Processing effort is a poor predictor of cross-linguistic word order frequency

41608

Authors

nan

Abstract

AbstractSome have argued that word orders which are more difficult to process should be rarer cross-linguistically. Our current study fails to replicate the results of Maurits, Navarro, and Perfors (2010), who used an entropy-based Uniform Information Density (UID) measure to moderately predict the Greenbergian typology of transitive word orders. We additionally report an inability of three measures of processing difficulty — entropy-based UID, surprisal-based UID, and pointwise mutual information — to correctly predict the correct typological distribution, using transitive constructions from 20 languages in the Universal Dependencies project (version 2.5). However, our conclusions are limited by data sparsity.

Keywords

nan

Full Content

Proceedings of the 24th Conference on Computational Natural Language Learning, pages 245–255 Online, November 19-20, 2020. c⃝2020 Association for Computational Linguistics https://doi.org/10.18653/v1/P17 245 �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`SOU$!!H!$-Rab^cY#Fd -PQR`SJ C!'!$#,D(") D! ! D#'!H!!"$)") !E )!e",# D "# !$- , E D!\%# !_$ N$ ",!- f!$HIKHW! g)E-#E D!]D!%!_$ N$I ",!O])%,-RahbSJY)F!'!$- D! !#$! )%!I FD# $ ,#$ D!$$!# )i D!(#$!%) '# !E H# !E) D!"$!'#,! !)* HW! IG$ ,# #! #E,# #! FD!$! D!'!$H#E)HW! #$!#EI W# ! J]D! !"$ ",! #$! D! !E )!e",# D!'!$( #%!F)$E)$E!$"# !$ D# %) '# !E D!% D!G$ ",# !JV D D!)#,)*# D!'# %)$!$)H !e",## )- )%!D#'!#$ !E%)$! $! ! F)$d D# "$) ! E*G , (OH$)#E,( )I $ !ES '!$ !,("$)")$ )#, ) ("),) #,*$!I j ! (OY#Fd -PQQ`-PQR`c_!Ek! Dd#! #,J- PQRhSJ# #! #$! ,#%!E )!'),'!*$)%F)$E )$E!$ D# #$!%)$!E*G , ( )"$) ! ))$E!$ D# #$!!# !$ )"$) ! J]D!$!*)$!-#F)$E)$I E!$ D#  %)$!E*G , )"$) ! D) ,EH!$#$!$ $) I,   #,,(JV D D ) !e -L# $ ! #,JOPQRQS# !%" !E ) ! D!&*)$%T*)$I %# ) !  (O&T SD(") D!  )!e",# D! E $H )#,"# !$ )* HW! I'!$HI)HW! )I $ ) JL# $ ! #,JOPQRQSOD! !*)$ DLMNS ) $ !E %",G!E ! )*!'! -$#d!# D ") H,!)$E!$)* HW! -'!$H-#E)HW! H( E!'# )*$)%&T E!#,J LMNl $! , D)F!E# )  ! DD$#d *)$fCK#EfKC-#E# )  ! ,)F$#d *)$CKf- ) $#$( ) D!)H !$'!E $) I,   *$!j ! (JVD,!LMN !E#! $)"(IH# !E&T %!# $!- D!$!#$!) D!$%!# $! )*"$) ! E*I G , ( D# ) ,EH!#"",!E ) D! E!$,(j ! I ))*FD! D!$,# #! !'),'!*$)%E*G , I )I"$) ! F)$E)$E!$ )!# (I )I"$) ! F)$E )$E!$ J]D "#"!$ D!$!*)$! !!d ) ! D! ,#% D# ,# #! !'),'! D F#(-H( )%"#$ D! "$!E !E"$) ! E*G , ()* D! e") H,! HW! I'!$HI)HW! F)$E)$E!$ J])E) D -F! ! D!*),,)F D$!!%! $ )*"$) ! E*G , (- E!G!E D!*),,)F …

Content truncated to 500 words

Labeled Text

AbstractSome have argued that word orders which are more difficult to process should be rarer cross-linguistically. Our current study fails to replicate the results of Maurits, Navarro, and Perfors (2010), who used an entropy-based Uniform Information Density (UID) measure to moderately predict the Greenbergian typology of transitive word orders. We additionally report an inability of three measures of processing difficulty — entropy-based UID, surprisal-based UID, and pointwise mutual information — to correctly predict the correct typological distribution, using transitive constructions from 20 languages in the Universal Dependencies project (version 2.5). However, our conclusions are limited by data sparsity.

Publication Details

Paper ID:
41608
Year:
2020
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan