PAC3 at JALT 2001 Conference Proceedings MENU Text Version Help & FAQ International Conference Centre Kitakyushu JAPAN November 22-25, 2001 Developing a One Million Word Spoken EFL Learner Corpus Yukio Tono Meikai University Tomoko Kaneko Showa Women’s University Hitoshi Isahara Communication Research Laboratory Emi Izumi Communication Research Laboratory Toyomi Saiga Communication Research Laboratory Emiko Kaneko ALC Press This paper will report on the progress of a project to compile a one-million-word spoken corpus of Japanese learners of English. In 1999, we launched a project to compile a new learner corpus in collaboration with ALC Press Inc. and Communications Research Laboratory. The major characteristic of the project is that the corpus data is entirely based upon the audio-recordings of an English oral proficiency interview test called ACTFL-ALC Standard Speaking Test. One of the unique features of the corpus is that each speaker’s data include his or her proficiency profile based on the SST evaluation schemes. This makes it possible for corpus users to study the learner language across different proficiency levels, a feature which has TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS not often been available for other learner corpora. In this paper, we will describe the project by summarizing the data collection procedure, text format, transcription guidelines, and annotation schemes (especially error tagging schemes), as well as the theoretical and pedagogical implications of the project. information฀on฀common฀learner฀errors.฀Projects฀such฀ as฀ICLE฀(International฀Corpus฀of฀Learner฀English;฀see฀ Granger฀1998)฀and฀JEFLL฀(Japanese฀EFL฀Learner)฀ Corpus฀(see฀Tono฀2000)฀both฀aim฀to฀compile฀learner฀ corpora฀to฀describe฀the฀interlanguage฀of฀particular฀L2฀ 本研究は１００万語規模の日本人英語学習者の話し言葉コ learner฀groups฀using฀a฀corpus-linguistic฀methodology.฀ ーパス構築プロジェクトの中間報告である。１９９９年に To฀date,฀however,฀most฀of฀these฀learner฀corpus฀projects฀ 我々は出版社アルクおよび通信総合研究所の協力を得て are฀composed฀of฀written฀data฀only,฀while฀the฀few฀spoken฀ 新しいコーパスを作成する試みを開始した。このプロジェ クトの最大の特徴は、コーパスがACTFL-ALC Standard learner฀corpora฀that฀do฀exist฀(e.g.฀LINDSEI฀project฀at฀ Speaking Test と呼ばれる英語会話能力面接試験の録音テ Louvain,฀described฀in฀De฀Cock,฀et฀al.฀1999),฀are฀rather฀ ープに基づいている点である。統一されたテスト判定結果 small฀in฀size.฀ がデータ内に話者属性として与えられているため、学習者 We฀launched฀our฀Standard฀Speaking฀Test฀(SST)฀corpus฀ グループを能力別に分けてサブコーパス検索などが可能に project฀in฀1999.฀This฀project฀is฀a฀joint฀collaboration฀ なる、という大きな特徴を持っている。本論では、データ 収集方法、フォーマット、書き起こしガイドライン、注釈 between฀Communication฀Research฀Laboratory฀ 付け（特にエラータグの仕様）などについて概説し、この and฀ALC฀Press,฀with฀a฀few฀advisory฀members฀from฀ プロジェクトの理論的、教育的意義について述べる。 universities.฀While฀English฀is฀taught฀over฀six฀years฀in฀ secondary฀schools฀in฀Japan,฀many฀of฀us฀feel฀that฀Japanese฀ he฀paper฀reports฀on฀an฀on-going฀project฀to฀ learners฀still฀cannot฀function฀properly฀in฀English฀for฀ compile฀a฀one-million-word฀spoken฀corpus฀ communication฀purposes.฀It฀has฀been฀argued฀that฀one฀ of฀Japanese฀learners฀of฀English฀and฀provides฀ of฀the฀problems฀of฀English฀language฀teaching฀in฀Japan฀is฀ possible฀implications฀for฀SLA฀and฀ELT฀research฀as฀well฀ that฀no฀serious฀attempt฀has฀been฀made฀to฀systematically฀ as฀language฀engineering.฀Recently,฀SLA฀researchers฀and฀ record฀and฀describe฀the฀acquisition฀process฀of฀Japanese฀ language฀teaching฀professionals฀have฀begun฀to฀realize฀ learners฀of฀English฀in฀our฀EFL฀context.฀It฀is฀very฀ the฀importance฀of฀learner฀corpora฀as฀resources฀for฀ important฀to฀know฀objectively฀how฀much฀English฀we฀ teaching฀and฀research,฀and฀major฀dictionary฀publishers฀ have฀acquired฀after฀six฀years฀of฀instruction,฀and฀the฀ such฀as฀Longman฀and฀Cambridge฀University฀Press฀ progression฀of฀standards.฀Mere฀adoption฀and฀application฀ have฀already฀compiled฀their฀own฀learner฀corpora฀in฀ of฀teaching฀methods฀from฀foreign฀countries฀will฀not฀ order฀to฀enrich฀their฀dictionary฀content฀by฀providing฀ always฀work฀in฀our฀country.฀We฀need฀to฀gather฀data฀ T PAC3 at JALT2001 872 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING on฀how฀Japanese฀learners฀learn฀English฀and฀to฀describe฀ the฀developmental฀path฀of฀their฀interlanguage.฀Thus,฀ one฀of฀the฀main฀purposes฀of฀this฀project฀is฀to฀identify฀ the฀features฀of฀interlanguage฀at฀different฀stages฀of฀L2฀ acquisition฀and฀construct฀a฀model฀of฀interlanguage฀ development.฀We฀hope฀to฀identify฀the฀mechanisms฀of฀ development฀from฀one฀stage฀of฀interlanguage฀to฀another,฀ which฀we฀hope฀will฀lead฀to฀the฀improvement฀of฀teaching฀ methods฀and฀more฀rigorous฀empirical฀research฀on฀the฀ effect฀of฀learning฀methods฀on฀the฀transition฀process. The฀Standard฀Speaking฀Test฀ The฀Standard฀Speaking฀Test฀(SST)฀is฀a฀collaboration฀ between฀the฀American฀Council฀on฀the฀Teaching฀of฀ Foreign฀Languages฀(ACTFL)฀and฀ALC฀Press.฀It฀is฀based฀ on฀the฀ACTFL฀Proficiency฀Guidelines฀for฀speaking฀and฀ the฀Oral฀Proficiency฀Interview฀(OPI).฀The฀ACTFL-OPI฀ was฀first฀developed฀in฀1982,฀and฀since฀then฀it฀has฀been฀ one฀of฀the฀most฀influential฀speaking฀tests฀in฀the฀world฀ despite฀the฀fact฀that฀there฀has฀been฀some฀criticism฀against฀ the฀empirical฀bases฀of฀the฀guideline฀(see,฀for฀example,฀ Bachman฀and฀Savignon฀1986;฀Chalhoub-Deville฀1997).฀ ACTFL฀and฀ALC฀Press฀worked฀together฀to฀develop฀a฀ new฀speaking฀test฀for฀Japanese฀learners฀of฀English.฀The฀ proficiency฀guideline฀defines฀9฀different฀proficiency฀levels฀ (Level฀1:฀‘Novice-Low’฀through฀Level฀9:฀‘Advanced’).฀ Each฀level฀is฀defined฀specifically฀in฀terms฀of฀the฀following฀ criteria:฀(a)฀context/content฀area,฀(b)฀text฀type,฀(c)฀global฀ PAC3 at JALT2001 A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS task฀&฀function,฀(d)฀accuracy,฀which฀includes฀grammatical฀ accuracy,฀fluency,฀and฀pronunciation.฀ The฀SST฀takes฀the฀form฀of฀a฀10฀to฀15-minute฀taperecorded฀conversation฀between฀a฀trained฀interviewer฀and฀ a฀test฀candidate.฀The฀SST฀utilizes฀interview฀techniques฀ and฀picture฀prompts฀to฀simulate฀natural฀conversation฀to฀ the฀maximum฀extent฀possible฀in฀a฀testing฀situation. The฀tape-recorded฀interview฀is฀scored฀by฀two฀different฀ Raters฀(in฀case฀of฀disagreement,฀the฀interview฀is฀rated฀by฀ a฀Master฀Rater).฀In฀the฀SST,฀the฀elicitation฀and฀scoring฀ of฀speech฀samples฀are฀separate฀procedures,฀unlike฀the฀ OPI฀where฀both฀tasks฀are฀performed฀on฀the฀spot฀by฀the฀ interviewer.฀The฀SST฀interview฀process฀elicits฀speech฀ samples฀through฀the฀application฀of฀the฀following฀fivestage฀format: 1.฀ Warm-up฀and฀initial฀assessment 2.฀ Single฀picture฀prompt฀with฀level฀checks฀and฀ probes 3.฀ Role-play฀with฀level฀checks฀and฀probes 4.฀ Single฀or฀picture฀sequence฀prompt฀with฀level฀ checks฀and฀probes 5.฀ Wind-down Although฀the฀interviewer฀is฀not฀responsible฀for฀the฀ formal฀rating฀of฀the฀test฀candidate,฀the฀interviewer฀must฀ be฀able฀to฀conduct฀an฀on-going฀informal฀assessment฀of฀ the฀speaker’s฀proficiency฀in฀order฀to฀tailor฀the฀questions,฀ 873 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING prompts,฀and฀role฀plays฀most฀suited฀to฀the฀test฀ candidate’s฀interests฀and฀level฀of฀speaking฀proficiency.฀ If฀the฀speech฀sample฀is฀poorly฀elicited฀via฀prompts฀ inappropriate฀to฀the฀speaker’s฀level฀of฀proficiency,฀the฀ rating฀of฀the฀speech฀sample฀may฀become฀invalid. The฀SST฀serves฀to฀discriminate฀spoken฀English฀ at฀Novice฀to฀Intermediate฀High฀levels฀of฀proficiency฀ utilizing฀a฀shorter฀interview฀than฀the฀OPI.฀The฀SST฀ discriminates฀more฀finely฀at฀Intermediate฀proficiency฀ levels฀than฀do฀other฀existing฀standardized฀measurement฀ instruments฀of฀speaking฀proficiency.฀The฀SST฀can฀also฀ serve฀as฀a฀potential฀screening฀interview฀for฀speakers฀who฀ might฀be฀ready฀for,฀and฀benefit฀from฀taking,฀the฀OPI. ALC฀Press฀possesses฀large฀archives฀of฀audio฀recordings฀ of฀this฀test.฀We฀saw฀this฀data฀as฀potentially฀very฀ useful฀spoken฀resources,฀as฀most฀learner฀corpora฀to฀ date฀consist฀of฀written฀data฀only.฀For฀this฀reason,฀we฀ decided฀to฀launch฀the฀present฀project฀to฀transcribe฀these฀ archived฀recordings฀and฀convert฀them฀into฀a฀spoken฀ learner฀corpus.฀The฀strength฀of฀this฀corpus฀project฀is฀ that฀each฀file/piece฀of฀data฀has฀specific฀information฀ on฀the฀examinee’s฀oral฀proficiency฀level,฀as฀assessed฀ by฀the฀professional฀examiner.฀Whilst฀there฀are฀some฀ developmental฀interlanguage฀corpora฀available฀(e.g.฀ JEFLL),฀the฀labelling฀or฀determination฀of฀learner฀ proficiency฀levels฀is฀often฀based฀on฀external฀factors฀such฀ as฀school฀years.฀Thus,฀a฀comparison฀between฀subcorpora฀ based฀on฀school฀years฀sometimes฀causes฀a฀problem.฀In฀ PAC3 at JALT2001 A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS the฀Longman฀Learner’s฀Corpus,฀learner฀proficiency฀ for฀each฀file฀is฀encoded฀in฀its฀header,฀but฀judgements฀ about฀the฀proficiency฀levels฀seem฀to฀be฀entirely฀up฀to฀ the฀teachers฀who฀donated฀the฀composition฀data,฀and฀are฀ thus฀not฀entirely฀reliable,฀since฀we฀have฀no฀information฀ about฀the฀possibly฀varied฀criteria฀or฀standards฀used฀ by฀the฀different฀teachers.฀Compared฀with฀these฀other฀ learner฀corpora,฀therefore,฀SST฀data฀have฀more฀reliable฀ information฀on฀learners’฀proficiency฀levels,฀which฀will฀ help฀to฀make฀comparative฀research฀based฀on฀proficiency฀ subsections฀of฀the฀corpus฀more฀valid.฀ The฀SST฀Corpus฀Project The฀SST฀Corpus฀has฀been฀compiled฀as฀part฀of฀a฀ larger฀project฀called฀‘Research฀&฀Development฀of฀ Congruent฀Communication฀Technologies’฀led฀by฀the฀ Telecommunications฀Advancement฀Organization฀of฀ Japan฀(TAO)฀and฀Communication฀Research฀Laboratory฀ (CRL).฀The฀primary฀goal฀of฀this฀project฀is฀to฀develop฀ natural฀language฀processing฀technologies฀that฀can฀handle฀ human฀errors฀in฀speech฀or฀writing฀and฀their฀application฀ in฀such฀areas฀as฀support฀systems฀for฀writing฀in฀English,฀ machine฀translation,฀and฀applications฀in฀education฀(e.g.฀ learner฀error฀databases,฀computer-assisted฀self-access฀ language฀learning฀systems). The฀SST฀learner฀corpus฀data฀have฀various฀types฀of฀ error฀(lexical,฀syntactic,฀phonetic,฀etc.).฀Information฀ on฀the฀types฀and฀rates฀of฀error฀that฀learners฀make฀ 874 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING at฀various฀proficiency฀levels,฀the฀contexts฀in฀which฀ particular฀learner฀errors฀occur,฀etc.฀will฀serve฀as฀the฀ input฀for฀machine฀learning฀of฀interlanguage฀grammar.฀ CRL฀has฀created฀an฀automatic฀machine฀learning฀system฀ (consisting฀of฀a฀lexicon฀and฀a฀grammar)฀which฀can฀be฀ potentially฀very฀useful฀for฀testing฀ and฀evaluating฀the฀interlanguage฀ grammar฀model฀that฀the฀machine฀ learns฀automatically.฀The฀results฀ can฀be฀used฀for฀NLP฀and฀ educational฀purposes. The฀initial฀phase฀of฀the฀project฀ includes฀the฀development฀of฀the฀ following฀tools฀and฀guidelines:฀ (1)฀transcription฀guidelines,฀(2)฀ tagging฀schemes,฀(3)฀a฀tag฀editor,฀ (4)฀error฀tagging฀schemes,฀and฀ (5)฀error฀tagging฀support฀tools.฀ At฀the฀time฀of฀writing฀(December฀ 2001),฀the฀second฀version฀of฀a฀ set฀of฀transcription฀guidelines฀has฀ been฀prepared฀for฀transcribing฀the฀ data฀of฀about฀510฀subjects,฀using฀a฀ “TagEditor”฀(see฀Figure฀1). A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS TagEditor฀has฀been฀developed฀specially฀for฀this฀ learner฀corpus฀project.฀Besides฀the฀usual฀functionalities฀ common฀to฀text฀editors,฀it฀has฀unique฀features฀such฀as฀ the฀ability฀to฀insert฀customized฀XML-like฀tags,฀validate฀ tags฀(see฀Figure฀2),฀show฀tags฀in฀colour,฀do฀simple฀grep฀ Figure฀1:฀A฀screenshot฀of฀ TagEditor฀Version฀1.2. PAC3 at JALT2001 875 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS searches฀and฀concordancing,฀use฀ file฀templates,฀check฀file฀formats,฀ do฀automatic฀‘search฀and฀replace’,฀ and฀so฀on.฀A฀similar฀tool฀was฀ developed฀at฀Louvain฀for฀the฀ ICLE฀project,฀but฀our฀tool฀has฀ several฀new฀features,฀including฀ tag฀validation฀and฀a฀simple฀ concordancer.฀ Figure฀2:฀Tag฀validation฀ function฀of฀TagEditor Currently,฀the฀team฀is฀working฀ on฀error฀tagging.฀It฀is฀extremely฀ difficult฀to฀define฀a฀generic฀error฀ tagset฀that฀can฀be฀applied฀to฀any฀ type฀of฀learner฀corpus.฀So฀far฀we฀ have฀developed฀a฀generic฀error฀ tagset฀and฀an฀accompanying฀error฀ tag฀manual.฀We฀focus฀mainly฀on฀ lexical฀and฀syntactic฀errors฀only฀ at฀this฀stage.฀At฀a฀later฀stage,฀ however,฀we฀will฀shift฀our฀focus฀to฀other฀types฀of฀errors,฀ e.g.฀prosodic,฀pragmatic,฀and฀discoursal฀errors.฀The฀ data฀will฀be฀made฀publicly฀available฀after฀the฀three-year฀ project฀ends฀in฀2003. PAC3 at JALT2001 SST฀Data,฀SLA/ELT฀Research฀and฀Beyond Using฀SST฀Corpus฀for฀SLA฀research The฀influence฀of฀corpus฀linguistics฀has฀spread฀rather฀ slowly฀in฀the฀fields฀of฀SLA฀research.฀Leech฀(1998)฀states฀ 876 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING two฀main฀reasons฀for฀this:฀firstly,฀it฀is฀extremely฀timeconsuming฀and฀labour-intensive฀to฀create฀a฀corpus.฀ Secondly,฀the฀intellectual฀climate฀in฀applied฀linguistics฀ has฀been฀such฀that฀theory-driven฀experimental฀studies฀ have฀prevailed฀for฀the฀last฀20฀years฀and฀the฀data-driven฀ descriptive฀methods฀of฀corpus฀linguistics฀have฀not฀ been฀preferred.฀Recently,฀however,฀there฀is฀a฀growing฀ awareness฀that฀a฀computational฀analysis฀of฀a฀large฀body฀ of฀language฀use฀data฀could฀provide฀useful฀data฀for฀ language฀researchers฀as฀more฀focus฀is฀now฀on฀usagebased฀linguistic฀models฀as฀well฀as฀lexico-grammatical฀ aspects฀of฀a฀language.฀ L2฀learner฀corpora฀have฀great฀potentials฀ for฀clarifying฀the฀L2฀acquisition฀process.฀ It฀serves฀us฀as฀better฀data฀than฀what฀ we฀have฀mostly฀relied฀on฀so฀far฀and฀ enables฀us฀to฀investigate฀the฀non-native฀ speaking฀learners’฀language฀not฀only฀ from฀a฀negative฀point฀of฀view฀(‘What฀did฀ the฀learner฀get฀wrong?’)฀but฀also฀from฀ a฀positive฀one฀(‘What฀did฀the฀learner฀ get฀right?’)฀It฀also฀makes฀it฀possible฀ to฀investigate฀L2฀learners’฀avoidance฀ behaviour฀by฀examining฀overuse/underuse฀ phenomena.฀Undeniably,฀all฀types฀of฀SLA฀ data฀have฀their฀strengths฀and฀weaknesses฀ and฀one฀cannot฀help฀but฀agree฀with฀ Ellis฀(1994:676)฀that฀‘Good฀research฀is฀ PAC3 at JALT2001 A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS research฀that฀makes฀use฀of฀multiple฀sources฀of฀data.’฀The฀ learner฀corpora,฀if฀properly฀compiled,฀serve฀well฀as฀a฀ valuable฀addition฀to฀current฀SLA฀data฀sources.฀They฀can฀ be฀used฀for฀(a)฀verification฀of฀theories฀and฀hypotheses฀in฀ SLA฀and฀(b)฀description฀of฀interlanguage฀development฀ (developmental฀stages฀of฀linguistic฀features฀(lexicogrammatical/฀discoursal/฀pragmatic),฀L1฀transfer฀effect,฀ overuse/฀underuse฀of฀particular฀features,฀universal/฀L1฀ specific฀errors,฀native-like/฀non-native-like฀performance฀ (Tono฀1998).฀ 877 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING Figure฀3:฀The฀potential฀of฀SST฀Corpus฀for฀SLA฀research Figure฀3฀illustrates฀this฀point.฀If฀the฀patterns฀of฀ use/misuse฀and฀overuse/underuse฀are฀identified฀across฀ different฀stages฀of฀acquisition฀and฀a฀proper฀comparison฀ is฀made฀between฀the฀SST฀data฀and฀different฀types฀of฀ learner฀corpora฀such฀as฀written฀or฀parallel฀versions฀(with฀ corrected฀errors),฀we฀will฀be฀able฀to฀identify฀various฀ aspects฀of฀interlanguage฀grammar฀development.฀Ellis฀ (2001)฀predicts฀an฀important฀role฀of฀corpora฀in฀SLA฀in฀ the฀future฀and฀distinguishes฀three฀types฀of฀ corpora฀as฀useful:฀(1)฀corpora฀of฀‘authentic’฀ native฀speaker฀language฀use,฀(2)฀corpora฀ of฀learner฀language฀use,฀and฀(3)฀corpora฀of฀ native฀speaker฀language฀use฀with฀learners฀ (i.e.฀samples฀of฀foreigner฀talk).฀The฀SST฀ data฀serves฀as฀a฀useful฀tool฀for฀the฀second฀ type. A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS syllabus฀and฀materials฀design,฀(b)฀test฀construction฀and฀ (c)฀teacher฀education.฀For฀direct฀use,฀SST฀Corpus฀can฀be฀ exploited฀as฀a฀part฀of฀on-line฀resources฀for฀L2฀learners’฀ data-driven฀learning฀or฀a฀part฀of฀research฀resources฀for฀ postgraduate฀students฀in฀applied฀linguistics.฀ Figure฀4:฀The฀potential฀of฀SST฀Corpus฀for฀English฀ language฀teaching The฀implications฀of฀the฀SST฀Corpus฀ project฀for฀English฀language฀teaching The฀exploitation฀of฀SST฀Corpus฀for฀ English฀language฀teaching฀can฀be฀either฀ direct฀or฀indirect.฀As฀shown฀in฀Figure฀4,฀as฀ indirect฀applications฀of฀the฀data,฀spoken฀ learner฀corpus฀data฀could฀provide฀the฀ information฀about฀L2฀learners’฀acquisition฀ process฀and฀possible฀error฀sources฀and฀ patterns,฀which฀can฀be฀invaluable฀resources฀for฀(a)฀ELT฀ PAC3 at JALT2001 878 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING Future฀technologies Another฀interesting฀possibility฀is฀the฀collaboration฀with฀ researchers฀in฀natural฀language฀processing฀(NLP)฀fields.฀ They฀have฀considerable฀expertise฀in฀using฀language฀ corpora฀for฀machine฀translation,฀information฀retrieval,฀ and฀language฀modelling.฀Some฀computers฀can฀learn฀ rules฀of฀a฀language฀statistically฀on฀the฀basis฀of฀a฀large฀ amount฀of฀natural฀language฀data.฀If฀a฀learner฀corpus฀ is฀fed฀into฀such฀a฀computer,฀it฀could฀possibly฀learn฀a฀ grammar฀of฀interlanguage.฀Since฀SST฀Corpus฀has฀nine฀ proficiency฀levels,฀it฀would฀be฀an฀interesting฀possibility฀ to฀explore฀how฀a฀computer฀can฀statistically฀learn฀the฀ grammars฀of฀interlanguages฀at฀different฀proficiency฀ levels.฀This฀approach฀is฀in฀line฀with฀the฀one฀that฀Dan฀ Jurafsky฀and฀his฀colleagues฀have฀taken฀recently฀in฀ what฀they฀call฀‘computational฀psycholinguistics.’฀We฀ could฀possibly฀develop฀an฀error฀detector฀or฀a฀language฀ A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS proficiency฀assessment฀system฀that฀can฀analyse฀the฀input฀ from฀learners฀and฀identify฀which฀developmental฀stage฀a฀ particular฀learner฀is฀at. Learner฀data฀can฀also฀be฀used฀for฀the฀development฀ of฀a฀noise-proof฀machine฀translation฀system.฀A฀future฀ translation฀tool฀could฀hear฀the฀non-native฀speakers’฀ erroneous฀sentences฀and฀interpret฀them฀properly.฀฀ The฀learner฀corpus฀data฀can฀be฀useful฀resources฀for฀ developing฀the฀knowledge฀base฀for฀such฀a฀system.฀Our฀ team฀is฀ideal฀in฀the฀sense฀that฀we฀have฀both฀SLA฀and฀ NLP฀researchers฀working฀together฀to฀fully฀exploit฀a฀ corpus฀of฀Japanese-speaking฀learners฀of฀English.฀ We฀are฀hoping฀to฀finish฀the฀data฀collection฀within฀a฀ year฀and฀release฀the฀corpus฀in฀the฀fiscal฀year฀2003฀so฀that฀ it฀will฀be฀freely฀available฀for฀research฀and฀commercial฀ purposes.฀We฀are฀looking฀forward฀to฀your฀input฀and฀ comments฀on฀this฀exciting฀project. References Bachman,฀L.F.฀and฀Savignon฀S.฀(1986).฀The฀evaluation฀of฀communicative฀language฀proficiency:฀A฀critique฀of฀the฀ ACTFL฀oral฀interview.฀Modern฀Language฀Journal฀70,฀380-390. Chalhoub-Deville,฀M.฀(1997).฀Theoretical฀models,฀assessment฀frameworks฀and฀test฀construction.฀Language฀Testing฀ 14,฀3-22. De฀Cock,฀S.,฀Granger,฀S.฀and฀Petch-Tyson,฀S.฀(1999).฀The฀Louvain฀International฀Database฀of฀Spoken฀English฀ Interlanguage฀(LINDSEI)฀Project.฀An฀internal฀report฀at฀Catholic฀University฀of฀Louvain. Ellis,฀R.฀(1994).฀The฀Study฀of฀Second฀Language฀Acquisition.฀฀Oxford฀University฀Press:฀Oxford. PAC3 at JALT2001 879 Conference Proceedings TONO, KANEKO, ISAHARA, IZUMI, SAIGA & KANEKO: DEVELOPING A ONE MILLION WORD SPOKEN EFL LEARNER CORPUS Ellis,฀R.฀(2001).฀Real฀Data฀and฀Real฀Pedagogy.฀Lecture฀given฀at฀the฀2nd฀Learner฀Corpus฀Workshop฀at฀Showa฀ Women’s฀University,฀Tokyo. Leech,฀G..฀(1998).฀Preface.฀In฀S.฀Granger฀(ed.)฀Learner฀English฀on฀Computer.฀฀London:฀Longman. Tono,฀Y.฀(1998).฀A฀computer฀learner฀corpus-based฀analysis฀of฀the฀acquisition฀order฀of฀English฀grammatical฀ morphemes.฀฀In฀TALC฀(Teaching฀and฀Language฀Corpora)฀98฀Conference฀Proceedings,฀Keble฀College฀Oxford.฀ Tono,฀Y.฀(2000)฀A฀corpus-based฀analysis฀of฀interlanguage฀development:฀analysing฀part-of-speech฀tag฀sequences฀of฀ EFL฀learner฀corpora.฀Lewandowska-Tomaszcyk฀&฀J.฀Melia฀(eds.)฀PALC’99:฀Practical฀Applications฀in฀Language฀ Corpora.฀Peter฀Lang฀GmbH,฀Frankfurt฀am฀Main,฀pp.฀323-342. PAC3 at JALT2001 880 Conference Proceedings