学習者コーパス研究におけるマルチレベル順序ロジットモデルの活用

ֶशऀίʔύεݚڀʹ͓͚Δ ϚϧνϨϕϧॱংϩδοτϞσϧͷ׆༻ খྛɹ༤Ұ࿠ -&5ϝιυϩδʔݚڀ෦ձ ೥݄೔ 1

ࣗݾ঺հ •  খྛɹ༤Ұ࿠ʢ͜͹΍͠ɹΏ͏͍ͪΖ͏ʣ •  ؔ৺ྖҬ •  ίʔύεݴޠֶ •  ࣗಈ࠾఺ ʢϥΠςΟϯάɺεϐʔΩϯάʣ
•  ڭҭσʔλ෼ੳ ɹFUD •  ϝιݚॳࢀՃ೥݄ 2

झࢫ ֶशऀίʔύεݚڀʹ ʢϚϧνϨϕϧʣॱংϩδοτϞσϧΛ ࢖͏ͱྑͦ͞͏ͳͷͰɺ ΈΜͳͰߟ͑ͯΈ͍ͨ ɹ
ͨͩ͠ɺࢲ͸ͦΕ΄Ͳ͜ͷख๏ʹৄ͍͠Θ͚Ͱ͸ͳ͍ͷͰɺ օ༷ͷϑΟʔυόοΫʹظ଴ʢଞྗຊئʣ 3

ൃදͷྲྀΕ •  ໰୊ҙࣝ •  ֶशऀίʔύεݚڀ •  ࣗಈ࠾఺ݚڀ •  ఏҊख๏ • 
ॱংϩδοτϞσϧ •  ϚϧνϨϕϧϞσϧ •  έʔεελσΟ •  ෼ੳσʔλ •  ෼ੳํ๏ •  ෼ੳ݁Ռ BOENPSF 4

໰୊ҙࣝ •  ֶशऀίʔύεݚڀͷࢹ఺ʢίʔύεΛهड़͠ɺൃୡࢦඪΛಛఆʣ •  ͲͷΑ͏ͳ౷ܭϞσϧΛ࢖͏͔ʁ •  ୯७ͳઢܗϞσϧͰΑ͍ͷ͔ʁ ɹˣ •  ίʔύεݚڀͷجຊతͳσʔλ୯ҐͰ͋Δʮස౓ʯ͸Χ΢ϯτσʔλ
ʢෛ஋ΛͱΒͳ͍ʣ •  ֶशऀίʔύεݚڀͰҰൠతͳ໨తม਺͸ΧςΰϦΧϧม਺ ʢʮֶ೥ʯ΍ʮशख़౓ϨϕϧʯͷΑ͏ͳॱংई౓ɺʮ฼ޠʯͷΑ͏ͳ໊ٛई౓ʣ ɹˣ •  ਖ਼͍͠ਪఆ͕Ͱ͖ͳ͍ख๏Λ࢖͏ͱɺਖ਼͍͠ൃୡࢦඪͷಛఆ͕Ͱ͖ͳ͍ 5

•  ࣗಈ࠾఺ݚڀͷࢹ఺ʢൃୡࢦඪΛ༻͍ͯɺशख़౓ϨϕϧΛ༧ଌʣ •  ࣗಈ࠾఺͸ɺൃୡࢦඪʹؔ͢Δ৘ใΛख͕͔Γͱͯ͠ར༻ʢઆ໌ม਺ʣ •  ଟ͘ͷγεςϜ͸ɺ఺਺ʢִؒई౓ʣͰ͸ͳ͘ɺशख़౓Ϩϕϧʢॱংई౓ʣ Λ༧ଌʢ໨తม਺ʣ •  ໨తม਺͕ॱংई౓Ͱ͋ͬͯ΋໊ٛई౓ͱΈͳ͠ɺճؼ໰୊Ͱ͸ͳ͘ɺ෼ྨ ໰୊Ͱղ͘Ξϓϩʔν͕ଟ͍ҹ৅
ʢ෼ྨ໰୊ͱͯ͠ղ͍ͨํ͕ߴ͍ਫ਼౓ΛಘΒΕΔ͜ͱ΋ଟ͍ͨΊʣ ɹˣ •  ͔͠͠ɺσʔλ෼ੳͷਖ਼߈๏ͱͯ͠͸ɺॱংม਺ͷճؼ໰୊ͱͯ͠ղ͘΂͖ Ͱ͸ͳ͍͔ʁ •  ໊ٛई౓ΑΓ΋ॱংई౓ͷํ͕୯७ʹղऍͰ͖ɺજࡏతʹ༧ଌਫ਼౓΋ߴ͍ "HSFTUJ 6

ఏҊख๏ ϚϧνϨϕϧॱংϩδοτϞσϧ ʢʹॱংϩδοτϞσϧʴϚϧνϨϕϧϞσϧʣ 7

ॱংϩδοτϞσϧ PSEFSFEMPHJUNPEFM •  ཭ࢄબ୒Ϟσϧ EJTDSFUFDIPJDFNPEFM ͷͭ •  ໨తม਺ɿॱংม਺ʢͭҎ্ʣ • 
આ໌ม਺ɿ਺஋ܕͰ΋ҼࢠܕͰ΋0, 8 ໨తม਺ͷई౓ ॱংई౓ ໊ٛई౓ ໨తม਺ͷ਺ ͭ ϩδεςΟοΫճؼ෼ੳ ϩδεςΟοΫճؼ෼ੳ ͭҎ্ ॱংϩδοτϞσϧ ॱংϓϩϏοτϞσϧ ଟ߲ϩδοτϞσϧ ଟ߲ϓϩϏοτϞσϧ ॱংϓϩϏοτϞσϧ͸ɺॱংϩδοτϞσϧͱ΄΅ಉ݁͡ՌΛฦ͕͢ɺϩδεςΟοΫؔ਺Ͱ͸ͳ͘ਖ਼ن෼෍ ؔ਺Λ༻͍Δɻσʔλʹۃ୺ʹେ͖͍஋ؚ͕·ΕΔ৔߹͸ɺॱংϩδοτϞσϧͷํ͕ଥ౰ FH -JBP ˗ΧςΰϦΧϧม਺Λѻ͏ओͳճؼ෼ੳ

•  ॱংϩδοτϞσϧΛ༻͍ͨݴޠݚڀ •  ాதɾ໦ଜɾґాɾീౡ •  ʮॱংϩδοτϞσϧʹج͍ͮͨӳ୯ޠͷशಘࠔ೉౓ͷਪఆͱͦͷཁҼͷ෼ੳʯ •  ଞʹݟ͚ͭΒΕͳ͔ͬͨͷͰɺԿ͔͝ଘ஌ͷํɺڭ͍͑ͯͩ͘͞N
@@ N 9

ϚϧνϨϕϧϞσϧ NVMUJMFWFMNPEFM •  ֊૚తσʔλΛ෼ੳ͢ΔͨΊͷϞσϧ •  -ֶशऀʢ-ͷࠩΛߟྀͨ͠෼ੳʣ •  λεΫֶशऀʢλεΫͷࠩΛߟྀͨ͠෼ੳʣ • 
ֶशऀଌఆ࣌఺ʢ࣌ܥྻ෼ੳʣ ɹɹFUD •  Ϩϕϧ͚ͩͰͳ͘ɺϨϕϧҎ্Ͱ΋0, •  ଟ͘ͷֶशऀίʔύε͸ɺ֊૚తͳߏ଄ •  ࠃʢ㲈-ʣ •  Ϟʔυʢॻ͖ݴ༿ʗ࿩͠ݴ༿ʣ •  λεΫ •  τϐοΫ FUD 10 ೔ຊ ֶ श ऀ ֶ श ऀ ֶ श ऀ ؖࠃ ֶ श ऀ ֶ श ऀ ֶ श ऀ தࠃ ֶ श ऀ ֶ श ऀ ֶ श ऀ ूஂϨϕ ݸਓϨϕ

•  ֊૚తσʔλʹ௨ৗͷճؼ෼ੳΛ༻͍Δ໰୊ ਗ਼ਫ •  αϯϓϧͷಠཱੑԾఆͷҧ൓ •  ճؼ෼ੳ͸αϯϓϧͷಠཱੑΛԾఆ͢Δ͕ɺ֊૚తσʔλͷαϯϓϧ͸ಠཱ͍ͯ͠ͳ͍ • 
ඪ४ޡࠩΛෆ౰ʹখ͘͞ݟੵ΋ͬͯ͠·͍ɺλΠϓᶗΤϥʔΛ൜͢Մೳੑ ɹɹˣ •  ਖ਼͍͠ൃୡࢦඪͷಛఆɺਖ਼͍͠-ͷӨڹͷಛఆͷ๦͛ •  ਪఆ஋ͷղऍͷ໰୊ •  ճؼ܎਺ͷதʹɺूஂͷੑ࣭ͱݸਓͷੑ࣭͕ࠞࡏ͍ͯ͠ΔͨΊɺղऍ͕ࠔ೉ ɹɹˣ •  ೔ຊਓ͔ͩΒɺݴޠ߲໨9ΛΑ͘࢖͏ͷ͔ʁ •  ೔ຊਓ͸ฏۉతʹशख़౓͕௿͍͔Βɺݴޠ߲໨9ΛΑ͘࢖͏ͷ͔ʁ •  शख़౓ͷ௿ֶ͍शऀ͸ɺ-ʹ͔͔ΘΒͣɺݴޠ߲໨9ΛΑ͘࢖͏ͷ͔ʁ 11

•  σʔλͷ֊૚ੑͷ֬ೝ •  ֊૚తσʔλͷಛ௃͸ɺूஂ಺ྨࣅੑ ʢFH ೔ຊਓ͸ݴޠ߲໨9ΛΑ͘࢖͏ʣ •  ूஂ಺ྨࣅੑͷେ͖͞͸ɺڃ಺૬ؔ܎਺ͳͲͰධՁʢޙड़ʣ 12

•  ༷ʑͳϚϧνϨϕϧϞσϧ (SJFT •  ྫʣճؼࣜͷ੾ยɾ܏͖ʹؔ͢ΔԾఆʢઢܗϞσϧͷ৔߹ʣ 13 άϧʔϓ͝ͱʹ੾ย͕ҟͳΔ ͱԾఆ
άϧʔϓ͝ͱʹ܏͖͕ҟͳΔ ͱԾఆ άϧʔϓ͝ͱʹ੾ยͱ܏͖ͷ ྆ํ͕ҟͳΔͱԾఆ

•  ϚϧνϨϕϧϞσϧΛ༻͍ͨݴޠݚڀ •  (SJFTBOE%FTIPST •  &'-BOEWT&4- "NVMUJMFWFMSFHSFTTJPONPEFMJOHQFSTQFDUJWFPOCSJEHJOHUIF QBSBEJHNHBQ
•  .VSBLBNJ •  .PEFMJOHTZTUFNBUJDJUZBOEJOEJWJEVBMJUZJOOPOMJOFBSTFDPOEMBOHVBHF EFWFMPQNFOU5IFDBTFPG&OHMJTIHSBNNBUJDBMNPSQIFNFT •  4UFQIFOT WBOEFS4MJL BOEWBO)PVU •  5IF-JNQBDUPOMFBSOJOH-%VUDI5IF-EJTUBODFFGGFDU •  ;IBOH (FFSBFSUT BOE4QFFMNBO •  /PO NFUPOZNJDFYQSFTTJPOTGPSHPWFSONFOUJO$IJOFTF"NJYFEFGGFDUT MPHJTUJDSFHSFTTJPOBOBMZTJT •  1BRVPU /BFUT BOE(SJFT •  6TJOHTZOUBDUJDDPPDDVSSFODFUPUSBDFQISBTFPMPHJDBMDPNQMFYJUZEFWFMPQNFOU JOMFBSOFSXSJUJOH7FSC PCKFDUTUSVDUVSFTJO-0/(%"-& ɹFUD 14

•  ίʔύεݴޠֶʹ͓͚ΔϚϧνϨϕϧͷԠ༻ʢՄೳੑʣ •  #BBZFOʹΑΔ࠷ॳͷಋೖ •  (SJFTʹΑΔҰ࿈ͷղઆ࿦จ FH
•  ྔతͳ໨తม਺Λର৅ͱ͢Δճؼ෼ੳɺ͋Δ͍͸ɺ஋ͷ໨తม਺Λର৅ͱ ͢ΔϩδεςΟοΫճؼ෼ੳ͹͔ΓͰɺ஋Ҏ্ͷॱংม਺Λѻ͏ճؼ෼ੳ ͷ࢖༻ྫΛݟ͚ͭΒΕͣʜ •  Կ͔͝ଘ஌ͷํɺڭ͍͑ͯͩ͘͞N @@ Nʢճ໨ʣ 15

έʔεελσΟ -ͷҧ͍Λߟྀͨ͠ ݴޠशಘաఔͷϞσϧԽ ʔϝλஊ࿩ඪࣝΛྫʹʔ 16

෼ੳσʔλ •  *$/"-& *OUFSOBUJPOBM$PSQVT/FUXPSLPG"TJBO-FBSOFSTPG&OHMJTI *TIJLBXB •  ࠃʢ㲈-ʣɿ೔ຊ
+1/ ɺλΠ 5)" ɺ୆࿷ 58/ •  शख़౓Ϩϕϧʢ$&'3ʣɿ"ɺ#ɺ#ɺ#ʢˠʙʹม׵ʣ •  τϐοΫɿ1BSUUJNFKPC 17 ɹ CEFR Total 1 2 3 4 L1 JPN 154 179 49 18 400 THA 119 179 100 2 400 TWN 29 87 61 23 200 Total 302 445 210 43 1000

•  ݴޠ߲໨ͷස౓ूܭʢˠઆ໌ม਺ʣ •  )ZMBOEͷϝλஊ࿩ඪࣝ •  ϝλஊ࿩ͱ͸ɺςΩετͷ໋୊಺༰ʹ௚઀ӨڹΛ༩͑ ͳ͍͕ɺटඌҰ؏ͨ͠࿦ཧతͳจষΛߏ੒ͨ͠Γɺಡ Έख໋͕୊಺༰΍จষͷల։ɺಡΈखʹର͢Δॻ͖ख ͷཱ৔Λཧղͨ͠Γ͢Δࡍʹ໾ཱͭ֓೦ • 
ϝλஊ࿩ඪࣝ͸ɺϝλஊ࿩ʹ͓͚Δ༷ʑͳػೳΛ୲͏ දݱͷ͜ͱͰɺΞΧσϛοΫɾϥΠςΟϯάͷݚڀͳ ͲͰ෼ੳର৅ͱ͞ΕΔ͜ͱ͕ଟ͍ݴޠ߲໨ •  ӳޠֶशऀͷϥΠςΟϯάͰ͸ɺॻ͖खͷ-΍शख़ ౓͕ϝλஊ࿩ඪࣝͷස౓ύλʔϯʹ৭ೱ͘ݱΕΔ FH ,PCBZBTIJ 18

•  ϝλஊ࿩ඪࣝͷछྨͷػೳΧςΰϦʔ )ZMBOE 19 &/% FOEPQIPSJDNBSLFST ͸ɺ ෼ੳσʔλதʹग़ݱͤͣ
छྨ͍ۙදݱ͕Ϧετʹؚ·Εͯ ͍ΔͨΊɺଟมྔղੳͳͲʹศར ɹˣ ਺͕ଟա͗Δ৔߹ʢଟॏڞઢੑͷ͍ٙʣ ౷ܭతͳม਺બ୒ͳͲΛ࣮ߦ

෼ੳํ๏ •  ༧උత෼ੳ •  εςοϓϫΠζ๏ʹΑΔม਺બ୒ •  બ୒͞Εͨม਺ͷՄࢹԽ •  બ୒͞Εͨม਺ͷڃ಺૬ؔ܎਺ • 
ॱংϩδοτϞσϧ •  -ͷҧ͍Λߟྀ͠ͳ͍Ϟσϧ •  -ͷҧ͍Λߟྀ͢ΔϞσϧ •  ϞσϧͷධՁ 20

༧උత෼ੳ •  εςοϓϫΠζ๏ʹΑΔม਺બ୒ʢॱংϩδοτϞσϧʣ •  "*$ʢ੺஑৘ใྔج४ʣʹجͮ͘ •  CEFR ~ TRA +
SEM + BOO ɹɹˣ •  53" 5SBOTJUJPOT •  4&. 4FMGNFOUJPOT •  #00 #PPTUFST 21 Step: AIC=2356.04 CEFR ~ TRA + SEM + BOO Df AIC <none> 2356.0 + HED 1 2357.4 + ENG 1 2357.7 + ATM 1 2357.8 + EVI 1 2357.9 + FRM 1 2358.0 + COD 1 2358.0 - TRA 1 2358.5 - BOO 1 2358.7 - SEM 1 2365.1

ՄࢹԽ •  ശͻ͛ਤ 22
$&'3 53" $&'3 4&. $&'3 #00 +1/ 5)" 58/ - 53" +1/ 5)" 58/ - 4&. +1/ 5)" 58/ - #00 આ໌ม਺ͷස౓͸ɺޠ͋ͨΓͷ૬ରස౓

$&'3 53" $&'3 4&. $&'3 #00 +1/ 5)" 58/ - 53" +1/ 5)" 58/ - 4&. +1/ 5)" 58/ - #00 આ໌ม਺ͷස౓͸ɺޠ͋ͨΓͷ૬ରස౓ 53"͸ɺ͋·Γ$&'3ؒ΍-ؒͷ͕ࠩͳ͍

$&'3 53" $&'3 4&. $&'3 #00 +1/ 5)" 58/ - 53" +1/ 5)" 58/ - 4&. +1/ 5)" 58/ - #00 આ໌ม਺ͷස౓͸ɺޠ͋ͨΓͷ૬ରස౓ 4&.͕$&'3΍-ͷ༧ଌʹ࠷΋د༩ͦ͠͏

$&'3 53" $&'3 4&. $&'3 #00 +1/ 5)" 58/ - 53" +1/ 5)" 58/ - 4&. +1/ 5)" 58/ - #00 આ໌ม਺ͷස౓͸ɺޠ͋ͨΓͷ૬ରස౓ #00͸ɺ$&'3ؒ΍-ؒͷ͕ࠩएׯ͋Γͦ͏

•  ରࢄ෍ਤ 26 -
$&'3 53" 4&. #00 ૬ؔ܎਺͸ɺ4QFBSNBOͷํ๏

•  ڃ಺૬ؔ܎਺ JOUSBDMBTTDPSSFMBUJPODPFGGJDJFOU •  ஋͕େ͖͍΄Ͳɺूஂ಺ͷྨࣅੑʢ͜͜Ͱ͸ɺ-ͷӨڹʣ͕େ͖͍ ʢ؍ଌ஋ͷಠཱੑ͕௿͘ɺϚϧνϨϕϧ෼ੳΛߦ͏ඞཁੑ͕ߴ͍ʣ •  ͭͷ໨҆ͱͯ͠ɺҎ্ͰޮՌྔখɺҎ্ͰޮՌྔதɺҎ্ ͰޮՌྔେ
FH )PY .PFSCFFL BOEWBOEF4DIPPU 3BVEFOCVTIBOE-JV 27 ICC TRA -0.00 SEM 0.25 BOO 0.11 4NJUIͷํ๏ 4&.ͱ#00ʹ͸ɺ-ͷӨڹ͕͋Γͦ͏

-ͷҧ͍Λߟྀ͠ͳ͍Ϟσϧ .PEFMʣ 28 Cumulative Link Mixed Model fitted with the
Laplace approximation formula: CEFR ~ TRA + SEM + BOO link threshold nobs logLik AIC niter max.grad cond.H logit flexible 1000 -1172.02 2356.04 5(0) 4.87e-07 2.7e+03 Coefficients: Estimate Std. Error z value Pr(>|z|) TRA -0.08146 0.03846 -2.118 0.034174 * SEM -0.05194 0.01565 -3.320 0.000901 *** BOO -0.11488 0.05357 -2.144 0.032005 * --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Threshold coefficients: Estimate Std. Error z value 1|2 -1.7216 0.2351 -7.323 2|3 0.2379 0.2280 1.044 3|4 2.2763 0.2641 8.618 ࢖༻ͨ͠3ͷؔ਺ͷઆ໌ Ϟσϧࣜ ܎਺ͷਪఆ஋ ઢܗϞσϧͱҧ͍ɺͦͷ··໨తม਺ʹର͢Δ આ໌ม਺ͷӨڹ౓ͱͯ͠ղऍ͢Δ͜ͱ͕Ͱ͖ͳ͍ ʢݶɹɹ ɹքޮՌΛผ్ܭࢉ͢Δඞཁ͕͋Δʣ ੾ยͷਪఆ஋ ͨͱ͑͹ɺc͸ɺ໨తม਺͕ҎԼ͔ɺҎ্͔Λ ༧ଌ͢Δࣜͷ੾ยʢ؍ଌม਺Ͱ͸ͳ͘જࡏม਺ͳͷͰɺ ஋͸ศٓతͳ΋ͷʣ Ϟσϧͷઃఆɺద߹౓ͳͲ ݁Ռͷݟํ

-ͷҧ͍Λߟྀ͠ͳ͍Ϟσϧ .PEFMʣ 29 Cumulative Link Mixed Model fitted with the
Laplace approximation formula: CEFR ~ TRA + SEM + BOO link threshold nobs logLik AIC niter max.grad cond.H logit flexible 1000 -1172.02 2356.04 5(0) 4.87e-07 2.7e+03 Coefficients: Estimate Std. Error z value Pr(>|z|) TRA -0.08146 0.03846 -2.118 0.034174 * SEM -0.05194 0.01565 -3.320 0.000901 *** BOO -0.11488 0.05357 -2.144 0.032005 * --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Threshold coefficients: Estimate Std. Error z value 1|2 -1.7216 0.2351 -7.323 2|3 0.2379 0.2280 1.044 3|4 2.2763 0.2641 8.618 ม਺ͱ΋ɺ शख़౓Ϩϕϧ্͕͕ΔʹͭΕͯɺ ༗ҙʹݮগ͍ͯ͘͠ આ໌ม਺͕ݴޠ߲໨ͷස౓ͳͷͰɺ ަޓ࡞༻͸૝ఆ͍ͯ͠ͳ͍

-ͷҧ͍Λߟྀ͢ΔϞσϧ .PEFM 30 Cumulative Link Mixed Model fitted with
the Laplace approximation formula: CEFR ~ TRA + SEM + BOO + (1 | L1) link threshold nobs logLik AIC niter max.grad cond.H logit flexible 1000 -1152.72 2319.45 424(863) 2.91e-04 1.6e+03 Random effects: Groups Name Variance Std.Dev. L1 (Intercept) 0.2369 0.4868 Number of groups: L1 3 Coefficients: Estimate Std. Error z value Pr(>|z|) TRA -0.07840 0.03874 -2.024 0.0430 * SEM -0.04044 0.01699 -2.380 0.0173 * BOO -0.09567 0.05466 -1.750 0.0801 . --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Threshold coefficients: Estimate Std. Error z value 1|2 -1.7850 0.3705 -4.818 2|3 0.2469 0.3660 0.674 3|4 2.3368 0.3891 6.005 ϥϯμϜ੾ยϞσϧ SBOEPNJOUFSDFQUNPEFM ϥϯμϜ܏͖ SBOEPNTMPQF ʹ͍ͭͯ͸ޙड़ .PEFMͱൺ΂ͯɺ܎਺ͷ༗ҙੑ͕௿͘ͳ͍ͬͯΔ ɹˣ -ͷҧ͍Λߟྀ͠ͳ͍ͱɺशख़౓Ϩϕϧͷ༧ଌʹର͢Δ د༩౓͕ʢෆ౰ʹʣߴΊʹݟੵ΋ΒΕͯ͠·͏ ϚϧνϨϕϧϞσϧͳͷͰɺ ϥϯμϜޮՌͷ෼ࢄͱඪ४ภࠩͳͲ͕දࣔ

ϞσϧͷධՁ 31 ===================================================== Model 1 Model 2 ----------------------------------------------------- TRA -0.08
* -0.08 * (0.04) (0.04) SEM -0.05 *** -0.04 * (0.02) (0.02) BOO -0.11 * -0.10 (0.05) (0.05) 1|2 -1.72 *** -1.78 *** (0.24) (0.37) 2|3 0.24 0.25 (0.23) (0.37) 3|4 2.28 *** 2.34 *** (0.26) (0.39) ----------------------------------------------------- AIC 2356.04 2319.45 BIC 2385.49 2353.80 Log Likelihood -1172.02 -1152.72 Num. obs. 1000 1000 Groups (L1) 3 Variance: L1: (Intercept) 0.24 ===================================================== *** p < 0.001; ** p < 0.01; * p < 0.05 "*$ΛݟΔͱɺ .PEFMͷํ͕ϕλʔ

•  ໬౓ൺݕఆ •  .PEFMͱൺ΂ͯɺ.PEFM͸༗ҙʹઆ໌ྗ͕ߴ͍ 32 Likelihood ratio tests of cumulative
link models: formula: link: threshold: Model 1 CEFR ~ TRA + SEM + BOO logit flexible Model 2 CEFR ~ TRA + SEM + BOO + (1 | L1) logit flexible no.par AIC logLik LR.stat df Pr(>Chisq) Model 1 6 2356.0 -1172.0 Model 2 7 2319.4 -1152.7 38.593 1 5.22e-10 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

•  .PEFMͰ༗ҙͩͬͨઆ໌ม਺ͷՄࢹԽ 53" 4&. 33 TRA effect plot TRA
CEFR (probability) 0.1 0.2 0.3 0.4 0.5 2 4 6 8 10 = CEFR 1 0.1 0.2 0.3 0.4 0.5 = CEFR 2 0.1 0.2 0.3 0.4 0.5 = CEFR 3 0.1 0.2 0.3 0.4 0.5 = CEFR 4 SEM effect plot SEM CEFR (probability) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0 5 10 15 20 = CEFR 1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 = CEFR 2 0.0 0.1 0.2 0.3 0.4 0.5 0.6 = CEFR 3 0.0 0.1 0.2 0.3 0.4 0.5 0.6 = CEFR 4 53"ɺ4&.ͱ΋ʹɺ௿͍शख़౓Ϩϕϧʹݦஶͳ߲໨ ͜ͷਤͰ͸ɺ͋ΔཁҼ͕มԽͨ͠৔߹ͷ$&'3Ϩϕϧͷ֬཰ʹ༩͑ΔӨڹΛՄࢹԽ ʢมԽ͢ΔཁҼҎ֎͸ɺฏۉ஋ʹݻఆʣ

݁Ռͷ·ͱΊʢʴ؆୯ͳߟ࡯ʣ •  शख़౓Ϩϕϧͷ༧ଌʹ͓͚Δ-ͷӨڹ͸ແࢹͰ͖ͳ͍ •  ಉ͡ϨϕϧͷֶशऀͰ͋ͬͯ΋ɺϝλஊ࿩ඪࣝͷ࢖༻܏޲͕ҟͳΔ •  σʔλͷ֊૚ੑΛద੾ʹߟྀ͠ͳ͍ͱɺਖ਼͍͠ൃୡࢦඪͷಛఆ͕Ͱ͖ͳ͍ •  शख़౓Ϩϕϧ্͕͕ΔʹͭΕͯɺ53"ͱ4&.͕༗ҙʹݮগ͍ͯ͘͠
•  ઀ଓදݱ 53" ͷݮগ͸ɺฏۉจ௕ͷൃୡ΍౷ޠߏ଄ͷෳࡶԽͱؔ܎ •  ਓশ୅໊ࢺ 4&. ͷݮগ͸ɺߏจͷଟ༷Խͱؔ܎ 34

ิ଍ •  ૝ఆ͠ಘΔϞσϧͷશύλʔϯ૯౰ͨΓൺֱ "*$ •  ม਺બ୒͞Ε߲ͨ໨͔ΒͳΔશͯͷ૊Έ߹Θͤ ɹº •  த৺Խʢूஂฏۉத৺Խʣͷ༗ແ
ɹº •  ϥϯμϜ܏͖ͷԾఆͷ༗ແ ɹɹɹˣ •  લܝͷ.PEFM͕࠷ྑͳϞσϧ ౷ܭతม਺બ୒ʴ౷ܭతϞσϧબ୒ͷ૯౰ͨΓൺֱ ɹͱ͍͏ࢥߟ์غͷྗٕʢসʣ 35

ࠓޙͷ՝୊ •  ΑΓଟ͘ͷ-ͷൺֱ •  $PNJOHTPPO •  -ͷӨڹͷ౓߹͍ •  ͨͱ͑͹ɺʮ೔ຊਓ͸ɺݴޠ߲໨9Λɺಉ͡ϨϕϧͷλΠਓͷ:ഒ࢖༻͢Δʯ ͳͲͷ৘ใΛ஌Γ͍ͨʢճؼ෼ੳͷ࿮૊ΈͷதͰʣ
•  ϚϧνϨϕϧͰ͸ͳ͍ॱংϩδοτϞσϧͷઆ໌ม਺ʹɺʮ˓˓ਓμϛʔʯม਺ΛೖΕ Δʁ •  ϚϧνϨϕϧଟ߲ϩδοτϞσϧͰɺशख़౓ϨϕϧΛߟྀͨ͠-൑ผΛߦ͏ʁ ʢͨͩ͠ɺ34UBOͩͱɺϕΠζܥͷϞσϦϯάʹͳΔ໛༷ʣ 36 Ͳͳ͔ͨΑ͍ΞΠσΞΛ͓࣋ͪͷํɺ͝ڭࣔ௖͚Ε͹޾͍Ͱ͢N @@ N

37 5IBOLZPVGPSZPVSUJNFBOEBUUFOUJPO 5IJTXPSLXBTTVQQPSUFECZ+414,",&/)*(SBOU/VNCFS,

学習者コーパス研究におけるマルチレベル順序ロジットモデルの活用

学習者コーパス研究におけるマルチレベル順序ロジットモデルの活用

More Decks by Yuichiro Kobayashi

Other Decks in Research

Featured

Transcript