Mapping  between  par;cles  and  inten;onality  in  Japanese  preferen;al  looking   Franklin  Chang,  University  of  Liverpool   Tessei  Kobayashi,  NTT  Communica;on  Sciences  Laboratories,  NTT  Corp.  

Overview   Preferen;al  looking  paradigms  have  been  used  to  examine  the   rela;onship  between  scenes  and  uferances  (Naigles,  1990).    But  ogen   scenes  can  vary  in  mul;ple  ways  (e.g.,  causa;vity,  synchronicity,   mo;on)  and  uferances  can  also  differ  in  mul;ple  ways  (e.g.,  number  of   noun  phrases,  structure,  role-­‐func;on  mapping).    Here  we  seek  to   disentangle  these  features  by  using  a  language  that  allows  us  to  match   the  number  of  noun  phrases  (Japanese)  and  scenes  which  are  similar  in   mo;on  and  configura;on  of  par;cipants.    Our  study  examines  the   rela;onship  between  Japanese  par;cles  and  whether  an  ac;on  is   inten;onality  caused.  

Results   Dependent  measure:  empirical  logit  of  looking  ;me  to  matching  screen   Mixed  models  with  age  (20,  24,  28),  sec;on  (baseline,  linguis;c),  and   par;cle  (SUBJ,  OBJ)  as  predictors  and  subject  and  verb  as  random   factors.    Found  significant  three  way  interac;on  (t=2.49,  p  <  0.02).      

Proper&es  of  Japanese   Japanese  is  a  pro-­‐drop  language  where  phrases  are  marked  by  par;cles   indica;ng  their  syntac;c  func;on.      apple  OBJ  hit  -­‐>  “someone/he  hit  the  apple”      apple  SUB  fell  -­‐>  “the  apple  fell”      dog  SUB  hit  -­‐>  “the  dog  hit  something/it”   Object  par;cle  is  used  to  mark  the  object  of  an  inten;onal  ac;on.  

20  months:  No  main  effects  or  interac;ons.   24  months:  Significant  mismatch  effect  (t  =  2.1,  p  <  0.04),  where   children  look  at  the  non-­‐inten;onal  video  when  they  hear  a  phrase  with   an  object  par;cle.       28  months:  Significant  main  effects  of  sec;on  (t  =  2.26,  p  <0.01),  main   effect  of  par;cle  (t  =  4.9,  p  <  0.001),  and  a  significant  interac;on  (t  =   4.6,  p  <  0.001).    This  is  due  to  a  significant  baseline  difference  (t  =  4.7,  t   <  0.001)  due  to  the  pairing  of  the  novel  verb  and  the  matching  screen  in   the  exposure  phase.  

Our  study   Matched  inten;onal  ac;on  and  non-­‐inten;onal  ac;ons  were  used.   Scenes  differed  in  the  par;cle  that  would  normally  be  used.                            “duck-­‐OBJ  give”                                                                                  “duck-­‐SUBJ  fall”  

Inten;onality  was  emphasized  by  showing  two  afempts  for  inten;onal   ac;ons.    A  trial  presented  the  scene  twice  and  then  a  test  phrase  with  a   baseline  looking  phase  and  two  presenta;ons  of  a  linguis;c  s;mulus.                              exposure  phase                                                              pa                                                                                                            “pateta”    (novel  verb)  

Another  mixed  model  by  par;cle  shows  that  the  mapping  between  the   SUBJ  par;cle  and  meaning  does  not  change  over  these  ages.    All   development  changes  seem  to  be  concentrated  in  the  OBJ  par;cle.  

Findings   These  results  did  not  conform  to  the  predic;ons  of  theories  of   preferen;al  looking.    The  significant  mismatch  effect  for  object  par;cles   is  consistent  with  a  number  of  noun  phrases  account,  but  we  did  not   find  a  strong  match  effect  for  subject  par;cles.    The  mismatch  effect  is   difficult  to  explain  in  theories  that  assume  that  preferen;al  looking   indexes  the  match  between  scenes  and  innate/learned  linguis;c   knowledge.        

Connec&onist  Model  of  Preferen&al  Looking   Chang,  Dell,  &  Bock  (2006)  proposed  that  error  in  predic;on  from  scene   meaning  could  be  used  to  explain  preferen;al  looking  behavior.                Error  =  Actual  sequence  –  Predicted  sequence  from  meaning            Match  Preference  =  Mismatch  error/(Mismatch  error  +  Match  error)   A  Japanese  version  of  the  Dual-­‐path  model  was  created  and  trained  on   intransi;ves,  transi;ves,  and  da;ves  (like  Chang,  2009).    Model  tested   with  transi;ve/intransi;ve  message  with  a  novel  verb  paired  with  single   noun  phrase  with  either  subject  or  object  par;cle.    

                                         baseline  phase  (4  secs)                                            linguis;c  phase  (4  secs)                                                                                                                                                                                    “duck  OBJ  pateta”                                                    “duck  OBJ  pateta”  

Object  mismatch  is  greater  than  subject  match  

90  20-­‐month,  24-­‐month,  and  28-­‐month  old  Japanese  children  were   tested.  

Predic&ons     If  the  number  of  noun  phrases  is  cri;cal  (Fisher,  2002),  then  since  both   phrases  have  only  one  noun  phrase,  there  should  be  a  strong  match   effect  for  subject  par;cles  and  a  mismatch  effect  for  object  par;cles.   If  the  children  have  a  link  between  the  par;cles  and  inten;onality,  then   we  would  expect  a  match  effect  for  both  par;cles.  

Model  shows  preference  to  match  subject  par;cle  and  transi;ve   message,  but  there  is  a  slight  mismatch  preference  for  object  par;cles   (below  0.5)  early  in  development.    Although  there  is  a  match  preference   at  the  par;cle,  it  is  swamped  by  a  mismatch  preference  at  the  noun,   where  two  argument  structures  have  more  trouble  predic;ng  the  first   noun  than  one  argument  structures.       This  suggests  that  mismatch  preference  in  this  study  could  arise  from   conflict  of  cues  at  DIFFERENT  sentence  posi;ons  as  is  suggested  in   Chang,  et  al.  (2006).  

References   Chang,  F.  (2009).  Learning  to  order  words:  A  connec;onist  model  of  heavy  NP  shig  and  accessibility  effects  in  Japanese  and  English.  Journal  of  Memory  and  Language,  61(3),  374–397.   Chang,  F.,  Dell,  G.  S.,  &  Bock,  J.  K.  (2006).  Becoming  syntac;c.  Psychological  Review,  113(2),  234–272.   Fisher,  C.  (2002).  Structural  limits  on  verb  mapping:  the  role  of  abstract  structure  in  2.5-­‐year-­‐olds'  interpreta;ons  of  novel  verbs.  Developmental  Science,  5(1),  55–64.   Naigles,  L.  (1990).  Children  use  syntax  to  learn  verb  meanings.  Journal  of  Child  Language,  17(02),  357–374.  

Mapping between parlcles and intenlonality in ...

All development changes seem to be concentrated in the OBJ parlcle. Mapping between parlcles ... apple OBJ hit -‐> “someone/he hit the apple” apple SUB fell ...

1MB Sizes 1 Downloads 142 Views

Recommend Documents

Inferring Semantic Mapping Between Policies and ... - Semantic Scholar
Legend: WRB: Whadverb, PRP: Personal pronoun, VB: Verb, JJ: Adjective, IN: Preposition,. NN: Noun (singular), NNS: Noun(plural), MD: Modal, DT: Determiner, VBG: Verb. Fig. 2. An example of tagged policy statement. This grammatical mapping process is

Mapping of the bilateral arrangements between the ICMRA Members
BR CA CH CN EU1 IE JP SG UK US / CA EU1 SG / NZ8 SG9 US10. Brazil (BR) ... Germany (DE) - The Paul-Ehrlich-Institute of the Federal. Republic of Germany ...

Mapping communication and collaboration in ...
ARCS, Austrian Research Centers, Seibersdorf (Austria). Address for .... downloaded the records of all articles, reviews, letters, and notes from the journals identified ..... need to have the appropriate data, preferably micro data about the positio

Inferring Semantic Mapping Between Policies ... - Research at Google
open-source social networking sites, namely Diaspora and Friendica. Our ..... Page 10 ... most popular social networks with a published privacy policy are closed ...

IOF/SEEOA Developing seminar for mapping Mapping and printing ...
IOF/SEEOA Developing seminar for mapping. Mapping and printing sprint and urban orienteering maps. 15-18 th. October 2015 Sofia, Bulgaria, hotel Legends.

PGIS and mapping for conservation in Namibia
data, mapping and GIS as tools to support CBNRM and for ... Competing land use activities take place within conservancies (for example, farming, wildlife, settlement, mining). ... use a variety of spatial data sets to develop appropriate maps.

Direct mapping of the temperature and velocity gradients in discs ...
(2017) recently intro- duced a method to map the thermal and density gas structure of ...... We use the Monte Carlo radiative transfer code MCFOST. (Pinte et al.

IOF/SEEOA Developing seminar for mapping Mapping and printing ...
Seminar fee: 30eur per person (includes seminar materials, lectures, practice work, free wi fi, using of hotel conference room, water and coffe/tea breaks with ...

Mapping-Open-Data-in-Armenia_Armenian.pdf
Page 3 of 10. Page 3 of 10. Mapping-Open-Data-in-Armenia_Armenian.pdf. Mapping-Open-Data-in-Armenia_Armenian.pdf. Open. Extract. Open with. Sign In.

FIRe seveRITy mAPPIng: mAPPIng THe - Bushfire CRC
financial support was provided by Bushfires Nt and student support from Charles darwin university. ... mapping system (www.firenorth.org.au) enhancing the ...

Mapping-Open-Data-in-Armenia_Armenian.pdf
Mapping-Open-Data-in-Armenia_Armenian.pdf. Mapping-Open-Data-in-Armenia_Armenian.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying ...

Mapping-Open-Data-in-Armenia_Armenian.pdf
Whoops! There was a problem loading this page. Mapping-Open-Data-in-Armenia_Armenian.pdf. Mapping-Open-Data-in-Armenia_Armenian.pdf. Open. Extract.

The Relationship Between Cholesterol and Survival in ...
teristic curves demonstrated a best cut-off value of 5.2 mmol/l (200.8 mg/dl) as being the best predictor of ... 11, 2003. © 2003 by the American College of Cardiology Foundation ... physician, or through the Hospital Information System in.

Relations between microhabitat use and limb shape in ...
Blackwell Science, LtdOxford, UKBIJBiological Journal of the Linnean ... 1Laboratory of Functional Morphology, Biology Department, University of ...... were captured under Arizona State Game and Fish ... Bartlett, Travis La Duc, Wayne Van Devender, B

Nonlinearities in the relationship between debt and ...
Jan 5, 2017 - tionship between public debt and growth employing long time series data from up to. 27 countries. My empirical approach recognises that ...

Relations between microhabitat use and limb shape in ...
ods that take into account the relationships between the groups ..... species into account, phylogenetic analyses were used. As these ...... access to females. ... We would also like to express our gratitude ... American Zoologist 23: 347–361.

Collaborations between NCREE in Taiwan and MAE ...
displacements and reacting forces. An Executor executes all ... Furthermore, there are some auxiliary modules other than the above essential modules ... camera control and instant visualization. The auxiliary modules in UI-SimCor and NCREE-ISEEdb wor

Mobile Mapping
Apr 24, 2013 - MOBILE MAPPING WITH ANDROID DEVICES. David Hughell and Nicholas Jengre .... 10. Windows programs to assist Android mapping .

Mobile Mapping
MOBILE MAPPING WITH ANDROID DEVICES. David Hughell and Nicholas Jengre. 24/April/2013. Table of Contents. Table of Contents .

pdf-62\theatre-making-interplay-between-text-and-performance-in ...
... of the apps below to open or edit this item. pdf-62\theatre-making-interplay-between-text-and-performance-in-the-21st-century-by-duska-radosavljevic.pdf.