modelling interventions for children with word …€¢...

38
Intervening to alleviate lexical retrieval difficul2es in children Challenges of research within an educa2onal neuroscience framework Michael Thomas

Upload: duongthu

Post on 05-May-2018

216 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Intervening  to  alleviate  lexical  retrieval  difficul2es  in  children  

Challenges  of  research  within  an  educa2onal  neuroscience  framework  

Michael  Thomas  

Page 2: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Michael(Thomas(

Anna(Fedor(

Wendy(Best(

Lucy(Hughes(

Jackie(Masterson(

Anna(Kapikian(

1"

Silvia(Roncoli(

•  Study  of  typical  development  of  produc2ve  vocabulary  •  Study  of  atypical  development  •  Interven2on  –  which  technique  works  best  •  Computa2onal  modelling  of  above  to  understand  mechanisms  

Hala  Alireza  

Page 3: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Word  finding  difficul2es  •  Developmental  problems  in  produc2ve  vocabulary  •  Characteris2c  behaviours:  

–  the  use  of  filler  words  (e.  g.,  um),  empty  words  (thing)  or  general  verbs  (doing)  instead  of  more  specific  words  

–  the  use  of  a  similar  sounding  word  (canister  for  camera)  –  the  use  of  a  word  with  a  similar  meaning  or  in  the  same  category  (2ger  for  lion)  

–  hesita2on  –  repe22on  of  words  or  phrases  –  rephrasing  what  they  are  saying  –  the  use  of  gesture  (miming  cleaning  teeth  for  toothbrush)  –  talking  about  their  difficulty  (“I  know  it,  but  I  can’t  think  of  it”)  

Affects  educa0onal  achievement,  self-­‐esteem  

Primary  but  not  necessary  sole  language  deficit  

Page 4: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Interven2ons  for  WFD  

Seman2c  therapy   Phonological  therapy  

Page 5: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Two  case  studies  

0

5

10

15

20

25

Num

ber c

orre

ct

Asessment

Amy Naming

Page 6: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Two  case  studies  

0

5

10

15

20

25

Num

ber c

orre

ct

Assessment

Magda naming

Page 7: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Challenges  •  What  is  the  link  between  mechanism  and  interven2on?  

Best,  W.  (2003)  Finding  the  right  approach:  how  do  SLTs  tackle  word-­‐finding  problems  in  children.  Bulle2n  of  the  Royal  College  of  Speech  and  Language  therapists,  September  p.5-­‐6.  

Page 8: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Challenges  •  Therapy  is  a  discovery  process  that  is  specific  to  the  child  –  

no  general  principles?  •  Unclear  whether  best  to  work  on  areas  of  weakness  or  use  

areas  of  strength  –  depends  on  child?  changes  over  2me?  

Best,  W.  (2003)  Finding  the  right  approach:  how  do  SLTs  tackle  word-­‐finding  problems  in  children.  Bulle2n  of  the  Royal  College  of  Speech  and  Language  therapists,  September  p.5-­‐6.  

Page 9: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Challenges  

•  Ask  the  child:  what  worked  for  you?  

Amy Magda How much did you enjoy taking part in WORD?

5 5

How helpful was it to think about the MEANING of words?

4 3

How helpful was it to think about the SOUNDS in words?

3 5

What helps you most when you are stuck?

Chunking out; doing the actions; sometimes spelling.

I show someone the action… Tell a teacher or friend.

Do you think finding words is easier now?

At the beginning 1 and now it is 3.

A little bit easier

(Phonological  worked  best)  

(Seman0c  worked  best)  

Page 10: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

CLINICIANS  

MECHANISMS  

A^empt  to  close  the  gap  from  both  direc2ons:    (1)  Talk  to  clinicians  about  implicit  causal  theory    (2)  Build  formal  models  of  atypical  neuro-­‐computa2onal  systems  and  simulate  effects  of  interven2on  

Page 11: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Computa2onal  modelling  as  a  way  to  inves2gate  mechanism  

Ar2ficial  neural  networks  Real  neural  networks  

Page 12: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Word Building Intervention

•  Does it matter when phonological problem resolves?

•  Compare with training on orthographic-phonology relationships

Othograhy

Semantics

Context

Orthograhy Phonology

Meaning

Context

Harm,  McCandliss  &  Seidenberg  (2003)  

Page 13: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Word Building Intervention

•  Does it matter when phonological problem resolves?

•  Compare with training on orthographic-phonology relationships

Othograhy

Semantics

Context

Orthograhy Phonology

Meaning

Context

Harm,  McCandliss  &  Seidenberg  (2003)  

A  good  model  but  the  ONLY  model  of  interven2on!  

Page 14: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Model  for  inves2ga2ng  the  principles  of  interven2on  

•  Simple  network  •  Easy  to  visualize  behaviour  

•  Esp.  forma2on  of  internal  representa2ons  

Page 15: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Input  space  

Range:  -­‐0.5,  0.5  10  000  items  

Page 16: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Simple  learning  problems  

Diagonal     Islands  •  Irregular  •  Regular  

Page 17: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Developmental  deficits  

Low  connec0vity  (C  =  0.3)   Low  temperature  (T  =  0.5)  

Page 18: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Interven2on  pa^erns  

Target  pattern  

Training  pattern  

Intervention  1  

Intervention  2  

Intervention  3  

Intervention  4  

Intervention  5  

Intervention  6  

 

 

Page 19: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Trajectories  

Page 20: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of
Page 21: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Sample  anima2ons…  

•  h^ps://www.youtube.com/watch?v=_RCSFhoFg6s  

•  h^ps://www.youtube.com/watch?v=WIrr0Jr6kfo  

•  h^ps://www.youtube.com/watch?v=WIrr0Jr6kfo  

Page 22: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Principles  

•  Importance  of  2ming  of  interven2on  •  Specificity  to  deficit  type?  •  Specificity  to  problem  domain?  

Page 23: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Back  to  word-­‐finding  difficul2es  •  Aim:  

– Model  individual  profiles  of  developmental  deficits  in  naming  

– Model  interven2ons  –  which  is  most  effec2ve?  –  Test  predic2ons  against  real  data  

•  Key  data:  compare  performance  on  four  tasks,  child  vs.  model  –  Picture  naming  –  Picture  comprehension  –  Seman2c  associa2ons  –  Phonological  ability  

Page 24: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Picture  naming  task  

“Square”  

“Cocoon...some  beach  thing”  

“Ghost”  (via  goat)  

Target:    Triangle  

Target:    Coconut  

Target:    Llama  

Page 25: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Picture  comprehension  

Time  1:    Is  it  pineapple?   YES  

Time  2:    Is  it  melon?   NO  

Combined  accuracy  score:    1  

Page 26: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Picture  comprehension  

Time  1:    Is  it  bu^erfly?   YES  

Time  2:    Is  it  wasp?   YES  

Combined  accuracy  score:    0  

26  

Page 27: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Seman0c  associa0ons  

Page 28: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Seman0c  associa0ons  

Page 29: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Phonological  ability  (nonword  repe00on)  

pennel  

commeecitate  

perplisteronk  

ski2cult  

ballop  

woogalamic    

29  

Page 30: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Two  case  studies  

-­‐1.8  

-­‐3.4  

-­‐3.1  

-­‐1.3  

-­‐1.8  

-­‐3.2  

-­‐4.2  

-­‐2.5  

Seman2cs   Phonology   Naming   Comprehension  

Comparison  to  controls  (SDs  from  mean)  

Amy   Magda  

AMY:  More  a  phonological  problem    MAGDA:  …  and  a  seman2c  problem  

Page 31: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Model  of  naming  development  

Word%meaning%

Word%sounds%

SEMANTICS   PHONOLOGY  

Page 32: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Amy  

Word%meaning%

Word%sounds%

SEMANTICS   PHONOLOGY  

Page 33: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Magda  

Word%meaning%

Word%sounds%

SEMANTICS   PHONOLOGY  

Page 34: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Amy$

Word%meaning%

Word%sounds%

SEMANTICS) PHONOLOGY)

!

Amy:

-1.8sd -3.4sd -3.1sd -1.3sd

!

Magda:

-1.8sd -3.2sd -4.2sd -2.5sd

Magda%

Word%meaning%

Word%sounds%

SEMANTICS) PHONOLOGY)

Page 35: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Simulate  the  interven2on  

!

Amy   Magda  

Page 36: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Simulate  the  interven2on  

!

Amy   Magda  

Data:            Phonological  works  best                                                                      Seman0c  works  best        

Page 37: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Conclusions  •  Link  interven2on  with  underlying  mechanism  

•  Use  computa2onal  models  to  advance  theory  /  generate  predic2ons  

•  Work  with  clinicians  to  understand  (a)  their  implicit  causal  theories  (b)  what  kinds  the  interven2ons  they  use  

•  Interven2on  studies  =  lots  of  challenges!  

CLINICIANS'

MECHANISMS'

Page 38: Modelling interventions for children with word …€¢ Unclear)whether)bestto)work)on)areas)of)weakness)or)use) areas)of)strength)–depends)on)child?)changes)over)2me?) Best,)W.)(2003))Finding)the)rightapproach:)how)do)SLTs)tackle)wordUfinding)problems)in)children.)Bulle2n)of

Acknowledgements  

•  Colleagues:  Wendy  Best,  Jackie  Masterson,  Anna  Fedor,  Lucy  Hughes,  Silvia  Roncoli,  Donna-­‐Lynn  Shepherd,  Liory  Fern-­‐Pollak,  Anna  Kapikian,  Hala  Alireza  

•  Project  website:  h^ps://sites.google.com/site/wordfinding/home