language Corpus. International Journal of Corpus Linguistics, 20(1), 103 121. http://doi.org/10.1075/ijcl.20.1.05mes Mesch, J., Wallin, L., & Björkstrand, T. (2012). Sign language resources in Sweden: Dictionary and corpus. In O. Crasborn, E. Efthimiou, E. Fotinea, T. Hanke, J. Kristoffersen, & J. Mesch (Eds.), Proceedings of the 5th Workshop on the Representation and Processing of Sign Languages: Interactions between Corpus and Lexicon [Language Resources and Evaluation Conference (LREC)] (pp. 127 130). Paris: European Language Resources Association (ELRA). Retrieved from http://www.lrec-conf.org/proceedings/lrec2012/ 76 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Challenges of creating a Sign Dictionary Johanna Mesch workshops/24.proceedings_signlanguage.pdf Mesch, J., Wallin, L., Nilsson, A.-L., & Bergman, B. (2012). Dataset. Swedish Sign Language Corpus project 2009--2011 (version 1). Sign Language Section, Department of Linguistics, Stockholm University. Retrieved from http://www.ling.su.se/teckensprakskorpus Österberg, O. (1916). Teckenspråket. Uppsala: P. Alfr. Persons förlag. Riemer Kankkonen, N., Björkstrand, T., Mesch, J., & Börstell, C. (2018). Crowdsourcing for the Swedish Sign Language Dictionary. In M. Bono, E. Efthimiou, S.-E. Fotinea, T. Hanke, J. Hochgesang, J. Kristoffersen, Y. Osugi (Eds.), Proceedings of the 8th Workshop on the Representation and Processing of Sign Languages: Involving the Language Community [Language Resources and Evaluation Conference (LREC)] (pp. 171 174). Paris: European Language Resources Association (ELRA). Schönström, K., & Mesch, J. (2017). Dataset. The project From speech to sign learning Swedish Sign Language as a second language. Department of Linguistics, Stockholm University. Svenskt teckenspråkslexikon. (1997). Leksand: Sveriges Dövas Riksförbund. Wallin, L., & Mesch, J. (2018). Annoteringskonventioner för teckenspråkstexter. Version 6, januari 2018. [Annotation guidelines for sign language texts]. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 77

주제 4 대한민국수어사전의현황과발전방향 최혜원, 이현화한국, 국립국어원 Hyewon Choi, Hyunhwa Lee National Institute of Korean Language, Korea

2 부발표 대한민국수어사전의현황과발전방향 최혜원 이현화한국, 국립국어원 1. 머리말 한국수어언어법 (2016) 이시행된지 2년여의시간이흘렀다. 한국사회에서한국수어가언어로인정되고한국수어의사용권리를보장받기위한농인의오랜염원을담은 한국수어언어법 은한국수어의발전을위한다양한제도와사업활성화의근거가되었다. 2016년 한국수어언어법 이제정된이래대한민국은한국수어의언어로서의발전을위해다양한연구와사업을시행하고있다. 5년마다한국수어발전기본계획을체계적으로수립하고, 한국수어사용환경을개선하고자한국수어의사용실태조사, 한국수어말뭉치구축, 문법에관한기초연구등을본격적으로가동했다. 또한한국수어의교육및보급을위해한국수어교원자격제도를시행하면서대상별수어교재를개발하고한국수어문화학교프로그램을운영하고있다. 이와함께기존의 한국수어사전 을정비하는과정에서한언어의총체를담은한국수어사전으로의도약을위한발전적인방향을모색하고있다. 수어사전의편찬은수어의독립적지위를알리기위한가장기초적인작업이다. 음성언어의종속적인언어, 보조적인의사소통수단이라는통념을딛고언어로서의수어를기술한수어사전은농인들이독립된언어공동체를이루고있다는증거가된다. 실제로도여러나라에서언어사전으로서의수어사전을편찬하여농인공동체의언어적, 문화적지위향상의계기를마련하였다. 이때, 수어사전의형태는그간의음성언어중심의관점으로음성언어에대응하는수어를보여주는단순한대역어형태의사전을뛰어넘어야함은물론이다. 조선이일본에주권을빼앗긴일제강점기시대, 국어학자들을중심으로한글로된최초의조선어사전을편찬하려하였으나결국세상에먼저나오게된것은조선총독부가편찬한 조선어사전 (1920) 이었다. 사전명칭과는달리이사전은조선어를대상화한일본어중심의사전일뿐이었다. 1) 조선어가독립국가의언어가아닌상태에서편찬된사전은일본어표제어와이에대응하는한글표기, 그리고일본어뜻풀이 1) 이후문세영이라는국어학자의치열한노력으로 조선어사전 (1938) 이라는한글로표기된한국어사전이만들어져우리민족에게큰희망을주었다. 80 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 로구성되어, 조선인의일본어교육, 일본인의조선어교육 이라는식민지지배정책에활용되었다 ( 최경봉, 2014). 위의사례와마찬가지로대한민국에서그간만들어진한국수어사전들도진정한의미의언어사전으로서틀을갖추지못하였다. 한국수어에대한이해가깊지않은청인교육자들이한국어텍스트로된교과목을가르치기위한수단으로만든대역어형태의수어사전은그태생적한계를극복하지못한채현재까지답습되고있는실정이다. 농교육을목적으로한사전의출발점은서구 2) 에서도동일하지만 90년대전후유럽을중심으로언어학적접근을통해수어의자연언어로서의체계를담은진일보한수어사전을편찬하고있다. 우리나라는 한국수어언어법 을통해한국수어를언어로선언하였지만, 한국사회에서명실공히언어로서인정되기위해서는선언을뛰어넘어한국수어의언어로서의체계를밝히보이는사전이필요하다. 수어의무한한가능성을보이고한국농사회의정체성과자부심을뒷받침할수있는사전편찬을위해이번학술행사는매우의미깊은역할을할것이라기대한다. 이글에서는그간한국에서편찬된수어사전들의현황과문제점을살펴보고향후농인이실제사용하는한국수어자료를충실히담은새로운한국수어사전편찬을위해국립국어원에서수립한중장기계획을밝히도록하겠다. 2. 한국수어사전의편찬현황 수어사전이없던시기에수어어휘목록은주로수어보급을위해만든수어교재를통해볼수있었다. 1982년에야한국에서는처음으로수어사전이편찬되었는데 표준수어사전 이라는사전의명칭에서도확인할수있듯이이사전은 수어표준화 3) 에기여하기위한목적에서출발하였다 ( 이준우, 남기현, 조준모, 2018). 이목적에따라농학교교사들을중심으로구성된 한국표준수어사전편찬위원회 는일상생활수어어휘, 초 중등국어교과서의국어어휘에대응하는수어어휘를사전에담아냈다. 아래의표에서보듯이한국에서는그간 표준수어사전 (1982) 을시작으로총 6종의수어사전이편찬되었다. 4) 연번사전 / 편찬기관 ( 편찬자 ) 출판연도표제어미시구조 1 표준수어사전 표준수어사전편찬위원회 1982 5,494 개 2) 1989 년호주수어사전에서도영어의문법형태 ( 과거형분사인 -ed ) 에대응시킨수어표현의흔적이남아있다. 3) 여기서말하는 표준화 는일반적인의미의언어표준화와는다르다. 광범위한수어자료의수집과정이없었기때문에지역적변이형등여러형태로나타나는어형가운데가장일반적이면서널리통용되는어휘를대표형으로선정하는방식은아니었다. 한국어대응어가없는것은새로만들어내고, 동형이의어는변별을위해보조사인을첨가하는등음성언어의관점에서수어를 규격화 하는작업이대거포함되었다. 4) 어원사전, 전문용어사전등은목록에서제외하였다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 81

2 부발표 연번사전 / 편찬기관 ( 편찬자 ) 출판연도표제어미시구조 2 수어사전 단국대학교출판부 1983 5,966 개 3 표준수어 김승국 1993 5,966 개 4 한국수어사전 국립국어원 한국농아인협회 2005 6,800 여개 5 한국수어사전 김삼찬외 2005 9,818 개 6 한국수어사전 국립국어원 ( 온라인 ) 2015 12,793 개 < 표 1> 국내에서편찬된한국수어사전 위에서보듯이한글자모가나다순으로배열된 6개사전은한국어표제어와이에대응하는수어를담은대역어사전의형태이다. 대역어조차도실제사용되는수어만담은것이아니라한국어어휘를표현하는데에필요하다는이유로수어표현에는없는것을새로만들어올린경우가적지않았다. 이러한결과로과도한수어표제어가산출되었고, 그중에는한국어어미와조사, 합성어에대응하는다수의인위적인수어가포함되었다. 5) 일반적으로대역어사전은 A 언어표제어의여러의미를 B 언어의어휘나대체표현으로다양하게제시한다. 예를들어표제어가한국어 상하다 일경우 상하다 의여러의미와이에대응하는한국수어대역어정보를제시해야한다. 5) 1998 년네덜란드정부는네덜란드수어의법적지위를인정하기위한선결과제로네덜란드기본어휘표준화 (STABOL, Standardisation of the Basic Lexicon of NGT) 를내세웠다. 농사회에수어연구자들의거센반대에부딪히기도한이프로젝트에서표준화된어휘수는약 14,000 개를기록한다. 이중 33% 가의료나법률, 학교교과서용어등새로운어휘이다.(Roland Pfau, Markus Steinbach & Bencie Woll, 2012) 82 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 한국어표제어 의미 / 한국수어대역어 다치다 야위다 망가지다 상하다 상하다 < 표 2> 한국어 상하다 에대응되는여러개의수어표현 그러나 한국수어사전 은한국어표제어의대응수어를하나로국한하여한국어표제어의뜻을한국수어단어하나로제시하거나 (1, 5), 한국어표제어의사전뜻풀이를거의단일하게압축하여보이고있다 (4, 5, 6). 이는사전이용자에게한국어에하나의수어가대응되고, 한국어와한국수어의의미갈래가일치한다는한국수어에대한잘못된이해를불러일으킬수있다. 상하다 ' 상하다 ' < 그림 1> 한국어표제어 상하다 와글로스가일치하는한국수어의제시 대역어사전의틀에서기술되면서사전에서다루는수어정보는수어의형태적특징에한정될뿐그외수어의언어적특징은기술되지못하였다. 이마저도한국어에한국수어를무리하게연결시키다보니한국수어가왜곡되는현상을초래하였다. 예를들어 5의 목적지나일정한수준에다다름. 의뜻을지니는 도달 에대응하는수어의동작을보자. 수향이수직방향으로위로향해있으나 목적지 는반드시높은곳에있지않을수있다. 목적지 의위치특성에따라 도달 을표현하기위해수어화자는수향의변화를다양하게주는데사전에서는이를반영하지못하고있다. 또한구체화된표현이많은한국수어의특성상한국어표제어의뜻풀이와수어가대응되더라도한국어용례와는불일치하는경우가발견된다. 한국어 가장자리 ( 둘레나끝에해당되는부분 ) 에대응되는수어는한국어의일반적인뜻에서크게벗어나지않지만 눈가장자리가빨갛게되었다. 와같은용례를표현할때는부적절한표현이된다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 83

2 부발표 < 그림 2> 표제어 가장자리 에대응되는수어표현 한국수어사전이처음만들어진시기나사전의총수, 표제어수는다른어느나라에비교해봐도크게뒤지지않는다. 6) 그러나사전콘텐츠의충실성은이처럼별다른개선없이답보상태에머물고있다. 7) 이는 2005년 한국수어사전 까지진행된사전편찬작업이 수어표준화 의기본취지에서크게벗어나지못하였기때문이다. 이러한한계는 2005년 한국수어사전 의형태를그대로반영한, 현재까지서비스되고있는온라인형태의사전인국립국어원의 한국수어사전 8) (2015) 에서도마찬가지이다. 3. 한국수어사전의발전방향 앞서살펴본바대로한국수어표준화사업의결과물인 한국수어사전 은한국수어자료를모아연구하고수어연구방법론에대한환기를시켰다는점에서의의를찾을수있다. 그러나구축당시의시대적한계로인하여이사전은사용자들에게지속적인비판을받아왔다. 이에국립국어원은그간의비판을겸허히수용하고기존사전의한계를극복한새로운한국수어사전의편찬을계획하고있다. 효과적인사전설계는사용자의목적과능력의현실적인이해에서부터시작된다고한 Atkins & Rundell(2008) 의주장처럼국립국어원은 한국수어사전 개선방향에대한올바른설계를위하여여러사업을통해농사회의의견을수렴하였다. 9) 2014년에구성된 한국수어연구자문위원회 에서는전국의한국수어사용자 553명을대상으로 수어전문용어제정및표준화수요조사 를진행하였다. 이조사에따르면참여자들은수어표준화가추진된전문분야수어를사용하지않는이유로 실제사용되는수어와다름 (28.1%) 에제일많이응답하였다. 수형기반한국수어사전구축 ( 국립국어원, 2016) 사업에서는농인 15명을대상으로초점집단인터뷰 6) 초기의뉴질랜드수어사전 (The Dictionary of New Zealand Sign Language, 1986) 은 1,200 개, 호주수어사전 (Auslan Dictionary, 1989) 은 3,500 개, 영국수어사전 (Dictionary of british sign language, 1992) 은 1,800 개의표제어가수록되어있다. 7) 1960 년대스토키를중심으로한갤러뎃대학연구자들의미국수어에대한언어학적연구방법론은 1970 년대중반영국과유럽전역의연구자들에게영향을주었고, 1980 년대이후수어는전세계적인연구분야의중심에섰다. 8) 한국수어사전 (2015) 에는 한국수어사전 (2005) 의어휘목록과함께이후동일한사전편찬방향하에연차별작업으로추가된 6 천여개의어휘가포함되어있다. 9) 이글에서는조사결과중에서기술에필요한의견일부를선택적으로제시하였다. 84 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 를실시하였는데그결과기존사전이 농사회에서자주사용되는수어가부재 한상태이며 한국어대응수어표현이중심이되다보니농인들로부터외면 받고있다고하였다. 그리고이사전이 수어통역사들의교육용으로활용 되고 청인들의수어학습용으로활용 되고있지만 농인조차이해하기어려운수어사전 이라 농인과청인수어통역사간소통의문제가발생 한다고하였다. 농인들은현사전을 필요할때이용하지만큰도움을주지않는수어사전 이며 농인이이해하기어려운표현의한국어정보 로한국어를이해하기위해사전을이용하는것에도제한이있는것으로평가하고있었다. 이어 2017년진행된 한국수어사전수어뜻풀이와용례구축및정비 ( 국립국어원 ) 사업을통해수어학계와농사회의수어관련전문가 30인을대상으로설문조사를실시하였다. 조사결과 한국수어사전 의단점으로거시구조측면에서는낮은접근성과인위적으로만들어진수어가등재되어있다는점이주로지적되었고, 미시구조측면에서는정보항목간의낮은일치도와수어를활용하기위해반드시필요한용례, 문법정보등의부재가지적되었다. 더불어 2016년과 2017년에각각진행된두조사에서는 한국수어사전 의발전방향에대해제언을받기도하였는데주요내용은다음과같다. ㅇ수형기반한국수어사전구축 ( 국립국어원, 2016) - 농인사용자중심의새로운사전제작 - 한국어정보를농인의관용적수어표현으로 10) 쉽게풀이해서설명 - 수어사전의목적과이용대상에대한명확한설정필요 - 접근성향상을위한방안ㅇ한국수어사전수어뜻풀이와용례구축및정비 ( 국립국어원, 2017) - 농인과청인이모두사용할수있는사전 - 현사전에서인위적인표제어의삭제및정비 - 농인이실제로사용하는수어를수록하고수어관련정보를충실히담은사전 - 수어의다양한변이형을담은사전 - 사전의접근성을높이기위한표제어분류및검색방법 ( 움직임인식등 ) 개발 이상의결과로볼때현재 한국수어사전 의이용자는한국수어학습자와한국어학습자로나뉘며각그룹은다른목적으로사전을이용하고있다. 또한이사전이한국수어표현을인위적으로만들어사전에등재하였으나이것이농사회에받아들여지지못해사전으로서의역할을하기에부족한상황임을알수있다. 따라서이러한의견을바탕으로 한국수어사전 을정비하여사전의일치도와완성도를높이는한편이용자들이목적에따라사전을이용할수있도록새로운사전을편찬할필요가있다. 10) Cokely, D. & Baker-Shenk, K, C. (1980) 는미국수어 (ASL) 의설명에서 관용어 (idioms) 로표현된것들의대부분은사실관용어가아니고관용어정의에맞지않는다고하며단순히몇개의다른용법을가진수어들이라고하였다. 이정옥, 이준우 (2005) 도한국의이와같은현상을정리하며 자연수어 혹은 농식수어 라는명칭으로관용표현을받아들이고있음을지적하고있다. 따라서본고에서인용한조사의결과도같은맥락으로해석할수있다. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 85

2 부발표 Hartmann(1983) 은외국어를학습하고교육하는보조도구로서이언어사전이유용하게사용될수있다고하였다. 또한한국수어기술사전편찬의필요성에대해논의한이현화 (2017) 는한국수어사전이인위적으로가공된자료가아닌한국농인의자연스러운발화를담고있는실제자료에근거하여표제어를추출하고이를바탕으로편찬되어야한다고하였다. 이러한논의들을바탕으로볼때한국수어사전의편찬방향은다음두가지로귀결된다. 첫째, 사전은사용목적을분명히하여 한국수어-한국어사전, 한국어-한국수어사전 으로이원화구축되어야한다. 둘째, 변이가많은수어의특성을담은사전은기술언어학에근거하여야하며따라서한국수어의실증적인자료를모은한국수어말뭉치를바탕으로한사전편찬이이루어져야한다. 3.1. 양방향이중언어사전중장기구축계획 한국수어사전 의사용자와전문가집단의의견과제언등을종합적으로고려하여국립국어원이수립한 새로운한국수어사전의중장기구축계획은다음과같다. 3.1.1. 양방향이중언어사전의구조 (1) 거시구조 한국수어-한국어사전ㅇ표제어의선정 - 한국수어말뭉치 ( 15~) 와추가로구축할사전용말뭉치 ( 예정 ) 를바탕으로표제어선정 - 한국수어의고정된어휘부 (frozen lexicon) 를중심으로생활용어, 지역어, 신어등수록ㅇ표제어표기방법 - 한국수어의형태를기록할수있는수어문자 (SignWriting) 로수어를표기하여한국수어가한국어와다른독자적인언어라는인식제고 - 글로스 (gloss) 사용으로인해서발생하는여러문제점개선ㅇ표제어의배열 - 수형기반한국수어사전구축 ( 국립국어원, 2016) 의수어소분류를참조하여배열 한국어-한국수어사전ㅇ표제어의선정 - 한국농인의한국어학습에필요한표제어 10,000개평정 - 한국수어-한국어사전에한국어대응표현으로수록된한국어표제어를중심으로수록ㅇ표제어표기방법 - 표제어는한글로표기하되한국어표제어에대응하는한국수어표현은섬네일 (Thumbnail) 로제시 86 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 ㅇ표제어의배열 - 가나다순 (2) 미시구조 한국수어-한국어사전ㅇ형식항목 - ( 표제어 ) 기본형과변이형모두확인할수있도록영상으로제시 - ( 마우스제스처 ) 수어를할때필수적으로나타나는마우스제스처를그림으로제시 - ( 수어문자 ) 표제어의형태를수어문자로기록 - ( 수형그림 ) 수어소검색시활용할수있도록수형그림제시 ㅇ내용항목 - ( 대응표현 ) 한국수어에대응하는복수의한국어표현을제시 - ( 뜻풀이 ) 한국수어말뭉치를바탕으로추출한한국수어의의미를한국어로쉽게이해할수있도록기술 - ( 용례 ) 표제어의다양한활용예를수어동영상으로제시 제시된용례를수어문자로표기하고이를클릭할경우해당표제어로이동할수있도록연동기능구현 정확하고자연스러운한국어로번역 - ( 관련어 ) 표제어와일정한의미관계를맺는한국수어를제시 - ( 형태는비슷하지만의미가다른수어 ) 한국수어학습자들이빈번하게나타내는학습오류유형으로, 수어학습의효과를높이기위해관련정보제공 - ( 문법정보 ) 수어의복수형, 부정형, 활용형등문법정보제시 - ( 수어지도 ) 지역변이가매우발달한수어의특성에따라수어지도를통해지역어정보제공 < 그림 3> 한국수어 - 한국어사전정보항목 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 87

2 부발표 < 그림 4> 한국수어 - 한국어사전시범화면 한국어-한국수어사전ㅇ형식항목 - ( 표제어 ) 1만개의규모로표제어제시 - ( 활용 ) 표제어의활용형을제시하여한국어학습효과제고 - ( 품사 ) 한국어표제어의품사제시 88 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 ㅇ내용항목 - ( 대응표현 ) 표제어에대응하는복수의한국수어표현을섬네일 (Thumbnail) 로제시 - ( 뜻풀이 ) 한국어기초사전의뜻풀이제시 - ( 용례 ) 한국어기초사전의용례제시 - ( 한국수어로풀이 ) 번역이아닌농인의이해를기반으로하여한국어표제어의뜻풀이와용례를하나의영상안에서설명 - ( 관련어 ) 한국어표제어와유의, 반의관계를맺고있는단어를관련어로제시 < 그림 3> 한국수어 - 한국어사전정보항목 < 그림 6> 한국어 - 한국수어사전시범화면 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 89

2 부발표 3.1.2. 양방향이중언어사전구축계획 구분 1 단계 ( 20~ 22) 2 단계 ( 23~ 25) 3 단계 ( 26~ 28) 4 단계 ( 29~ 31) 한국수어말뭉치구축 한국수어자료수집 한국수어자료수집및가공 한국수어말뭉치구축 - 서울외지역말뭉치구축 - 기본전사 - 의미전사 사전용말뭉치기초연구및자료구축 - 의미전사 - 변이형전사 - 문법정보전사 - 마우스제스처전사 사전용말뭉치구축 - 기본전사 - 의미전사 한국수어말뭉치구축 - 기본전사 ( 계속 ) 한국수어말뭉치구축 - 기본전사 ( 계속 ) 멀티미디어자료구축 한국수어 - 한국어사전편찬기개발및기능고도화 한국수어지식관리시스템구축 한국수어자료통합관리시스템고도화 한국수어 - 한국어용례구축시스템개발및기능고도화 한국수어 - 한국어사전서비스시스템구축및기능고도화 - 한국어어휘망활용검색도입및수형검색기능고도화 - 사용자화면편집기능개발 한국어 - 한국수어사전편찬기개발및기능고도화 한국어 - 한국수어사전서비스시스템구축및기능고도화 한국수어뜻풀이영상관리시스템개발및기능고도화 한국어 - 한국수어사전서비스시스템구축및기능고도화 ( 이원화된사전간연결 ) - 모바일웹인터페이스구축 < 한국수어 - 한국어사전 > 양방향 사전 편찬 < 한국수어 - 한국어사전 > - 사전시범제작 - 한국수어표제어구축및지침마련 - 한국수어의미구축및지침마련 - 한국수어뜻풀이구축및지침마련 - 한국수어용례구축및지침마련 - 한국어대응표현구축및지침마련 - 한국수어뜻풀이구축및지침마련 - 수어동영상제작및감수 < 한국수어 - 한국어사전 > - 한국수어표제어추가구축 - 한국수어변이형구축및지침마련 - 수어문자구축및지침마련 - 수형정보구축및지침마련 - 한국수어의미구축확장 - 한국수어뜻풀이확장구축 - 한국수어용례확장구축 - 한국어대응표현구축 - 관련어구축및지침마련 - 수어동영상제작및감수 - 문법정보구축및지침마련 - 수어지도구축및지침마련 - 마우스제스처정보구축및지침마련 - 사용성평가및오류정비 - 수어문자교육및보급 - 오류정비 < 한국어 - 한국수어사전 > - 한국어표제어 1 만개평정 - 한국수어뜻풀이, 용례선정및지침마련 - 한국어의뜻풀이와용례를한국수어로풀이 - 한국수어대응표현구축및지침마련 < 한국수어 - 한국어사전 > - 수어문자교육및보급 - 오류정비 < 한국어 - 한국수어사전 > - 한국어의뜻풀이와용례를한국수어로풀이 - 한국수어로한국어의뜻풀이확장구축 - 한국어의용례수어번역확장구축 - 한국수어대응표현구축 - 한국수어대응표현동영상제작 - 관련어구축 < 표 3> 양방향이중언어사전중장기구축계획 90 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 오랜시간과대규모의작업범위로농사회의자원이총투입되는이계획에따르면새로운한국수어사전 은 2031 년에완성될예정이다. 향후편찬될사전은다음과같은부분에서이전의사전과는차별화된모습 으로도약할것이라고기대해볼수있다. ㅇ이용자의사용목적에따른사전이원화편찬수어사전을편찬한대부분의나라들이이중언어사전을편찬하였지만현실적인제한 ( 사업기간, 예산등 ) 으로인해하나의사전으로두목적을이루는방향을취하고있다. 이러한사전은주로수어사용자에게는수어에해당하는음성언어대역어를제공하고, 수어학습자에게는수어와관련된정보를제공한다. 물론이것은두가지목적을동시에이룬다는장점이있으나여전히하나의사전으로모든정보를제공하기에는제한적일수밖에없다. 따라서학습자사전으로서의효용성을높이기위해서는두목적을분리한사전의편찬이필요하다. 국립국어원에서계획하고있는사전은목적에따라이원화한것으로 한국수어-한국어사전 은한국수어사용자를위한표현용사전이며, 한국수어학습자를위한이해용사전이다. 반면에 한국어-한국수어 사전은한국어학습자를위한이해용사전이며한국수어학습자를위한표현용사전이다. ㅇ증거가있는사전지금까지한국에서편찬된 6종의수어사전은대체로비슷한방법으로자료를수집하였는데그방법은다음과같다. 첫째, 한국어표제어목록을선정하고이에대응하는한국수어를수집하거나수어표현이존재하지않는경우에는수어를만든다. 둘째, 기존에나와있는수어교재또는사전에서수어단어를표집한다. 이렇게수집한수어를사전으로엮었지만이용자입장에서는어느수어가만들어진것인지, 어느수어가농인이실제로사용하는것인지등을확인할수없어혼란이가중되었고결국사전의신뢰도가하락하였다. 이에수어를기술하는사전으로편찬방향을설정하고말뭉치에근거하여표제어등사전정보를추출하고자한다. 수어를정의하는부분에서중요한것은수어를제1언어로사용하는사람 (native signer) 이해당수어를맥락안에서어떻게사용하는지를보는것이다 (Johnston,1998). 따라서새로운사전이이용자에게단어의적절한용법을가르쳐줄수있도록용례를제공하고자한다. 용례는말뭉치에서추출한용례를부분적으로가공한가공례와농인의직관에의해제작한작성례로구분한다. 이를제시할경우그출처를표시하여사전이용자들이참고할수있도록할예정이다. ㅇ변이수용 Johnston(2003) 은수어에관해서, 어휘적변이 (lexical variation, 다양한수어사용자들이같은의미를완전히다른수어로나타내는것 ) 는몇몇의수어사전편찬자들에게주요한문제가된다고하였다. 그러나이러한변이를없애기위해추진하는수어표준화가얼마나실현가능성이낮은지여러이유를근거로제시하였다. 역사적으로우리나라뿐만아니라다른나라에서도시도되었던음성언어중심의수어표준화는농사회로부터외면을받아왔다. 그러므로이러한변이를없애려는시도보다변이를모두수용하고기록하 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 91

2 부발표 여사전이수어언어의총체적자원이되도록해야할것이다. ㅇ수어문자 (SignWriting) 도입수어는문자체계가없는언어로이것은사전편찬을더어렵게만드는요인이된다. 수어사전편찬자들은이러한어려움을극복하고자여러접근을하고있는데 Kristoffersen & Troelsgård(2010) 에따르면편찬된수어사전들은수어표제어를나타내기위해동영상, 그림, 사진, 형태표기 (HamNoSys, SignWriting 등 ), 글로스 (glosses), 숫자, 대역어, 어떻게수어가산출되는지글로묘사하기와같은다양한방법을사용하고있다. 대부분은글로스를사용하는데이것은수어의의미를나타낸음성언어의어휘이며, 수어를기억하기위해사용된다. 음성언어의어휘를하나의라벨처럼수어표제어에부여하고관리하는것은사전편찬자와이용자모두에게상당한편의를제공하지만해당수어가음성언어어휘의의미와완전히동일하다거나혹은그어휘가가지고있는여러의미중에서수어는단하나의의미만을가지고있다는오해를일으키기도한다. 또한음성언어라는라벨을보고수어에접근하는것은결국음성언어를통해수어를인지하도록해서수어학습에방해요소가된다. 수어를정확하게기술하고, 학습자로하여금이를시각적으로인식할수있도록하기위해서향후나올사전에서는수어문자 (SignWriting) 를사용해수어를표기하고자한다. 이미미국, 호주, 벨기에, 브라질, 캐나다등여러나라에서수어문자를사용하고있으며 11) 체코수어사전 12) 에서는수어문자를사용하여표제어를표기하고있다. 수어문자는한국의국가사업으로는처음도입되는것으로이를계기로하여수어로생산되는많은정보가기록, 전승되는발판을마련하게될것이다. ㅇ번역이아닌설명한국수어가하나의언어로인정받기이전의시기에는이에대한이해부족으로한국어에한국수어어휘만대응시킨일대일번역이주를이루었다. 그러나 청각장애학생을위한국어교육용한국어-수어동영상사전개발 ( 윤점룡외, 2009) 에서지적한바처럼시각언어인수어는청각음성언어인한국어와는다른문법적특성을가지고있으므로단어를일대일로대응시켜번역하는것만으로는의미를제대로전달할수없다. 이와같은오류를피하기위하여국립국어원은 2017년에한국어의뜻풀이를한국수어로시범번역하였고 13) 문헌고찰과전문가의견수렴을통해지침을수립하였다. 그러나번역시한국수어가한국어에강하게견인되고, 한국어와한국수어간의설명방식차이가상당하여수어사용자들이번역문을이해하지못하는경우가발생하였다. 이와같은경험에근거하여수어사용자가사전의뜻풀이를직관적으로이해할수있도록번역이아닌설명방식의뜻풀이를제공하고자한다. 또한대부분의경우예시를들어설명하는농인의어법을충실히따라뜻풀이와용례를분리하지않고하나의흐름으로제작하여상호보완적역할을하도록하며이를통해수어사용자의접근성과활용도를높일것이다. 11) http://www.signwriting.org/about/who/ 12) http://www.dictio.info/czj?action=page&page=front&lang=en 13) 한국수어사전수어뜻풀이와용례구축및정비사업 92 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

대한민국수어사전의현황과발전방향 최혜원, 이현화 4. 결론 지금까지한국수어사전의편찬현황을개관하고발전방향을제시하였다. 현시점에서그간편찬된한국수어사전은진정한의미의언어사전으로서틀을갖추지못하였다고할수있다. 그러나국립국어원은 한국수어언어법 제정 (2016) 이후에관련정책과연구를위한기틀을마련하고있다. 더불어한국수어사전이언어지식의총체로거듭날수있도록다각도로접근하고있다. 이러한계획이적절히실현되기위해서는대규모의인력이필요하나그간농사회에서연구인력양성과그에대한투자가충분하지못했다. 이러한제한으로사전구축초기에는다소시행착오를겪을수도있겠으나지속적으로연구인력양성이병행된다면농인의언어그리고농인과청인이공존하는방향을담은사전에한발더다가갈수있을것이라기대해본다. 참고문헌 김삼찬, 이선호외 36인 (2005), 한국수어사전, 형설출판사. 김승국 (1993), 표준수어, 오성출판사, 서울. 국립국어원 (2014), 한국수어연구자문위원회회의결과보고서, 국립국어원. 국립국어원 (2016), 수형기반한국수어사전구축, 국립국어원. 국립국어원 (2017), 한국수어사전수어뜻풀이와용례구축및정비, 국립국어원. 국립국어원, 한국농아인협회 (2005), 한국수어사전 김승국 (1983), 수어사전, 단국대학교출판부윤점룡, 정희선, 김선영, 원성옥, 김지숙, 장은숙, 김경진, 임진이, 하길종, 양국종 (2009), 청각장애학생을위한국어교육용한국어-수어동영상사전개발, 한국연구재단. 이정옥, 이준우 (2005), 한국수어관용표현에관한탐색적연구, 특수교육저널 : 이론과실천 제6권제3 호, 한국특수교육문제연구소, 335-354쪽이준우, 남기현, 조준모 (2018), 한국수어사전편찬방향을위한제언, 특수교육저널 : 이론과실천 제19 권제1호, 한국특수교육문제연구소, 137-156쪽이현화 (2017), 한국수어정비사업 : 한국수어사전 ( 한국수어의기록 ), 새국어생활, 제27권제2호, 67-83쪽 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 93

2 부발표 최경봉 (2014), 원고로남은최초의우리말사전, 말모이, 새국어생활 제24권제3호, 국립국어원, 20-37 표준수어사전편찬위원회 (1982), 표준수어사전, 금옥학술문화재단허일 (2003), 한국농인의언어사전편찬에있어서의쟁점연구, 제5회한국수어학회학술대회. Atkins, B.T.S. & M. Rundell. (2008), The Oxford Guide to Practical Lexicography, Oxford/New York: Oxford University Press. Cokely, D, & Baker-Shenk, K, C. (1980), American Sign Language: A Teacher s Resource Text on grammar and Culture, Gallaudet University Press. Johnston, T.(1998). Signs of Australia. A new dictionary of Auslan(Australian Sign Language) (the sign language of the Australian deaf community), North Rocks, NSW, Australia: North Rocks Press, Royal Institute for Deaf and Blind Children. Johnston, T.(2003). Language Standardization and signed Language Dictionaries, Sign Language Studies, vol 3, Published by Gallaudet University Press, 431-468. Kristoffersen, J. H., & Troelsgård, T. (2010), Compiling a Sign Language Dictionary: Some of the Problems faced by the Sign Language Lexicographer In M. Mertzani(Ed.), Sign Language Teaching and Learning: Papers from 1st Symposium in Applied Sign Linguistics(Applied Sign Linguistics), Bristol: University of Bristol, Centre for Deaf Studies. Roland Pfau & Markus Steinbach & Bencie Woll(2012), Sign Language: An International Handbook, de Gruyter Mouton R.R.K. Hartmann (1983), 사전편찬의원리와실제 ( 서태길, 김양진, 도원영, 이상혁, 권오희, 옮김 ). 서울 : 제이앤씨.( 원서출판 1983) 국립국어원한국수어사전 (http://sldict.korean.go.kr/front/main/main.do) 수어문자누리집 (http://www.signwriting.org/about/who/) 한국민족문화대백과사전 (http://encykorea.aks.ac.kr/) 체코수어사전 (http://www.dictio.info/czj?action=page&page=front&lang=en) 94 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee National Institute of Korean Language, Korea 1. Introduction It has been approximately 2 years since the <Korean Sign Language Act> (2016) was implemented. The <Korean Sign Language Act>, embodying a long-standing desire of the deaf who had hoped Korean sign language to be recognized as a language in Korean society and to be granted with the right to use Korean sign language, became the basis for various systems and projects that support the development of Korean sign language. Since the enactment of the <Korean Sign Language Act> in 2016, the Republic of Korea has carried out a wide range of research and projects for the development of Korean sign language including systematic establishment of Korean sign language development plan in every 5 years; research on the use of Korean sign language to improve the environment for the use of Korean sign language; creation of Korean sign language corpus; and full-scale basic research on grammar. In addition, a qualification system for Korean sign language teachers has been put in place along with development of sign language textbooks for different targets and Korean sign language cultural school programs for education and distribution of Korean sign language. The Republic of Korea is also seeking for the ways to improve Korean Sign Language Dictionary to a dictionary that conveys the summation of the language. The compilation of a sign language dictionary is the most fundamental project to publicize the independent status of the sign language. A sign language dictionary that describes sign language as a language, against a common notion that sign language is a sub-language of a spoken language and a subsidiary communication means, proves that the deaf form an independent linguistic community. In fact, many countries have provided opportunities to enhance the status of the deaf community culturally and linguistically by publishing a sign language dictionary. Of course, the form of a sign language dictionary in this case should be more than just a translation dictionary that simply presents sign language equivalents to spoken language words from a spoken language-oriented perspective. When Joseon, an old Korean dynastic kingdom, was a colony of Japan, Korean linguists tried to compile the 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 95

Session 2 first Joseon language dictionary in Korean alphabet. However, Joseon Language Dictionary (1920) became the first dictionary published by the Government-General of Joseon. Unlike the title, the dictionary was centered around Japanese language, targeting Joseon language. The Joseon language 1) dictionary, which was compiled while Joseon was not an independent country, consisted of Japanese entries, Korean equivalents in Korean alphabet and definition of Japanese entries. The dictionary was used for a colonization policy to teach Japanese language to Koreans and Korean language to Japanese (Choi Kyeong-bong, 2014). Just as the case above, Korean sign language dictionaries published in the Republic of Korea so far did not have the right frame as a true language dictionary either. Sign language dictionaries in the form of a target language dictionary were made by hearing teachers without deep understanding of Korean sign language as a means to teach a curriculum in Korean texts. This type of dictionaries still continue to follow the old format without being able to overcome their inherent limitations. Sign language dictionaries in western countries 2) had a similar starting as they were for education of the deaf. However, more advanced sign language dictionaries began to be compiled mostly in Europe around the 1990s to present the system of sign language as a natural language from a linguistic approach. The Republic of Korea also announced the Korean sign language as a language through the <Korean Sign Language Act>. Nevertheless, to ensure that a sign language is recognized as a legitimate language in Korean society, the country needs to go beyond publicizing and should compile a dictionary that clearly shows the system of Korean sign language as a language. This conference is expected to play a pivotal role in presenting infinite potential of sign language and creating a dictionary that supports the identity and pride of the Korean deaf community. This article will look into the current status and challenges of the sign language dictionaries published in the Republic of Korea and the mid- & long-term plan established by the National Institute of Korean Language for the compilation of a new Korean sign language dictionary that faithfully includes actual Korean sign language data used by the deaf. 2. Current Status of Korean Sign Language Dictionary When there was no sign language dictionary, a list of sign language words was included in sign language textbooks created for the distribution of sign language. It was not until 1982 that the first sign language dictionary was published in the Republic of Korea under the title of Standard Sign Language Dictionary. As the title 1) After the publication, a Koren dictionary titled as Joseon Language Dictionary (1938) in Korean alphabet was created thanks to the hard work by Moon Se-young, a Korean linguist, and it gave a great hope to the Koreans. 2) An Australian sign language dictionary from 1989 has a trace of applying English language grammar ( -ed for past particle) to sign language expressions. 3) The 'standardization' here is different from general standardization of a language. As there was no collection of a broad range of sign language documents, the standardization was not about selecting the most widely used form as the standard from variations. Rather, it focused on regularizing sign language from a perspective of spoken language by creating new sign language equivalents of Korean language if there was no equivalent and adding supplementary signs to distinguish same signs with different meanings. 96 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee suggests, this dictionary was created to contribute to standardizing sign language 3) (Lee Jun-woo, Nam Ki-hyun & Cho Jun-mo, 2018). The Korean Standard Sign Language Dictionary Compilation Committee consisting of school teachers was formed to this end and the committee compiled the dictionary which contained sign language words for daily life and the sign language equivalents of Korean words used in elementary and middle school textbooks. As seen in the table below, a total of 6 sign language dictionaries have been published in the Republic of Korea starting from <Standard Sign Language Dictionary> (1982). 4) No. Dictionary/ Institution(compilerr) Year No. of entries Micro structure 1 Standard Sign Language Dictionary Standard Sign Language Dictionary Compilation Committee 1982 5,494 2 Sign Language Dictionary Dankook University Press 1983 5,966 3 Standard Sign Language Kim Seung-kuk 1993 5,966 4 Korean Sign Language Dictionary NIKL Korea 2005 Around 6,800 5 Korean Sign Language Dictionary Kim Sam-chan and others 2005 9,818 6 Korean Sign Language Dictionary National Institute of Korean Language (online) 2015 12,793 <Table 1> Korean sign language dictionaries published in the Republic of Korea 4) Dictionaries for etymology and terminology are excluded. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 97

Session 2 As seen above, the 6 dictionaries presented in Korean alphabetical order are translation dictionaries that contain Korean entries and sign language equivalents. Even the translation in these dictionaries included artificially created signs in addition to actually used sign language simply because it was necessary in expressing Korean words. As a result, an excessive number of sign language entries were compiled including many artificial sign language expressions corresponding to suffixes, prepositions and compound words of Korean language. 5) A translation dictionary in general provides various meanings of an A language entry in B language as words or alternative expressions. For instance, if an entry is Sanghada (go bad) in Korean, a translation dictionary should propose different meanings of Sanghada along with the information on Korean sign language translations. Korean language entry Meaning / Korean sign language translation Get hurt Become thin Get ruined Go stale Go bad <Table 2> Various sign language expressions equivalent to Sanghada in Korean language However, Korean sign language dictionaries either limit the sign language equivalent of a Korean language entry to one and suggest only one Korean sign for a Korea entry (1,5) or simplify the definition of a Korean language entry (4,5,6). This could cause a misunderstanding about Korean sign language among the dictionary users that one sign is the equivalent of one Korean word and that Korean language and Korean sign language share the identical semantic domain. Sanghada Sanghada <Picture 1> Korean language entry Sanghada and Korean sign language with the same gloss 5) The government of the Netherland in 1998 has set Standardisation of the Basic Lexicon of NGT (STABOL) as a priority task to recognize the legal status of Dutch sign language. The project, opposed greatly by sign language researchers in the deaf community, standardized approximately 14,000 words. 33% of them were new words including terminologies for medical and legal sectors as well as for school textbooks. (Roland Pfau, Markus Steinbach & Bencie Woll, 2012) 98 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee As these dictionaries were based on the frame of a translation dictionary, the information on sign language was limited to the morphological features of sign language rather than embracing linguistic characteristics. Furthermore, the unnatural linkages of Korean sign language with Korean language in these dictionaries resulted in distorting Korean sign language. For instance, the sign language for Dodal, meaning reaching a destination or a certain level in Korean in dictionary 5 shows that the hand orientation is vertically upward although destination is not always above. Sign language users change the hand orientation to express Dodal depending on the locational features of a destination, but the dictionary does not reflect such usage. Also, due to the characteristics of Korean sign language that has a great number of detailed expressions, a sign language gesture may not correspond to Korean language examples even if the gesture is equivalent to the definition of the Korean entry provided in the dictionary. For example, the sign language for Gajangjari (circumference or edge) in the dictionary is not particularly wrong in general meaning, but is inadequate to be used in an expression as, The edge of an eye is red. <Picture 2> A sign language expression equivalent to an entry Gajangjari Korean sign language dictionaries are not far behind compared to the dictionaries of other countries in terms of the time it was first compiled, the number of dictionaries and the number of entries. 6) Nevertheless, Korean sign language dictionaries are in a stalemate without much improvement when it comes to how faithfully the content reflects the actual language. 7) This is because the compilation process that had continued up to the creation of Korean Sign Language Dictionary in 2005 did not develop much from the basic intention of the standardization of sign language. This limitation is also seen in Korean Sign Language Dictionary 8) (2015) compiled by the National Institute of Korean Language, which is still in service online, as it reflects the format of Korean Sign Language Dictionary created in 2005. 6) The first sign language dictionary in New Zealand(The Dictionary of New Zealand Sign Language, 1986) included 1,200 entries while the first sign language dictionaries in Australia(Auslan Dictionary, 1989) and Britain(Dictionary of british sign language, 1992) contained 3,500 and 1,800 entries, respectively. 7) The linguistic research methodology for American sign language by researchers of Gallaudet University led by Stokoe in the 1960 affected the researchers in the UK and Europe in the mid-1970s, and sign language became the major subject of research after the 1980s. 8) Korean Sign Language Dictionary (2015) includes the lexicon from Korean Sign Language Dictionary (2005) and 6,000 other words that were added each year under the same direction of compiling dictionary. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 99

Session 2 3. Development Direction of Korean Sign Language Dictionary As has been described, Korean Sign Language Dictionary, which was a result of the Korean sign language standardization project, is meaningful in that it collected and studied Korean sign language materials and called attention to the methodologies of sign language research. However, because of the limitations of the time when it was published, the dictionary has continued to receive criticism from its users. The National Institute of Korean Language has humbly accepted the criticisms and is now planning to compile a new Korean sign language dictionary that overcomes the limitations of the previous dictionary. According to Atkins & Rundell(2008), designing an effective dictionary begins from the realistic understanding of the objectives and capabilities of users. As such, the National Institute of Korean Language has collected opinions from the deaf community through various projects to set the right direction for the improvement of Korean Sign Language Dictionary 9). The Korean Sign Language Research Advisory Committee, established in 2014, conducted a Survey on the demand for enactment and standardization of sign language terminology. According to the survey, respondents answered that the biggest reason for not using standardized sign language terminologies was because it was different from the sign language actually being used (28.1%). The Creation of Korean sign language dictionary based on hand orientation project (National Institute of Korean Language, 2016) performed a targeted group interview with 15 deaf, and the result showed that the existing dictionaries were lacking sign language expressions frequently used in the deaf community and disregarded by the deaf as they are centered around sign language expressions that are equivalents to Korean language. The interviewees also answered that the existing dictionaries were used for educating language interpreters and teaching sign language to hearing people, but were difficult even for the deaf to understand which could cause communication problems between the deaf and hearing sign language interpreters. The deaf saw the existing dictionaries as dictionaries they use when necessary but not of a great help and evaluated that even using the dictionaries to understand Korean language was limiting because the information provided in Korean language was difficult for the deaf to understand. The Establishment and modification of definitions and examples of sign language in Korean sign language dictionaries project (National Institute of Sign Language) in 2017 performed a survey targeting 30 sign language experts in the sign language academia and deaf community. The result suggested that the drawbacks of Korean Sign Language Dictionary were poor accessibility and artificially created sign language expression from a macro-structure perspective. For a micro structure, poor matching of information items and lack of examples and information on grammar that are absolutely necessary for the use of sign language. The two studies performed each in 2016 and 2017 also collected suggestions for the development direction of Korean Sign Language Dictionary, and the major details are as follows. 9) This article selectively addresses the opinions that are necessary from the survey results. 100 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee Creation of Korean sign language dictionary based on hand orientation (National Institute of Korean Language, 2016) - Creation of a new dictionary centered around the deaf users - Easy explanation of Korean language information in sign language expressions commonly used by the deaf 10) - Clear designation of the objectives and target users of a sign language dictionary - Measures to improve accessibility Establishment and modification of definitions and examples of sign language in Korean sign language dictionaries (National Institute of Korean Language, 2017) - Dictionaries that can be used by both the deaf and hearing people - Deletion and modification of artificial entries in existing dictionaries - A dictionary that contains sign language actually used by the deaf and faithfully includes sign language-related information - A dictionary that contains diverse variations of sign language - Development of methods to categorize entries and search for words (motion recognition, etc.) in order to enhance accessibility of dictionary Considering the results described above, the users of the current Korean Sign Language Dictionary can be divided into Korean sign language learners and Korean language learners, and the two groups are using the dictionary with different objectives. Also, the dictionary created and included artificial Korean sign language expressions, but these were not accepted by the deaf community. As such, the dictionary can not perform its role as a dictionary. It is necessary to revamp Korean Sign Language Dictionary based on these opinions to ensure that it reflects the actual language better and it is more complete while compiling a new dictionary so that the users can use a dictionary according to their objectives. According to Hartmann(1983), a bilingual dictionary can be useful as a supportive tool to learn and teach a foreign language. In addition, Lee Hyun-hwa (2017), who discussed the need for publishing a descriptive dictionary for Korean sign language, argued that a Korean sign language dictionary should be compiled based on the entries extracted from the materials that contain natural dialogue of the Korean deaf rather than artificiallymade materials. These suggestions lead to a conclusion that the direction of a Korean sign language dictionary should follow the two below. First, the Korean sign language dictionary should have a clear objective and should be dualized into Korean sign language-korean language dictionary and Korean language-korean sign language dictionary. 5) Cokely, D. & Baker-Shenk, K, C. (1980) noted, in explaining American sign language (ASL), that most of expressions that are described as idioms are actually not idioms and do not fit the definition of idiom and that they are simply sign language gestures that have a few different uses. Lee Jeong-ok & Lee Jun-woo (2005) also summarize the same case in Korea, pointing out that idiomatic expressions are accepted as natural sign language or sign language of the deaf. The result of the survey mentioned in this article can be interpreted within the same context. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 101

Session 2 Second, the dictionary that embodies the characteristics of a sign language with a wide range of variations should be based on descriptive linguistics, and, as such, it should be centered around the Korean sign language corpus that has a collection of empirical materials of Korean sign language. 3.1. Mid- & long-term plan to establish a two-way bilingual dictionary The mid- & long-term plan to publish a new Korean sign language dictionary established by the National Institute of Korean Language in consideration of the opinions and suggestions by the users of Korean Sign Language Dictionary and expert groups is as follows. 3.1.1. Structure of two-way bilingual dictionary (1) Macro structure Korean sign language Korean language dictionary Entry selection - Select entries from the Korean sign language corpus ( 15~) and an additional corpus to be established exclusively for dictionary (tentative) - Include daily vocabularies, dialects and new words focusing on the frozen lexicon of Korean sign language Notation of entry - Notate sign language in sign writing (SignWriting) that could document the form of Korean sign language to raise an awareness that Korean sign language is an independent language from Korean language - Address various issues caused by the use of gloss Entry arrangement - Refer to the classification of chereme in Creation of Korean sign language dictionary based on hand orientation (National Institute of Korean Language, 2016) for arrangement Korean language Korean sign language dictionary Entry selection - Evaluate 10,000 entries necessary for the Korean deaf to learn Korean language - Focus on Korean entries included in the Korean sign language-korean language dictionary as Korean language equivalents Notation of entry - Notate entries in Korean letters (Hangul) but suggest Korean sign language expressions equivalent to Korean language entries in thumbnails Arrangement of entry - In Korean alphabetical order 102 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee (2) Micro structure Korean sign language - Korean language dictionary Format items - (Entry) Provide via a video to show both default and variations - (Mouth gesture) Present mouth gestures, essential for sign language, in a picture - (Sign writing) Document the form of entries in sign writing - (Hand shape picture) Provide pictures of hand shape to be used for searching for chereme Content items - (Equivalent expression) Propose multiple Korean language expressions that are the equivalents of a Korean sign language expression - (Definition) Describe the meaning of Korean sign language extracted from the Korean sign language corpus in easy Korean language - (Example) Suggest various examples of entries via a video Notate the provided examples in sign writing and link the notation with the entry concerned so that users can click the link to find the entry Translate examples into accurate and natural Korean language - (Relevant words) Suggest Korean sign language gestures that are semantically related to an entry - (Morphologically similar signs with different meanings) Provide information on the signs that are frequently mistaken by Korean sign language learners for more effective sign language learning - (Information on grammar) Present information on grammar such as plural, negative, conjugation, etc. of sign language - (Sign language map) Include information on dialects via sign language map considering the characteristics of a sign language that has a wide range of regional variations. KSL-KL dictionary information items Format items Entry Mouth gesture Sign writing Picture of hand orientation Content items KL equivalentsmouth Definition Examples Relevant words Information on grammar Similar signs with different meanings Multi-media materials Sign language map <Picture 3> Korean sign language-korean language dictionary information items 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 103

Session 2 <Picture 4> A demonstration of Korean sign language-korean language dictionary Korean language-korean sign language dictionary Format items - (Entry) Suggest around 10,000 entries - (Usage) Propose usages of entries for more effective Korean language study - (Word class) Provide word class of Korean language entries 104 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee Content items - (Equivalent expression) Propose multiple Korean sign language expressions that are the equivalents of an entry in thumbnail - (Definition) Present definitions from a basic Korean language dictionary - (Examples) Provide examples from a basic Korean language dictionary - (Explanation in Korean sign language) Explain the definitions and examples of Korean language entries via a video based on the understanding by the deaf rather than translation - (Relevant words) Suggest vocabularies that are synonyms or antonyms of Korean language entries as relevant words KL-KSL dictionary information items Format items Entry Usage Word class Content items KSL equivalents Definition Example Explanation in KSL Relevant words <Picture 5> Korean language-korean sign language dictionary information items <Picture 6> A demonstration of Korean Language-Korean Sign Language Dictionary 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 105

Session 2 3.1.2. Publication plan for two-way bilingual dictionary Phase 1 ( 20~ 22) Phase 2 ( 23~ 25) Phase 3 ( 26~ 28) Phase 4 ( 29~ 31) Collect KSL materials Collect and modify Korean sign language materials Establish KSL corpus - Create corpus for Seoul and other regions - Basic transcript - Semantic transcript Conduct research for the corpus for dictionary and accumulate materials Establish KSL corpus - Semantic transcript - Transcript of variations - Transcript of grammatical information - Mouth gesture transcript Create corpus for dictionary - Basic transcript - Semantic transcript Accumulate multi-media materials Create KSL corpus - Basic transcript (continued) Create KSL corpus - Basic transcript (continued) Establish KSL Knowledge management system Improve KSL material integration and management system Develop a system to build examples of KSL KL and enhance its functions Develop KSL-KL dictionary compilation system and enhance its functions Establish KSL-KL dictionary service system and enhance its functions - Introduce search functions utilizing Korean language lexical network and enhance search functions based on hand orientation - Develop user interface editing function - Establish mobile web interface Develop KL-KSL dictionary compilation system and enhance its functions Establish KL-KSL dictionary service and enhance its functions Develop a management system for KSL definition videos and enhance its functions Establish KL-KSL dictionary service system and enhance its functions (connect the two dictionaries) <KSL-KL dictionary> Publish two-way dictionaries <KSL-KL dictionary> - Create a dictionary for trial - Establish KSL entries and guidance - Establish KSL meanings and guidance - Establish KSL definitions and guidance - Establish KSL examples and guidance -Establish equivalent KL expressions and guidance - Establish KSL definitions and guidance - Create and edit sign language videos <KSL-KL dictionary> - Establish additional KSL ' entries - Establish KSL variations and guidance - Establish sign writing and guidance - Establish information on hand shape and guidance - Establish and expand meaning of KSL - Expand KSL definitions - Expand KSL examples -Establish equivalent KL expressions - Establish relevant words and guidance - Create and edit sign language videos - Establish grammatical information and guidance - Establish sign language map and guidance - Establish information on mouth gesture and guidance - Evaluate usability and correct errors - Teach and distribute sign writing - Correct errors <KL-KSL dictionary> - Evaluate 10,000 KL entries - Select KSL definitions and examples and establish guidance - Explain definitions and examples of KL in KSL - Establish equivalent KSL expressions and guidance - Create relevant words <KSL-KL dictionary> - Teach and distribute sign writing - Correct errors <KL-KSL dictionary> - Explain definitions and examples of KL in KSL - Expand KL definitions in KSL - Expand KSL translations of KL examples - Establish equivalent KSL expressions - Create videos for equivalent KSL expressions <Table 3> Mid- & long-term plan for compilation of two-way bilingual dictionary 106 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee The plan, a long-term large-scale project resourced by the deaf community, states that the new Korean sign language dictionary will be completed in 2031. The dictionary that will be published in the future can be expected to be differentiated and more advanced from the previous dictionaries with the following reasons. Dualized compilation of dictionary according to objectives of users The majority of the countries that have published sign language dictionaries have bilingual dictionaries, but these dictionaries attempt to fulfill two different objectives with one dictionary due to realistic limitations (project period and budget). Such dictionaries provide sign language users with the spoken language equivalents of sign language and sign language learners with sign language-related information. This satisfies two objectives simultaneous, but there are limitations in providing all information with only one dictionary. Therefore, to make it more useful as a dictionary for learners, two separate dictionaries with different objectives should be compiled. The dictionary planned by the National Institute of Korean Language is dual dictionaries - Korean sign language-korean language dictionary supports Korean sign language users for expression and Korean sign language learners for understanding. Meanwhile, Korean language-korean sign language dictionary is to support Korean language learners for understanding and Korean sign language learners for expression. Dictionary with evidence The 6 dictionaries published in Korea collected materials in a similar method. First, a list of Korean entires was selected and Korean sign language equivalents were collected or created if there were none. Second, sign language words were sampled from existing sign language textbooks or dictionaries. The collected sign language words were included in a dictionary, but users could not tell which words were artificially created and which were actually used by the deaf. This has led to more confusion and undermined the credibility of a dictionary. Therefore, the National Institute of Korean Language is to compile a dictionary that describes sign language and extract information such as entries based on a corpus. One important thing in defining a sign language is to observe how native signer who use the sign language as the first language use the sign language concerned in context (Johnson, 1998). Therefore, the new dictionary is to provide examples so as to teach adequate usage of a word to the users. Examples will be divided into adapted examples that are extracted from a corpus then partially processed and editorial examples that are intuitively created by the deaf. Examples will have their original sources marked so that the users can refer to them. Acceptance of variations According to Johnston(2003), lexical variation (expression of a same meaning in completely different sign language gestures by sign language users) is a major issue for some sign language dictionary compilers. However, it has been proposed that standardization of sign language that aims to eliminate the variation is unrealistic due to a number of reasons. Historically, sign language standardization based on a spoken language 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 107

Session 2 has been disregarded by the deaf community not only in Korea but in other countries. As such, it is more desirable to accept and document all variations rather than remove them to ensure that a dictionary can be a comprehensive resource for sign language. Introduction of SignWriting Sign language is a language without a writing system, which makes it more difficult to publish a dictionary. Compilers of sign language dictionaries are approaching this issue in various ways to overcome the difficulty. According to Kristoffersen & Troelsgård (2010), the existing sign language dictionaries use different ways to present sign language entries such as video, picture, photo, transcript of form (HanNoSys, SignWriting, etc.), gloss, number, translation and written description of how a sign language gesture is created. In most cases, these dictionaries use gloss, which is a spoken language word that represents the meaning of a sign and is used to remember sign language. Assigning a spoken language word to a sign language entry as if a label offers a great convenience to both compilers and users, but it can cause a misunderstanding that the sign language entry concerned is completely identical to the meaning of the spoken language word or that the sign language entry only means one particular meaning out of the various meanings of the spoken language word. Also, approaching a sign language based on the label of a spoken language makes learners recognize the sign language through the spoken language which impedes sign language learning. For accurate description of sign language and visual recognition of the description by learners, the new Korean sign language dictionary will be using SignWriting to transcribe the sign language. It is already being used in many countries including the US, Australia, Belgium, Brazil and Canada 11), and the Czech sign language dictionary 12) presents entries in SignWriting. SignWriting will be introduced for the first time in Korea s national project and it will lay a foundation for a wide range of information created in sign language to be documented and passed down. Explanation, not translation When Korean sign language was not recognized as a language, Korean language words were paired with the sign language lexicon for one-on-one translation due to the lack of understanding. However, as mentioned in Development of a Korean language sign language video dictionary for Korean language education of students with hearing impairment (Yoon Jeom-ryong and others, 2009), Sign language, which is a visual language, has different grammatical characteristics from Korean language, which is a spoken language. Therefore, one-on-one translation of the lexicon can not fully convey the meaning of the sign language. In order to avoid such errors, the 11) http://www.signwriting.org/about/who/ 12) http://www.dictio.info/czj?action=page&page=front&lang=en 13) Establishment and modification project of sign language definitions and examples in Korean sign language 108 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee National Institute of Korean Language translated the definition of Korean language into Korean sign language in 2017 as a trial 13) and established a guidance after studying documents and collecting opinions from experts. However, the result showed that Korean sign language was strongly led by Korean language in translation, and there were some cases where sign language users could not understand translation due to a significant disparity in the ways Korean language and Korean sign language explain meanings. Based on this experience, the new dictionary will contain definitions that are explanations, not translations, so that sign language users can intuitively understand definitions in the dictionary. Also, it will ensure that definitions and examples are provided coherently, rather than separately, in order to faithfully follow the way the deaf present explanation with examples and enhance its accessibility and usability for sign language users. 4. Conclusion This article has looked into the status of Korean sign language dictionary compilation and proposed its development direction. The Korean sign language dictionaries that have been published up to date are not equipped with the appropriate frame as true language dictionaries. However, the National Institute of Korean Language has established a foundation for relevant policies and research since the enactment of the <Korean Sign Language Act> (2016). The institute is in need of a large-scale manpower to materialize these plans, but the deaf community has not been able to generate a sufficient number of researchers or make investment in fostering researchers. Due to such limitations, the project to compile a dictionary may come across some difficulties in the beginning. Nevertheless, continued fostering of researchers in the future will allow us to take a step closer to the dictionary that embodies the real language of the deaf for coexistence of the deaf and hearing people. Reference Kim Sam-chan, Lee Seon-ho & 36 others (2005), Korean Sign Language Dictionary, Hyungseul Publishing Co. Kim Seung-kuk (1993), Standard Sign Language, Osung Publishing House, Seoul. National Institute of Korean Language(2014), Korean Sign Language Research Advisory Committee Report, National Institute of Korean Language. National Institute of Korean Language(2016), Creation of Korean sign language dictionary based on hand orientation, National Institute of Korean Language. 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 109

Session 2 National Institute of Korean Language(2017), Establishment and modification of definitions and examples of sign language in Korean sign language dictionaries, National Institute of Korean Language. National Institute of Korean Language, Korea Association of the Deaf (2005), Korean Sign Language Dictionary Kim Seung-kuk (1983), Sign Language Dictionary, Dankook Press Yoon Jeom-ryong, Jeong Hee-seon, Kim Seon-young, Won Seong-ok, Kim Ji-sook, Jang Eun-sook, Kim Kyeong-jin, Lim Jini, Ha Gil-jong & Yang Kook-jong (2009), Development of a Korean language sign language video dictionary for Korean language education of students with hearing impairment, National Research Foundation of Korea. Lee Jeong-ok & Lee Jun-woo (2005), An Exploratory Study on the Idiomatic Expressions of Korean Sign Language, Special Education Journal: Theory and Practice Vol. 6, No. 3, Research Institute of the Korea Special Education, p.335-354 Lee Jun-woo, Nam Ki-hyun & Cho Jun-mo (2018), A Proposal for the Direction of Korean Sign Language Compilation, Special Education Journal: Theory and Practice Vol. 19, No. 1, Research Institute of the Korea Special Education, p.137-156 Lee Hyun-hwa (2017), Korean Sign Language Alignment Project : Korean Sign Language Dictionary (Documentation of Korean Sign Language), Sae kugo saenghwal, Vol. 27, No. 2, p.67-83 Choi Kyeong-bong (2014), The First Korean Language Dictionary in Manuscript, Malmoi, Sae kugo saenghwal Vol. 24, No. 3, National Institute of Korean Language, 20-37 Standard Sign Language Dictionary Compilation Committee (1982), Standard Sign Language Dictionary, Kumok Research and Culture Foundation Heo Il (2003), A Study on Challenges in Compiling Language Dictionary for the Korean Deaf, 5th Symposium of Korea Sign Language Society Atkins, B.T.S. & M. Rundell. (2008), The Oxford Guide to Practical Lexicography, Oxford/New York: Oxford University Press. Cokely, D, & Baker-Shenk, K, C. (1980), American Sign Language: A Teacher s Resource Text on grammar and Culture, Gallaudet University Press. Johnston, T.(1998). Signs of Australia. A new dictionary of Auslan(Australian Sign Language)(the sign language of the Australian deaf community), North Rocks, NSW, Australia: North Rocks Press, Royal Institute for Deaf and Blind Children. Johnston, T.(2003). Language Standardization and signed Language Dictionaries, Sign Language Studies, vol 3, Published by Gallaudet University Press, 431-468. 110 2018 국립국어원국제학술대회 보이는언어의기록, 수어사전

Current Status and Development Direction of Korean Sign Language Dictionary Hyewon Choi, Hyunhwa Lee Kristoffersen, J. H., & Troelsgård, T. (2010), Compiling a Sign Language Dictionary: Some of the Problems faced by the Sign Language Lexicographer In M. Mertzani(Ed.), Sign Language Teaching and Learning: Papers from 1st Symposium in Applied Sign Linguistics(Applied Sign Linguistics), Bristol: University of Bristol, Centre for Deaf Studies. Roland Pfau & Markus Steinbach & Bencie Woll(2012), Sign Language: An International Handbook, de Gruyter Mouton R.R.K. Hartmann (1983), Lexicography : principles and practice (translated by Seo Tae-gil, Kim Yang-jin, Doh Won-young, Lee Sang-hyuk, Kwon Oh-hee). Seoul:J&C.(Wonseo Publishing Co. 1983) National Institute of Korean Language s Korean Sign Language Dictionary (http://sldict.korean.go.kr/front/main/main.do) SignWriting website (http://www.signwriting.org/about/who/) Encyclopedia of Korean Culture (http://encykorea.aks.ac.kr/) Czech sign language dictionary (http://www.dictio.info/czj?action=page&page=front&lang=en) 2018 NIKL Conference Records of Visible Language: Sign Language Dictionary 111

주제 5 미래의수어사전 토마스항케독일, 함부르크대학교 Thomas Hanke Hamburg University, Germany

2024 © docsplayer.org 개인정보보호 | 약관 | 지원