Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’...

37
Nega%ve Polar Ques%on Types in English Maribel Romero, Anja Arnhold, Be?na Braun and Filippo Domaneschi NELS 47 Umass at Amherst, October 1416, 2016

Transcript of Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’...

Page 1: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Nega%ve  Polar  Ques%on  Types    in  English  

Maribel  Romero,  Anja  Arnhold,  Be?na  Braun  and  Filippo  Domaneschi  

 NELS  47    

Umass  at  Amherst,  October  14-­‐16,  2016  

Page 2: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  PQ  Forms  

•  Different  Polar  Ques%on  (PQ)  forms:  

(1)  Did  John  drink?            (2)  Did  John  not  drink?  (3)  Didn’t  John  drink?  (4)  DID  John  (really)  drink?  

 •  All  these  forms  raise  the  same  issue  {p,  ¬p},  i.e.,  they  induce  a  

choice  between  p  and  ¬p.    •  But  the  forms  cannot  be  used  interchangeably.  The  choice  of  form  

depends  (perhaps  among  other  things)  on  the  epistemic  bias  of  the  speaker:  –  Original  bias  –  Contextual  evidence  bias  

PosQ  LowNQ  HiNQ  really-­‐PosQ  

2  

Page 3: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  Bias  

•  Original  bias:  e.g.  PosQ  vs.  HiNQ  

(5)    Original  bias  (for  a  proposi%on  p):    Speaker’s   belief   or   expecta%on   –possibly   private–   that   p   is   true,  based  on  her  epistemic  state  prior  to  the  current  situa%onal  context  and  conversa%onal  exchange.    

(6)      Scenario:  Lawyer  asking  unbiased  ques%ons  at  court:    a.  Did  you  see  the  culprit  hit  the  vic%m?    b.  #  Didn’t  you  see  the  culprit  hit  the  vic%m?  

(7)      A  PosQs  p?  doesn’t  necessarily  express  original  bias  (for  p  or  ¬p).    A  HiNQ  n’t  p?  necessarily  conveys  original  bias  for  p.    

 

3  [Ladd  1981,  Romero  &  Han  2004]  

Page 4: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  Bias  

•  Contextual  evidence  bias:  e.g.  PosQ  vs.  LowNQ  (8)  Contextual  evidence  bias  (for  a  proposi%on  p)    

Expecta%on  that  p  is  true  (possibly  contradic%ng  prior  belief  of  the  speaker)  induced  by  evidence  that  has  just  become  mutually  available  to  the  par%cipants  in  the  current  discourse  situa%on.    

(9)  Scenario:  A  enters  S’  windowless  computer  room  wearing  a  dripping  wet  raincoat  (contextual  evidence  for  p=  it  is  raining).    a.  S:  What’s  the  weather  like  out  there?  Is  it  raining?    b.  S:  #  What’s  the  weather  like  out  there?  Is  it  not  raining?  /  Is  it  

sunny?    

(10)  A  PosQ  p?  is  compa%ble  with  contextual  evidence  for  p.                  A  LowNQ  Not  p?  is  incompa%ble  with  contextual  evidence  for  p.  

4  [Büring  and  Gunlogson  2000]    

Page 5: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  Bias  and  Ambiguity  

•  While  expressing  original  bias  for  p,  HiNQs  of  form  n’t  p?  allow  for  two  intui%ve  interpreta%ons:    (11)  Didn’t  John  drink?  

 

•  Disambiguated  by  the  presence  of  a  PPI  vs.  an  NPI:    (12)  Didn’t  John  drink  somePPI  beer?  

   (13)  Didn’t  John  drink  anyNPI  beer?  

 

 

   Outer  reading  only  

   Inner  reading  only  

   OUTER  nega%on  reading  double-­‐checking  p  

   INNER  nega%on  reading  double-­‐checking  ¬p  

5  [Ladd  1981,  Romero  &  Han  2004,  Sudo  2013,  a.o.]  

Page 6: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  Goal  

•  In  this  talk,  we  mostly  concentrate  on  nega%ve  PQ  forms:  (14) Did  John  not  drink?  (15) Didn’t  John  drink  a  /  any  beer?  (16) Didn’t  John  drink  a  /  some  beer?    

•  Blatant  disagreement  as  to  what  nega%ve  PQ  forms  share  a  joint  bias  profile  and  what  forms  have  separate  bias  profiles:  (17)  Four  different  splits  in  the  literature:  

     

•  OUR  GOAL:  To  provide  experimental  evidence  towards  split  (iv)  

 

LowNQ  Inner  HiNQ  Outer  HiNQ  

6  

Page 7: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Introduc%on:  Roadmap  

•   STUDY  1  on  splits  (i)  and  (ii):  Bias  condi%ons        The  predic%ons  of  lines  (i)  and  (ii)  are  falsified  in  several  bias  

condi%ons.    

•   STUDY  2  on  splits  (iii)  and  (iv):  Prosody        Disambigua%on  of  inner  vs.  outer  HiNQ  correlates  with  

prosody.          Explainable  both  under  split  (iii)  and  split  (iv).  

•   STUDY  3  on  splits  (iii)  and  (iv):  Reversed  polarity        Selec%on  of  really-­‐PosQ  vs.  stacked  nega%on  HiNQs  directly  

follows  from  line  (iv)  but  is  unexpected  under  line  (iii)  .    

7  

Page 8: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

   

 Study  1  

8  

Page 9: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1  

•  Four  relevant  pragma%c  condi%ons:  (18).  –  x-­‐axis  on  original  bias:  for  p  vs.  n(eutral)  –  y-­‐axis  on  contextual  evidence  bias:  n(eutral)  vs.  for  ¬p  –  Cell  n/n  as  control  cell:  PosQ  

   

 (18)    Pragma%c  condi%ons:  

Original  bias  

Contextual    Evidence  bias    

9  

Page 10: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1:  Predic%ons  

•  SPLIT  (i):  van  Rooij  and  Safarova  (2003)  –  All  three  nega%ve  PQ  forms  –LowNQs,  inner  and  outer  HiNQs–  are  treated  

uniformly  as  requiring  that  the  u%lity  value  of  the  pronounced  proposi%on  ¬p  be  higher  than  that  for  p.  

–  Concentra%ng  on  informa%vity-­‐based  u%lity  (i.e.,  no  desirability),  the  u%lity  value  of  ¬p  is  higher  than  that  of  p  when  the  speaker’s  expecta%on  is/was  towards  p,  since:    •  learning  the  unexpected  ¬p  would  be  highly  informa%ve  and  •  learning  the  expected  p  would  be  less  informa%ve.  

•  Predic%ons  of  split  (i)  for  the  relevant  cells:    (19)  

   

10  

Original  bias  

Contextual    Evidence  bias    

Page 11: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1:  Predic%ons  

•  SPLIT  (ii):  Krira  (2012)  –  LowNQs  and  inner  HiNQs  (double-­‐checking  ¬p)  are  treated  uniformly  as  

having  the  underlying  representa%on  (20):    (20)    [REQUEST  [  ASSERT  ¬p]]  

–  This  underlying  representa%on  is  argued  to  be  felicitous  with  contextual  evidence  bias  for  ¬p.  

 

•  Predic%ons  of  split  (ii)  for  the  relevant  cells:                  (21)  

11  

Contextual    Evidence  bias    

Original  bias  

Page 12: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1:  Design  

Par1cipants:  42  English  students  (U.  College  London)  -­‐  [M  =  24.7;  SD  =  2.9;  15  males,  27  fem]  S1muli:  42  writen  scenarios  with  two  cap%ons  and  two  pictures.  Sample:  CAPTION  1:  “If  it  doesn't  rain  tomorrow,  you  will  surely  go  to  the  beach.  The  forecast  for  the  next  morning  indicates:”    

p   neutral   ¬  p  CAPTION  2:  “The  day  ayer  your  flat  mate  Sam  comes  from  the  outside  and  enters  in  your  bedroom..”  

¬  p   neutral   p  

Task  –  Select  &  Pronounce:  

12  

Page 13: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1:  Results  

0

20

40

60

80

n/n n/not−p p/n p/not−pCondition

Perc

enta

ge o

f occ

urre

nce

Polar question typeHiNQ

LowNQ

PosQ

Fig.  1:  Results  of  Study  1.  

13  [Domaneschi,  Romero  &  Braun,  to  appear]  

Page 14: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  1:  Discussion  

0

20

40

60

80

n/n n/not−p p/n p/not−pCondition

Perc

enta

ge o

f occ

urre

nce

Polar question typeHiNQ

LowNQ

PosQ

Neither  the  general  parallelism  between  LowQs  and  HiNQs  predicted  by  split  (i)  nor  the  restricted  parallelism  predicted  by  split  (ii)  is  borne  out.    Instead,  LowNQs  and  HiNQs  are  selected  as  preferred  choice  in  different  condi%ons.  

14  

Split  (i)   Split  (ii)  

Page 15: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

   

 Background  for  studies  2  &  3  

15  

Page 16: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Ladd's  "ambiguity"  

•  Ladd’s  “ambiguity”:  two  intui%ve  interpreta%ons  of  HiNQs:  (22)  Isn’t  there  a  vegetarian  restaurant  around  here?  

–  OUTER  nega%on  reading:  S  originally  believes  p  and  (with  or  without  new  evidence  to  the  contrary)    wants  to  confirm  p    by  means  of  double-­‐checking  p.  

(23)  A:  You  guys  must  be  starving.  You  want  to  get  something  to  eat?    S:  Yeah,  isn’t  there  a  vegetarian  restaurant  around  here?  

 –  INNER  nega%on  reading:  S  originally  believes  p,  has  now  tenta%vely  inferred  the  opposite  and  wants  to  double-­‐check  ¬p,  the  new  inference.  

(24)  S:  I’d  like  to  take  you  guys  out  to  dinner  while  I’m  here  –  we’d  have  %me  to  go  somewhere  around  here  before  the  evening  session  tonight,  don’t  you  think?    A:  I  guess,  but  there’s  not  really  any  place  to  go  in  Hyde  Park.    S:  Oh,  really,  isn’t  there  a  vegetarian  restaurant  around  here?    

16  [Ladd  1981]  

Page 17: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Analysis  in  split  (iv)  

•  SPLIT  (iv):  Romero  &  Han  2004:  –  Preposing  of  nega%on  introduces  the  VERUM  operator:  (25) [[VERUM]]  =  λp<s,t>.λws.  ∀w'  ∈  Epix(w)  [  ∀w"  ∈  Convx(w')  [  p  ∈  CGw"]  ]      –  Nega%on  may  scope  over  VERUM  ,  as  in  (26),  or  under  VERUM  ,  as  in  (27):  

(26) [Q  [¬  VERUM    [p]]]  

 

(27) [Q  [VERUM  [¬  p]]]  

–  The  two  readings  are  disambiguated  by  PPIs/NPIs  and  possibly  by  other  means.  

 S(peaker)  asks  A(ddressee)  for  doubts  about  p.    The  prejacent  –the  proposi%on  the  ques%on  is  

about–  is    p.    S  is  s%ll  leaning  towards  her  original  belief    p.  

 S  asks  A  for  conclusive  evidence  for  ¬p.    The  prejacent  is  ¬p.    S  is  considering  switching  to  the  new  inference  ¬p.  

 

Outer  reading  

Inner  reading  

17  [Romero  &  Han  2004,  see  also  Romero  2015]  

Page 18: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Analysis  in  split  (iii)  

•  SPLIT  (iii):  AnderBois  2011  –  A  special  lexical  entry  for  high  nega%on,  abbreviated  ¬hi,  includes  

VERUM  as  one  of  its  ingredients  

–  HiNQs  unambiguously  have  the  structure  (28):  (28)  [  Q  ¬hi    p  ]  

–  By  default,  S(peaker)  has  a  tendency  towards  retaining  her  original  belief  p.  

 –  Devia%ng  from  this  tendency  requires  an  NPI  –whose  pragma%c  

proper%es  overwrite  the  default–  or  other  overt  marking.              

  The  prejacent  is    p.    S  s%ll  leaning  towards  her  original  belief  p.  Outer  reading  

  The  prejacent  is  p.    S  is  considering  switching  to  the  new  inference  ¬p.  Inner-­‐like  interpreta%on  

18  [AnderBois  2011,  see  also  Northrup  2014]  

Page 19: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

   

 Study  2  

19  

Page 20: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2  

•  The  ques%on  arises,  what  other  overt  marking  may  help  dis%nguish  the  outer  and  inner  interpreta%ons.  

•  More  concretely:  

–  Does  some  prosodic  cue  correlate  with  the  outer/inner  dis%nc%on?  

–  If  so,  does  the  meaning  contribu%on  of  this  cue  –to  the  extent  that  this  meaning  can  be  pinned  down–  allows  us  to  favor  one  analysis  over  the  other?  

•  Caveat!   Study   2   was   conducted   in   University   of   Alberta,  where  an  uptalk  variety  of  English  is  spoken.  Thus,  the  shape  of  prosodic  cues  may  differ  from  those  in  other  varie%es.  

20  

Page 21: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  Design  

•  4  condi%ons:  Ownership  of  checked  proposi%on  (mine/Speaker's  vs.  other’s/Addresee's)  x  Certainty  about  checked  proposi%on  (high  vs.  low)  

•  Task:  –  Read  dialogue  se?ng  up  S’s  belief  that  p,  A’s  belief  that  ¬p  –  Dialogue  describes  S’s  wish  to  check  p  or  ¬p  and  her  degree  of  certainty  about  

the  checked  proposi%on  (90%  as  high,  60%  as  low).  –  Choose  HiNQ  or  LowNQ  to  resolve  conflict  and  uter  it  out  loud  

•  24  par%cipants  (15  female,  8  male,  1  other):  Na%ve  speakers  of  Canadian  English,  students  at  U.  Alberta  

•  16  experimental  items,  each  appeared  twice  per  list  in  different  condi%ons  (2  blocks)  

•  15  fillers,  each  appeared  twice  per  list  in  different  condi%ons  →  62  dialogues  per  list  

•  Analysis  with  linear  mixed-­‐effects  models  (Baayen  et  al.  2008)  

21  

Page 22: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  Results  (2)  

Fig.  2:  Results  of  prosody  of  HiNQs.  

       (29)  Aren't  they  plan%ng  a  waterlily  ↑?                          (30)  Aren't  they  plan%ng  a  waterlily  ↑?    

22  

Page 23: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  Discussion  

•  A  related  cue  in  falling  and  rising  declara%ves?  (31)    It  is  raining  ↓.  (32)    It  is  raining  ↑?  

•  Gunlogson  (2003):  locus  of  commitment/authority  –  Declara%ve  syntax  signals  commitment  to  the  prejacent  (=  ‘it  is  raining’).  –  A  final  fall  ↓  signals  that  the  locus  of  the  commitment/authority  is  Speaker.  –  A  final  rise↑  signals  that  the  locus  of  the  commitment/authority  is  Addressee.  

•  Adap%ng  this  idea  to  HiNQs  under  Romero  &  Han's  analysis:  –  A  HiNQ  n't  p?  expresses  original  bias  for  p  and  (re-­‐)opens  the  issue  {p,  ¬  p}.  –  A  final  low  rise  ↑  signals  that  the  locus  of  the  authority  is  Speaker.  

 –  A  final  high  rise  ↑  signals  that  the  locus  of  the  authority  is  Addressee.    

 

 Outer  reading,  with  S  retaining  her  original  belief  p    

 Inner  reading,  with  S  considering  switching  to  ¬p    23  

Page 24: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  Discussion  

•  A  related  cue  in  reversed  polarity  tags?  (33)    Amalia  ley,  didn't  she  ↓?  (34)    Amalia  ley,  didn't  she  ↑?  

•  Farkas  &  Roelofsen  (to  appear),  building  in  Northrup  (2014):  –  A  final  fall  ↓  signals  that  S's  condi%onal  commitment  to  the  prejacent  proposi%on  p  

(=  'that  Amalia  ley')    is  moderate  to  high.  –  A  final  rise↑  signals  that  S's  condi%onal  commitment  to  the  prejacent  proposi%on  p  

is  low.  

•  Adap%ng  this  idea  to  HiNQs  under  AnderBois'  analysis:  –  A  HiNQ  n't  p?  always  has  p  as  the  prejacent:  [Q  ¬hi      p].  –  A  final  low  rise  ↑  signals  that  S's  condi%onal  commitment  to  prejacent  p  is  high(ish).  

–  A  final  high  rise  ↑  signals  that  S's  condi%onal  commitment  to  prejacent  p  is  low.    Outer  reading,  with  S  retaining  her  original  belief  p    

 Inner  reading,  with  S  considering  switching  to  ¬p    24  

Page 25: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  Discussion  

•  A  prosodic  cue  has  been  found  to  correlate  with  the  two  interpreta%ons  of  HiNQs:  –  Outer  reading            =  final  low  rise  ↑    –  Inner  reading/interpreta%on    =  final  high  rise  ↑  

•  Different  meanings  may  be  assigned  to  this  cue,  …  

•  …  making  the  finding  in  principle  compa%ble  both  with  split  (iii)  and  with  split  (iv).    

25  

Page 26: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  2:  A  second  result  

26  

•  A  HiNQ  is  preferred  over  a  LowNQ  no  mater  whether  the  Speaker's  original  belief  p    is  checked  or  the  new  inference  ¬p  is  checked.    

Page 27: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

   

 Study  3  

27  

Page 28: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3  

•  Back  to  Ladd’s  “ambiguity":  –  OUTER  nega%on  reading:  

S  originally  believes  p  and  (with  or  without  new  evidence  to  the  contrary)    wants  to  confirm  p    by  means  of  double-­‐checking  p.  

–  INNER  nega%on  reading:  S  originally  believes  p,  has  now  tenta%vely  inferred  the  opposite  and  wants  to  double-­‐check  ¬p,  the  new  inference.    

•  Idea:  Start  with  original  belief  in  a  nega%ve  proposi%on  –  OUTER  nega%on  reading:  

S  originally  believes  ¬p  and  ...  –  INNER  nega%on  reading:  S  originally  believes  ¬p  and  ...  

28  

Page 29: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Predic%ons  

•  Original  speaker  belief  for  a  posi%ve  p  (=  ‘John  drank’)  

 

•  Original  speaker  belief  for  a  nega%ve  ¬p  (=  ‘John  did  not  drink’)  

•  Caveat:  The  stacked  nega%on  form  is  somewhat  marked.  So,  even  in  the  condi%ons  where  both  splits  predict  it  to  be  correct,  it  may  not  be  always  selected.    

 

 

 

 

In  split  (iv):  [Q  ¬  VERUM  ¬p]  [Q  VERUM  ¬¬p]  

   Didn’t  John  not  drink?      DID  John  drink?  

In  split  (iii):  [Q  ¬hi  ¬p]    

   Didn’t  John  not  drink?  

In  split  (iv):  [Q  ¬  VERUM  p]  [Q  VERUM  ¬p]  

In  split  (iii):  [Q  ¬hi  p]    

   Didn’t  John  drink?      Didn’t  John  drink?      Didn’t  John  drink?  

29  

Page 30: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Design  

•  4  condi%ons:  Ownership  of  checked  proposi%on  (mine/Speaker's  vs.  other’s/Addressee's)  x  Certainty  about  checked  proposi%on  (high  vs.  low)  

•  Task:  –  Read  dialogue  se?ng  up  S’s  belief  that  ¬p,  A’s  belief  that  p  –  Dialogue  describes  S’s  wish  to  check  p  or  ¬p  and  her  degree  of  certainty  about  

the  checked  proposi%on  (90%  as  high,  60%  as  low).  –  Choose  (really-­‐)PosQ  or  StackedNega%on  to  resolve  conflict    

•  30  par%cipants  (23  female,  7  male):  Na%ve  speakers  of  Canadian  English  

•  16  experimental  items,  each  appeared  once  per  list  in  one  of  the  condi%ons  

•  16  fillers  with  choice  between  other  ques%on  types  (HiNQ,  LowNQ,  declara%ve  ques%on,  really-­‐PosQ)    32  scenarios  per  list  

•  Analysis  with  linear  mixed-­‐effects  models  (Baayen  et  al.  2008)  

30  

Page 31: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Results  

31  

Page 32: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Discussion  

32  

SPLIT    (iv)  checking  S's  proposi%on  [Q  ¬  VERUM  ¬p]    Didn't  John  not  drink?  

•  In  the  "mine"  condi%ons  (outer  reading),  the  stacked  nega%on  form  Didn't  John  not   drink?   is   in  principle  predicted   to  be  preferred  over   the   (really-­‐)PosQ   form  DID  John  drink?  by  both  approaches.  The  fact  that  the  stacked  nega%on  form  is  preferred   only   in   approx.   50%   of   the   cases   may   be   due   to   its   inherent  markedness.  

SPLIT    (iii)  [Q  ¬hi  ¬p]  Didn't  John  not  drink?  

Page 33: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Discussion  

33  

SPLIT    (iv)  Checking  A's  proposi%on  [Q  VERUM  ¬¬p]    DID  John  drink?  

•  In   the   "other"   condi%ons   (inner   reading),   the   split   (iv)   approach   predicts   the  (really-­‐)PosQ   form  DID   John   drink?   to   be   preferred   over   the   stacked   nega%on  form.  Since  markedness  does  not  play  against  the  preferred  form,  the  preference  is  expected  to  be  significant.  This  predic%on  is  borne  out.  

Page 34: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Study  3:  Discussion  

34  

•  Split   (iii)  does  not  predict  a  difference  between  the  "mine"  condi%ons   (genuine  outer   reading)   and   the   "other"   condi%ons   (inner-­‐like   interpreta%ons).   In   both  cases,  to  express  the  combina%on  of  biases   in  the  context  (original  bias  for  ¬p,  contextual  evidence  for  p),   the  structure   [Q  ¬hi  ¬p]  should  be  equally  available  (with  the  a  small  rise  in  the  "mine"  ↑  condi%ons  and  a  high  rise  ↑  in  the  "other").  

•  It  is  unclear,  at  this  point,  why  the  (really-­‐)PosQ  form  is  a  stronger  compe%tor  in  one  condi%on  than  in  other.  

SPLIT    (iii)  [Q  ¬hi  ¬p]  Didn't  John  not  drink?  

Page 35: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

Conclusion  

•  Four  different  splits  of  nega%ve  polar  ques%ons  in  the  literature:  

     •  STUDY  1:  LowNQs  and  HiNQs  are  selected  across  different  bias  condi%ons.  

   Against  splits  (i)  and  (ii).  

•  STUDY    2:  Final  rise  ↑  vs.  ↑  correlates  with  outer  vs.  inner  reading  of  HiNQs.      Explainable  both  under  split  (iii)  and  (iv).  

•  STUDY    3:  Really-­‐PosQ  are  significantly  preferred  to  stacked  nega%on      HiNQs  in  certain  condi%ons  with  reversed  polarity.        Tenta%vely  favors  split  (iv)  over  split  (iii).  

35  

Page 36: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

 Thank  you  for  your  aten%on…  

       also  from…  

   

36  

Be?na  Braun  University  of  Konstanz  

Filippo  Domaneschi  University  of  Genoa  

Anja  Arnhold  University  of  Alberta  

Page 37: Negave’Polar’Ques%on’Types’’ inEnglish€¦ · Negave’Polar’Ques%on’Types’’ inEnglish Maribel’Romero,’Anja’Arnhold,’BenaBraun’and’ Filippo’ Domaneschi’

References  [A]  AnderBois,  S.  2011.  Issues  and  Alterna%ves.  Ph.D.  disserta%on,  UCSC.  [B]  Baayen,  R.  H.,  Davidson,  D.   J.,  &  Bates,  D.  M.  2008.  Mixed-­‐effects  modeling  with  crossed  random  

effects  for  subjects  and  items.  Journal  of  Memory  and  Language,  59(4),  390–412  [B&G]  Büring,  D.  &  C.  Gunlogson.  2000.  Aren’t  posiBve  and  negaBve  polar  quesBons  the  same?      [DRB]  Domaneschi,  F.,  M.  Romero  &  B.  Braun.  To  appear.  Bias  in  polar  ques%on:  Evidence  from  English  

and  German  produc%on  experiments.  Glossa.  [F&R]  Farkas,  D.  &  F.  Roelofsen.  To  appear.  Division  of   labor   in  the   interpreta%on  of  declara%ves  and  

interroga%ves.  Journal  of  SemanBcs.  [G]   Gunlogson,   C.   2003.   True   to   From.   Rising   and   Falling   DeclaraBves   as   QuesBons   in   English.   NY:  

Routledge.  [Kr]  Krira,  M.  To  appear.  Negated  polarity  ques%ons  as  denega%ons  of  asser%ons,  ms.    [L]  Ladd.  R.  1981.  A  first  look  at  the  seman%cs  and  pragma%cs  of  nega%ve  ques%ons  and  tag  ques%ons,  

CLS  17.    [N]  Northrup,  O.  2014.  Grounds  for  commitments,  UCSC  diss.    [R]  Repp,  S.  (2013).  Common  ground  management:  modal  par%cles,  illocu%onary  nega%on  and  VERUM.  

In  Beyond  Expressives:  ExploraBons  in  Use-­‐CondiBonal  Meaning.  Pp.  231-­‐274.  Leiden:  Brill.    [Ro]  Romero,  M.  2015.  High  Nega%on  in  Subjunc%ve  Condi%onals  and  Polar  Ques%ons,  SuB  19.    [R&H]  Romero,  M.  &  C.-­‐h.  Han  2004.  On  nega%ve  yes/no  ques%ons.  L&P  27.    [S]  Sudo,  Y.  2013.  Biased  polar  ques%ons  in  English  and  Japanese.  In  Beyond  Expressives:  ExploraBons  in  

Use-­‐CondiBonal  Meaning.  Pp.  275-­‐296.  Leiden:  Brill.    [vR&S]  van  Rooij,  R.  &  M.  Šafárová  2003.  On  polar  ques%ons,  SALT  13.    

  37