University of Iowa
Iowa Research Online
Theses and Dissertations
Spring 2009
L2 acquisition of Chinese wh-questions by English-speaking
learners
Binnan Gao
University of Iowa
Follow this and additional works at: https://ir.uiowa.edu/etd
Part of the First and Second Language Acquisition Commons
Copyright 2009 Binnan Gao
This dissertation is available at Iowa Research Online: https://ir.uiowa.edu/etd/236
Recommended Citation
Gao, Binnan. "L2 acquisition of Chinese wh-questions by English-speaking learners." PhD (Doctor of
Philosophy) thesis, University of Iowa, 2009.
https://doi.org/10.17077/etd.gzf5i2gw
Follow this and additional works at: https://ir.uiowa.edu/etd
Part of the First and Second Language Acquisition Commons
!
"
#)
#$
% & $' (
) "%)" ))&$
" #$ $*% $&$ " #$ + !
"
(# '$, $$
$! '
,% ,$ !*%
#$
'% $ ))$,$ "
#$
-$
" .
/
#$
%($ -
2
001
! $
"$
$)$
#$
3
# '
# $ $ )$
)
$
))
$
)$
! &()$&$
$
*%$
9
; $
$
,
,$4
) ' ""$ $
%#
% $' #
#
& -$&$
) ,% ,$
' )$
$
(
5 , &&
''
' ()
!(
'$-$) (&$
"
" #$ < .#> "$ % $
#
)
!
$' "
8%',&$
9 :;
% -$ .#
$
8$! .# *%$
$'
% ) ,% ,$ ,
.$ $
&$ 3
! ! &$!#
!)$
! )
)$
& #
%' $ 6
)$
# %,#6 #$
$ 5
, &&
" % '
'
$
)
( $ $ !$ " <=.#> "$ % $
$ ,#
'
8$! .# *%$
$(
#$ ' ""$ $
!(
# , &&
'
!(
! ) "% !
' $! *%$
$ 6 #$ (
#
) .# *%$
#$ (
"
%' 3 $(
!(
-$ , $
, &&
!*%
" # $ $6
#$ (
' '-$
$ , "!
' #$
.$ $ % $'6
( $' & $ ' "" !%)
#$ $,
,
" ' ""$ $
#$ $ ' " #$ "
,) # ($ 7 , )$
) ( '%!
%' $ 4 .# !# $ ($! -$)
" &()$ .# *%$
-$ #$ "
.# *%$
'
" .
$ 5 !*%
, %'
3"
!
"
,$ "
$ , #4
.
' #$
) $ 4 %
# ($'6
%'
$-$ ) # )$
8$! ! &()$&$
#$ $ ' " "
, &&
$
.
)$
# .$'
! &()$&$
4
.# !#
,6
- %
''
#$ ($ " & !$
' $! *%$
3
# .
' / '% $ $6 $
,$ #$ . # #
&
#$ ($ " & !$
- '$ !$ "
!) % $
, '
' $! *%$
4 #$
)
!
" .# *%$
$!
")%$ !$4 #$ &
$ 5 ($ " & !$
! &()$&$
. #
$& $''$' <=.#>
(
! &()$&$
4 .# !# .
($
$ $'
$ $
" ! &()$? .# *%$
&()$ .# *%$
$, ' , #$ , &&
5
$& $''$' <=.#>4
&()$ .# *%$
" #$ .
#$ ($ " & !$
!(
.# *%$
? <=.#>4 %
" .# *%$
$& $''$' !) % $
$ .$$
,$ " (
")%$ !$4 #$ $& $''$' <=.#>6
3
? !) % $
' .# *%$
&()$ .# *%$
,$ ) 7$ % $ $& )$ . '$ ! ($ .# *%$
/ )
#
$ 5 ($ " & !$
'&
6
?
! ) "$ % $ # .# . '
$
"
)) .$'
!
$!
!) % $
$ .$$
! &()$? .# *%$
!*%
"
)(
.
#$ ($ " & !$
6
%! % $
%,,$
" ' ""$ $
!
# )$
$-$ )$'4 % #$ $ .
, $- '$ !$
$& $''$' !) % $
($
$ & ,#
$%
" #$ .
, ' ""$ $
"
$, $ "
%$6
(( -$'2 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
#$
%($ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
)$ ' +$( &$
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
+ $
! ! !" # $ % & ' ("'( ''#
(
& )" & # ! ( &
* + & (
' ! % $ ,&
+ $
," , +)" !
& $"
'' , (
- &!
( .
/
! ! "% &- ! &2
!! +
001
& ( !! &
'
% &,
001
''
, !
! &- $
& $"
'' ,
- &!
( .
.
3 .
4
55555555555555555555555
6*6
555555555555555
! !
!
( &
$ ,&
!!
+ & (
6*6
!! (
%%& - $
7 #
, ##
! ! & )" & # ( &
* + & (
' ! %
+ $
," , +)" !
/
001 ,& $"
##
2 55555555555555555555555555555555555
'
3
! ! "% &- ! &
55555555555555555555555555555555555
'$
6
'
55555555555555555555555555555555555
'' # *6 * - !
55555555555555555555555555555555555
"& 6
' #
55555555555555555555555555555555555
" &
6
,& $# 6
* /
/ $ % !
& ",
''
! "$ !
8! ,
! %!
+ '' +
,$
3
& ! &+ 6
.
$
#"+ 6
#
+ #%'
' !( &#
"
&3
#"' %'
! &-
.&
!
' ," ! +
& !3
$ ",
' ," ! +! %& - $ $ # .
! . )" !
. !- &
"+ $
+ ##
$ !%
$- +
(
%& ( !! &! .
!
$'
'
$ $'
!
.
",
#
!
.
!
- & , - "%
$$
, ("'' # :
#%& -
! %
! %6
#% !! ' 6
,
+& $ ' ,
& "!
!
'
!
'& $
$6
!
$
+ #
&3 "
+ #
# # &
"
! ! # ! &3
'
",
$
$ . ! %& !! $ ( & # 6 & ( !! &
#
! #
&%&
!3
(
(
& !"' !
$.
, - # %'
$ & ! &+
.
(
6
+
,& $
!
7
+ "'$ !
& $ &+
+
(
! 7% & !
& ", . +
! . )" !
'' #
$ ! - $ # (& # ! - & '
& " ' ,$
"! ("' & ! "&+ !3
' #
! "
&
- & ( &,
$ $& . %& + + '
# # &! ( &
8$ % &
& ( !! &
$ ,
' ! ! ( . )" !
! #
! ''
+ % &!
. "'$
'
$ +)" !
&
! + '
$ & - ! , %& + !!6 & ( !! &
( &
+ +
'! %"! $ #
$!
, $ !! &
7 & ' + ##
!
#% & + ' & ! &+
& + ##
,
$& (
* - ! ," $ $ #
& '
+ $ # + & !%
( (& #
,
& !"' !
. $ " $ &!
+ ##
- & & $ # $ !! &
% ( ''!6
!
8 ''
# & # "& & ! &+ &
,& $
$3 !
! $ !! &
' $ $ # $ !! &
8 $'
.&
(
& # $ !%
3 (& # $ ! ,
$ !+"!! ,
$
$& ( ( &!
+ #%'
'! .
.
!
$- ! $ #
$ & ! '- $ ' ! ( # $ " !6
# !.
"& $ "
6
! $ !! &
' 9 ,$
+ "& , ,3 %
)"
'
(
& 7% & !
'%("'
. ! !"+
# .
+ #%'
$ (& #
' !!. ! 7& # '
!
$- ! & & ( !! &
+ !! & ( &
( $ ,! ( +"&&
. !
#
"
76
! &-
! $ & $ , # $ !! &
( #
!
& -! 6
. ,#
.
!
(
, (& # & ! &+
)" !
,
!
!
& ! &+ $ ! , 3
- &
&%&
$- + !"+
+
7
!
$
,
.
.
$
& % &
(
!% && 6
& -!
$ !! &
+ "& , #
+ ##
3!
"
+
$#
#
!
& !! !
! +!6
& ",
+
&
!'
&
'
+ '' +
% !
'+ % &
- '
3
$ (&
%"
. ! !"+
&
-
,#
!
#!
!.
$
. &$!
!8
&!6 '
$# $ #
", !
!",, !
8 $
.
!
,3
$ ,<
,
$ "
$;
' ,
$!3 . ! ' -
, %& + !!6
. &
& ( !! & ;"$
#
$+ $ ,$
$
+ ! % # $
8 ; , "3
'! ( # ' # # &!
.
,("' . 6
' + % & . "'$
,
#!3
8! '! ,
8 $' & $ # (
&3 # (
!!
'! , - # - & "! 8 $! (
+ !
#
,& "'
(
" &
( $ ,!
. "'$ '! ' 8
( &
- '$
6 & ( !! &
6/ $ %
&
!6
,
- $ # 8 , %& ' #
+ "& , , %& ( !! &
+ #%'
7 #
"( &
&
!
!
.6
,< "
'% .
$ !"%% &
'% $ #
! $ !! &
+ ! ! ! ( . ! "$ !3 . + & !% + - '
! ' & &!= +)" !
'
, "$
''
. )" !
!
> ( &!
- &
& ( &!
+ #%' #
&
&' &
!
?
!, (+
&
% & +%
& " $
!
# - #
' ," ,
% & +%
!= ,& ## & $ $
$ - ' %#
$ (( &
$ ' & &! (
(
(
&6
+ ' (" +
+ '
'
$
:"$,#
! #
+ +# +
? ;@
!
!
!. &
>
$
!#
( " $
! "$ !6
' & &! (
! " ' ," , ,
& ", 6
'! ,
+' & %& ! +
( AB. C (
"& ! & , 3
"& ! & ,
' & &!=
$
& " - .
: + . )" !
! !& % & $
!
!+ !! $ (
: + . )" !
!
!
$ & + )" !
% & +%
,& ##
!
!
$ !%'
A . C(
& +)" !
"$ > & % & !
' . )" !
' $ (( &
,& ##
& ( &!
% & +%
$ $- &
! ( $ (( &
! 6
@ ! ! . & "! $6
,! , 3
! "
$
$ (
! % ! $ # & $ (( +"'
,
,
,' ! !% 8 , ' & &! (
$ & ' %& $"+
)" !
( ! #%' . )" !
- ! ,
,& ## & . !
'
(
$
&3 "
! % $6
"$
& - '!
: + + #%' #
$ ( ( &!
. $
+ #%' #
!3
!
- "! >
$$
! .
&.
& % &( &# +
$ & + )" !
('" + 3
>
!3
('" + 3
& &!= % &( &# +
'! ,
.
!
+ #%' #
.
& % &( &# +
! & , &$ ,
-
!=
& . )" !
!
# $$ $ AB. C3
!.
+ #%' #
# $$ $ AB. C
!%
& $
!3 . + . !
% ! ( + #%' 7 . )" !
! #%' . )" !
,& ##
!
!
(% & +%
! #%' . )" !
!
$ . )" !
! #%' . )" !
# & 7 AB. C3 "
# $$ $ +' "! ! (
+
!
# $$ $ AB. C6
# & 7 +' "! ! ( . )" !
& % &( &# +
+' "! ! ( . )" !
&
$ & + )" !
" & ! # ' ! . $ !+ % . )" !
$ / $"& ! 6
-$ + ( &+
!, $ !
,6
&, ' 8
/ '
,
!. !
&' &
,& ## & !
. +
' & &!= % &( &# +
+ '(
!
"&
&
!6
$# &7
. . &$!
&
( &+
'' . $
+
.
+ #%' 7 . )" !
+)" !
+' "!
'% !
. ! & - ' $3 "
& % &( &# +
!6
!",, ! !
( ! &"+ "& ! ( $ (( &
& . !
# $$ $ +' "! ! (
' & &! # ,
"& !6
-
"! , $ (( &
!&
, -$ +
.
% ! (
!&
, !( &
DDDDDDDDDDDDDDDDDDDDDDDDDDD6667
DDDDDDDDDDDDDDDDDDDDDDDDDD667 >
*
66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>
>6>6
," ! + * (( & +
$ ! &+
% +!66666666666666666666666666666666666666666666666>
>6 6
+)" !
& ! $ ! &+ 666666666666666666666666666666666666666666666666666666E
>6F6 ! &+ " !
!666666666666666666666666666666666666666666666666666666666666666666666666666666666666G
>6E6 %%& + ! (
"$ 66666666666666666666666666666666666666666666666666666666666666666666666666661
4
6666666666666666666666666666666666666666666666666666666666666666666666666666>0
6>6 * !+& %
(
,' !
$
! /
+' "! '
" !
! $
" !
!
: + ' "! ! 66666666666666666666666666666666666666666666>0
6 6 * (( &
++ " !
7%'
% ' , + ' * (( & + ! 6666666666666666666>
6 6>6
++ "
" , ?>1H @ 666666666666666666666666666666666666666666666666666666>
6 6>6>6 & + %' ! $ & # &! 666666666666666666666666666666666666666666666666666>
6 6>6 6
% & ! ! ! "
(+
' % & &!66666666666666666666666666>E
6 6>6F6
7 ( ,' !
/ - # 666666666666666666666666666666666666666>I
6 6>6E6 ++ " ( & ,' !
$
! /
+' "! '
)" !
!6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>H
6 6>6J6 ++ " ( & ,' !
$
!
#%' 7
" !
!.
)" !
! ! : + ' "! ! 66666666666666666666666666666 0
6 6>6I6
7%'
&
. & ( " ,=! & % ! ' 666666666666666666666
6 6 6
++ "
! ?>11E3 >111@ $ '
$ &#
?>11H@ 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666 F
6 6 6>6 / # ' ! & ,& #6666666666666666666666666666666666666666666666666666666666666 F
6 6 6 6
. &$! !
/ &, & ( % & &! $ 4 & ' !
,' !
$4 & ' !
! 66666666666666666666666666666666666666666666666666666 E
6 6 6F6 ++ "
(
,' !
$
! /
+' "! '
)" !
!6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666 I
6 6 6E6 ++ "
(
,' !
$
!
#%' 7
)" !
!6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666 H
6 6 6J6 7%'
&
. & ( !
$ '
$ &# =!
& % ! ' 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666 1
6F6 ' + ,
," ! + / $ ' $ ! &+
% +!666666666666666666666666666666666F0
6E6 - . ( / : &
& ! $ "$ ! , &$ , +)" !
(
" !
!
& + %'
$ & # & & # . &8 6666666666666666666666666666F
6E6>6 / : & !!" ! 7 # $ 66666666666666666666666666666666666666666666666666666666666666666F
6E6 6 "'' & !( & $ "'' +)" !
$/ # '
'
7 %
! ! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666FF
6E6F6 &,"#
$ $:" + ! ## & 6666666666666666666666666666666666666666666666FG
6E6E6 #%' 7
" !
! 6666666666666666666666666666666666666666666666666666666666666666E>
6E6J6 "## & ( ! &+
% +!66666666666666666666666666666666666666666666666666666666E
6J6 ! &+ " !
!6666666666666666666666666666666666666666666666666666666666666666666666666666666666EF
-
F
*
666666666666666666666666666666666666666666666666666666666666666666666666666666666EJ
F6>6 %% ,
," ! + #%
+ 2 ' + $ & ' & $"+
!8
$ & ## + ' ;"$,#
!8 666666666666666666666666666666666666666666666666666666666666666EJ
F6 6 ! &+ * ! , (
.
"$ ! 6666666666666666666666666666666666666666666666666666666EG
F6 6>6 "$ >2
, "$ ' "$ 666666666666666666666666666666666666666666666666666666EG
F6 6 6 "$ 2
"$ 66666666666666666666666666666666666666666666666666666666666EH
F6F6 ! / & '! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666J
F6F6>6 ' + $ & ' & $"+
! 66666666666666666666666666666666666666666666666666666666J
F6F6 6 & ## + ' ;"$,#
!866666666666666666666666666666666666666666666666666666JJ
F6E6 & + $"& !666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666I0
F6J6 & + % ! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666I
F6I6 *
& # 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666I
F6I6>6 $ , $ & , "
* 666666666666666666666666666666666666666666666I
F6I6 6 ; * 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666IF
F6I6F6 /
$ (
! + '
' ! !6666666666666666666666666666666666666666666666666666666IE
E
*
666666666666666666666666666666666666666666666666666666666666666666666II
E6>6
! &+ " !
>2
+)" !
( #%'
)" !
! 666666666666II
E6>6>6
'
' ! ! ( * 666666666666666666666666666666666666666666666666666666666666666666IG
E6>6>6>6
'
' !! (
* 6666666666666666666666666666666666666666666IG
E6>6>6 6
'
' !! (
; * 66666666666666666666666666666666666666666666G
E6>6 6
' !! (
!"' ! 666666666666666666666666666666666666666666666666666666666666666666GE
E6>6 6>6 ! &+ " !
! >6>2 +)" !
( #%'
" !
!.
% & ! ! '
,
' ( * (( &
& ## + ' " +
! 66666666666666666666666666666666666666666666666666666666666666666666GE
E6>6 6 6 ! &+ " !
>6 2 * - ' %#
'
6666666666666666666666H
E6>6 6F6 ! &+ " !
! >6F2 +)" !
&
2 & "%
!"' ! -!6 $ - $" ' !"' ! 6666666666666666666666666666666666666666666666666666666666HJ
E6 6 ! &+ " !
2
+)" !
( .
% ! ( #%' 7
" !
! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666610
E6 6>6
'
' !! (
* 66666666666666666666666666666666666666666666666666666666666610
E6 6>6>6
'
' !!
* 66666666666666666666666666666666666666666666666610
E6 6>6 6
'
' !! (
; * 666666666666666666666666666666666666666666661I
E6 6 6
' !! (
!"' ! 66666666666666666666666666666666666666666666666666666666666666666611
E6 6 6>6 ! &+ " !
6>2
'
(
A. C
"&
& ,
+ !! & ( & +)" !
(
" !
!
(
$
)" !
! (
$
& ,
!"' ! ! #% & $
( #%'
" !
! 666666666666666666611
E6 6 6 6 ! &+ " !
6 2 +)" !
&
2 & "%
& !"' ! -!6 $ - $" ' & !"' ! 66666666666666666666666666666666666666666666666666666666666>0E
E6 6 6F6 ! &+ " !
6F2
& &!= #%
+
* (( &
' "! ! , &$ ,
! + (
&$!
' "!
' !
66666666666666666666666666666666666666666666666666666666666666666666>0H
E6 6 6E6 / : & ! &+ " !
6E2
+
.
& + % != &( &# +
* (( &
' "! ! , &$ ,
! + (
&$!
' "!
' !
66666666666666666666666666>>I
J*
66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>
-
J6>6
$ ,!
+)" !
(
A. C
"&
& ,
#%'
)" !
!666666666666666666666666666666666666666666666666666666666666666666666666666666666666>
J6>6>6 * (( +"'
& &+
(
)" !
!.
. &$!
'
,
' ( * (( &
& ## + ' " +
!66666666666666666666666666>
J6>6>6>6 & " ' ! #
& " " !
! 666666666666666666666666666> F
J6>6>6 6 &,"#
$ $:" + ! ## & 6666666666666666666666666666666666> I
J6>6 6
$ - ' %#
(
A. C
"&
& , 3 "''
& !( & $ "'' ++ !! %
! !3 $ / # '
'
7
%
! ! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>F0
J6>6 6>6
'
, 2 / -!6
666666666666666666666666666666666666666>F>
J6>6 6 6
* - ' %#
'
(
A. C
"&
& , 666>FF
J6>6F6 +)" !
&
2 & "% !"' ! -!6 $ - $" ' !"' ! 6666666666>FI
J6 6
$ ,!
+)" !
(
A. C
"&
& ,
#%' 7
)" !
! 66666666666666666666666666666666666666666666666666666666666666666666666666666666>FH
J6 6>6
'
(
A. C
"&
& ,
+ !! & ( &
+)" !
(
" !
! (
$
)" !
! (
$
& , !"' ! ! #% & $
( #%'
" !
! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>FH
J6 6 6>6 * !+& % +
.
;"$,# ! ( & ## + '
+ ! $ ,& ## '
+ ! 6666666666666666666666666666666666666666>F1
J6 6 6 6 # $$ $ AB. C
" !
! ( ;" $ 66666666666666666666>E0
J6 6 6 +)" !
&
66666666666666666666666666666666666666666666666666666666666666666666666>E
J6 6F6
& &!= #%
+
* (( &
' "! ! , &$ ,
! + (
&$!
' "!
' !
666666666666666666666666666666666>EF
J6 6F6>6 * - ' %#
'
(
" !
! (
666666666666666>EE
J6 6F6 6
"& (
&$!
# $$ $ ' "! ! (
" !
! (
$ (
&$!
# $$ $
' "! ! (
" !
! (
666666666666666666666666666666666666666666666666>EJ
J6 6E6
+
.
& + % != &( &# +
* (( &
' "! ! , &$ ,
! + (
&$!
' "!
'
!
66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>EI
J6F6 "## & (
$ ,!66666666666666666666666666666666666666666666666666666666666666666666666>EG
J6E6 #
! (
"$ 666666666666666666666666666666666666666666666666666666666666666666666666>EH
6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>J>
* K6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>JG
*
*
66666666666666666666666666666666666666666666666666666666666666>JG
! > 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>JG
& ' & $"+
!8 >6>66666666666666666666666666666666666666666666666666666666666666666666666>JG
& ## + ' ;"$,#
!8 >6 6666666666666666666666666666666666666666666666666666666>JH
! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>IF
& ' & $"+
!8 6>66666666666666666666666666666666666666666666666666666666666666666666666>IF
& ## + ' ;"$,#
!8 6 6666666666666666666666666666666666666666666666666666666>IE
! F 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>I1
& ' & $"+
!8 F6>66666666666666666666666666666666666666666666666666666666666666666666666>I1
& ## + ' ;"$,#
!8 F6 6666666666666666666666666666666666666666666666666666666>G0
! E 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>GJ
& ' & $"+
!8 E6>66666666666666666666666666666666666666666666666666666666666666666666666>GJ
& ## + ' ;"$,#
!8 E6 6666666666666666666666666666666666666666666666666666666>GI
7
' >6
A. C
'
* (( & +
.
,' !
$
!
" !
!
&#! (
"&
& , 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666 >
6
% &
&- &
' ! &"+ "&
(
' F6
% &
&- &
' ! &"+ "&
(/ $ &
' E6
&$!
' J6 - &- . (
! $
!
"&
,
# 3
,' ! . . &$! 6666666666666666666666666666666666666666 1
$ ;
&# 3
$
. . &$! 6666666666666666666666666666666666666F0
! !66666666666666666666666666666666666666666666666666J>
," ! +
# 666666666666666666666666666666666666I>
' I6
! ( % ! (
+ !
& + % ! & $"+ $ .
# '
&$! $ &+
, ( ++"&& + ( A
C $ AB C
"&
! !6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666IH
' G6
! ( % ! (
+ !
& + % ! & $"+ $ .
$: + - '
&$! $ &+
, ( ++"&& + ( A
C $ AB C
"&
! !6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666I1
' H6
! ( % ! (
+ !
& + % ! & $"+ $ .
$- & '
&$! $ &+
, ( ++"&& + ( A
C $ AB C
"&
! !6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666I1
' 16
&+
, (A
C $ AB
'' &
& ## + ' " +
' >06 ++"&& +
&$!
' >>6
(
A. C
- , * (( &
C
" !
"& ;
!
!.
&$! '
,
! ! 66666666666666666666666666666666666666666666GF
"&
& ,
#%'
& ## + ' " +
!
" !
!.
! ! 66666666666666GE
%
$ / !"&
4 ( & & + % != &( &# + - & "&
! ! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666GJ
' > 6
( &&
% ! (
&.!
#%'
#% & !
(
" !
! - &
& + % != &( &# +
&
"& ! ! 6666666666666666666666666666666666666666666666GI
' >F6
( &&
% ! (
&.!
#%'
#% & !
" !
!
(
' >E6
( &&
% ! (
&.!
#%'
#% & !
" !
!
(
' >J6
( &&
% ! (
&.!
#%'
#% & !
" !
!
(
& + % != &( &# +
&
! F 666666666666666666666666666666666666666666666666666666666GG
' >I6
( &&
% ! (
&.!
#%'
#% & !
" !
!
(
& + % != &( &# +
&
! E 666666666666666666666666666666666666666666666666666666666GG
& + % != &( &# +
&
! > 666666666666666666666666666666666666666666666666666666666GI
!
& + % != &( &# +
&
666666666666666666666666666666666666666666666666666666666GG
' >G6 ++"& +
#%'
" !
!.
&$! - ,
* (( &
& ## + ' " +
!
& ## + '
+ ! ( "& ;
! !6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666GH
7
' >H6
%
$ / !"&
& ## + '
4 ( & & + % != &( &# +
+ ! - & "& ; ! !666666666666666666666666666666666666666666666666666666G1
' >16 ++"& +
#%'
" !
!.
&$! - ,
* (( &
& ## + ' " +
!
,& ## + '
+ ! ( "&
; ! ! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666H0
'
06
%
$ / !"&
,& ## + '
4 ( &
+ ! - &
'
>6
( &&
% ! (
#% & !
" !
!
'
6
'
F6
'
E6 "# &!
"&
'
J6
++"&& +
& $"+ $
"&
I6
'
G6
H6
'
16
4 ( &
#%' 7
& + % != &( &# +
&
! E 666666666666666666666666666666666666666666666666666666666H>
!
& ## + ' $ ,& ## + '
! ! 6666666666666666666666666666666666666666666666666666666666666666666666666666666H1
! (
! (
& +%
&.!
#%'
( &&
#%'
+ !
(
' F06
4 ( &
#%' 7
( &&
% ! (
(
'
$
(
!= &( &# +
! !66666666666666666666666666666666666666666666666666H>
! (A
C $ AB C
" !
! & +% !
! !66666666666666666666666666666666666666666666666666666666666666666666666666666666H
& + % != ++"& +
+ !
"& ;
(
'
&.!
#%'
& +%
"& ;
& +%
++"&& + ! ( & , A. C
.
% ! (
" !
$
8 A. C
! 666666666666666666666666666661I
!= &( &# +
#%'
$ .
% !
! E 6666666666666666666666666666666666666666666666666666666666666666666666666666>0>
#% & !
" !
!
(
& + % != &( &# +
&
! E 6666666666666666666666666666666666666666666666666666666>0>
!= &( &# +
& ## + '
#%'
+ !
$ .
% !
! E 666666666666666666666666666666666>0
&.!
#% & !
( & + % != &( &# +
$ .
% ! ( #%' 7
& ## + '
! E 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>0F
4 ( &
#%' 7
( &&
#%'
& +%
!= &( &# +
,& ## + '
&.!
#% & !
$ .
% ! (
' F>6 "# & $ &+
, ( & +%
& ,
( # $$ $ ' "!
(
#%'
+ !
& +%
#%' 7
!=
$ .
% !
! E 66666666666666666666666666666>0F
&( &# +
! E666666666666666666666666666>0E
! +)" & ,
" !
! (K
8 A. C
"&
, < $ 66666666666666666>0I
' F 6 "# & $ &+
, ( & + % ! +)" & ,
8 A. C (
/
' "!
$
8 A. C ( # $$ $ ' "!
" !
! (
;" $ 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>0I
' FF6
' FE6
4 ( & & +%
' "!
' !
( &&
! +
E
!=
.' $, ( ! + (
* (( &
' "! !
E
&.!
(
&$!
! 666666666666666666666666>>0
#% & !
( & + % != &( &# +
&$!
' "!
' !
* (( &
' "! !
! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>0
7
' FJ6 "# &
&$!
$ &+
' "!
' FI6 "# &
&$!
E
$ &+
, ( & +% !
+)" & $ ! + (
' "!
' !
/ & 7 ' "! (
( ;" $
! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>>
' FG6 "# &
&$!
9 $
$ &+
' "!
E
' FH6 "# &
&$!
E
$ &+
, ( & +% !
+)" & $ ! + (
' "!
' !
# $$ $ ' "! (
( ;" $
! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>
' F16
,
'
,
4 ( & & +%
' "!
' !
' E06
(
!
& +%
!
#%'
+)" & $
! +
E
(
! 66666666666666>>>
( & +% !
+)" & $ ! + (
' !
# $$ $ ' "! (
(K ,
! 666666666666666666666666666666666666666666666666666666666666666666666666666666666>>>
!=
.' $, ( ! + (
* (( &
' "! !
E ;
&$!
! 6666666666666666666666666>>F
( &&
&.!
#% & !
( & + % != &( &# +
! + (
&$!
' "!
' !
* (( &
' "! !
E ; ! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>E
' E>6 "# &
' E 6 "# &
E
$ &+
,& ##
, (
+ '
& +% !
+ ! ( #%'
$
- H0L ++"& +
E ; ! 66666666666>>E
$ &+
, ( & +% !
$
- H0L ++"& +
,& ## + '
+ ! ( / & 7 ' "! (
( ;" $
; ! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>J
' EF6 "# &
K
$ &+
, ( & +% !
$
- H0L ++"& +
,& ## + '
+ ! ( # $$ $ ' "! (
(
, 9 $ 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>J
' EE6 "# &
$ &+
, ( & +% !
$
- H0L ++"& +
,& ## + '
+ !
# $$ $ ' "! (
(
;" $ 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>J
' EJ6
!! +
.
' "!
' !
/
' "! ! (
' EI6
!! +
' "!
$ +)" !
E
' EG6
&& '
.
& + % != &( &# +
* (( &
' "! !
, &$ ,
! + (
&$
' "!
' !
E
! 666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>H
' EH6
& +%
!= +)" !
#%'
" !
" !
! ( ;" $
(
!
! + (
&$!
$ +)" !
(
E
! 6666666666666666666666>>G
.
& + % != +)" !
( ! + (
&$!
' !
# $$ $ ' "! ! (
(K ,9 $
(
# $$ $ ' "! ! (
" !
! ( ;" $
! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>G
!! +
.
& + % != H0L ++"& +
" !
! $ /
' "! ! (
" !
! ( ;" $
! + (
&$!
' "!
' !
E
7
#%'
, &$ ,
; ! 666666666666666>>1
' E16
' J06
' J>6
!!
' "! !
9 $
E
+
.
& + % != H0L ++"& +
# $$ $
(
( ;" $
$
# $$ $ ' "! ! (
(K ,
, &$ ,
! + (
&$!
' "!
' !
; ! 6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666>>1
&& '
.
& +%
, &$ ,
! + (
,& ## + '
+ !
!=
&$
E
&( &# +
* (( &
' "! !
' "!
' !
; ! 66666666666666666666666666666666666666666666666666> 0
& $"+
&$ & (
&$! ! $
"&
$ ;
! !66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666> J
' J 6 "# & (
& &!
&
" !
! ( * (( &
7% ! $
7
& ## + ' " +
! . +
8 66666666666666666666666666666666666666666666666666666666> I
7
,"& >6 / $ ' ( & ## &3 & + %' !
,"&
6/ $ ' (/
,"& F6 ++"& +
& +%
,"& E6
& #
&! 6666666666666666666666666666666666666666666666666>F
& ,& # 6666666666666666666666666666666666666666666666666666666666666666666666666666 E
( & ## + ' ;"$,#
! ! ( #%'
" !
!
! - &
&!
& ? M >@6666666666666666666666666666666666666666666666666666666HJ
$ - $" ' !"' ! ( ++"&& + ( A . C
"&
& , +& !! "&
! ! 66666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666HG
,"& J6 ++"& +
" !
!
,"& I
# '!
$
++"& +
! ( & ##
& +% !
(
&
+ ' ;"$,#
$ ( &!
% ! (
!
" !
,"& G6 "# & ( & + % ! - , H0L ++"& +
& ## + ' $ ,& ## + '
+ ! (
,"& H6 "# & ( & + %
$ ,& ## + '
!
( &
% ! (
& 6666666666666666666666666666666666666666666661G
!
E
( &
(K
! 666666666666666>00
,9 $
6666666666>0G
- , H0L ++"& +
( &
& ## + '
+ ! (
" !
! (
6666666666666666666666666666666>0H
7-
1
CHAPTER 1 INTRODUCTION
This study on the second language (L2) acquisition of Chinese wh$questions by
English$speaking learners is motivated both by the contribution that this research can
make to understanding the acquisition of parameters under the Principles and Parameters
framework (PP), and by the lack of empirical studies along this line.
1.1. Linguistic Difference and Research Topics
Wh$questions are questions headed by wh$words, such as
,
,
,
, and so forth. There are two major linguistic differences between English and
Chinese wh$questions. A study of L2 English$speaking learners’ acquisition of Chinese
wh$questions can make an interesting contribution to the L2 acquisition theories from the
Principles and Parameters paradigm.
One difference is that in wh$questions, English words are fronted to sentence
initial positions, while Chinese wh$words remain in$situ, i.e., in the position where they
originated in the deep structure, as shown by the English and Chinese sentences in (1).
That is why traditionally English is categorized as a movement language and Chinese is
categorized as a wh$in$situ language. In indirect questions, as shown by (2) a, and in wh$
questions in object complement, as shown by (2) b, the same syntactic difference remains
for English and Chinese. For easy reference, in this dissertation the author will call the
latter two types of wh$questions complex wh$questions.
(1) Monoclausal wh$questions
English: What did XiaoWang buy?
Chinese: XiaoWang maile shenme?
Little Wang buy$Perf what?
(2) Complex wh$questions
a. Indirect questions
English: He wondered what Little Wang bought.
2
Chinese: Ta xiang zhidao Xiao Wang maile shenme.
He want know Little Wang buy$Perf what
b. Wh$Question in object complements
English: What does he think Little Wang bought?
Chinese: Ta juede Xiao Wang maile shenme?
He feel Little Wang buy$Perf what
If we adopt Huang’s movement model (1982), this difference between English
and Chinese wh$questions is due to the different timing of the wh$movement. English
wh$questions are the result of wh$movement prior to the spell$out, thus the fronted wh$
words, while in Chinese the wh$movement takes place after the spell$out, thus the in$situ
wh$questions. Later on, Rizzi (1990) attributed the reason for the different timing to the
settings of a single parameter—the strong and weak settings of the [wh] feature. The
strong setting of a [wh] feature in English attracts a wh$movement prior to spell$out. Due
to the weak [wh] feature of C, the wh$movement in Chinese is delayed after the spell$out.
Thus the L2 acquisition by English$speaking learners of Chinese simple wh$
questions and complex wh$questions can be regarded as the acquisition of a single
parameter—the weak feature strength of [wh], by learners whose L1 has a strong [wh]
feature strength.
The other difference is that Chinese wh$words or $phrases can be extracted out of
such constructions as Sentential Subjects, Complex NP, Adjunct Clause, and so forth,
while in English such extraction is illegitimate. By stipulating that the Subjacency
Principle, the principle governing long$distance movement, does not apply after spell$out,
Huang attributed the difference to the Subjacency Principle applying in English, where
the wh$movement happens before the spell$out, but not in Chinese, where the wh$
movement takes place after the spell$out (Huang, 1982). Cole and Hermon’s account
(1998) is that the Subjacency Principle is vacuous, since Chinese is not a movement
language. If this difference between English and Chinese is regarded as the instantiation
3
of the Subjacency Principle in English and uninstantiation of it in Chinese, then whether
L2 English$speaking learners can automatically uninstantiate this principle once they
realize that Chinese wh$movement takes place after the spell$out or Chinese wh$words
stay in$situ, can disclose whether this principle is indeed accessible to L2 learners and
thus does not need to be learned.
(3) a. Questioning out of a Relative Clause:
English: *Who do you like the flower that mum gives t?
Chinese: Ni xihuan mama gei shui de hua?
1
You like mum give who de flower?
To whom did mum give a flower X, which you like the best?
b. Questioning out of a Wh$question:
English: *Who do you want to know whether t will come?
Chinese: Ni xiang zhidao shui lai bu lai.
You want know who come not come.
Who is the person X, whom you wonder whether he will come or not?
c. Questioning out of a Sentential Subject:
English: *How I did it t will make him happy?
Chinese: Wo zenme cai neng rang ni gaoxing?
I how do only will make him happy?
What is the way, by which if I do it, he will be happy?
d. Questioning out of an Adjunct Clause:
English: *What are you unhappy about because I said t?
Chinese: Ni yinwei wo shuole shenme er bu gaoxing t?
You because I say$le what then not happy?
1
De is the particle introducing relative clause in Chinese. It occurs between the relative
clause and the noun phrase it modifies.
4
What is X (X=something I said) that made you unhappy?
(Huang, 1982)
In this dissertation, I choose to examine the L2 acquisition of Chinese [$wh]
feature strength in simple wh$questions and the [$wh] feature strength in complex wh$
questions by English$speaking learners.
1.2. L2 Acquisition Theories and Research
There have been several hypotheses proposed regarding the process of L2
acquisition of a parameter, for example, the Full Transfer and Full Acquisition hypothesis
(Schwartz & Sprouse, 1994, 1996a, b), the Full Acquisition and No Transfer hypothesis,
and the Minimal Tree hypothesis (Vainikka & Young$Scholten, 1994, 1996). The Full
Transfer and Full Access (FTFA) hypothesis has generally been supported by research in
the Principles and Parameters framework (White, 1985, 1986; Yuan, 1998; Slabakova,
2000). It says that learners start with the L1 setting, and then switch to the L2 setting by
accommodating to the L2 input. The problem with FTFA is that some studies show that
the L2 properties emerge very early without an obvious stage of L1 influence, and L1
influences vary from zero to different degrees (Yuan, 2001; Gess & Herschensohn, 2001;
Paradi, Schwartz, & Clahsen, 1997). The reason why there are such phenomena remains
a puzzle.
For the acquisition of the [$wh] parameter, Kim’s (2003) and Yuan’s (2007)
studies reported an even more drastic result: there is no initial transfer of the L1 English
setting [+wh] in the interlanguage of English$speaking learners of Chinese. Kim proposed
the Minimal Initial Syntax hypothesis (MIS) (Platzack, 1996) under the minimalist
framework, which says that since movement is more costly than merger, and movement
is procrastinated as long as possible, L2 learners would start with the default and minimal
5
setting, the weak feature strength, and in the case of acquiring wh$questions, the [$wh]
feature strength. Yuan (2007) resorted to a similar explanation in his study.
However, the MIS hypothesis does not seem to be able to provide a final solution
to the mixed results of L2 acquisition of weak feature strength by learners whose L1 has
strong feature strength. For example, in Parodi et al. (1997), the beginning L1 romance
language learners of L2 German, in which the N$feature is weak, produced a strong N$
feature as high as 20% of the time, influenced by their L1. In addition, the research
methods used in the studies by Kim and Yuan made the results less than convincing. For
example, in Kim’s study, the subjects were those who had learned the target language for
periods ranging from six to sixteen months, well past the stage when the initial state of
wh$questions is manifested in the learners’ L2 grammar. In Yuan’s study, only a
grammatical judgment task is used, which might not be able to provide a full picture of
learners’ real ability of [$wh] feature strength without the oral production task data.
With neither of the hypotheses having been adequately supported, and the
research results concerning the acquisition of the [$wh] feature strength being unclear, it
is worthwhile to carry out a study detailing the learning process of the [$wh] feature
strength and to test which hypothesis gives a closer account of the process.
Studies along the line of argument and adjunct asymmetry and of subject and
object asymmetry also draw our attention. There are two competing hypotheses. One is
Stromswold’s hypothesis (1995), the major claim of which is that a trace that is θ$
governed is more local and thus easier to learn than a trace that is antecedent$governed. A
fronted wh$word θ$governs the trace left by the object, and an antecedent governs the
trace of the subject, as well as that of the adjunct. Thus the learning order predicated by
this hypothesis is object > subject, adjunct. The second hypothesis is O’Grady’s
Computational Complexity Hypothesis (1997). The major claim here is that learning
difficulty is decided by the structural distance between the moved element and the trace.
According to this hypothesis, the learning order should be adjunct > subject > object.
6
Studies generally support the claim that argument wh$questions are easier than
adjunct wh$questions (Lee, 2004; Santelmann, 2003), which is correctly predicted by
Stromswold’s hypothesis. In terms of subject and object asymmetry, some studies
support Stromswold’s hypothesis (Stromswold, 1988, 1995; Chueng & Lee, 1993), and
some studies support O’Grady’s hypothesis (Kim, 1999; Seidle & Hollich, 2002;
Yoshinaga, 1996; Fahn, 2003).
Wh$questions with wh$words functioning as attributives (attributive wh$questions
henceforward) were not discussed in these two hypotheses. Since these wh$words are
fronted together with the nouns they modify as a whole noun phrase (NP), their difficulty
rating should be slightly higher than the subject or object if it were not the same,
according to O’Grady’s hypothesis. From Stromswold’s point of view, the same
conclusion can be drawn regarding the difficulty of attributive wh$questions in relation to
subject or object wh$questions, since the trace left by the attributive and the noun that it
modifies is either θ$governed or antecedent governed.
Inspired by these studies, this author would like to see whether grammatical
functions are a factor affecting the L2 acquisition of [$wh] feature strength among
English learners of Chinese. I would also like to see which of the two hypotheses can
correctly predict the learning order among these same learners.
Learners’ acquisition of object attributive, adverbial wh$questions will be
explored in this dissertation. Subject wh$questions will not be considered, since Chinese
and English subject wh$questions are exactly the same in sentence order, and if learners
produce a correct subject wh$question, there is no way for us to know whether they are
copying their English way of saying it, or if there is real change taking place in the
learners’ L2 grammar. Thus, the difficulty hierarchy relevant to our study predicted by
Stromswold’s theory should be object > = object attributive > adverb. The difficulty
hierarchy predicted by O’Grady’s theory should be adverbial > object > = object
attributive.
7
Studies on the L1 or L2 acquisition of complex wh$questions have not been
located. If the subcategorization of verbs linking indirect questions,
(wonder), and the subcategorization of verbs linking wh$questions in object
complements,
(think), is respectively assumed to be the same as their English
counterparts by English$speaking learners of Chinese, a logical question to ask is whether
learners who acquire the [$wh] feature strength of simple wh$questions would follow the
English setting or the L2 Chinese setting of the [wh] feature strength in indirect wh$
questions and in wh$questions in object complements. Also, would the matrix [$wh]
feature strength of indirect wh$questions be learned earlier than the embedded [$wh]
feature strength of wh$questions in object complements as a result of the transferring
effect of the [$wh] feature strength of simple wh$questions, since the syntactic behaviors
of simple sentences and main clauses are normally the same, as proven by many
linguistic phenomena?
2
1.3. Research Questions
In a pilot study, 75% of first$year learners of Chinese had acquired the [$wh]
feature strength of simple wh$questions at the end of the first year’s classes. Thus this
study will use first$year learners of Chinese as research subjects in order to trace the
development of the acquisition of the [$wh] feature strength.
2
One of the examples is that the yes$no question transformation rule in simple sentences
is the same as that for matrix clauses of complex sentences, but not the same as the embedded
clauses.
(i) a. Does he know it?
b. Does he know that she would not come?
c. He knows whether she would come.
Another example is that, in German, the “verb second” word order rule applies in simple
sentences and in matrix clauses of complex sentences, but not in the embedded clauses.
8
The research questions can be summarized as below. The first set of research
questions pertains to the L2 acquisition of simple Chinese wh$questions by English$
speaking learners of Chinese.
1) Is the grammatical function a factor affecting learners’ acquisition of wh$
questions? If so, would the difficulty hierarchy be object > = object attributive > adverb,
with the easiest on the left as predicated by Stromswold’s proposal, or adverbial > object
> = object attributive as predicated by O’Grady’s proposal?
2) Does learners’ interlanguage initially manifest [+wh] feature due to the
influence from their L1 English and then accommodate to the [$wh] feature in L2 Chinese
as predicated by the Full Transfer Full Acquisition Theory, or does it begin with the [$
wh] feature strength as predicated by Minimalist Initial Syntax hypothesis?
3) Do first$year learners acquire the [$wh] feature in simple in$situ wh$questions?
If so, when?
The second set of research questions pertains to indirect wh$questions and wh$
questions in object complements:
1) What is the initial stage of indirect wh$questions and wh$questions in object
complements? What are the learning results of these questions in comparison to simple
wh$questions?
2) Are indirect wh$questions or wh$questions in object complements successfully
acquired at the end of first$year Chinese learning?
3) At the end of first$year Chinese learning, what are the learning results of
learners’ acquisition of the grammatical feature that wh$words are not allowed in clause$
initial position in different clauses? Does their acquisition of this grammatical feature
vary between different clauses?
4) What is the relationship between learners’ acquisition of this grammatical
feature in different clauses?
9
1.4. Approaches of the Study
To answer the two sets of research questions, two studies were designed to obtain
the data. Study 1 is a longitudinal study to track the development of the [$wh] feature
strength of simple wh$questions by first$year English$speaking learners of Chinese. It
consists of four elicited oral production (OP) tests and four grammaticality judgment (GJ)
tests. Study 2 was intended to discover the learners’ acquisition of the [$wh] feature
strength of indirect wh$questions and wh$questions in object complements at the end of
the first year. It also consists of an OP test and a GJ test. The OP and GJ test items
designed to reveal learners’ acquisition of these [$wh] feature strength were respectively
mixed together with the fourth OP and GJ items on simple wh$questions.
This dissertation consists of five chapters. Following the introduction (Chapter 1),
Literature Review (Chapter 2) is divided into two major parts. The first part concerns the
description and theoretical account of the English and Chinese wh$questions, which leads
to the topic of the dissertation. The second part is the literature review of previous studies
related to the topic, which also leads to the research questions. Chapter 3 spells out the
research design by which the research questions can be answered. Chapter 4 proceeds to
the report of the experimental results and analysis. Chapter 5 is devoted to a discussion of
the results, and also the limitations of this study.
10
CHAPTER 2 LITERATURE REVIEW
Studies of L2 acquisition of Chinese wh$questions by L1 English$speaking
learners are motivated by the theoretical contributions that these two typologically
different languages can make to the study of L2 acquisition under the Principles and
Parameters framework. In this chapter, I will first describe the differences between
Chinese and English in simple wh$questions and complex wh$questions, including
indirect wh$questions and wh$questions in object complements, provide linguistic
accounts for those differences, and then lay out the research topics. Next, I will give a
brief review of the important issues that have been examined so far regarding acquisition
of wh$questions, and then review the L2 acquisition theories and major works related to
our research topics, which will lead to the detailed research questions in this dissertation.
2.1. Description of the English and Chinese Monoclausal
Wh$Questions and Wh$Questions in Object Clauses
I have chosen to focus on the L2 acquisition of simple Chinese wh$questions and
complex wh$questions with wh$questions as object clauses in this dissertation. Therefore,
I will describe in detail how English and Chinese differ in these two kinds of questions in
their surface structures.
First of all, the type of Chinese wh$questions, the acquisition of which I am
3
investigating in this dissertation, is non$focalized wh$questions. Traditionally, according
3
Focalized questions are used to question a piece of previously heard information when
the listeners did not hear it clearly or understand it. In Chinese focalized wh$questions, wh$
phrases can be pre$posed to the sentence$initial positions. They are excluded from the study, since
I want to eliminate the possibility that L2 English learners wrongly accepted wh$questions with
wh$words at a sentence$initial position because they took them to be focalized sentences.
A: Jinwan yeye he nainai yao da lanqiu.
This evening grandpa and grandma want play basketball.
Grandpa and grandma are going to play basketball this evening.
B: (did not hear it clearly or understand it)
jinwan tamen yao zuo?
What this evening they will do?
11
to the position of wh$words in non$focalized wh$questions (wh$questions hereafter),
languages can generally be categorized as wh$in$situ languages or movement languages.
Wh$in$situ languages refer to those languages in which wh$words stay where they are
generated in the deep structure. Chinese, Japanese, and Korean belong to wh$in$situ
languages. Note that, in Chinese, wh$questions can also optionally take a sentence final
. English, along with Spanish and Italian, belongs to the movement languages, in which
wh$words invariably occur at the sentence$initial position. This name was given since
this kind of question is traditionally accounted for as being the result of the movement of
the wh$words from their generated position to the beginning of sentences. The contrast
between Chinese and English wh$questions can be exemplified by the Chinese sentences
4
and their English counterparts in (4).
(4) Non$focalized questions in Chinese and English
a. Jinwan
hui da lanqiu (ne)?
This evening who will play basketball (ne)?
Who will play the basketball this evening?
b. Tamen jinwan hui zuo
(ne)?
They tonight are going to do what (ne)?
What are they going to do in the evening?
c. Tamen
da lanqiu (ne)?
They when play basketball (ne)?
When are they going to play basketball?
What about the distribution of wh$phrases in indirect questions and wh$questions
in object complements? In English indirect questions, the wh$phrases occur at the
What will they do this evening?
4
Some languages, such as French, have a mixed system, in that wh$words can either stay
or move to the front in the matrix clauses.
12
sentence initial position of the embedded clauses. In English wh$questions in object
complements, the wh$phrases occur at the sentence initial position of the matrix clause.
In Chinese, wh$phrases stay in$situ in both cases. This difference can be exemplified by
the English and Chinese sentences in (5) a and b.
(5) a. Ta xiang zhidao Xiao Wang maile shenme.
He want know Little Wang buy$le what.
He wondered
Little Wang bought.
b. Ta juede Xiao Wang maile shenme?
He think Little Wang buy$le what?
What does he think Little Wang bought?
2.2. Different Accounts to Explain the Typological
Differences
There are two major accounts to explain the typological differences between
English and Chinese wh$questions. The traditional idea, as proposed by Huang (1982),
maintains a movement account for wh$questions. English wh$questions are formed by
overt wh$movement, while Chinese wh$questions are formed by covert wh$movement.
Later, Tsai (1994, 1999) and Cole and Hermon (1998), under the Minimalist Program,
proposed that only English wh$questions are formed by wh$movement, while no wh$
movement happens at any level in Chinese wh$questions.
2.2.1. The Account by Huang (1982)
2.2.1.1. Principles and Parameters
Huang’s proposal was originated under the Principles and Parameters framework
of Chomsky’s universal grammar (UG). Thus it is necessary to examine the basic ideas of
this framework first.
13
Chomsky’s universal grammar (UG) (1965) was based on the idea that human
beings are endowed with an innate language faculty, which enables children to acquire
languages rapidly and efficiently. The rule$based phase of UG grammar, in which
language faculty is formalized in language$specific rules, was the first phase to develop
out of UG theory. In the 1980s, to meet the need to efficiently explain as well as
adequately describe the languages of the world, the Principles and Parameters (PP)
paradigm was developed, which is the second phase of UG grammar. Under this
framework, grammar consists of lexicon, D$structure, S$structure, and two interpretative
components: Phonetic Form (PF) and Logical Form (LF). Language facility consists of
two constraints—principles and parameters. Principles are the features of natural
languages that are universal; parameters are the values of a property that are subject to
cross$linguistic variation. While principles explain how languages are the same,
parameters are used to explain how languages are different.
Syntax
S$Structure
PF
LF
Figure 1. Model of Grammar, Principles and Parameters
Source: From
Dordrecht: Foris.
(p. 17), by Noam Chomsky, 1981,
From the point of view of language acquisition, principles are predetermined and
need no learning. Parameters have limited values, the choice of which has to be fixed
according to input in the learning process. Hence, the acquisition of the syntax of a
foreign language consists of setting the parameters, which vary across languages.
14
What parameters need to be reset in the L2 acquisition of Chinese wh$questions?
Let us discuss this further after we give a linguistic account of Chinese wh$questions.
2.2.1.2. Wh$phrases as Quantificational Operators
Huang’s account (1982) was built upon the claim that wh$words in both English
and Chinese are operators, just like the common quantificational operators
and
From (6) and (7) below we can see the characteristics manifested by
quantificational operators. First, semantically, they do not pick out specific referents.
Rather, their interpretation is variable. For example, (6) means: “for every x, x is human,
it is the case that Louis saw x.” Second, the trace that
is a variable; it is bound by
or
or
leaves behind
for interpretation. Third, it takes scope over the
entire sentence by being raised to the beginning of the clause, which is called Quantifier
Raising.
(6) Louis saw everyone.
a. Deep structure: [vp Louis v’[saw everyone]]
b. Surface structure: [IP Louisi sawj I’[ti ji everyone]]
c. Logical Form: [For all x, x is human,] [Louise saw x]
(7) Louis saw someone.
a. Deep structure: [vp Louis v’[saw someone]]
b. Surface structure: [IP Louisi sawj I’[ti ji someone]]
c. Logical Form: [For some x, x is human,] [Louise saw x]
These quantificational sentences also constitute one type of evidence for the
existence of logical form (LF). Deep structure (DS) encodes the thematic and selectional
properties of a sentence, and surface structure (SS) encodes the properties of the sentence
for its “spell$out,” for the way that the sentence is spoken. For example, (6) a and (6) b
represent the DS and SS of “Louis saw everyone,” respectively. However, the
interpretation of sentence (6) cannot be derived from its S$structure, (6) b, but is more
15
appropriately represented by (6) c, which is the logical form of the sentence. The
quantificational sentences show us the necessity of proposing LF as another layer of
representation of a sentence. Thus, LF is defined as the logic, or interpretative
representation, of a sentence.
With these concepts introduced, let us look at why Huang regards wh$words as
operators. According to him, a wh$question can be considered as “consisting of a
presupposition of having the form of a quantificational sentence and a focus indicating
the speaker’s request for a specification on the value of the quantificational element in the
presupposition” (1982, p. 176). Take the English wh$question (8) as an example. It can be
decomposed into
Since
and the request for
.
can be taken as [For some x, x is human,] [x arrived], the LF
representation of
can then be [[which x; x a person] [x arrived]] in (8) c, as
suggested in Chomsky (1975). Note that the LF representation of the English wh$
question is directly obtainable from its S$structure by a simple algorithm.
(8) Who arrived?
a. Deep structure: [+wh] Somebody arrived. (Note: In early transformational
grammar, the D$structure of a wh$question is “somebody arrived,” with somebody
containing the feature [+wh].)
b. Surface Structure: [Whoi [ti arrived]]?
c. Logical form: [[which x; x a person] [x arrived]]?
(Huang, 1982, p. 252)
The above analysis leads to the conclusion that interrogative wh$words are also
quantificational operators because 1) their interpretations are also variable, for example,
the interpretive structure of “who” is “which x, x a person”; 2) wh$words also have a
binding relationship with the traces they leave behind, i.e., variables; 3) they take scope
over the elements to their right, and render them questions.
16
Because of the semantic similarity between English and Chinese wh$words,
Chinese wh$words are assumed by Huang to be quantificational operators. Since
quantificational operators are subject to QR, it makes sense to assume that Chinese wh$
questions have a similar quantifier$variable representation in LF, as shown by example
(9) c. Note that the LF representation of the Chinese wh$question is not directly
obtainable from its surface representation.
(9) Ni kanjian$le shei?
You see$Asp who
Who did you see?
a. Deep structure: [+wh] [Youren lai le]
Someone come$Asp.
b. Surface structure: [Nii kanjianj$le [ti ji shei]?
You see$Asp who
Who did you see?
c. Logical form: [[neige x; x yige ren] ni kanjian$le x]
which x, x a man you see$Asp
(Huang, 1982, p. 253)
2.2.1.3. Syntax of English Wh$Movement
What is the syntax and motivation of wh$movement? First, wh$movement is a
movement of maximal projection to Spec of CP. Since, in simple wh$questions, the
interrogative wh$words are in an initial position and the auxiliary verbs must precede the
subjects—which are believed to occupy a Spec, IP position—interrogative wh$words and
auxiliary verbs must land in a position further up to IP. Spec, CP is proposed as the
landing position of wh$words, and C is believed to be where the auxiliary verbs move to.
For example, in all the sentences in (10), the wh$words, functioning as subject, object, or
17
adverbials, are maximal projections by virtue of being the largest expression headed by
wh$words, and are all moved to the Spec of CP.
(10) a. [CP
is [IP t [VP [VP going to play basketball] in the evening]]]?
CP
spec C’
who C
IP
is spec
t
I’
I
VP
VP
V
PP
NP in the evening?
going to play basketball
b. [CP
are [IP they [VP [VP going to do t] in the evening]]]?
c. [CP
are [IP they [VP [VP going to play basketball] t]]]?
Second, there are at least two reasons that motivate the wh$movement. One is to
check the features of C, more specifically, the Q feature of C (Chomsky, 1993, 1995).
Under the feature checking theory, functional projections, i.e., IP or CP, have abstract
5
features and, to be interpretable according to Full Interpretation , they have to be checked
by the checkers, i.e., person, number and tense, which are believed to be generated within
the VPs. The Q feature of C is one of these abstract features, and it has to be checked by a
wh$word which has the Q feature of C. Rizzi (1990) utilizes the features for wh$
movement in wh$questions and relative clauses in C position, and calls them the [wh]
feature. Thus for Rizzi, it is the [wh] feature that motivates the operator movement, in
5
Full Interpretation is a principle that requires that every element of PF and LF must
receive an appropriate interpretation, that is, must be licensed in the relevant sense (Chomsky,
1986).
18
relative clauses as well as in wh$questions. The other is for semantic reasons. The
movement of wh$operators to sentence$initial positions, or quantifier raising, renders the
interrogative scope to the whole sentence. Quite similarly, Cheung (1992) proposes that
every clause must be typed as either declarative or interrogative, and that a clause is
typed as interrogative if it contains an interrogative head or specifier. The movement of
wh$phrases to Spec of CP in English then serves to type the sentence as interrogative.
2.2.1.4. Account for English and Chinese Monoclausal Wh$
questions
After we become familiar with the important concepts of wh$words as
quantificational operators and wh$movement, let us see how the difference between
English and Chinese wh$questions is accounted for.
Since wh$words, as quantificational operators, are subject to QR, it makes sense
to believe that both English and Chinese wh$questions involve wh$movement, with
English wh$movement taking place in the surface structure and Chinese wh$movement
taking place in the LF. An advantage of this conceptualization is that “it allows for a
simple statement of the fact that all languages have the same semantics of wh$questions,
though they may each have a different syntax of such sentences” (Huang, 1982, p. 178).
In the English sentences in (9), though
,
, and
are originated within
the different layers of verb phrases in deep structures, that they are located at the
sentence$initial position in these monoclausal wh$questions in the surface structure
indicates that the movement of wh$words must have taken place in the surface structure.
If Chinese wh$words move to Spec of CP to take scope, as we have assumed to be
the universal nature of wh$words, the mismatch between the starting point of the
interrogative scope (the beginning of the sentence) as shown in their LF representations
in (11), and the in$situ position of wh$words in surface structure as shown in (10), is the
evidence that wh$words move to the Spec of CP after spell$out, i.e., in LF.
19
(11) a. Jinwan
hui da lanqiu (ne)?
This evening who will play basketball (ne)?
Who will play the basketball this evening?
b. Tamen jinwan hui zuo
(ne)?
They tonight are going to do what (ne)?
What are they going to do in the evening?
c. Tamen
da lanqiu (ne)?
They when play basketball (ne)?
When are they going to play basketball?
(12) a. (CPshuii (IPjinwan ti hui VP (da lanqiu)))
Who tonight will play basketball?
b. (CPshenmei (IPtamen jinwan hui VP (zuo ti)))
What they tonight will do?
c. (CPshenme shihoui (IPtamen (VP ti da lanqiu)))
When they play basketball?
If both English and Chinese wh$words move to the front to either check the [wh]
feature or to take scope, what is the reason that English wh$words move in the surface
structure while Chinese wh$words move in LF? Rizzi (1990) attributed it to the strength
of the [wh] feature. The [wh] feature strength in English is strong, while the [wh] feature
strength in Chinese is weak, as a result of which wh$words are attracted to Spec, CP in
English before the spell$out, while wh$words are attracted to Spec, CP in Chinese only
after the spell$out. Looked at in terms of parameters, the parameter at which English and
Chinese wh$questions differ is in the [+wh] feature of C in English and the [$wh] feature
of C in Chinese.
20
2.2.1.5.Account for English and Chinese Complex Wh$
Questions with Wh$questions as Object Clauses
What is the syntactic explanation for the difference between indirect questions
and wh$questions in object complements in English? Can the different timing of the wh$
movement apply in the explanation of the differences between the two languages in
complex wh$questions?
According to Huang (1982), the difference between indirect questions and wh$
questions in object complements, as shown by (2) a and (2) b, repeated here as (13) a and
(13) b, is due to the different subcategorization of the main verbs. Since
is
subcategorized to take an interrogative clause, to satisfy the selectional requirement of
the main verb,
can move up to the Spec of the lower CP only, but not further up.
Another way to look at it is that since (13) a belongs to indirect questions, where the
interrogative force lies within the embedded clauses cross$linguistically,
in the lower CP to render it a question. In contrast, since
statement,
has to stay
is subcategorized to take a
has to move further up to avoid the selectional clash with the main verb.
Another way to look at it is that if the entire sentence is planned as a question,
has
to move to the higher CP to render it so.
The difference between the Chinese and English complex wh$questions can be
accounted for by different timing of movement as well. In (13) a,
as a result of the
movement to Spec of CP of the lower clause, takes scope over the lower clause, as to
how it is interpreted; in (13) b,
takes scope over the entire clause as a result of the
wh$word moving to the Spec of CP of the matrix clause. If the semantics of wh$questions
is assumed to be the same cross$linguistically, then for the Chinese cases, the mismatch
between the position of in$situ wh$words in (14) and the starting points of the
interrogative force rendered by these wh$words, i.e., the positions of wh$words in LF in
example (15), indicates that the movement of wh$words is delayed in LF, just as it is in
simple wh$questions.
21
(13) a. [CP He wondered [CP
b. [CP
Xiao Wang bought t]].
does he think [CP Xiao Wang bought t]]?
(14) a. Ta xiang zhidao Xiao Wang maile shenme.
He want know Xiao Wang buy$le what.
He wanted to know what Xiao Wang bought.
b. Ta juede Xiao Wang maile shenme?
He feel/think Little Wang buy$le what?
What does he think Little Wang bought?
(15) a. (IP Ta xiang zhidao (CP shenmei (IP Xiao Wang maile ti))).
b. (CP Shenmei (IP ta juedei Xiao Wang maile ti)).
If we attribute the motivation for the different timing of movement to the strength
of the [wh] feature, then the difference between English and Chinese indirect questions is
due to the embedded [wh] being strong in English and weak in Chinese. Similarly, the
difference between English and Chinese wh$questions in object complements lies in the
matrix [wh] feature strength being strong in English and weak in Chinese.
Table 1. The Difference between English and Chinese WH$Questions in Terms of [wh]
Feature Strength
Wh$Q in Object
Language
Simple Wh$Q
Indirect Q
Complements
Position
CP
Embedded CP
Matrix CP
English
[+Wh]
[+Wh]
[$Wh]
Chinese
[$Wh]
[$Wh]
[+Wh]
Note: Q = Questions
22
2.2.1.6. The Explanatory Power of Huang’s Proposal
Huang’s proposal—that English wh$questions involve movement in SS, while the
movement in Chinese wh$questions is delayed in LF—also has the power to explain the
illegitimacy of wh$extraction out of islands in English and the legitimacy of wh$
extraction out of
in Chinese.
are structures such as relative clauses, wh$
questions, sentential subjects, and adjunct clauses. Since wh$extraction out of these
structure is impossible in English, they are metaphorically referred to as
.
Huang’s explanation is that subjacency6 is observed only in surface structure but
not in LF. Accordingly, this principle applies in the English cases but is irrelevant in the
Chinese cases, which results in ungrammatical movement out of islands in English but
grammatical sentences in Chinese.
(16) a. Questioning out of a Relative Clause:
English: *Who do you like the flower that mum gives t?
Chinese: Ni xihuan mama gei shui de hua?
You like mum give who the flower?
To whom did mum give a flower X, which you like the best?
b. Questioning out of a Wh$question:
English: *Who do you want to know whether t will come?
Chinese: Ni xiang zhidao shui lai le bu lai.
You want know who come$perf not come.
6
Chomsky proposed subjacency (1973, 1981 a, b, 1986a) as a principle for wh$
movement constraints which are observed in English and other Western languages such as French
and Italian. In its original form (Chomsky, 1973), subjacency says that “in a configuration
…X…[a…[b…Y…]…]…X... where a and b are bounding nodes, no rule may move an element
from the position Y to either the position of X or conversely.” In simple words, subjacency says
that no wh$movement can cross more than one bounding node. Since the bounding nodes are
taken to be NP and IP in English, this explains why extraction out of the various kinds of islands
is bad.
23
Who is the person X, whom you wonder whether he will come
or not?
c. Questioning out of a Sentential Subject:
English: *How I did it t will make him happy?
Chinese: Wo zenme cai neng rang ni gaoxing?
I how do only will make him happy?
What is the way, by which if I do it, he will be happy?
d. Questioning out of an adjunct clause:
English: *What are you unhappy about because I said t?
Chinese: Ni yinwei wo shuole shenme er bu gaoxing t?
You because I say$le what then not happy?
What is X (X=something I said) that made you unhappy?
(Huang, 1982)
2.2.2.The account by Tsai (1994, 1999) and Cole and
Hermon (1998)
2.2.2.1. Minimalist Program
The account given by Tsai (1994, 1999) and Cole and Hermon (1998) for English
and Chinese wh$questions was originated under the framework of the Minimalist
Program (MP). Let us look at some of the basic assumptions of MP. MP was an updated
version of UG after the rule$based phase in the 1950s and the PP phase in the 1980s. It
was proposed by Chomsky in the 1990s as a reaction to the increasingly complex
structures and principles of UG. Under the MP paradigms, grammar is pared down to
three parts only: the lexicon, the interpretative Phonetic Representation (PF) and Logical
Representation (LF), and two operations— move$α, and merge. The underlying ideas of
the Minimalist Program are as follows: 1) There really is only one grammatical system
for all languages. That is, languages are not subject to syntactic parameterization. 2)
24
Differences between languages are due to differences in the lexicon, where lexical items
are fitted out with different features. Hence, it is the features of the lexicon that lead to
the different surface manifestation of languages, rather than the parametric value inherent
in the syntax of the languages as viewed in the Principles and Parameters framework.
Under the MP paradigm, the acquisition of the syntax of a language is viewed as a
process of learning the features of the lexicon. What features of the English and Chinese
wh$words decide the difference between English and Chinese wh$questions? What
features of Chinese wh$questions does an English$speaking learner need to learn to
acquire Chinese wh$questions? These questions will be answered in the next section.
Lexicon
Spell$out
PF
LF
Figure 2. Model of Minimalist Program
Source: From !
Press
"
#
, by Noam Chomsky, 1995, Cambridge: MIT
2.2.2.2.Wh$words as the Merger of Operators and
Variables in English and Variables in Chinese
Tsai (1994, 1999) and Cole and Hermon (1998) proposed a binding model for wh$
questions. In this model, English wh$words are viewed as the merger of operators and
variables (OP+VAR), while Chinese wh$words are mostly variables.7
7
Adverbial wh$words are not variables, but are a merger of operator and wh$words, just
like English wh$words. This proposal is raised based on the fact that the extraction of nominal
wh$words is not sensitive to island effects, but the extraction of adverbial wh$words is sensitive to
island effects, just like the English wh$words.
25
The assumption for the claim that Chinese wh$words are variables is that wh$
words should uniformly be either operators or variables. Since Chinese wh$words with
non$interrogative meanings are variables, wh$words in wh$questions are variables as
well.
The evidence that has been used to argue that Chinese non$interrogative wh$
words are variables is that they are used in a variety of non$question situations where
their interpretation can be achieved only when they are regarded as bound variables (Li,
1995; Huang, 1982; Aoun and Li, 1993). Sentences in (16) indicate that Chinese wh$
words are indefinite variables bound by the universal quantifier
receive their interpretation from
(all), since wh$words
.
(17) a. Ta shei dou renshi.
He who all know.
He knows everyone.
b. Ni shenme shihou lai dou keyi.
You what time come all ok.
Whenever you come will be ok.
c. Wo shenme dou bu zhidao.
I what all not know.
I do not know anything.
By the same line of reasoning, the fact that the wh$word in (17) a receives its
interpretation from the operator
, which functions to turn the statement into a yes$and$
no question, and that the wh$word in (17) b receives its interpretation from the operator
, which negates a finished event, indicates that they are bound variables in these two
cases as well.
(18) a. Qiaofeng mai$le shenme ma?
Qiaofeng buy$asp what Q$Y/N?
Did Qiaofeng buy anything?
26
b. Qiaofeng mei$you mai shenme.
Qiaofeng not$have buy what.
Qiaofeng did not buy anything.
Another piece of evidence is wh$words (always two identical) in Bare
Conditionals. From (18), and its LF representation in (19), we can see that
(who) is a
variable, bound by a phonologically$null universal quantifier (Tsai, 1994).
(19) Chinese Bare Conditional
Shei xian lai, shei jiu xian chi.
Who first comes, who then first eat.
If x comes first, x eats first.
(20) For all (x) (x comes first x eats first)
Notice that Huang (1982) does not agree that wh$words uniformly fall into either
the operator or variable category. Thus, though he believes that non$interrogative Chinese
wh$words are variables, interrogative wh$words are operators just like their English
counterparts.
English wh$words are never bound by non$interrogative operators as in Chinese,
which Tsai (1994, 1999) and Cole and Hermon (1998) use as evidence to prove that
English wh$words are not variables. English wh$words are assumed to be a merger of
operator and variable because of their explanatory power for such phenomena as the
overt movement and the island effect in English, which will be talked about in detail in
the next sections.
2.2.2.3. Account of the English and Chinese Monoclausal
Wh$questions
Under the MP paradigm, the difference between wh$questions in different
languages should not lie in the different strengths of the [wh] feature, since the syntax of
27
wh$questions should be the same. Rather, it is the different properties of wh$words that
cause the difference.
Tsai (1994, 1999) and Cole and Hermon (1998) assumed that wh$questions cross$
linguistically have the following operator/variable structure, where the OP binds a
variable x. In addition, the [wh] feature is strong cross$linguistically.
(21)
OPx [… x …]
With the properties of English and Chinese wh$words specified in the previous
section and the syntax of wh$questions laid out as above, let us see how the difference
between English and Chinese wh$questions can be accounted for. As the base$generated
wh$words are the fusion of the OP and VAR in English, they are forced to move with the
OP movement and they move to Spec, CP. The OP movement is motivated by the need to
check the strong [wh] feature of C, which is realized through the Spec and head
relationship. As a result, the OP binds the trace, x, it left behind (See (22) a). In contrast,
since in Chinese the base$generated wh$nouns are VAR, and the phonologically null OP
is generated in the Spec of CP, the binding relationship between the OP and VAR is
established without the need for wh$word movement. The strong [wh] feature of C can be
regarded as being checked by the base$generated null OP. The null operator can bind a
number of different variables (See (22) b). That is why the relationship between the null
operator and variables in Chinese wh$questions are regarded as Unselective Binding by
Tsai.
(22) a. English Wh$Questions
OPx [… x …]
whoi
ti
whati
ti
wheni
ti
28
b. Chinese Wh$Questions
OPx [… x …]
Null
Shei (who) (VAR)
Shenme (what) (VAR)
Shenme shihou (when) (VAR)
……
2.2.2.4. Account of the English and Chinese Complex Wh$
questions
How would the accounts of Tsai and Cole and Hermon cope with English and
Chinese wh$questions in object clauses? They never explicitly addressed this in their
articles. However, it can be predicated that they would say that, due to the different
subcategorizations of the main verbs, the two kinds of wh$questions have OPs generated
in different scope positions. When the verb is subcategorized to take a question, such as
, OP is generated in the lower Spec, CP both in English and Chinese wh$questions
as shown in (23) a and (24) b. When the verb is subcategorized to take a statement, such
as
, OP is generated in the higher Spec, CP both in English and Chinese, as shown in
(23) b and (24) b. The different positions of wh$words in English and Chinese are again
decided by the nature of the wh$words. Since English wh$words are the merger of OP and
8
VAR, they move to check the [wh] feature of C in OP, whether they are in the lower
clause or upper clause. Chinese wh$nouns stay in$situ since they are VAR, and thus they
can form a binding relationship with the null OP without any movement.
(23) a. He wondered (OPi what Xiao Wang bought i).
b. (OPi What does he think (that Xiao Wang bought i))?
8
In Cole and Hermon (1998), the feature that English wh$word fronts to check in Spec,
CP position and form wh$questions is called Q. Rizzi (1990) believes that [wh] feature motivates
wh$movement in relative clauses as well as in wh$questions.
29
(24) a. Ta xiang zhidao (OPi Xiao Wang maile shenmei).
He want know Xiao Wang buy$le what.
He wonders what Little Wang bought.
b. (OPi Ta juede (Xiao Wang maile shenmei))?
He think Little Wang buy$le what?
What does he think little Wang bought?
2.2.2.5. Explanatory Power of Tsai and Cole and Hermon’s
Proposal
The binding model can equally well explain the island effects in English and
Chinese. English wh$extraction out of islands is illicit, since the wh$movement violates
Subjacency. Chinese wh$questions do not involve wh$movement, as a result of which the
wh$extraction out of islands is not governed by Subjacency. In other words, that
Subjacency applies in English wh$questions, but is vacuous in Chinese, accounts for the
different island effects in the two languages.
Tables 2 and 3 summarize the structure of English and Chinese wh$words
respectively (adapted from Cole & Hermon, 1998).
Table 2. The Operator$variable structure of English wh$words
Question word
Matrix C
Who, what,
Q
when, where,
Wh$ word
OP+VR
Consequence
Movement before spell$out
Island effects
how and why
Note: How and why in this table are not adverbial but nominal.
30
Table 3. The Operator$variable structure of Mandarin wh$words
Question word
Matrix C
Wh$word
who, what,
Q/OP
VAR
Consequence
Binding; no island effects
when, where,
how, why
2.3. Selecting a Linguistic Model and Research Topics
The strongest criticism of Huang’s model is that the claim that the Subjacency
Principle applies in Surface Structure but not in Logical Form is a stipulation. In addition,
partial wh$movement languages challenge the claim that Subjacency Principle does not
apply in Logical Form.9 The binding model (Tsai, 1994, 1999; Cole & Hermon, 1998) is
superior in this sense. However, as far as the difference between Chinese and English
simple wh$questions and complex wh$questions with wh$questions as objects are
concerned, the two linguistic models make little difference. According to Huang’s
9
The strongest criticism of Huang’s account is that the claim that Subjacency cannot be
observed at LF is a stipulation. More seriously, partially$moved wh$questions found in some
languages prove that subjancency is observed in LF, and thus challenges Huang’s claim.
The Malay sentence in example (i) a is an example of a partially moved wh$question, in
that it involves an overt movement (i.e., in surface form) of a wh$phrase to the intermediate Spec,
followed by a covert movement (i.e., in LF) out of object clauses to the upper Spec (Cole &
Hermon, 1998). That sentence (i) b, which involves an overt movement of wh$phrase and then a
covert movement out of a relative clause in LF, is ungrammatical casts doubt on whether
Subjacency is really not obeyed at LF cross$linguistically:
(i) a. Partially$moved wh$questions out of object clause:
Ali memberitahu kamu tadi ( $ i yang (Fatimah baca ti))?
Ali told you just now what that Fatimah read.
What did Ali tell you just now Fatimah was reading?
b. Partially$moved wh$phrases questioned out of relative clause:
*Kamu sayang (perempuan yang Ali fikir (dengan $ i yang telah jumpa ti))?
Kamu love woman that Ali thinks with who that already meet.
Which is the x, that Kamu loves the woman that Ali thinks meets x?
(Cole & Hermon, 1998)
According to Tsai (1994) and Cole & Hermon’s model, the ungrammaticality of (i) b is
explainable. When $ crosses over the relative clause island in LF, it violates Subjacency,
causing the sentence to be ungrammatical.
31
movement model in the PP framework, the Chinese and English wh$questions differ in
one single parameter, the [wh] feature strength, which is strong in English but weak in
Chinese. According to the binding model in the MP framework, the Chinese and English
wh$questions differ in one single parameter as well, the lexicalization of the wh$words—
in English it is [operator+variable] and in Chinese it is [variable] in most cases. Most of
the previous studies on wh$questions adopted Huang’s movement model. I will adopt
Huang’s model in this dissertation as well, not because I believe it provides a sounder
linguistic explanation of wh$questions, but because it would be easier to compare this
study with others due to the common terms used.
Two theoretically interesting topics arise if the acquisition of Chinese wh$
questions by English$speaking learners is regarded as resetting the [wh] feature strength
from strong to weak. One regards the acquisition of simple wh$questions. Will English$
speaking learners of Chinese initially manifest the [+wh] feature due to the influence
from their L1 English and then reset it to [$wh] according to the L2 Chinese? The other
regards the acquisition of complex wh$questions, including indirect questions and wh$
questions in object complements, as shown by (23), quoted here as (24). These two kinds
of complex wh$questions will hereafter be respectively termed wh$questions of
and wh$questions of
for brevity. If learners are shown to have mastered the
[$wh] feature strength of simple wh$questions, what path would learners follow when
they learn the two kinds of complex wh$questions? Would they start afresh and follow
the English setting, [+wh], or the L2 Chinese setting, [$wh]? Also, would the [$wh]
feature strength of the main clause of wh$questions of
wh] feature strength of wh$questions of %
be learned earlier than the [$
, since the syntactic behavior of
simple sentence and main clause are normally the same, as proven by many linguistic
phenomena?
(25) a. Ta xiang zhidao Xiao Wang maile shenme.
He want know Xiao Wang buy$le what.
32
He wanted to know what Xiao Wang bought.
b. Ta juede Xiao Wang maile shenme?
He feel/think Little Wang buy$le what?
What does he think Little Wang bought?
Next we are going to examine theories and studies that are most relevant to our
research topics, and then lay out the detailed research questions of this dissertation.
2.4. Review of Major Theories and Studies Regarding
Acquisition of Wh$Questions in the Principle and
Parameter Framework
2.4.1. Major Issues Examined
The study of wh$questions has always been a hot area. Research topics include
such issues as the acquisition order of wh$words in both L1 and L2, and the factors
contributing to the differential difficulty associated with acquisition order (Cairns & Hsu,
1978; Merkin, Wootten, & Bloom, 1980; Bloom, Merkin, & Wootten, 1982; Clancy,
1989; Saunders, 1983), inversion errors associated with different grammatical functions
(Rowland & Pine, 2000; Van Valin, 2002; Park, 2003; Lee, 2004), and argument/adjunct
asymmetry and subject/object asymmetry (Ervin$Tripp, 1970; Tyack & Ingram, 1977;
Cairns & Hsu, 1978; Stromswold, 1988; Sarma, 1991; Wilhelm & Hanna, 1992; Cheung
& Lee, 1993; Kim, 1995; Stromswold, 1995; Stromswold, 1998; Yoshinaga, 1999; Seidl
& Hollich, 2002; Lee, 2004).
Inspired by the Government and Binding syntactic theory, researchers used the L1
and L2 acquisition data of wh$questions to study the early stage of learners’
interlanguage (Radford, 1994; Bromberg & Wexler, 1995; Hulk, 1996; Santelmann,
1997; Grinstead, 2001; Kim, 2001; Vainikka & Young$Scholten, 2002; Kim, 2003;
Rojina, 2004; Gruter, 2006) and the syntax of wh$questions (Grinstead, 2004). Works
33
under the Principles and Parameters framework covered such topics as the availability of
UG principle to L2 learners or that of the uninterpretable feature (Schachter, 1988;
Johnson & Newport, 1989; White, 1990; Martohardjono, 1993; Hattori, 2004; Hawkins
& Hattori, 2006; Yuan, 2007) and the resetting of the movement/non$movement
parameter in L2 acquisition (Hulk, 1996; Kim, 2001; Kim, 2003; Yuan, 2007).
2.4.2. Full Transfer and Full Acquisition and Minimal
Initial Syntax Hypotheses
Since one of our research topics is regarding the resetting of the [+wh] feature
strength from strong to weak, we are going to first review the hypothesis regarding the
parameter resetting in L2 acquisition. There are several hypotheses regarding whether a
parameter can be reset from its L1 setting to the its L2 setting and the path of this
resetting, such as the Full Transfer Full Access (FTFA) Hypothesis (Schwartz & Sprouse,
1994, 1996), the No Parameter Resetting Hypothesis (Hawkins, 1998; Hawkins & Chan,
1997), and the Valueless Feature Hypothesis (Eubank, 1993, 1994, 1996; Beck, 1998).
Not only were there conceptual and empirical problems with the No Parameter Resetting
and Valueless Feature Hypotheses (White, 2003), but the results of many empirical
studies (White, 1985, 1986; Yuan, 1998; Slabakova, 2000) are generally consistent with
FTFA.
The major claim of the FTFA hypothesis is that L2 learners will start with the
parameter value of their L1 in their L2 grammar and then shift to the L2 value,
accommodating the L2 input. However, this hypothesis is not without problems. Some
studies show that the L2 properties emerge very early without an obvious stage of L1
influence, and the L1 influences vary from zero to different degrees (Yuan, 2001; Gess &
Herschensohn, 2001; Paradi, Schwartz & Clahsen, 1997). Supporters of the FTFA
hypothesis can argue that it was because the data of the study was not collected early
34
enough. However, the immediate problem, then, is how early is early enough, and is the
FTFA hypothesis falsifiable? Or, why are some parameters acquired earlier than others?
The Minimal Initial Syntax hypothesis (MIS) (Platzack, 1996) might provide an
answer for why some parameters are acquired earlier than others. Under the minimalist
framework, movement is constrained by the Procrastination Principle, which states that
movement should be avoided as long as possible, since movement costs. Platzack claims
that minimalist syntax provided new insight into the concept of markedness: strong
features motivating movement before spell$out are marked, and weak features causing
movement after spell$out are unmarked. He proposes that L2 learners start with the initial
minimal syntax, i.e., the unmarked features, irrespective of their L1, and then switch to
the marked features. It can be inferred that it could be easier for learners to acquire a L2
parameter, which is weak or unmarked, since it is the default. This seems to explain the
results of Yuan’s study (2001), in which the beginning French$speaking learners of
Chinese succeeded in learning the weak verb movement parameter of Chinese from the
beginning. However, it cannot explain why, in Parodi et al. (1997), the beginning L1
romance language learners of L2 German still produced the strong N$feature 20% of the
time, as a result of the L1 influence.
Studies that examine the resetting of the [wh] feature strength from strong to
weak are quite limited. There are two so far, both of which seem to show that there is not
an initial stage where the strong [wh] feature strength of L1 is manifested in their L2
interlanguage.
One of these studies was done by Kim (2003), who was trying to see which of the
above two hypotheses correctly predicts the initial state of the L2 interlanguage grammar.
He conducted a bi$directional study, which looked at the L2 acquisition of English wh$
questions by beginning$level Korean$speaking learners of English, and the L2 acquisition
of Korean wh$questions by English$speaking learners of Korean. Korean is an in$situ
language which has [$wh] feature strength, while English is a movement language with
35
[+wh] feature strength. The subjects in the study had been taught the target language for
at least six months but no more than sixteen months. Data was collected through two
elicited written production tasks. The results were that the English$speaking group
produced the target$like Korean wh$questions 77% of the time, and the non$target wh$
questions 16.95% of the time. On the other hand, although the Korean group produced
wh$questions with fronted wh$words 84.6% of the time, only 15.96% was target$like with
subject$auxiliary inversion, only the latter of which was deemed an indication of a fully$
developed CP projection, and thus evidence of the acquisition of [+wh] feature strength.
Kim’s conclusion is that the results of the study are supportive of the MIS
hypothesis, according to which both the English$speaking beginner group and the
Korean$speaking group would assume the [$wh] feature strength as the initial stage.
Consequently, the Korean group would experience more difficulty than the English
group, since the target form for them is the opposite of the form they assumed but is
exactly the same as the one assumed by the English group.
The other study was done by Yuan (2006), who researched whether the
Subjacency Principle can be uninstantiated in English$speaking learners’ L2 grammar,
and also whether there is L1 transfer when learners acquire the [$wh] feature strength.
Since the second one is directly related to the research topic, only that part of his study
will be reviewed. The instrument used to test the resetting of the [$wh] feature strength
was a GJ test, which contained eight grammatical in$situ wh$questions and eight
ungrammatical wh$questions with the wh$words moved. Half of these test items were
nominal wh$words, such as
adverbial wh$words, such as
(who) and
(what), and half of them were
(when),
(where),
(how) and
(why). The subjects included a native Chinese group and some English$
speaking groups with various levels of Chinese. The beginner group consisted of learners
who learned Chinese for four months. The results showed no difference between the
native Chinese group and the beginner English group in their judgment of the
36
grammatical sentences and the ungrammatical sentences of nominal wh$questions. The
conclusion Yuan drew from the study is that beginners did not start with their L1 English
setting, the [+wh] feature strength, in their L2 grammar, as predicted by the FTFA
hypothesis.
Yuan attributed the results to Minimalist grammar as well, which says that
moving is more costly than merging. Since his study also showed that learners’
acceptance of optional wh$questions particle
, and yes$no question particle
, and
their acceptance of in$situ questions all occurred at the same time, he hypothesized that
the use of
and
inform the learners that the C position is valued and that any
movement of a wh$word to Spec, CP will become unnecessary and impossible.
We can see from the above that the FTFA hypothesis has trouble explaining the
non$obvious L1 transfer effect at the early stage of L2 acquisition. Claiming that the
weak feature strength is the default of L2 learners’ grammar, and thus L2 learners would
not be subject to L1 influence if they were learning a weak feature strength, MIS theory
can only account for the results of some studies but not of all.
Kim’s study and Yuan’s study on L2 acquisition of the [$wh] feature strength lend
a certain support to MIS theory. However, the initial developmental stage of the [$wh]
feature strength was not clearly revealed. First, Kim’s study at best showed that the L2
English group had an easier time than the L2 Korean group in learning the wh$questions
in the target language at a certain point, and the initial stage of the in$situ wh$questions in
the L2 English group’s grammar was unclear. The reason was that the subjects were
those who learned the target language for from six to sixteen months, and might have
passed the initial stage of the development of this grammar. Second, what was
problematic in Yuan’s study is that the task he used was the GJ test alone. It has been
proved that the L2 GJ task may not elicit the true representation of learners’ L2 grammar
knowledge, since when making grammatical judgments, L2 learners are reported to use
strategies other than intuitive judgment, which is used by native speakers (Ellis, 1991;
37
Goss, Zhang, & Lantolf, 1994; Davies & Kaplan, 1998). We will discuss this in detail in
Chapter 3.
An interesting result in Yuan’s study is that in judgment of wh$questions with
adverbial wh$words, the beginner group had significantly lower scores than all the other
groups in both the grammatical and ungrammatical sentences. Their judgment of wh$
questions with nominal wh$words did not show such patterns. Yuan did not give an
explanation for this phenomenon in his study. Studies along the line of argument and
adjunct asymmetry might give us a clue to why such a difference occurred.
2.4.3. Argument and Adjunct Asymmetry
The results in Yuan’s study (2006) and the discovery that wh$questions headed by
wh$words with different grammatical functions are acquired at different stages (Ervin$
Tripp, 1970; Tyack & Ingram, 1977; Cairns & Hsu, 1978; Stromswold, 1988; Sarma,
1991; Wilhelm & Hanna, 1992; Cheung & Lee, 1993; Kim, 1995; Stromswold, 1995;
Stromswold, 1998; Yoshinaga, 1999; Seidl & Hollich, 2002; Lee, 2004) make us wonder
whether grammatical functions would also be an influencing factor in the adult English$
speaking learners’ path of resetting the [+wh] to [$wh]. Thus, it would be beneficial to
examine the theoretical explanations for and empirical works on the learning hierarchy of
wh$questions with different grammatical functions to see whether they can be used to
predict the learning order of Chinese wh$questions.
With the possibility of the input frequency eliminated, the different order of
acquisition was believed to be due to the syntactic structures of grammatical functions.
Stromswold (1995) proposed, based on generative and transformational (GB) grammar,
that the different manners by which the trace of object and the trace of adjunct are
governed (Aoun, 1985; Chomsky, 1986, Kayne, 1984; Lasnik and Saito, 1984, 1992;
Rizzi, 1990; Manzini, 1992) can lead to the different order of acquisition of wh$
questions. The details of these accounts vary, but let us look at Lasnik and Saito’s theory
38
(1984) only. There are two kinds of government of the traces: θ$government and
antecedent government. θ$government means that the trace is θ$marked, i.e., it is assigned
a θ role and is M$commanded. Antecedent government means that the moved element C$
commands the trace, and the moved element and trace are also co$indexed. In object wh$
questions, the verb both assigns the object a θ role and M$commands it, and thus the trace
of the object is properly governed by the verb. In subject wh$questions, the moved wh$
word C$commands the trace, and thus the trace of the subject is regarded as properly
antecedent$governed. Adjunct trace is also not θ$governed but antecedent$governed.
Stromswold believed that θ$government is more local and direct than antecedent
government. Consequently, the acquisition order should be object > subject, adjunct:
(26) a. [CP Whati [IP are they going to [VP do ti in the evening]]]?
b. [CP Whoi [IP ti will play the basketball this evening]]?
c. [CP Wheni [IP are they going to [VP play basketball ti ]]]?
The second hypothesis is O’Grady’s Computational Complexity Hypothesis
(1997). According to this hypothesis, the computational complexity rating for a moved
object is higher than that for a moved subject, since the subject crosses one IP boundary
before it lands in the Spec, CP position, while the object crosses two maximal
projections—VP and IP, as illustrated by sentences (28). Consequently, object wh$
questions are acquired earlier than subject wh$questions, according to the Developmental
Law in (27).
(27)
Computational Complexity Hypothesis:
X is cumulatively more complex than Y if X involves everything that Y does plus
something else.
(O’Grady, 1997, p. 349)
(28)
Developmental Law (weak version):
If X is cumulatively more complex than Y, X cannot emerge before Y (all other
things being equal).
39
(O’Grady, 1997, p. 353)
(29)
CP
Spec
C’
IP
VP
VP
V
NP
Adj
When do they have a
good
N
lunch?
O’Grady (1997) did not discuss the case of adverbial wh$words. Along the same
line of reasoning, the moved adverbial wh$words should be between subject wh$words
and object wh$words in terms of computational complexity, i.e., subject > adverb >
object. The logic is as follows: Though the linear distance of the moved adverbial wh$
word is the farthest away from the Spec of CP, as illustrated in example (25) c, it is
embedded in a deeper layer than the moved subject but in a higher layer than the moved
object, as illustrated in (28). Thus, according to the Developmental Law, the acquisitional
order should be subject > adverb > object.
What about the attributive wh$questions? In neither of the two proposals is the
status of attributive wh$questions talked about. Since the wh$words that function as the
attributive of the nouns move together with the nouns that they modify as a whole noun
phrase (NP) to the Spec, CP, their difficulty rating should be slightly higher than the
subject or object if it is not the same.
40
Thus, according to Stromswold’s theory, the difficulty hierarchy should be object
> attributive object > subject > attributive subject, adverbial. According to O’Grady’s
theory, the difficulty hierarchy should be subject > attributive subject > adverbial >
object > attributive object.
The previous studies on the L1 or L2 acquisition of wh$questions of different
languages did not reveal uniform patterns. As for the subject and object asymmetry, the
results of some studies are more compatible with Stromswold’s proposal (Stromswold,
1988, 1995; Chueng & Lee, 1993), whereas the results of some studies support
O’Grady’s proposal (Kim, 1999; Seidle & Hollich, 2002; Yoshinaga, 1996; Fahn, 2003).
What is more complicated is that, even in the same study, the comprehension task and
production task yield different results (Hanna & Wilhelm, 1992; Kim, 1995).
As for the argument and adjunct asymmetry, the studies seem to uniformly show
that the argument wh$questions are easier than the adjunct wh$questions. This is where
O’Grady’s hypothesis fails. However, in terms of the subject and object asymmetry,
Lee’s study (2004), which looked at the English$speaking children’s and adult Korean
learners’ acquisition of English wh$questions, indicates that, in general argument, wh$
questions are easier than adjunct wh$questions for both groups. In Santelmann’s study
(2003), which examined Swedish children’s acquisition of wh$questions, a similar pattern
was revealed.
For attributive wh$questions, the results are complicated as well. For both the L1
Chinese children subjects in Fahn’s study (2003) and the L1 English children’s group in
Yoshinaga’s study (1996), the attributive object wh$questions were shown to be more
difficult than object wh$questions. However, in Yoshinaga’s study, there was no such
difference for the L1 Japanese children’s group.
Most studies reviewed tested either L1 learners or L2 learners of movement
languages. Only Yuan’s (2006) study indicated that adverbial wh$questions were more
difficult than object wh$questions for L2 English$speaking learners of Chinese. Would
41
wh$questions of different grammatical functions pose different degrees of difficulty for
L2 learners of in$situ wh$questions such as Chinese? Neither Stromswold’s nor
O’Grady’s proposals made reference to L2 acquisition of in$situ wh$questions. We
wondered whether their proposals could be extended to this area. In other words, can
Stromswold’s hypothesis be extended to the acquisition of Chinese wh$questions so that
the movement from the Spec, CP position to an argument$governed position is easier than
that to an antecedent$governed position trace? Or can O’Grady’s hypothesis be extended
to the acquisition of Chinese wh$questions so that movement back to the most embedded
layer is the most difficult?
2.4.4. Complex Wh$Questions
There are no studies on L2 Chinese acquisition of the [$wh] feature strength in
complex wh$questions, i.e., wh$questions of
the subcategorizations of
and
and wh$questions of
. If
are assumed to be the same as their
English counterparts by English$speaking learners, a logical question is what is the initial
stage of the matrix [$wh] feature strength of wh$questions of
wh] feature strength of wh$questions of
, and the embedded [$
? Since learners have had some
knowledge of the [$wh] feature strength in simple wh$questions by the time they deal
with complex wh$questions, would they follow the L1 English setting or the L2 Chinese
setting? What is the relationship between their knowledge of the simple wh$questions and
their competence in the [$wh] feature strength in these two types of complex wh$
questions?
Also, would learners acquire the matrix [$wh] feature strength of wh$questions of
earlier than the embedded [$wh] of wh$questions of
as a result of
transferring the effect of the acquisition of [$wh] feature strength in simple wh$questions?
This question is inspired by the fact that the syntactic behaviors of simple sentences and
matrix clauses of complex sentences are normally the same. For example, the English
42
yes$no question formation rule for simple questions applies to the matrix clauses of
complex sentences, as shown by (30) a and b, but does not apply for embedded clauses,
as shown by the contrast between (30) a, b, and c. Another example is that, in German,
the verb second word order applies in simple sentences and in matrix clauses of complex
sentences, but not in the embedded clauses, as shown by (31).
(30) a. Does he know it?
b. Does he know that she would not come?
c. He knows whether she would come.
(31) a. Er
nur wenige Artikel gelesen.
He has only few articles read.
He has read only a few articles.
b. Seine Freunde glauben, er
nur wenige Artikel gelesen.
His friends believe he has only few articles read.
His friends believe that he has read only a few articles.
2.4.5. Summary of Research Topics
Kim’s study and Yuan’s study indicate that there is no L1 transfer in learning the
[$wh] feature strength in simple wh$questions by L2 English learners of Chinese. Their
studies support the MIS hypothesis. Since their studies might not have revealed a true
representation of the initial stage of the [$wh] feature strength in L2 learners’ grammar
due to the flaws with the design, the initial stage of and also the developmental path of
the [$wh] feature strength in simple wh$questions in L2 learners’ grammar will be
explored in this dissertation. The author would also like to see whether grammatical
functions will be an influencing factor for L2 learners acquiring the [$wh] feature
strength, and if so, what is the difficulty hierarchy for the learners.
The initial stage of the [$wh] feature strength in the two types of complex wh$
questions will be examined. In addition, such questions as the acquisitional order of the
43
grammatical feature that clause$initial positions are not allowed for wh$words in simple
wh$questions, main and embedded clauses of wh$questions of
clauses of wh$questions of
and embedded
, and the relationship between their acquisitions
will be studied.
2.5. Research Questions
A pilot study done by the author indicated that 75% of learners of Chinese were
shown to have acquired the [$wh] feature strength of simple wh$questions by the end of
the first year; therefore, in this study first$year learners of Chinese were used to test the
developmental path of the acquisition of the [$wh] feature strength. Due to the scarcity of
input of the two types of complex wh$questions for first$year learners, theoretically
speaking, the initial stage of the [$wh] feature strength in the complex wh$questions can
be any time within learners’ first$year learning. The details will be talked about in
Chapter 3. However, since most of the learners are supposed to handle complex wh$
questions better by the end of first year, the author chose the end of the first year as a
time point to test learners’ initial stage of the [$wh] feature strength in complex wh$
questions.
Learners’ acquisition of subject wh$questions will not be considered in this
dissertation, since Chinese and English subject wh$questions are in exactly same sentence
order, and there is no way for us to tell how they are represented in learners’ L2 grammar
in Spec, IP as in Chinese or in Spec, CP as in English. In other words, if learners produce
a correct subject wh$question, there is no way to know whether they are copying their
English way of saying it, or if there is real change taking place in learners’ L2 grammar.
The detailed research questions pertaining to the two major research topics are as
follows.
44
The first set of research questions (hereafter Major Research Question 1) pertains
to the L2 acquisition of simple Chinese wh$questions by English$speaking learners of
Chinese.
1) Is the grammatical function a factor affecting learners’ acquisition of wh$
questions? If so, would the difficulty hierarchy be object > = object attributive > adverb,
with the easiest on the left as predicated by Stromswold’s proposal, or adverbial > object
> = object attributive as predicated by O’Grady’s proposal?
2) Does learners’ interlanguage initially manifest the [+wh] feature strength due
to the influence from their L1 English and then accommodate to the [$wh] feature in L2
Chinese, as predicated by the Full Transfer Full Acquisition Theory, or does it begin with
the [$wh] feature strength as predicated by the Minimalist Initial Syntax hypothesis?
3) Do first$year learners acquire the [$wh] feature in simple in$situ wh$questions?
If so, when?
The second set of research questions (hereafter Major Research Question 2)
pertains to the acquisition of wh$questions of
and wh$questions of
at
the end of first$year learning.
1) What is the initial stage of indirect wh$questions and wh$questions in object
complements? What are the learning results of these questions in comparison to simple
wh$questions?
2) Are indirect wh$questions or wh$questions in object complements successfully
acquired at the end of first$year Chinese learning?
3) At the end of first$year Chinese learning, what are the learning results of
learners’ acquisition of the grammatical feature that wh$words are not allowed in clause$
initial position in different clauses? Does their acquisition of this grammatical feature
vary between different clauses?
4) What is the relationship between learners’ acquisition of this grammatical
feature in different clauses?
45
CHAPTER 3 RESEARCH DESIGN
This chapter introduces the two studies that were used to investigate the two
major research questions laid out in Chapter 2. In study one, which is to answer major
research question 1, participants’ development in [$wh] feature strength of simple wh$
questions was tracked in their first$year learning period. In study two, in which major
research question 2 was explored, the participants’ acquisition of the [$wh] feature
strength of wh$questions of
and wh$questions of
is examined. In both
studies, participants’ oral production (OP) and grammaticality judgment (GJ) data was
collected.
3.1. Tapping Linguistic Competence: Elicited Oral
Production Task and Grammaticality Judgment Task
The concepts of linguistic competence as contrasted with linguistic performance
were originally proposed by Chomsky (1965) to define what the generative grammar
seeks to describe, and they apply in studies of L1 and L2 learning as well. Linguistic
competence refers to the internalized knowledge of a language in an ideal situation
unaffected by memory limitations, distractions, shifts of attention and interest, and so
forth. In contrast, linguistic performance refers to the learners’ actual use of the language,
which is decided by speakers’ competence and is subject to the influence of external
factors.
Most L1 and L2 research studies will inevitably deal with attempting to reveal
learners’ underlying linguistic competence. Since there is no way of tapping it directly,
researchers have to resort to various kinds of performance tasks to indirectly determine
the characteristics of underlying mental representation. According to White (2003), the
methodologies developed so far can fall into three categories: comprehension tasks,
including act out and picture$identification tasks, production tasks, including spontaneous
46
and elicited production tasks, and decision$making tasks, including grammaticality
judgment tasks, truth$value judgment tasks, and more recently, online matching tasks.
Since each task has its constraints, to ensure the accurate reflection of the
learners’ competence in a language phenomenon a common practice is to use more than
one task to triangulate data. Data triangulation refers to the multiple, independent use of
methods of obtaining data in a single observation (Macky & Gass, 2005). As Johnson
(1992) noted, “the value of triangulation is that is reduces observer or interviewer bias
and enhances validity and reliability of the information.”
Specifically, GJ data has been questioned as an indicator of the true representation
of L2 learners’ grammar compared to native speakers (Ellis, 1991; Goss, Zhang, &
Lantolf, 1994; Davies & Kaplan, 1998). Davies and Kaplan (1998) reported that L2
English$speaking learners of French used three more strategies—learned, translation, and
analogy—in addition to those used by the native speakers. Furthermore, not only did the
L2 learners use more strategies per item than the native speakers, they used the feel
strategy (intuitive judgment) significantly less than the native speakers.
The OP task alone cannot always provide an accurate representation of L2
learners’ grammar, either. Failure to produce a grammatical item can be due to learners’
general low production ability or problems with areas other than the target linguistic
phenomenon. Furthermore, failure to produce a grammatical item does not necessarily
indicate that there is no underlying representation. For example, White (2002) found that
a Turkish learner who failed to supply determiners in spontaneous production was highly
accurate in other tasks.
For this dissertation, GJ task and OP task were both adopted for the following two
reasons. One is that both tasks were traditionally used in the studies examining the
acquisition of wh$questions. The other is that data from both tasks could provide more
perspective on L2 learners’ acquisition of wh$questions.
47
3.2. Research Design of the Two Studies
3.2.1. Study 1: A Longitudinal Study
Study one was a longitudinal study. In this study, this author collected data about
the development of L2 acquisition of the [$wh] feature strength of simple wh$questions
over a one academic year period.
Data of participants’ initial stage of [$wh] feature strength started the second
week after they were taught the explicit grammar rules of Chinese wh$questions.10 By
definition, the initial state refers to the kind of unconscious linguistic knowledge that the
L2 learner starts out with in advance of the L2 input and/or characteristics of the earliest
grammar (White, 2003). Data was not collected before the Chinese wh$questions
formation rule was taught, because this grammar rule was introduced the same week as
the first set of wh$words, and it was hard to test learners’ competence in this grammatical
phenomenon when they had zero knowledge of Chinese wh$words.
Data were collected four times at half$semester intervals, i.e., in the 6th week and
14th week in the first semester, and in the 7th week and 14th week in the second semester
over their first year Chinese class. The rationale behind this was that data should be
collected often enough to capture the change in learners’ interlanguage, but not so often
that it would cause the linguistic item under investigation to be noticed by the participants
or lead to a practice. Collecting data every half$semester was, thus, an appropriate time
interval, both for the sake of the experiment and in order not to disturb students’ regular
study.
The tests given to the participants in each interval were different, in order to
minimize a possible practice effect, but were also kept as similar as possible in terms of
10
It was when learners were six weeks into their Chinese class and four weeks into the
instruction of grammar. During the first two weeks, learners were introduced to the Pinyin
system, strokes orders, and basic radicals of Chinese.
48
the difficulty level to ensure that participants’ performance in the tests would not change
as a result of the difficulty level of the tests, but would reflect their real competence in the
[$wh] feature strength as outlined below. All the test items were designed to have the
same syntactic complexity: they all contained only one clause, whether they were
sentences to be judged in GJ or to be produced in OP. Although the length of the test
items in each test was not strictly controlled, the participants were given as much time as
they needed to finish each test. During the test, participants were asked to respond to each
item as quickly as they could and not to compare the answer with the previous ones, to
ensure that their conscious use of the grammar was used as little as possible.
To test participants’ competence in wh$words with different grammatical
functions—object, attributive, and adverbial—wh$words of all three parts of speech were
selected: nouns, adjectives, demonstrative pronouns, and adverbs. However, the number
of the three types of simple wh$questions was not the same and neither was it the same
from test to test. For example, in Test 1, there were four object wh$questions, four
adjectival wh$questions, and two adverbial wh$questions, while in Test 2, there were two
object wh$questions, five adjectival wh$questions, and three adjectival wh$questions, as
shown in Table 4. The reason was that each test included all wh$words that first appeared
in the textbook for the learners, and then the old wh$words. This could create a problem
of comparability of the three types of wh$questions, which will be addressed in Chapter
5.
3.2.2. Study 2: A One$Shot Study
In order to test the initial state of learners’ acquisition of the embedded [$wh]
feature strength of wh$questions of
wh$questions of
and the matrix [$wh] feature strength of
, when to collect data representing the initial state of the
development of these feature strengths needs to be decided.
49
Theoretically speaking, due to the scarce input provided in the first$year learning,
it can be anytime according to the definition of initial stage given by White (2003). The
linguistic phenomenon that wh$words stay in$situ in indirect questions and wh$questions
in object complements was not introduced as a grammar rule anywhere in the textbook at
any level. Neither was it taught as a grammar rule by the instructors. The input from the
textbook, from which learners might be able to generalize such a rule, is very scarce and
not very clear as to the position of wh$words in the embedded clauses. Sentence (32) B is
the only sentence where the learners might get a hint of what wh$questions of
is like. It is introduced by
instead of
the embedded clause is omitted, the wh$word,
, and since the subject of
&
' is at sentence initial
position, which might mislead learners to believe that Chinese wh$questions of
have wh$words located at the initial position of the embedded clauses just as in
English. This occurs in lesson 6, which was taught in the eighth week of the class. As for
wh$questions with
, there is not a single sentence in the textbook upon which
learners could model their attempts.
(32)
A: ta shenme shihou huilai?
He went come back.
When will he come back?
B: Bu zhidao shenme shihou huilai.
Not know when come back.
I do not know when he will come back.
The input provided in the first$year classroom setting was very limited as well.
The first$year instructors I consulted with said that they rarely used the two types of
complex wh$questions in the classroom, and students did not often have the chance to use
the two types of sentences due to the purpose of first$year Chinese language training.
First$year Chinese students were expected to finish basic real$life tasks, but were not yet
asked to express or ask each other’s opinions upon certain things. Thus, wh$questions of
50
and wh$questions of
were rarely used. In addition, first$year Chinese
students were asked mainly to deal with simple sentences but not complex sentences.
Thus instead of using “I am wondering when/what…” and “when/what…do you
feel/think…”, teachers and students normally both used a more straightforward pattern
“when/what …?”
To avoid an extra burden on the students, the data regarding these feature
strengths was collected along with the data on the [$wh] feature strength of simple wh$
questions in one set of tests, i.e., in the fourth OP and GJ test.
Since this study was intended to compare learners’ acquisition of the two types of
wh$questions, the difficulty level of the test items for the two types of wh$questions was
supposed to be identical. There were no new words for either of them in the OP test, and
in the GJ test there were three new words in the test items for wh$questions of
, and they were all provided with the English meaning in parenthesis. The length of
the test items in the GJ test and of the sentences they were supposed to produce in the OP
test were not strictly controlled, given that learners were allowed sufficient time to finish
all test items. It turned out, however, that the average length of test item for the two types
of wh$questions in the GJ test was almost the same. It was 11.8 characters per sentence
for wh$questions of
and 12.4 per sentence for wh$questions of
.
51
Table 4. Wh$Words Tested in the Four OP and GJ Tests
GF
O
A
OP Test 1
OP Test 2
OP Test 3
OP Test 4
GJ Test 1
GJ Test 2
GJ Test 3
GJ Test 4
Simple
Simple
Simple
Simple
Wh$Q of
Wh$Q
Wh$Q
Wh$Q
Wh$Q
Wh$Q
Xiang
Of Juede
shenmea
shenme
shenme a
shenme
shenmea
(what)
(what)
(what)
(what)
(what)
shui a
shui
shui
(who)
(who)
(who)
shenme a
shenme
na
(what)
(what)
(which)
(which)
ji a
ji
ji
ji
ji
(how
(how many)
(how
(how
(how
many)
na
many)
many)
many)
(which)
duoshao
duoshao
(how
(how much)
many)
duoda
(how big)
na
na
(which)
52
Table 4. Continued
nar a
Adv (where)
nar
nar
(where)
genshui
(where)
genshui
(with whom) (with
nar
nar
(where)
(where)
duojiu
genshui
(how
(with
long)
whom)
zenmeyang
whom)
(how)
zenmeyang
zenme
(how)
(how)
zenme
(how)
duojiu
(how long)
duojiu
(how long)
Note: GF = Grammatical functions, O = Object, A = Attributive, Adv = Adverbial
a
Number of test items is 2. Number of test items for the rest is 1.
3.3. Test Materials
3.3.1. Elicited Oral Production Test
Altogether there were four elicited OP tests— the first three for investigating
Major Research Question 1, and the fourth for investigating both Major Research
Questions 1 and 2. In each OP test, the participants were asked to perform two role$plays
with the testers. Participants played the role of interviewer to obtain the specified
information by asking wh$questions of the testers, who played the role of interviewees.
The reason for adopting a role$play on a real$life topic instead of using translation was to
direct learners’ attention to the task, so that they would less consciously use the explicit
grammar rules they had learned concerning wh$questions.
53
For each of the first three OP tests, learners were instructed to ask five simple wh$
questions in each role$play, and thus a total of 10 simple wh$questions in each test; in the
fourth OP test, there were 15 test items altogether, five for each of the three kinds of wh$
questions under investigation. These 15 test items were spread out in two role$plays as
well. The reason for choosing to use five instead of 10 test items for each kind of tested
wh$question was that participants would otherwise end up producing 30 sentences in the
OP test and judging at least 60 sentences in the GJ test. This overloaded task could affect
their performance in the tests in a negative way.
During the data collection, participants were first presented with the topic related
to all the questions they were going to ask. For example, in (33) below, learners would
know that they were to ask some questions about learning Chinese in China or life in
China. In addition, they were asked to produce a complete question to obtain each piece
of information. The wh$words they needed to use were provided. For one thing, this
reduced their burden of searching for the right vocabulary. For another, this could make
learners use only wh$questions, rather than other sentence patterns, such as tag questions,
to obtain the information.
The information they needed to obtain was provided in the form of English wh$
questions. This made it look like a set of Chinese$to$English translation exercises. The
reason that a less obvious way was not used to prompt learners’ use of wh$questions was
that they would sometimes use patterns other than wh$words to obtain the same piece of
information, resulting in the problem of lack of comparison between learners, and more
seriously, preventing us from seeing whether they knew how to use wh$questions. For
example, in the pilot study, facing the prompt for obtaining the information about “the
food you like (what),” instead of asking “You like what food?” some learners would ask
questions such as “What is the food you like?” For the prompt “the time you saw him
(when),” instead of asking “You when saw him?” learners would produce sentences such
as “You saw him was when?” The questions they asked were correct Chinese sentences;
54
however, when they simply linked the wh$words and the information to be obtained with
“to be” so that the wh$words were always in the subject or object position, it was not
certain that learners knew how to treat adjectival and adverbial wh$words. Moreover, this
also created the problem of how to compare their data with others. Presenting the prompt
in English wh$questions could evoke their conscious use of the grammar rules concerning
Chinese wh$questions, thus diluting our initial purpose of using a real$life task. However,
compared with the disadvantages of using the less obvious forms, this was judged to be a
better choice.
(33)
You are interviewing Thomas for a special TV program. Thomas is a student who
has just returned from an intensive study$abroad Chinese program in China. You need
to find out the following information. You can add more questions if you want.
. Please also
make sure to use the words in the brackets.
1) Which city did you go to? (which = 哪
)
2) How long did you stay in China? (how long = 多久
)
3) I am wondering who you went with. (wonder = 想知道
4) I am wondering where you visited. (wonder = 想知道
)
)
5) I am wondering which city you liked the best. (wonder = 想知道
)
6) How do you think we should study Chinese? (think/feel = 认为/觉得
)
7) What gift do you think we should give to the Chinese people? (think/feel =
认为/觉得
)
8) What do you think that you have learned? (think/feel = 认为/觉得
)
The other thing worthy of mention is that the wh$questions that learners were
asked to produce had wh$words playing the role of every grammatical function except the
subject for the reason discussed in Chapter 2.
55
To minimize the possibility that learners’ production of wh$questions would be
affected by their insufficient knowledge of vocabulary, the words they needed to use to
produce the wh$questions were strictly controlled. Most of them were taken from their
learned vocabulary. The words unknown to them were very limited and were provided in
both characters and Pinyin behind each piece of information to be obtained, together with
the particular wh$word that they needed to use, the Pinyin and meaning of which were
also provided. There were only three unknown words, two in Test 2 and one in Test 1, for
the participants in the four tests.
3.3.2. Grammaticality Judgment Task
Altogether there were four GJ tests. As was the case with the four OP tests, the
first three GJ tests were to gather data for Major Research Question 1, and the fourth one
was to gather data for both Major Research Questions 1 and 2. Since the knowledge of a
grammatical item (functional category, feature, feature strength) includes both the
knowledge of what it allows and the knowledge of what it disallows in a language, both
grammatical and ungrammatical wh$questions were included in the GJ tests. Fillers were
introduced in the tests to make learners less aware of the tested grammar item.
The first three GJ tests each consisted of 20 test items, with half being
grammatical wh$questions and half being ungrammatical wh$questions, and 20 fillers,
with half being grammatical and half being ungrammatical as well. In the fourth GJ test,
there were 30 test items: 10 for simple wh$questions, 10 wh$questions of
and 15 for wh$questions of
. For both simple wh$questions and wh$questions of
, grammatical and ungrammatical items took up half and half each. For wh$
questions of
, each of the 5 grammatical sentences corresponded to 2 ungrammatical
sentences, one with the wh$word at sentence initial position of the main clause
(ungrammatical sentence type 1), and one with the wh$word at sentence initial position of
the embedded clause (ungrammatical sentence type 2), as shown by (38) b, and c.
56
Though ungrammatical sentence type 2 does not manifest either the L1 English setting or
the L2 Chinese setting of the [wh] feature strength, it was included to test whether
learners would make this type of error in the learning process anyway. Only 15 fillers
were included. The reason for not introducing as many fillers as in the previous tests was
that this would result in too long a test.
The test items used in the GJ tests were from their corresponding OP test. The
purpose of doing so was to make the words used in these two tests as similar as possible
in aspects other than the test format.
As mentioned in Chapter 2, Chinese wh$words can exist at the beginning of a wh$
question, if the speaker is asking this question to confirm a piece of learned information
or to show surprise at a piece of information heard orally. Although this rule was not
taught in the textbook at all, such wh$questions might have existed in the input provided
by the language instructors in the classroom when the communication between teachers
and students was getting more complicated. Thus, it made sense to convey to the subjects
very clearly that the test items were all questions to obtain information never talked about
before. Thus each tested wh$question was inserted into a mini$dialogue between two
speakers, the context of which made it clear that the wh$question was not asked to
confirm or to show surprise about a piece of information but to obtain new information.
Here are two examples.
(34)
小李:我想请你们星期五到我家玩儿。
Xiaoli: Wo xiang qing nimen xiangqiwu dao wo jia wanr.
Xiaoli: I want invite you Friday go my home play.
Xiaoli: I would like to invite you to play at my home this Friday.
小王:太好了。我们怎么到你家?
Xiaowang: Tai haole. We how go your home?
Xiaowang: Too good. We how go your home?
Xiaowang: That is very good. How shall we go to your home?
57
A: 听说你去了中国。 你去了哪个城市?
(35)
Tingshuo ni qu$le Zhongguo. Ni qu$le na ge chengshi?
Heard you go$le China. You go$le which MW city?
I heard that you went to China. Which city did you go?
B: 我去了北京和上海。
Wo qu$le Beijing he Shanghai.
I go$le Beijing and Shanghai.
I went to Beijing and Shanghai.
For each wh$question to be judged, participants have three options to choose:
, and
. The reason for including
,
as one of the options is to
ensure that subjects did not decide the grammaticality of a sentence based on guessing, as
a participant could have marked a sentence—the grammaticality of which they either did
not know or were not sure of—randomly as correct or incorrect.
Since our interest is to discover whether a wh$word is located in$situ as in
subjects’ L2 Chinese or in clause$initial position as in their L1 English in their
interlanguage, the ungrammatical sentence is identical with the grammatical sentence in
every way, except that in the ungrammatical sentence the wh$word is in clause$initial
position, as shown by (36) a and b.
(36) a. A: 听说你去了中国。 你去了哪个城市?
Tingshuo ni qu$le Zhongguo. Ni qu$le nage chengshi?
Hear you go$le China. You go$le which$MW city?
I heard that you went to China. Which city did you go to?
B: 我去了北京和上海。
Wo qu$le Beijing he Shanghai.
I go$le Beijing he Shanghai.
I went to Beijing and Shanghai.
b. A: 听说你去了中国。 *哪个城市你去了?
58
Tingshuo ni qu$le Zhongguo. *Nage chengshi ni qu$le ?
Hear you go$le China. Which$MW city you go$le?
I heard that you went to China. Which city did you go to?
B: 我去了北京和上海。
Wo qu$le Beijing he Shanghai.
I go$le Beijing he Shanghai.
I went to Beijing and Shanghai.
To test the acquisition of the embedded [$wh] feature strength of wh$questions of
, as shown by (37) a and b, the wh$words were shifted to the beginning of
the embedded clause in the ungrammatical sentences.
(37)
a. A: 听说你去了中国。
我想知道你还参观(visit)了什么地方。
Tingshuo ni qu$le Zhongguo. Wo xiangzhidao ni hai canguan$le shenme
difang.
Hear you go$le China. I wonder you also visit$le what place.
I heard that you went to China. I wondering what other place you visited.
B: 我还参观了南京。
Wo hai canguan$le Nanjing.
I also visit$le Nanjing.
I also visited Nanjing.
b. A: 听说你去了中国。 *我想知道什么地方你还参观(visit)了。
Tingshuo ni qu$le Zhongguo. *Wo xiangzhidao shenme difang ni hai
canguan$le.
Hear you go$le China. I wonder what place you also visit$le.
I heard that you went to China. I wondering what other place you visited.
B: 我还参观了南京。
Wo hai canguan$le Nanjing.
I also visit$le Nanjing.
59
I also visited Nanjing.
As shown by (38) a, b and c, every grammatical sentence for testing the matrix [$
wh] feature strength of wh$questions of
corresponds to two ungrammatical
sentences. In one ungrammatical sentence, the wh$word is located at the sentence initial
position of the main clause as shown by (38) b, and in the other, the wh$word is located at
the sentence initial position of the embedded clause as shown by (38) c.
(38) a. A:你在中国旅行了一个月。 你觉得你学到了什么呢?
Ni zai Zhongguo lvxing$le yige yue. Ni juede xuedao$le shenme ne?
You in China travel$le one (MW) month. You feel learn$le what particle?
You traveled in China for a month. What did you feel you learned?
B: 我觉得我学到了很多中国的文化。
Wo juede wo xuedao le henduo Zhongguo de wenhua.
I feel I study$le a lot China the culture.
I feel that I learned a lot about Chinese culture.
b. A:你在中国旅行了一个月。 *什么你觉得你学到了呢?
Ni zai Zhongguo lvxing$le yige yue. Shenme ni juede xuedao$le ne?
You in China travel$le one (MW) month. What you feel learn$le particle?
You traveled in China for a month. What did you feel you learned?
B: 我觉得我学到了很多中国的文化。
Wo juede wo xuedao le henduo Zhongguo de wenhua.
I feel I study$le a lot China the culture.
I feel that I learned a lot about Chinese culture.
c. A:你在中国旅行了一个月。*你觉得什么你学到了呢?
Ni zai Zhongguo lvxing$le yige yue. *Ni juede shenme ni xuedao$le ne?
You in China travel$le one (MW) month. You feel what
ni learn$le particle?
You traveled in China for a month. What did you feel you learned?
B: 我觉得我学到了很多中国的文化。
60
Wo juede wo xuedao le henduo Zhongguo de wenhua.
I feel I study$le a lot China the culture.
I feel that I learned a lot about Chinese culture.
3.4. Procedures
As explained previously, the OP and GJ tests were administered four times over
one academic year, i.e., in week 6 and week 14 of the first semester, and week 7 and
week 14 of the second semester. The OP and GJ test items in study two were respectively
mixed together with OP and GJ items on simple wh$questions in Test 4 of study one, and
administered to the same group of learners at the end of their first$year class.
In the OP test, participants participated in two role$plays with this author and two
other first$year Chinese language instructors who helped administered the tests. We
played the role of interviewees and also recorded subjects’ speech. The GJ tests were in a
paper$and$pencil format, and were administered three days after the OP test was taken. In
theory, if the test items are the same, no matter which test is administered first, there will
be some carry$over effect into the second one administered. It seemed better to have the
participants construct the wh$questions from scratch in the OP test and then decide in the
GJ test which wh$question, probably including those they produced, is correct, instead of
the other way around. The reason is that if the GJ test is taken first, it might give the
students a clue as to how to construct a wh$question. The three$day interval between the
two tests was intended to reduce the possible carry$over effect to a minimum.
For clarification, Table 5 summarizes all of the tests that the subjects took, in the
order of sequence of the tests being administered. It also indicates what linguistic items
these tests were used to investigate.
61
Table 5. Overview of Testing Time, Format, and Linguistic Item
Testing Time
First Semester
Second Semester
Task type
Linguistic item
6th Week
14th week
7th week
14th week
OP Test 1
OP Test 2
OP Test 3
OP Test 4
(n=10)
(n=10)
(n=10)
(n=5+5+5)
GJ Test 1
GJ Test 2
GJ Test 3
GJ Test 4
(n = 20)
(n = 20)
(n = 20)
(n = 10+10+15)
[$wh] of
[$wh] of
[$wh] of
1)[$wh] of Simple Wh$
Simple Wh$
Simple Wh$
Simple Wh$Q
Q
Q
Q
2) …[$wh] of Wh$Q of
Xiang zhidao
3) [$wh] …[$wh] of
Wh$Q of Juede
Note: n = number of test items
Not only in designing the test items, but also in administering the tests,
procedures were taken to minimize the possibility that other confounding factors could
affect the subjects’ performance. In the OP test, the participants could be told the words
that they did not know when they asked for help, but would not be provided with any
information concerning grammar. In addition, the participants did not have a time limit in
which to finish the task, so that there did not exist a single case where participants failed
to produce a sentence because they did not have time. In the GJ test, participants were
62
asked to finish the test as fast as possible, and they were also not allowed to compare the
answers they gave with the previous ones. This was to ensure that learners chose their
answer according to their first intuition instead of retrieving explicit grammar knowledge
or considering the pattern of the test items before they make a decision. However, the
participants were all given time to finish all of the test items.
3.5. Participants
The participants who were involved in this study were first$year L1 American
English learners of Chinese with no background in Chinese prior to entering the first$year
Chinese class at the University of Iowa. Students who spoke some Chinese at home with
their parents or who had learned some Chinese before were excluded from the study. In
order to eliminate the L1 transfer effect in participants’ learning process because of its
resemblance to Chinese in terms of wh$questions, learners whose first language belonged
to in$situ languages, such as Japanese or Korean, or whose first language allowed in$situ
wh$questions, such as French, were excluded from participation in the study.
Since it was a longitudinal study, retaining participants in the study over time was
a challenge. Some of the participants who agreed to participate dropped Chinese in the
second semester, and some of them missed one of the tests, especially the last one. Thus
the study ended up with complete data for 21 participants.
3.6. Data Treatment
3.6.1. Coding and Sorting out the OP Data
For the oral production data, the correctly produced sentences were coded as 1,
and the incorrectly produced sentences were coded as 0. A coding system was developed
upon close examination of the incorrectly produced sentences, and used to code all the
problematic sentences produced by the participants. The sentences that the participants
incorrectly produced fell into the following categories: Wh$word in initial position, Zero
63
Production and Failure to produce a comprehensible sentence, Wh$word in second
position, Misplaced adjectival wh$words, Misplaced adverbial wh$words, and Wh$words
after objects.
Since the research interest was in the position of wh$words, as long as the
participants placed the wh$word in the correct position in the produced sentence, that
sentence was regarded as a correct sentence. Thus sentences with such mistakes as the
wrong choice of words, and omission of
(in, at) before
11
(where), or no duplication
of verbs when followed by complement of degree, were coded as correctly produced
sentences.
Two native speakers of Chinese with a Chinese language teaching background
were asked to recode learners’ oral production data using the coding system mentioned
above. One of them coded the first two OP tests, and the other did the last two OP tests.
Then the results of coding were compared: interrater agreement in the first two tests was
95.83%, and for last two tests it reached 96.67%. The sentences over which there was
disagreement in coding were picked out and relistened. The final decision concerning
each difference was made by discussion.
3.6.2. GJ Data
Recall that participants have three options to choose from:
,
, and
, for both the grammatical and ungrammatical sentences in the GJ tests. Participants’
responses in all four GJ tests were entered into an Excel spread sheet. Their answers to
fillers were identified and were not involved in the data analysis.
11
Unlike English “where,” its Chinese counterpart
needs to be preceded by a
preposition zai (=in, at) unless the verb is intransitive. For example:
(i) Ni zai nar chifan?
You at where eat?
Where do you eat?
(ii) Ni yao qu nar?
You want go where?
Where are you going?
64
First, the test items for simple wh$questions in the four tests were sorted out.
Participants’
,
, and
answers for their grammatical sentences were
respectively categorized into groups called
$,
. For ungrammatical sentences, participants’
categorized into groups called
,
, and
, and
$,
answers were
, and
.
These groups were then further categorized according to the grammatical functions that
these wh$words played.
Similar procedures as above were repeated for learners’ responses to the
grammatical and ungrammatical sentences of the two types of complex wh$questions that
appeared in the fourth test, with the only difference being that learners’ responses were
not further grouped according to their grammatical functions.
3.6.3. Method of Statistical Analysis
To address Major Research Question 1, in dealing with the OP data, the wh$
questions that the participants produced were analyzed in terms of [wh] feature strength,
and decisions were made about what were definitely surface expressions of the [+wh]
feature strength, and what actually manifested the [$wh] feature strength. For simple wh$
questions with wh$words/phrases playing the role of different grammatical functions, the
accuracy rates for the occurrence of the [$wh] feature strength for all three were
calculated. Since the test items for these three types of questions were not in proportion in
these four tests, and the difficulty that they posed to the participants regarding the [$wh]
feature strength could be different, the average of these accuracy rates was calculated and
taken as the accuracy rate of participants’ performance on the [$wh] feature strength of
simple wh$questions. In dealing with the GJ data, the participants’ accuracy rates for
simple wh$questions with wh$phrases playing different grammatical functions in both the
grammatical and ungrammatical sentences was calculated. The same procedure was
repeated in calculating the accuracy rate for the [$wh] feature strength of simple wh$
65
questions.
To find out whether participants had a significant gain in the [$wh] feature
strength of simple wh$questions over time, an analysis of variance (ANOVA) of
repeated$measure design was run on participants’ accuracy rates for the [$wh] feature
strength of the four OP tests and also the four GJ tests. To discover how learners
developed in their use of wh$words in wh$questions of different grammatical functions
regarding this feature strength, an ANOVA of two factor repeated$measure design was
run on participants’ accuracy rates for simple wh$questions with wh$words playing all
three different grammatical functions. An 80% accuracy rate was used as the criterion for
success of acquisition. Both the group results and the individual results are reported.
For Major Research Question 2, to find out how learners performed on the
embedded [$wh] feature strength of wh$questions of
feature strength of wh$questions of
and the matrix [$wh]
e in comparison with the [$wh] feature strength of
simple wh$questions at this point, an ANOVA was conducted on participants’ accuracy
rates for all these feature strengths.
To see whether there was any correlation between learners’ acquisition of these
feature strengths and their acquisition of the [$wh] feature strength of simple wh$
questions, a correlational analysis was performed on participants’ result of learning.
66
CHAPTER 4 ANALYSIS AND RESULTS
This chapter analyzes the data collected and reports the results of participants’
acquisition of simple wh$questions over their first year of Chinese learning, as well as
their acquisition of the two types of complex wh$questions taking wh$questions as
embedded object clauses, as revealed by the participants’ oral production (OP) data and
grammaticality judgment (GJ) data. In addition, to give a full picture of whether
participants have acquired the different types of wh$questions, both group results and
individual results will be analyzed.
4.1. Research Question 1: The Acquisition of Simple Wh$
questions
Based on the analysis of the data of the four oral production tests and that of the
four grammaticality judgment tests, we will address the following three sets of research
questions concerning participants’ acquisition of simple wh$questions:
1) Is the grammatical function a factor affecting learners’ acquisition of wh$
questions? If so, would the difficulty hierarchy be object > = object attributive > adverb
with the easiest on the left as predicated by Stromswold’s proposal, or adverbial > object
> = object attributive as predicated by O’Grady’s proposal?
2) Does learners’ interlanguage initially manifest the [+wh] feature due to the
influence from their L1 English and then accommodate to the [$wh] feature in L2 Chinese
as predicated by the Full Transfer Full Acquisition Theory, or does it begin with the [$
wh] feature strength as predicated by the Minimalist Initial Syntax hypothesis?
3) Do first$year learners acquire the [$wh] feature in simple in$situ wh$questions?
If so, when?
67
4.1.1. Initial Analysis of Data
4.1.1.1. Initial Analysis of the OP Data
In this section, the types of errors12 participants made in terms of the location of
wh$words in the OP data and the percentage that each type of error comprised among the
total number of test items are reported. Then these types of errors will be discussed in
terms of the [wh] feature strength.
Based on the initial observation of the emergent patterns, the types of errors
participants made are of two kinds (See Tables 6, 7 and 8). One kind is found to occur
across all parts of speech, such as Zero Production or Failure to Produce a
Comprehensible Sentence (ZPro/F$To$ProCompreS) and Wh$Words in Initial Position
(Wh$1st). The former refers to the instances where participants failed to produce a
sentence or failed to produce a comprehensible sentence. The latter refers to those with
wh$words at the sentence$initial position.
The other kind is found to be peculiar to certain parts of speech. As shown in
Tables 6 and 7, for nominal and adjectival wh$words alone, participants produced wh$
questions with Wh$Words in Second Position (Wh$2nd). For example, participants
produced such sentences as *
(
) (You whom invite?), where the nominal
wh$word occurs in the second position, and *
? (You what see a
movie?), where an adjectival wh$word occurs in the second position.
For nominal wh$words, participants produced a small number of sentences with
wh$words after Object (Wh$A$O). For example, for the English sentence “What would
you like to eat?” participants produced *
12
) (You want eat meal
Corder (1967) makes a distinction between error and mistake. Error is a breach of the
language’s code, resulting in an unacceptable utterance; with L2 learners this might occur
because “the learners have not yet internalized the formation rules of the code” (1973, p. 259).
Errors are systematic and not random. Mistakes or lapses are the result of some failure of
performance” (1967, p. 18). Error in this dissertation is mostly what is defined by Corder.
68
what?) instead of
) (You want eat what?)
The type of common errors participants made for wh$questions with adjectival
wh$words is Misplaced Adjectival Wh$Words. This refers to those wh$questions with
adjectival wh$words misplaced after the modified nouns. An example sentence is
) (*You saw movie what?).
The type of common errors participants made for wh$questions with adverbial
wh$words is Misplaced Adverbial Wh$Words (MPAdvWh). This refers to incorrect wh$
questions in which participants wrongly placed adverbial wh$phrases of time, place,
manners, or co$agents after the main verb, as influenced by their L1 English, in which
adverbials of time, place, manners, and co$agents occur after the main verb. An example
is that participants produced *
of the correct
) (You saw a movie where?) instead
) (You where saw movie?).
Table 6. Rates of Types of Sentences That Participants Produced with Nominal Wh$
Words and Percentage of Occurrence of [$Wh] and [+Wh] in Four OP Tests
Feature
Types of
Test 1
Test 2
Test 3
Test 4
Strength
Sentences
[$wh]
CS
73.81%
88.10%
71.3%
80.95%
Wh$A$O
3.57%
0%
0%
0%
Total
77.38%
88.10%
71.43%
80.95%
[+wh]
Wh$1st
1.19%
2.38%
7.14%
4.76%
?
Wh$2nd
11.90%
9.52%
14.29%
14.29%
Unknown
ZProd/F$To$
9.52%
0%
7.14%
0%
ProdCompreS
69
Table 7. Rates of Types of Sentences That Participants Produced with Adjectival Wh$
Words and Percentage of Occurrence of [$Wh] and [+Wh] in Four OP Tests
Feature
Type of
Test 1
Test 2
Test 3
Test 4
strength
sentences
[$wh]
CS
47.62%
61.90%
59.82%
80.95%
MP$AdjWh
15.48%
6.67%
2.38%
2.38%
Total
63.10%
68.57%
62.20%
83.33%
[+wh]
Wh$1st
4.76%
4.76%
23.81%
11.90%
?
Wh$2nd
27.38%
20.95%
9.52%
2.38%
Unknown
ZProd/F$To$
4.76%
5.71%
4.76%
2.38%
ProdCompreS
Table 8. Rates of Types of Sentences That Participants Produced with Adverbial Wh$
Words and Percentage of Occurrence of [$Wh] and [+Wh] in Four OP Tests
Feature
Type of
Test 1
Test 2
Test 3
Test 4
strength
sentences
[$wh]
CS
61.90%
61.90%
59.52%
85.71%
MP$AdvWh
11.90%
22.22%
20.24%
0%
Total
73.80%
84.12%
79.76%
85.71%
[+wh]
Wh$1st
9.52%
11.11%
16.67%
7.14%
Unknown
ZProd/F$To$
17%
5%
3.57%
7.14%
ProdCompreS
The next step is to give an analysis of these types of errors and to categorize them
according to the setting of the [wh] feature strength. Let me start from the most apparent
70
one. Wh$Words in Initial Positions (Wh$1st) is obviously the surface expression of the
[+wh] feature strength. Zero Production or Failure to Produce Comprehensible sentences
(ZPro/F$To$ProCompreS) could be caused by several factors, and insufficient knowledge
of the setting of the [wh] feature is just one of them. However, it is hard to judge whether
this is caused by participants’ insufficient knowledge of the [wh] feature or by their
limited production ability. Therefore, the best that can be said is that these sentences
reveal no information about the setting of the [wh] feature.
Wh$Words After Object (Wh$A$O) may be due to participants’ unfamiliarity with
the structure of a special type of Chinese verb called VO compounds. English intransitive
verbs such as
,
$
(eat meal),
(meal),
(words), and
, and
(talk words),
correspond to Chinese VO compounds such as
(write characters). In these verbs,
(characters) are dummy objects in the sense that, when there
are real objects in a sentence, the dummy objects should be replaced. Thus the translation
for the English sentence “What would you like to eat?” should be
in which the real object
replaces the dummy object
),
, but not
(You want eat (meal) what?). We can see that the cause of this type of error is not
that participants do not know where to place object wh$words in a wh$questions, but that
they do not know that
is one of the VO compounds. Thus, though this is a type of
error, Wh$Words After Object can be regarded as a manifestation of [$wh] feature
strength.
Misplaced Adjectival Wh$Words can be regarded as the surface manifestation of
[$wh] feature strength as well. There are two possible explanations for why participants
placed adjectival wh$words after the noun. One is that the participants used Chinese VO
compounds as intransitive verbs, and also simplified wh$questions in which we would
have liked them to use adjectival wh$words in sentences with nominal wh$words. For
example, they first of all simplified “what dinner will we have” to “what will we eat,”
and since they did not know that the Chinese correspondent for “eat”—
(eat
71
meal)—is a VO compound, and
(meal) should be replaced by
produced the incorrect sentence
(what), they
(we eat meal what). The other
explanation is that they probably did not know the correct position of adjectives in
relation to the nouns they modify. However, given that adjectival modifiers go before the
nouns both in English and Chinese, this explanation is less plausible. No matter which is
the actual case, however, we can see that the source of this kind of error is not the wrong
setting of the [wh] feature, but either participants’ unfamiliarity with Chinese VO
compounds or the wrong setting of the parameters governing the position of the
adjectival wh$words in relation to the modified nouns.
Along the same line of reasoning, Misplaced Adverbial Wh$Words, such as
) (You saw a movie where?), show that participants knew that the wh$
adverbials should be in$situ; however, they had trouble putting them in a correct within$
sentence position. In other words, participants had problems with the position of
adverbial phrases in relation to the modified verbs in Chinese, rather than with resetting
the [wh] feature into [+wh]. Thus, the error of Misplaced Adverbial Wh$Words can be
regarded as the surface expression of [+wh] feature strength.
What about Wh$Words in Second Position, such as
what play?) and
) (We
) (We what see movie?)? First of all, why
did this kind of error not occur when wh$words were adverbs? The answer is that when
wh$words are adverbs, the second position is actually a correct position for them.
Wh$words in Second Position are unlikely to be random, as shown by the fact that
they persisted through all four tests, and the occurrence rates in the first two tests were
not minimum, 11.90% and 9.52% respectively, for object wh$questions in the first two
tests, and 27.38% and 20.95% for attributive wh$questions. The individual results show
that it was not a strategy used by only a few participants. It was used by 28.57% and
61.90% of participants for object wh$questions in the first two tests, and 61.90% of
participants for attributive wh$questions.
72
There are two possible reasons why participants produced this kind of sentences.
One is that it is a result of the strong [wh] feature and the weak [wh] both trying to take
control in participants’ interlanguage. That wh$words in sentence$initial positions in
Chinese are ungrammatical (unless the question is asked about the subject) is very salient
for the learners, and thus they could easily avoid producing sentences like that. In the
meantime, their L1 English did not, all of a sudden, stop exerting influence on them so
that they tended to put wh$words at a frontal place in a sentence. The other reason is that,
influenced by the position of adverbials in adverbial wh$questions, which are in the
second position in Chinese, they might have thought this was another option for the
[+wh] feature strength of all grammatical functions.
If the above analysis regarding the nature of the error types is accepted, then we
can easily calculate the percentage of [+wh] feature strength and [$wh] feature strength in
the four OP tests for each part of speech, as shown by the first column of Tables 6, 7, and
8.
4.1.1.2. Initial Analysis of the GJ Data
The GJ data was sorted out in the following way. As explained in Chapter 3, for
grammatical sentences, we categorized participants’
into groups entitled
$,
ungrammatical sentences, participants’
groups called
$,
,
, and
,
, and
, and
, and
answers
. For
answers were put into
. For both the
grammatical and ungrammatical sentences, these groups were further divided according
to the parts of speech.
In the GJ task, participants can reject a sentence either correctly or incorrectly
because of linguistic factors other than the target grammatical form. If such factors are
assumed to be well controlled, then participants’ correct acceptance of grammatical
sentences and correct rejection of ungrammatical sentences is the surface expression of
73
the [$wh] feature strength, and their incorrect rejection of grammatical sentences and
incorrect acceptance of grammatical sentences is the manifestation of the [+wh] feature
strength.
Table 9. Percentage of [$Wh] and [+Wh] in Wh$Questions with Wh$Words Playing All
Three Grammatical Functions in Four GJ Tests
Kinds of Test
Items
Feature
Strength
Test 1
Test 2
Test 3
Test 4
Object Wh$Questions
Gram
Ungram
[$wh]
64.29%
78.57%
61.90%
76.19%
[+wh]
28.57%
16.67%
31.25%
19.05%
Unknown
7.14%
4.76%
10.42%
4.76%
[$wh]
72.62%
90.48%
85.71%
95.24%
[+wh]
13.10%
4.76%
9.52%
4.76%
Unknown
14.29%
4.76%
4.76%
0
Attributive Wh$Questions
Gram
Ungram
[$wh]
73.81%
75.95%
73.81%
73.81%
[+wh]
21.43%
20.24%
14.53%
14.29%
Unknown
4.76%
3.81%
13.58%
4.76%
[$wh]
82.14%
80.95%
77.38%
64.29%
[+wh]
17.86%
14.29%
19.79%
30.95%
Unknown
0
4.76%
8.33%
4.76%
Adverbial Wh$Questions
Gram
Ungram
[$wh]
78.57%
74.60%
71.43%
80.95%
[+wh]
21.43%
12.70%
21.88%
14.29%
Unknown
0
12.70%
9.38%
4.76%
[$wh]
66.67%
73.02%
71.43%
88.10%
[+wh]
33.33%
22.22%
22.62%
11.90%
Unknown
0
4.75%
5.95%
0
74
4.1.2. Analysis of the Results
4.1.2.1. Research Questions 1.1: Acquisition of Simple
Wh$Questions with Wh$phrases Playing the Role of
Different Grammatical Functions
4.1.2.1.1. The OP Data
Participants’ acquisition of the [$wh] feature strength of simple wh$questions with
wh$words playing the role of different grammatical functions has the following
characteristics, as shown in Table 11, which is derived from Tables 6, 7, and 8. First off,
participants started by having relatively high accuracy rates in Test 1 (77.38% in object
wh$questions, 63.13% in attributive wh$questions, and 73.01% in adverbial wh$
questions), followed by a rise in Test 2, a drop in Test 3 to a level similar to that in Test
1. They ended up by having the greatest accuracy rate in Test 4. Second, participants’
accuracy rates in these three types of simple wh$questions were very different in the first
three tests, but leveled off in Test 4.
Table 10. Occurrence of the [$wh] Feature Strength in Simple Wh$Questions with Wh$
Words Having Different Grammatical Functions in the OP Tests
GF
O
A
Adv
Test 1
Test 2
Test 3
Test 4
77.38%
88.10%
71.43%
80.95%
(0.28)
(0.21)
(0.37)
(0.40)
63.10%
68.57%
61.90%
83.33%
(0.29)
(0.26)
(0.35)
(0.29)
73.81%
84.13%
79.76%
85.71%
(0.34)
(0.20)
(0.28)
(0.23)
Note: The number in parenthesis is Standard Deviation. GF = Grammatical Functions, O
= Object, A = Attributive wh$questions, Adv = Adverbial wh$questions
75
Did participants’ performance on the three types of simple wh$questions all
develop in a similar pattern, as shown by the descriptive data, or differently over these
four time periods? Did they perform statistically significantly better on one type of wh$
questions than on the other? If so, in which one?
To answer these questions, a 3 X 4 two$way repeated$measure ANOVA test with
time and different grammatical functions as variables were conducted. The results
indicate that there is no interaction between these two variables at the 0.05 alpha level (F
(6, 120) = 0.45, p > 0.05). This means that, overall, participants’ performance on the
three types of simple wh$questions developed at a certain time point showed no
statistically significant differences.
However, there is a main effect of grammatical functions (F (2, 40) = 6.66, p <
0.05). This means that, overall, participants performed better on one type of simple wh$
question than the other. The post$hoc test indicates that, on average, participants
performed statistically significant better on adverbial wh$questions than on attributive
wh$questions. There is no statistically significant difference between participants’
performance on object and adverbial wh$questions. Neither is there a difference between
their performance on object and attributive wh$questions.
Table 11. Repeated$Measure ANOVA for Participants’ Performance over Four OP Tests
Source
"
*
$
2.30
0.08
6.66**
0.00
1.13
0.34
Time
3
0.25
Time X S
60
0.11
GF
2
0.34
GF X S
40
0.05
Test X GF
6
0.06
Test X GF X S
120
0.05
Note: S = subjects, GF = grammatical functions, ** p < 0.01
76
Table 12. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions over Four Tests
Wh$Q of different grammatical functions
Accuracy Rate
SD
A
Attributive wh$Q
69.22%
0.31
AB
Object wh$Q
79.46%
0.33
B
Adverbial wh$Q
80.88%
0.27
Note: The same letter indicates that the difference is not statistically significant at 0.05.
In which particular test did participants perform better on one type of wh$question
than the other? Post$hoc Bonferroni pair$wise comparison of participants’ performance
on wh$questions of different grammatical functions in every single test was performed to
yield the answer. The results showed that learners did not perform statistically
significantly differently on wh$questions of different grammatical functions in Tests 1, 3
and 4. Only in Test 2 did participants perform statistically significantly better on object
wh$questions than on attributive wh$questions.
Table 13. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 1
Wh$Q of different grammatical functions
Accuracy Rate
SD
A
Attributive wh$Q
63.10%
0.29
A
Adverbial wh$Q
73.81%
0.34
A
Object wh$Q
77.38%
0.28
Note: The same letter indicates that the difference is not statistically significant at 0.05.
77
Table 14. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 2
Wh$Q of different grammatical functions
Accuracy Rate
SD
Attributive wh$Q
68.57%
0.22
AB
Adverbial wh$Q
84.13%
0.26
B
Object wh$Q
88.10%
0.20
A
Note: The same letter indicates that the difference is not statistically significant at 0.05.
Table 15. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 3
Wh$Q of different grammatical functions
Accuracy Rate
SD
A
Attributive wh$Q
61.90%
0.35
A
Adverbial wh$Q
79.76%
0.28
A
Object wh$Q
71.43%
0.20
Note: The same letter indicates that the difference is not statistically significant at 0.05.
Table 16. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 4
Wh$Q of different grammatical functions
Accuracy Rate
SD
A
Attributive wh$Q
80.95%
0.40
A
Adverbial wh$Q
83.33%
0.29
A
Object wh$Q
85.71%
0.23
Note: The same letter indicates that the difference is not statistically significant at 0.05.
78
4.1.2.1.2. The GJ Data
Participants’ performance on attributive wh$questions and adverbial wh$questions
in the grammatical sentences of the GJ tests did not vary much in the four tests. For
attributive wh$questions, there was no increase in accuracy rate between Tests 1 and 4,
and the biggest difference in accuracy rate between two consecutive tests is 2.14%. For
adverbial wh$questions, there was no more than a 2.38% increase in the accuracy rate
between Tests 1 and 4, and the biggest difference between two consecutive tests in
accuracy rate was 7.14%. In contrast, for object wh$questions, the accuracy rate rose
from 64.29% in Test 1 to 78.57% in Test 2, dropped by more than 17% in Test 3, and
ended up rising to 76.19% in Test 4.
In addition, there was not much difference in participants’ accuracy rate for
attributive and adverbial wh$questions in each of the four tests. However, there is more
than a 12.00% difference between object wh$questions and either adjectival or adverbial
wh$words in Tests 1 and 3.
Table 17. Accuracy Rate in Simple Wh$Questions with Wh$Words Having Different
Grammatical Functions in Grammatical Sentences of Four GJ Tests
GF
Test 1
Test 2
Test 3
Test 4
64.29%
78.57%
61.90%
76.19%
(0.26)
(0.30)
(0.38)
(0.44)
73.81%
75.95%
73.81%
73.81%
(0.36)
(0.19)
(0.28)
(0.30)
78.57%
74.60%
71.43%
80.95%
(0.30)
(0.30)
(0.31)
(0.29)
O
A
Adv
Note: Number in parenthesis is Standard Deviation. GF = Grammatical Functions, O =
Object, A = Attributive wh$questions, Adv = Adverbial wh$questions
79
Did participants perform in a uniform pattern in all three different types of simple
wh$questions over the four tests? Did participants perform better on one type of simple
wh$question than on the other? The 3 X 4 two$way repeated$measure ANOVA analysis
with time and grammatical function as variables indicates that there is no interaction
between these two variables (F (6, 120) = 0.52, p > 0.05). Neither did there exist main
effects of tests and words. This indicates that, according to participants’ judgment of
grammatical sentences, participants’ performance on these three different types of simple
wh$questions did not vary according to the tests, and they did not perform differently on
simple wh$questions of different grammatical functions in general.
Table 18. Repeated$Measure ANOVA for Participants’ Performance on Grammatical
Sentences over Four GJ Tests
Source
"
*
$
3.01
0.27
0.73
0.49
0.52
0.79
Time
3
0.09
Time X S
60
0.07
GF
2
0.08
GF X S
40
0.11
Test X GF
6
0.05
Test X GF X S
120
0.10
Note: S = subjects, GF = grammatical functions
Did participants’ judgment of ungrammatical sentences yield the same results? As
shown in Table 20, which is derived from Table 9, their performance on object,
attributive, and adverbial wh$questions followed a similar pattern—the accuracy rates for
both rose in Test 2, dropped in Test 3, and rose to the highest level in Test 4. In contrast,
80
their performance on adjectival wh$questions, surprisingly, dropped incrementally over
time.
Table 19. Accuracy Rate in Simple Wh$Questions with Wh$Words Having Different
Grammatical Functions in Ungrammatical Sentences of Four GJ Tests
GF
Test 1
Test 2
Test 3
Test 4
72.62%
90.48%
85.71%
95.24%
(0.23)
(0.20)
(0.31)
(0.21)
82.14%
80.95%
77.38%
64.29%
(0.23)
(0.21)
(0.23)
(0.38)
66.67%
73.02%
71.43%
88.10%
(0.39)
(0.24)
(0.28)
(0.26)
O
A
Adv
Note: Number in parenthesis is Standard Deviation. GF = Grammatical Functions, O =
Object, A = Attributive wh$questions, Adv = Adverbial wh$questions
The 3 X 4 two$way ANOVA analysis with time and word as variables indicate
that there is an interaction between these two variables (F (6, 120) = 3.65, p < 0.01),
which means that participants’ performance on object, attributive and adverbial wh$
questions did not develop in a uniform pattern over the four tests.
81
Table 20. Repeated$Measure ANOVA for Participants’ Performance on Ungrammatical
Sentences over Four GJ Tests
Source
df
MS
F
p
Time
3
0.09
1.30
0.28
Time X S
60
0.07
GF
2
0.08
3.81*
0.03
GF X S
40
0.11
Test X GF
6
0.05
3.65**
0.00
Test X GF X S
120
0.10
Note: S = subjects, GF = grammatical functions. *p < 0.05,
**
p < 0.01
The next step is to see in which tests participants’ performance on wh$questions
of different grammatical functions differ and if so how. To find out, simple effect tests of
participants’ performance on wh$questions of different grammatical functions in each test
were carried out. The results are that there was only a simple effect of Test 4 (F (2, 120)
= 9.52, p < 0.05). This means that only in Test 4 did participants’ performance on wh$
words of different parts of speech differ. According to the post$hoc test, participants’
performance on adverbial and nominal wh$words was significantly better than that on
adjectival wh$words, as shown in Table 16.
Table 21. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 4
Wh$Q of different grammatical functions
Accuracy Rate
SD
A
Attributive wh$Q
64.29%
0.39
B
Adverbial wh$Q
88.10%
0.27
B
Object wh$Q
95.24%
0.22
Note: The same letter indicates that the difference is not statistically significant at 0.05.
82
4.1.2.2. Research Question 1.2: Developmental Path
4.1.2.2.1. The OP Data
Tables 6, 7 and 8 in Section 4.1.1.1 show the rates of occurrence of [+wh] and [$
wh] feature strength in participants’ production data according to the different
grammatical functions that wh$phrases act as. How can participants’ performance on the
[$wh] and [+wh] feature strength in general at these four different time points be
represented? Since object, attributive and adverbial wh$questions are not in numerical
proportion with each other in the four tests, and wh$words of different parts of speech
might contribute different weight to the difficulty level of the tests, participants’
knowledge of the [$wh] or [+wh] feature strength at each time point should be the
average of the rates of the occurrences of [$wh] or [+wh] feature strength of all
grammatical functions at each time point, which is shown in Table 22.
Table 22. The Occurrence Rates of [$Wh] and [+Wh] in Wh$Questions Participants
Produced in Four OP Tests
[wh] feature strength
Test 1
Test 2
Test 3
Test 4
[$wh]
71.42%
80.26%
71.13%
83.33%
[+wh]
5.16%
6.08%
15.87%
7.93%
?
13.09%
10.16%
7.94%
5.56%
ZProd/ProdCompreS
10.43%
3.57%
5.16%
3.16%
As can be seen from Table 22, the initial stage of participants’ interlanguage
shows very little presence of the [wh] feature strength of their L1 English setting. In
addition, the developmental path of their acquisition of the [$wh] feature strength over the
four tests is not linear: it can be characterized as an immediate entrance into the stage of
83
switching from the [+wh] feature strength to the [$wh] feature strength, followed by a
relatively slow progress with rise and drop, and an above 80% accuracy rate in Test 4.
The fact that participants gained an accuracy rate of as high as 71.42% for [$wh]
feature strength in Test 1, and that the rate for [+wh] feature strength in this test is only
5.16%, indicates that participants’ interlanguage did not initially display the [+wh]
feature strength as predicted by FTFA. However, participants as a group did not succeed
in acquiring the [$wh] feature strength or in completely getting rid of the [+wh] feature
strength at this stage, either. The supporting evidence for this claim is that the rate for the
occurrence of [$wh] feature strength (71.42%) did not exceed 80%, which was usually
used as a criterion for successful acquisition.
The progress is slow because there is only an 11.91% increase from Test 1 to Test
4 over a one$year span. Participants did not gain the 11.91% increase in accuracy rate
steadily, but experienced variation, as shown by an up$to$10% increase in accuracy rate
from 71.42 % in Test 1 to 80.26% in Test 2, a temporary drop in Test 3 to 71.13%, and
an increase to 83.33% in Test 4. The ANOVA test indicates that there is no statistically
significant difference among participants’ performance on [$wh] feature strength in these
four tests (F (3. 60) = 2.30, p > 0.05), as shown in Table 12. This means that though the
descriptive data shows a rise and drop in the accuracy rate in the learning process, no
statistically significant difference was detected.
The developmental path of participants’ acquisition of [$wh] feature strength can
also be reflected by the rates of the occurrence of [+wh] feature strength in the four tests.
As discussed before, the rate was surprisingly low in Test 1; what was more surprising
was that it did not decrease over time but increased steadily from Test 2 (6.08%) to Test 3
(15.87%), and then dropped in Test 4 (7.93%).
84
4.1.2.2.2. The GJ Data
In the same way how the OP data was dealt with, the average of the rates for the
[$wh] feature strength of all three different grammatical functions and the average of the
rates for the [+wh] feature strength of all three different grammatical functions was used
as the rate for [$wh] feature strength and the rate for [+wh] feature strength, respectively,
which yielded Figure 3.
As can be seen from Figure 3, the developmental path is as follows: participants
started out having a fairly good judgment of grammatical and ungrammatical wh$
questions, experienced some small rises and drops along the way, and ended up gaining
the highest accuracy rate by the end of their first$year class. The participants’ accuracy
rate for correctly accepting grammatical sentences, and that for correctly rejecting
ungrammatical sentences, in Test 1 was as high as 72.22% and 73.81% respectively;
these two rates rose up to 76.83% and 81.48% respectively in Test 2, dropped in Test 3 to
69.05% and 78.17%, and reached 76.98% and 82.54% respectively in Test 4.
The ANOVA on the 21 participants’ judgment of grammatical sentences of wh$
questions over the four tests indicates that there is no statistical difference between their
performance on this feature over time (F (3, 60) = 3.01, p > 0.05), as shown in Table 19.
The ANOVA test run on their performance on ungrammatical wh$questions over the four
tests indicates the same results (F (3, 60) = 1.30, p > 0.05), as shown in Table 21. Thus
the participants’ developmental path of acquisition of the [$wh] feature as revealed by the
GJ data is similar to that revealed by the OP data: Participants moved into the transitional
stage of switching from the [+wh] to the [$wh] feature strength as early as in Test 1.
However, the transitional period for them to completely get rid of the [+wh] feature
strength and to reset it to [$wh] feature strength in their interlanguage is relatively long.
85
!
"
#
$
Figure 3. Accuracy Rate of Grammatical Judgment Tests of Simple Wh$Questions by
Participants over the First Year (N = 21)
4.1.2.3. Research Questions 1.3: Acquisition or not: Group
results vs. Individual results
4.1.2.3.1. The OP Data
For the third sub$question in Research Question 1, when participants acquired the
[$wh] feature, the group results did not yield a conclusive answer. If an 80% accuracy
rate is used as the indicator for acquisition, then there seem to be two time points when
participants acquired the [$wh] feature strength, Test 2 and Test 4. However, since the
accuracy rate dropped in Test 3, it suggests that participants’ knowledge of [$wh] feature
strength was not retained, and participants as a whole group did not acquire [$wh] feature
strength by Test 2. Using the same line of reasoning, it is safe to say that there is a
likelihood that participants acquired this feature strength by the time they took Test 4, but
a follow$up test is needed to verify this.
The problem with only looking at group results is that it cannot be decided
whether each individual in the group performs at around the calculated accuracy rates in
the four tests, or whether the calculated rates are the result of some participants
86
performing at a level far above the average and some participants performing at a level
far below it. Applied to the data in this dissertation, there are two problems with only
looking at the group data. First, the developmental path revealed by the group results can
conceal the fact that there are some participants who might have acquired this feature
strength as early as Test 1 and retained it across all four tests. Second, even though the
participants at the group level were shown to have acquired the feature strength by Test 4,
it is possible that not all of the participants acquired this feature strength by Test 4.
These problems cannot be solved until the individual results are examined. Before
doing that, it is necessary to decide the criterion of acquisition for individual participants’
longitudinal data. As stated before, if a learner had an 80% accuracy rate in one test and
then scored lower than 80% in the following test, it means that this learner did not retain
the knowledge. Thus it is sensible to say that if from a certain test (the latest would be
Test 3), a learner has an 80% accuracy rate consecutively to Test 4, this learner can be
regarded as having successfully acquired this feature strength.
As shown in Figure 4, six of the participants (28.57%) had an above$80%
accuracy rate across all four tests, meaning that they succeeded in acquiring this feature
strength, as defined in this study, as early as Test 1. Four of them (19.05%) had an above$
80% accuracy rate from Test 2 to Test 4, meaning that they succeeded in acquiring this
feature strength as defined in this study by Test 2. One of them (4.76%) did not acquire it
until Test 3.
There are also some special cases. Two participants (9.52%) did not have an 80%
accuracy rate until Test 4. Another 2 participants had 80% accuracy rates in Tests 2 and
4, but the accuracy dropped a bit to 75% in Test 3.
Overall, of the 15 participants who had an 80% accuracy rate in Test 4, 11
acquired [$wh] feature strength, which is 52.38% of all the participants. Four are highly
likely to have acquired it, though further evidence is needed to prove this, and that makes
up 19.05% of all the participants. Again it should be noted that acquisition as defined in
87
this study was an 80% correctness rate.
There are still 6 participants (28.57%) who did not acquire [$wh] feature strength.
Three of them had a below$80% accuracy rate across all four tests, two had an 80%
accuracy rate only in Test 1, but could not maintain it in the remaining tests, and one had
an 80% accuracy rate in Test 1 and Test 2, but did not maintain it in Test 3 and Test 4.
Figure 4. Individual Results of Occurrence of [$wh] Feature Strength across Four OP
Tests
The individual results, as compared with the group results, indicate that the [$wh]
feature is not a grammatical feature that each individual performed at a level around the
average accuracy rate of the group. Thus the group results cannot reveal a full picture of
the successful acquisition of [$wh] feature strength by the students. Only by looking at
the individual results did we find that there were participants who acquired it at any stage
of the learning, rather than that the participants did not acquire it by taking Test 1 and
acquired it by taking Test 4, as might otherwise be said from the group results.
88
4.1.2.3.2. The GJ Data
If an 80% accuracy rate in both the grammatical and ungrammatical sentences is
used as the criterion for successful acquisition, the GJ data shows that participants as a
group did not learn [$wh] feature strength even by the end of their first$year class.
Participants’ accuracy rate in ungrammatical sentences reached above 80% (82.54%) and
their accuracy rate in grammatical sentences is very close to that, but did not reach the
80% criterion (76.98%).
For individual results, similarly to how the OP data was dealt with, an 80%
accuracy rate in both grammatical and ungrammatical sentences in consecutive tests is
used as the criterion for successful acquisition.
The individuals’ results across all four tests revealed the following findings: only
one of the participants (4.76%) had an 80% accuracy rate across all four tests, meaning
that 1 participant succeeded in acquiring the [$wh] from Test1. One participant (4.76%)
succeeded in acquiring the [$wh] at Test 2. One participant (4.76%) succeeded in
acquiring it at Test 3, and 2 participants (9.53%) had an 80% accuracy rate in all four
tests except in Test 2. Two participants (9.53%) did not have an 80% accuracy rate until
Test 4. Thus we can say that 5 of the participants (23.81%) succeeded in acquiring [$wh]
feature strength. Another 2 participants (4.76%) may have acquired it by taking Test 4;
however, further evidence is needed to prove that.
The special case is that there were 7 participants who had an 80% accuracy rate at
one point or another; however, they could not maintain this accuracy rate throughout all
the tests, losing it either earlier or later in the first$year learning process.
89
Table 23. Participants’ Accuracy Rates in both Grammatical and Ungrammatical
Sentences in Four GJ Tests
Participants
Test 1
Test 2
Test 3
Test 4
1
n
n
n
n
2
y
n
y
y
3
y
y
y
y
4
y
n
y
y
5
y
n
n
n
6
y
n
n
n
7
n
n
n
n
8
y
y
n
n
9
n
y
n
n
10
n
n
n
n
11
n
n
n
y
12
n
n
n
n
13
n
n
y
y
14
n
n
y
n
15
n
y
y
y
16
n
n
n
y
17
n
n
n
n
18
n
n
n
n
19
n
y
n
n
20
n
n
n
n
21
n
y
y
n
Note: y indicates an 80% accuracy rate in both grammatical and ungrammatical
sentences; n indicates an 80% accuracy rate in either grammatical sentences or
ungrammatical sentences or neither.
90
4.2. Research Question 2: The Acquisition of Two Types of
Complex Wh$Questions
Recall that the participants’ acquisition of wh$questions as object clauses
introduced by
and
was examined at the end of their first$year Chinese
learning. The test items of the OP and GJ tests were mixed together with those of the
fourth OP test and the fourth GJ test for simple wh$questions respectively.
In this section, we address the second set of research questions regarding the
acquistion of
and wh$questions with
. This set of question includes the
following sub$questions:
1) What is the initial stage of indirect wh$questions and wh$questions in object
complements? What are the learning results of these questions in comparison to simple
wh$questions?
2) Are indirect wh$questions or wh$questions in object complements successfully
acquired at the end of first$year Chinese learning?
3) At the end of first$year Chinese learning, what are the learning results of
learners’ acquisition of the grammatical feature that wh$words are not allowed in clause$
initial position in different clauses? Does their acquisition of this grammatical feature
vary between different clauses?
4) What is the relationship between learners’ acquisition of this grammatical
feature in different clauses?
4.2.1. Initial Analysis of the Data
4.2.1.1. Initial Analysis the OP Data
In this section, the author will first categorize errors that participants made in
terms of the location of wh$words for the two types of wh$questions, report their
percentages, and then determine the setting of [wh] feature that the correct sentences
express and the setting of [wh] that the different kinds of incorrect sentences express.
91
From the participants’ responses, the author identified three types of errors that
participants made for wh$questions of
: Wh$Words in Clause$Initial Position
of the Embedded Clauses (Wh in C$1st$EC), Misplaced Adverbial Wh$Words in the
Embedded Clause (MP$Adv$Wh$EC), and Zero Production or Failure to Produce
Comprehensible Sentences (ZProd/F$To$ProdCompreS). There were six types of errors
that they made for wh$questions of
: Wh$Words in Clause$Initial Position of the
Embedded Clauses (Wh in C$1st$EC), Wh$Words in Clause$Initial Position of the Main
Clause (Wh in C$1st$MC), Misplaced Adverbial Wh$Words in the Embedded Clause
(MP$Adv$Wh$EC), Wh$Words in Second Position of the Embedded Clause (Wh$2nd$
EC), Adverbial Wh$words within Main Clause (Adv$Wh$MC), and Zero Production or
Failure to Produce Comprehensible Sentences (ZProd/F$To$ProdCompreS).
(39)
Wh$Questions of %
a. Wh$Words in Clause$Initial Position of the Embedded Clauses (Wh in C$1st$
EC)
Wo xiang zhidao jige
I
ren
ni kanjian.
want know how many (measure word) people you see.
I wonder how many people you have seen.
b. Misplaced Adverbial Wh$Words in the Embedded Clause (MP$Adv$Wh$EC)
Wo xiang zhidao ni
I
faxian de rushiqiangjie zenme?
want know you discover particle13 burglary how
I wonder how you discovered this burglary.
(40)
Wh$Questions of Juede
a. Wh$Words in Clause$Initial Position of the Embedded Clauses (Wh in C$1st$
13
(Shi)…de is a particle for a known past event.
92
EC)
Ni juede zenme women yinggai xuehao hanyu?
You think how we should learn Chinese well
How do you think that we should learn Chinese well?
b. Wh$Words in Clause$Initial Position of the Main Clause (Wh in C$1st$MC)
Zenmei ni juede women yinggai xuehao hanyu?
How you think we should learn Chinese well?
How do you think that we should learn Chinese well?
c. Misplaced Adverbial Wh$Words in the Embedded Clause (MP$Adv$Wh$EC)
(i) Wo xiangzhidao ta likai shenme shihou.
I want know he leave when.
I wonder when he left.
(ii) Ni juede ta jin
fangzi zenme?
You feel he enter house how
How do you feel that he entered the house?
d. Wh$Words in Second Position of the Embedded Clause (Wh$2nd$EC)
Ni juede ni shenme xuedao le?
You feel you what learn (completed aspect particle)
How did you feel that you learned?
e. Adverbial Wh$words within Main Clause (Adv$Wh$MC)
Ni zenme juede women yinggai xuehao zhongwen?
You how feel we should learn Chinese well?
Next, the author analyzed both the accurate sentences and the kinds of incorrect
sentences that participants produced to find out what feature strength of [wh] the accurate
sentences appear to express, and what feature strength of [wh] or what combination of
feature strengths of [wh] lead to the kinds of incorrect sentences identified above.
For wh$questions of
, the accurate wh$questions that participants
93
produced indicate the underlying embedded [$wh] feature strength. The most common
error participants made is Wh$Words in Clause$Initial Position of the Embedded Clause,
which indicates the embedded [+wh] feature strength.
For wh$questions of
, in terms of the [wh] feature strength, the accurate
sentences that participants produced is a manifestation of the matrix [$wh] feature
strength in the main clause, as shown by (41) a. Participants misplaced wh$words in the
clause intial position of the main clauses (Wh$Words in Clause$Initial Position of the
Main Clause) and also of the embedded clauses (Wh$Words in Clause$Initial Position of
the Embedded Clause). The first kind of error is a manifestation of the [wh] feature
hosted in Spec, CP of the main clause being strong, as a result of the direct influence
from participants’ L1 English, as shown by (41) b. The seond kind of error, as shown by
(41) c, indicates that the matrix [wh] feature strength is weak, just as their L2 Chinese is.
Since those learners who produced (41) c deemed it as the Chinese counterpart of
+
, they must have a wide scope interpretation
of the sentence they produced. That is, the matrix has a [wh] feature. As the wh$word is
not moved, this feature strength is weak. In addition, it also seems to show that the
embedded [wh] is strong. What is puzzling is that it manifests neither their L1 English
nor the L2 Chinese setting.
(41)
a.
[$wh]
(CP (IP ni
juede (CP (IP women yinggai zenmei ti xuehao
you think
we
should how
hanyu))))
learn$well Chinese
94
b.
[+wh]
(CP zenmei (IP ni juede (CP ti (IP women yinggai ti xuehao
How
you think
we
should
hanyu?))))
learn$well Chinese
c.
[$wh]
(CP (IP ni
[+wh]
juede (CP zenmei (IP women yinggai ti xuehao hanyu))))
you think
how
we
should
learn$well Chinese
It remains questionable how to categorize the error of Adverbial Wh$words within
Main Clause (Adv$Wh$MC) in terms of the [wh] feature. In (40) e, the fact that the
adverbial wh$word is moved to the matrix clause indicates that the matrix [wh] is strong.
However, that
is actually placed in in$situ position within the main clause indicates
that the [wh] feature strength must be weak. The possible reason why the participants
made this type of error with regard to the [wh] feature strength is unclear to the author.
Regarding Wh$Words in Second Position of the Embedded Clause (Wh$2nd$EC),
the fact that wh$words are not present in the main clause indicates that the [wh] feature
strength in the main clause has been reset to weak. The [wh] feature strength in the
embedded clause is puzzling if the same line of reasoning found in section 4.1.1.1 is
adopted.
There are also shared errors that participants made in producing both types of wh$
questions. The first one was the Misplaced Adverbial in Embedded Clause (MP$Adv$
Wh$EC). As detailed in section 4.1.1.1, the Misplaced Adverbial, though still an error, is
actually a manifestation of the [$wh] feature strength being located in the clause
95
containing the adverb. Therefore, the Misplaced Adverbial in the Embedded Clause for
wh$questions of
, as shown by (40) b (i), can be regarded as an instance of
the embedded [$wh] feature strength. Similarly, for wh$questions of
, the embedded
[wh] feature strength is weak as well. Since the sentence is acknowledged as a question
with wide$scope, as evidenced by the fact that it is produced as a counterpart to the
English sentence ,
, and the wh$word is not moved to
the matrix clause, it can be inferred that the matrix sentence has a [$wh] feature strength.
For both types of questions, there was Zero Production or Failure to Produce
Comprehensible Sentences (ZProd/F$To$ProdCompreS), which can tell us nothing about
the setting of the [wh] feature as explained in section 4.1.1.1.
According to the above analysis, the rates of occurrences of the [$wh] feature
strength and of the [+wh] feature strength in the two types of complex wh$questions in
participants’ production data are then calculated. The results are summarized in Table 19.
For wh$questions of
, the total number of the occurrences of the ……[$wh]
feature strength in participants’ production data is the sum of the number of the accurate
sentences and that of Misplaced Adverbs in Embedded Clause; this amount totals
59.05%. The total number of the occurrences of the embedded [+wh] feature strength of
wh$questions of
is the number of the Wh$Words in Clause$Initial Position
in the Embedded Clause, which is 21.90%. For wh$questions of
, the total number
of the occurrences of the [$wh] feature strength in both matrix and embedded clauses is
the sum of the number of the accurate sentences participants produced and that of
Misplaced Adverbs in Embedded Clause; this amount is 60.95%. As illustrated above,
there are two kinds of errors regarding wh$words in the clause$initial position—the [+wh]
feature strength in both matrix and embedded clauses, which is the number of the Wh$
Words in Initial Position of Embedded Clause, and the [+wh] feature strength in
embedded clauses, which is the sum of the number of the Wh$Words in Clause$Initial
Position of the Embedded Clause and the number of Misplaced Adverbs in MC; these
96
amounts are 2.86% and 16.19%, respectively.
Table 24. Numbers and Rates of the Occurrences of Strong [wh] and Weak [wh] Feature
in the OP Test of the Two Types of Wh$Questions
Types of Produced Wh$Q
Wh$Q$OC of %
Wh$Q$OC of -
Embedded [$wh] for Wh$Q of Xiang zhidao
Matrix [$wh] and No Embedded [wh] for wh$Q of Juede
Accuracy Rate
51 (48.57%)
59 (56.19%)
MP$Adv$EC
11 (10.47%)
5 (4.76%)
[+wh] for Wh$Q of %
[$wh]…[+wh] for Wh$Q of Wh$1st$EC
23 (21.90%)
14 (13.33%)
Matrix [$wh] for Wh$Q of Wh$2nd$EC
0
5 (4.76%)
Matrix [+wh] for Wh$Q of Wh$1st$MC
3 (2.86%)
?
Adv$MC
3 (2.86%)
Unknown
Zrod/F$To$ProdCompreS
20 (18.09%)
16 (15.23%)
Note: Empty Cells indicate that they are irrelevant. Total N = 5 x 21 = 105
4.2.1.2. Initial Analysis of the GJ Data
Just as how GJ data of the simple wh$questions was sorted, participants’
and
,
,
answers for the grammatical sentences were categorized into three
groups, called
$,
, and
, and their
,
, and
answers for ungrammatical sentences were categorized into groups called
97
$,
, and
. The descriptive data yielded the
following chart (Figure 5).
%&
'
()
*
,&
'
+' + )
-.&+
()
)
'
/
!
"
*
$ &
#
+'
+
$
Figure 5. Accuracy Rates of Grammatical Judgment Test of Three Types of Wh$
Questions by Participants at the End of First Year
What represents participants’ knowledge of the various kinds of [$wh] feature
strength under investigation, namely embedded [$wh] of wh$questions of
and matraix [$wh] feature strength of wh$question of
has yet to be determined.
Of course, participants’ knowledge of the embedded [$wh] feature strength for
,
98
wh$questions of
, and the matrix [$wh] feature strength for wh$questions of
should be measured by their performance on both the grammatical and the
ungrammatical sentences on which they were tested.
The case is slightly more complicated for wh$questions of
wh$questions of
. Recall that for
, two kinds of ungrammatical sentences were included in the test.
These were included to determine whether the participants could reject both
ungrammatical sentence type 1 (those with wh$words at sentence$initial position of main
clauses) and ungrammatical sentence type 2 (those with wh$words at sentence$initial
position of embedded clauses) at the 80% accuracy rate, which was chosen as the
criterion for acquisition; if they were able to do so, they were regarded as having the
matrix [$wh] feature strength successfully established in their internal grammar and not
having embedded [+wh], as shown by some learners’ production data. The rationale is as
follows: If participants reject the ungrammatical sentence type 1 at an 80% accuracy rate,
participants could have either matrix [$wh] and zero embedded [wh] or embedded [+wh]
feature strength in their interlanguage grammar, the latter of which is exactly the
ungrammatical sentence type 2. If participants reject the ungrammatical sentence type 2
at an 80% accuracy rate, participants can have in their internal grammar either matrix [$
wh] and zero embedded [wh] or matrix [+wh], the latter of which is exactly the
ungrammatical sentence type 1. Only when participants are able to reject both sentence
types can they be regarded as having the correct setting successfully set up, i.e. matrix [$
wh] feature strength and zero embedded [wh] feature strength. However, when each
individual’s data is examined, it is found that there is not a single learner who rejected
ungrammatical sentence type 2 at an 80% accuracy rate but did not reject ungrammatical
sentence type 1 at 80%; however, the reverse does not apply. Thus, the possibility that
participants who rejected ungrammatical sentence type 2 because they preferred
ungrammatical sentence type 1 (in other words, matrix [+wh] existed in the participants’
interlanguage) is eliminated. Thus, participants’ rejection rate of ungrammatical sentence
99
type 2 and their acceptance rate of grammatical sentences can serve to adequately
measure their knowledge of the matrix [$wh] and zero embedded [wh].
Participants’ rejection rates of ungrammatical sentence type 1 can be used to
measure their knowledge of [$wh] in the main clause. The reason is that if participants
reject ungrammatical sentence type 1 at an 80% accuracy rate, participants must have
either the matrix [$wh] and zero embedded [wh] or embedded [+wh] feature strength
established in their interlanguage grammar. No matter which one it is, it is apparent that
the [wh] in the main clause is set to weak in participants’ internal grammar.
4.2.2. Analysis of the Results
4.2.2.1. Research Question 2.1: Initial State of the [wh]
Feature Strength Necessary for Acquisition of Wh$
Questions of %
and Wh$questions of -
and
Learning Results as Compared to That of Simple Wh$
Questions
4.2.1.1.1. The OP Data
The [wh] feature strength necessary for the acquisition of wh$questions of
is embedded [$wh] feature strength. For wh$questions of
, learners need to be
aware that the matrix clause has [$wh] feature strength and the embedded questions does
not bear the [wh] feature, since the scope of question is in the entire sentence rather than
in the embedded clauses.
As can be seen from Table 24, for wh$questions of
, there was a
21.90% occurrence rate of the embedded [+wh] feature strength and a 59.05% occurrence
rate of the correct [$wh] feature strength, apparently indicating a considerable L1
influence upon the initial stage of participants’ L2 grammar.
100
For wh$questions of
, there was only a 4.76% occurrence rate of matrix
[+wh] feature strength, suggesting very little direct L1 influence. But an error occurred
with 19.05% of the occurrence rate of the wh$words in embedded clauses, which
manifests neither L1 influence nor L2 influence. In other words, very early in their
acquisition of wh$questions of
, learners acquired the matrix [$wh] feature strength,
but unexpectedly made errors of creating a medial [+wh] feature strength, the nature of
which will be discussed in Chapter 5.
How is participants’ performance on wh$questions of
questions of
and of wh$
compared to their performance on simple wh$questions? The
descriptive data indicates that they did better in simple wh$questions than in wh$questions
of
and of wh$questions of
, the performances of which were very
similar, as shown in Figure 6.
Figure 6 Accuracy Rate of Three Types of Wh$Questions in the 4th OP Test
A one$way ANOVA shows that participants performed differently on simple wh$
questions, wh$questions of
, and wh$questions of
(F (2, 60) = 6.54, p <
0.05). A post$hoc Bonferroni pair$wise comparison test reveals that participants
performed significantly better on simple wh$questions than on both types of complex wh$
questions, and that their performance on these two types of wh$questions was not
101
significantly different.
Table 25. ANOVA for Participants’ Performance on Simple Wh$Q and Two Types of
Complex Wh$Q in Test 4
Source
Df
MS
F
P
Between Groups
2
13.40
6.54**
0.0027
Within Groups
60
2.04
Note: S = subjects, GF = grammatical functions.
**
p < 0.01
Table 26. Bonferroni Pair$wise Comparison of Participants’ Performance on Three Types
of Simple Wh$Questions in Test 4
Wh$Q of different GF
Accuracy Rate
A
Wh$Q of Xiang zhidao
59.05%
SD
1.53
A
Wh$Q of Juede
60.95%
1.72
B
Simple wh$Q
83.33%
0.92
Note: The same letter indicates that the difference is not statistically significant at 0.05.
4.2.2.1.2. The GJ Data
What does the GJ data reveal about the participants’ initial state of L2 grammar
for the two types of wh$questions? For wh$questions of
, the initial state
shows a considerable influence from their L1 English, as shown by the 61.90% accuracy
rate for correctly accepting the [$wh] feature strength and the 45.71% accuracy rate for
correctly rejecting the [+wh] feature strength, both of which are around chance level.
For wh$questions of
, participants had a high accuracy rate for correctly
accepting the matrix [$wh] feature strength, 78.10%, and a high accuracy rate for
correctly rejecting the [+wh] feature strength, 85.71%. But they only correctly rejected
102
the [$wh]…[+wh] feature strength at a chance level, 50.48%. This is consistent with the
OP data: the participants’ initial L2 grammar does not indicate a direct influence from
their L1, but shows a kind of error which is not like their L1 or L2 situation.
To reveal participants’ competence in embedded [$wh] of wh$questions of
and matrix [$wh] of wh$questions of
, as compared to their competence in [$
wh] of simple wh$questions, their performance on the judgment of grammatical
sentences of these three types of wh$questions and their judgment of the ungrammatical
sentences are respectively examined.
As shown in Figure 5, participants’ accuracy rates for grammatical sentences of
wh$questions of
(56.67%) are much lower than those for simple wh$
questions (76.98%). There does not seem to be a big difference between participants'
accuracy rate in wh$questions of
(70.83%) and simple wh$questions; however,
statistical analysis will more clearly illuminate participants’ performance in these tests.
A one$way ANOVA test shows that participants’ performance on their judgment
of the grammatical sentences is statistically significantly different (F (2, 60) = 3.16, p =
0.05). The post$hoc Bonferroni pair$wise test reveals that participants’ performance on
grammatical sentences was significantly better on simple wh$questions and wh$questions
of
than on grammatical sentences of wh$questions of
.
Table 27. ANOVA for Participants’ Performance on Simple Wh$Q and Two Types of
Complex Wh$Q in Grammatical Sentences in Test 4
Source
Df
MS
F
Sig.
Between Groups
2
4.33
3.16
0.05*
Within Groups
60
1.37
Note: S = subjects, GF = grammatical functions. * p = .05
103
Table 28. Bonferroni Pair$wise Comparison of Participants’ Performance on Simple Wh$
Q and Two Types of Complex Wh$Q in Grammatical Sentences in Test 4
Wh$Q of different GF
Accuracy Rate
A
Wh$Q of Xiang zhidao
61.91%
SD
1.57
A
Wh$Q of Juede
76.98%
0.73
A
Simple wh$Q
78.10%
1.04
Note: The same letter indicates that the difference is not statistically significant at 0.05.
As shown in Figure 5, participants’ accuracy rate for ungrammatical sentences of
(45.71%) and that for ungrammatical sentences type 2 of wh$question with
(50.48%) is much lower than the accuracy rate for ungrammatical sentences of
simple wh$questions (82.54%). A one$way ANOVA shows that participants performed
differently on their judgment of these different feature strengths (F (2, 60) = 7.68, p <
0.01). A post$hoc Bonferroni pair$wise test yields that participants performed better on
simple wh$questions than wh$questions of
and wh$questions of
. In
addition, participants' performance on the latter two is not significantly different.
Table 29. ANOVA for Participants’ Performance on Simple Wh$Q and Two Types of
Complex Wh$Q in Ungrammatical Sentences in Test 4
Source
Df
MS
F
P
Between Groups
2
18.11
7.68**
0.00
Within Groups
60
2.36
Note: S = subjects, GF = grammatical functions.
**
p < 0.01
104
Table 30. Bonferroni Pair$wise Comparison of Participants’ Performance on Simple Wh$
Q and Two Types of Complex Wh$Q in Test 4
Wh$Q of different GF
Accuracy Rate
A
Wh$Q of Xiang zhidao
45.71%
SD
1.64
A
Wh$Q of Juede
50.48%
1.72
B
Simple wh$Q
82.54%
1.83
Note: The same letter indicates that the difference is not statistically significant at 0.05.
The grammaticality judgment data did not yield a consistent picture of
participants' knowledge of the [$wh] feature strength necessary for the acquisition of the
two types of complex wh$questions as compared to that which was necessary for simple
wh$questions. As shown above, the data on their judgment of grammatical sentences
shows that participants’ performance on embedded [$wh] feature strength of wh$
questions of
and matrix [$wh] and zero medial [wh] of wh$questions of
was of the same level, but it did not reach the level of [$wh] feature strength of
simple wh$questions. However, according to participants’ performance on ungrammatical
sentences, neither their performance on wh$questions of
performance on wh$question of
nor their
regarding the [wh] feature strength reaches the
same level as their performance on [$wh] feature strength of simple wh$questions.
4.2.2.2. Research Question 2.2: Acquisition or not: Group
results vs. individual results
4.2.2.2.1. The OP Data
When the accuracy rates for the embedded [$wh] feature strength of wh$questions
of
and of the matrix [$wh] feature strength of wh$questions of
(59.05% for wh$questions of
, 60.95% for wh$questions of
) are
105
measured against an 80% accuracy rate, which is assumed as an indication of acquisition,
the conclusion concerning sub$question 2 of Major Research Question 2 is that at the end
of their first year of learning Chinese, participants did not acquire the correct setting of
the embedded [wh] of wh$questions of
, nor did they acquire the correct
setting of the matrix [wh] of wh$questions of
.
The next question is whether some particpants could have succeeded in acquiring
the embedded [$wh] of wh$questions of
questions of
and the matrix [$wh] of wh$
, although the participants as a group did not. In other words, are the
accuracy rates of the group representative of each learner in the group? To answer this
question, the individual data is examined. Again, an 80% accuracy rate is used as the
criterion for successful acquisition.
As demonstrated by Table 32, at the end of first$year of Chinese learning, 47.62%
of participants had acquired the [$wh] feature strength in the embedded clause for wh$
questions of
, as compared to the 57.37% accuracy rate for the group. As
shown in Table 33, for wh$questions of
, 42.86% of participants had acquired it, as
compared to the 60.95% accuracy rate for the group. Thus, not only are the accuracy rates
for the two types of wh$questions very similar, but there are also almost as many students
who had learned the embedded [$wh] feature strength of wh$questions of
those who had learned the matrix [$wh] feature strength of wh$questions of
as
.
When the group results and individual results for the two types of wh$questions
are examined together, it is found that, much like the data regarding the participants’
acquisition of simple wh$questions, participants’ fairly high accuracy rates for the two
types of wh$questions (59.05% and 60.95%) at this stage are not the result of every
learner performing equally well at that level; rather, they are the result of some
participants performing so well that they actually succeeded in learning it, while some
participants performed poorly and did not learn it.
106
Table 31. Number and Percentage of Participants Acquiring the Weak [wh] Feature
Strength of Embedded Clause in Wh$Questions of Xiang Zhidao
Accuracy Rate
Test 4 (n = 5)
80% or above
10 (47.62%)
Below 80%
11 (52.38%)
Note: n = number of test items
Table 32. Number and Percentage of Participants Acquiring the Weak [wh] of Main
Clause and Weak [wh] of Embedded Clause in Wh$Questions of Juede
Accuracy Rate
Test 4 (n = 5)
80% or above
9 (42.86 %)
Below 80%
12 (57.14%)
Note: n = number of test items
4.2.2.2.2. The GJ Data
If the criterion for acquisition is an 80% accuracy rate in correctly accepting the
grammatical and correctly rejecting the ungrammatical sentences at the same time,
participants as a whole group clearly did not succeed in acquiring either the embedded [$
wh] feature strength of wh$questions of
of wh$questions of
or the matrix [wh] feature strength
. Participants’ accuracy rates in the grammatical (61.90%) and
ungrammatical sentences of wh$questions of
(45.71%) were both below the
80% criterion. Their accuracy rate for rejecting ungrammatical sentences type 2 (50.48%)
and their accuracy rate for accepting grammatical sentences (76.98%) of wh$questions of
were both 80%.
Are there participants who acquired these two features, even though the
participants as a group did not? Participants who had an 80% or above$80% accuracy rate
107
for both the grammatical and ungrammatical sentences were regarded as having
successfully acquired the specific [$wh] feature strength. As shown in the following two
charts, 23.80% of participants (5/21) had acquired the embedded [$wh] feature strength of
wh$questions of
, and 42.86% of them (9/21) had acquired the matrix [$wh]
and zero embedded [wh] feature strength of wh$questions of
.
5
5
1
1
Figure 7. Number of Participants Having 80% Accuracy Rate for Both Grammatical and
Ungrammatical Sentences of Wh$Q of Xiang zhidao
Note:
Grammatical sentences
Ungrammatical sentences
108
5
3
1
9
Figure 8. Number of Participants Having 80% Accuracy Rate for both Grammatical and
Ungrammatical Sentences of Wh$Questions of -
Note:
Grammatical sentences
Ungrammatical sentences Type 1
Ungrammatical Sentences Type 2
4.2.2.3. Research Question 2.3: Learners’ Competence in
Different Clauses Regarding the Absence of Wh$Words in
Clause$Initial Position
The above sections show that participants face two tasks when learning wh$
questions of
. The first task is to learn that wh$words are absent in matrix clause$
initial positions, because in Chinese, the matrix [wh] is weak. The other task is to know
that wh$words are absent in embedded clause$initial positions and are in$situ of the
embedded clauses, because Chinese wide$scope questions do not have [wh] feature in
embedded clauses.
Participants’ accuracy rates in putting wh$words correctly in the embedded
clauses of wh$questions of
and wh$questions of
were very similar in
both OP and GJ tasks. So were the error rates of placing them in the clause$intial position
(See Table 24 and Figure 5). Will the statistical analysis indicate that participants’
competence of the two clauses is similar regarding the grammatical feature that wh$words
are not allowed in clause$initial position?
109
Recall also that for wh$questions of
, participants made very few mistakes in
putting wh$words in matrix clause$initial position in both the OP and GJ tasks (See Table
24 and Figure 5). Meanwhile, they made very few mistakes in fronting wh$words in
simple wh$questions. Is their competence of matrix clause of wh$questions of
similar as that of simple wh$questions regarding the grammatical feature according to the
statistical analysis?
4.2.2.3.1. The OP Data
As explained in section 4.2.2.1, when learners put wh$words in the embedded
clause$initial position for wh$questions of
, they realize that the matrix clause$initial
position is bad, but they somehow very unexpectedly created a medial [+wh] feature
strength. Participants’ correct production of wh$questions of
indicates that they are
aware that both embedded and matrix clause$initialpositions are bad for wh$words. Thus,
participants’ knowledge of this grammatical feature in the matrix clause$initial positions
should be represented by correctly produced sentences, by the error of Wh$words in
Clausal Initial Position of Embedded Clause, and by the error of Wh$Words in Second
Position of Embedded Clause.
Participants’ performance on this grammatical feature in the embedded clause of
wh$questions of
questions of
(59.05%) is similar to that on the embedded clauses of wh$
(60.95%), both of which are inferior to their performance on this
grammatical feature in matrix clause of wh$questions of
(79.05%) and that in
simple wh$questions (83.33%).
To obtain the statistical support for this observation, a one$way ANOVA was
conducted. The results of the test confirmed our observation: there is a statistically
significant difference in participants’ performance on this grammatical feature among
different clauses (F (3, 80) = 4.92, P < 0.01, see Table 33). The post$hoc Bonferroni pair$
wise test indicates that participants’ performance on simple wh$questions and main
110
clause of wh$questions of
is better by a statistically significant margin than their
performance on the two types of wh$questions concerning this grammatical feature.
However, there is no statistically significant difference between participants’
performance on the embedded clause of the two types of wh$questions regarding this
feature. Neither is there any difference between participants’ performance on simple wh$
questions and the main clause of wh$questions of
.
Table 33. ANOVA for Participants’ Knowledge of Absence of Wh$Words in Clause$
initialPosition in Different Clauses in the 4th OP Test
Source
df
MS
F
P
Questions
3
10.20
4.92**
0.00
Residual
80
2.07
Note: S = subjects, GF = grammatical functions.
**
p < 0.01
Table 34. Bonferroni Pair$wise Comparison of Participants’ Performance on Absence of
Wh$Words in Clause$initialPosition in Different Clauses in the 4th OP Test
Wh$Q of different GF
Accuracy Rate
A
EC of Wh$Q of Xiang zhidao
59.05%
SD
1.53
A
EC of Wh$Q of Juede
60.96%
1.71
B
Simple wh$Q
81.90%
1.47
B
MC of Wh$Q of Juede
83.33%
1.53
Note: The same letter indicates that the difference is not statistically significant at 0.05.
If an 80% accuracy rate is still used as the criterion of acquisition, the individual
results show that there are as many participants who acquired the feature of the absence
111
of the wh$word in clause$initial position in simple wh$questions as those who acquired
this grammatical feature in matrix clause of wh$questions of
. There are as many
participants who acquired this grammatical feature in embedded clauses of wh$questions
of
as those who did for wh$questions of
.
Table 35. Number and Percentage of Participants Who Acquired Absence of Wh$Words
in Clause$initialPosition in Simple Wh$Q in the 4th OP Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
15 (71.43%)
Below 80%
6 (28.57%)
Table 36. Number and Percentage of Participants Who Acquired Absence of Wh$Words
in Clause$initialPosition in Matrix Clause of Wh$Q of Juede in the 4th OP Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
15 (71.43%)
Below 80%
6 (28.57%)
Table 37. Number and Percentage of Participants Who Acquired Absence of Wh$Words
in Clause$initialPosition in Embedded Clause of Wh$Q of Xiang zhidao in the 4th OP
Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
9 (42.86%)
Below 80%
12 (57.14%)
112
Table 38. Number and Percentage of Participants Who Acquired Absence of Wh$Words
in Clause$initialPosition in Embedded Clause of Wh$Q of Juede in the 4th OP Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
9 (42.86%)
Below 80%
12 (57.14%)
Thus both the ANOVA test and individual results support the initial observation
that participants’ performance on simple wh$questions and matrix clauses of wh$
questions are in the similar developmental stage, which is better than their performance
on embedded clauses of wh$questions of
questions of
and embedded clauses of wh$
.
4.2.2.3.2. The GJ Data
What does the participants’ GJ data show? Since grammatical sentences of wh$
questions of
would tell participants’ performance on both the main clauses and
embedded clauses, but would not reveal specific information about participants’
judgment of one of them, their performance on ungrammatical sentences will be used to
compare their performance on these two clauses. To be more specific, participants’
performance on ungrammatical sentence type 1 and ungrammatical sentence type 2 will
be used to reveal their performance on main clauses and embedded clauses of wh$
questions of
respectively, and their performance on ungrammatical sentences of
wh$questions of
questions of
to reveal their performance on embedded clauses of wh$
.
According to the data on their judgment of ungrammatical sentences, participants
performed much better on main clauses of wh$questions of
, the accuracy rate of
which reaches 85.71%, than on embedded clauses of the two types of wh$questions,
113
which are 45.71% and 50.48% respectively. Participants’ accuracy rate for main clauses
of wh$questions
is slightly higher even than that for simple wh$questions, which is
82.54%.
A one$way ANOVA test was conducted to reveal whether the difference is
statistically significant. It is shown that participants’ performance on different clauses of
these different types of wh$questions regarding the grammatical feature under
investigation is significantly different (F (3, 80) = 10.07, p < 0.01, see Table 39). The
post$hoc Bonferroni pair$wise comparison test indicates that participants’ performance on
main clauses of wh$questions of
was at the same level of wh$questions of simple
wh$questions, which is significantly better than that of the embedded clauses of the two
types of wh$questions (see Table 40). This result is consistent with that yielded by the
production data.
Table 39. ANOVA for Participants’ Knowledge of Absence of Wh$Words in Clause$
initialPosition in Different Clauses in the 4th GJ Test
Source
df
MS
F
p
Questions
3
21.63
10.07**
0.00
Residual
80
2.14
Note: S = subjects, GF = grammatical functions.
**
p < 0.01
114
Table 40. Bonferroni Pair$wise Comparison of Participants’ Performance on Absence of
Wh$Words in Clause$initialPosition in Different Clauses in the 4th GJ Test
Wh$Q of different GF
Accuracy Rate
A
EC of Wh$Q of Xiang zhidao
45.71%
SD
1.65
A
EC of Wh$Q of Juede
50.40%
1.72
B
Simple wh$Q
82.54%
1.18
B
MC of Wh$Q of Juede
85.71%
1.23
Note: The same letter indicates that the difference is not statistically significant at 0.05.
The individual results indicate that there are almost as many participants who had
above 80% accuracy rate in simple wh$questions as those for matrix clauses of wh$
questions of
regarding the grammatical feature that clause$initialposition is not
allowed for wh$words. There are more participants who had above 80% accuracy rate
regarding this grammatical feature in embedded clauses of wh$questions of
embedded clauses of wh$questions of
than in
.
Table 41. Number and Percentage of Participants Who Had Above 80% Accuracy Rate in
Ungrammatical Sentences of Simple Wh$Q in the 4th GJ Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
14 (66.67%)
Below 80%
7 (33.33%)
115
Table 42. Number and Percentage of Participants Who Had Above 80% Accuracy Rate in
Ungrammatical Sentences of Matrix Clause of Wh$Q of Juede in the 4th GJ Test
Accuracy Rate
Number of Participants (Percentage)
80% or above
16 (76.19%)
Below 80%
5 (23.81%)
Table 43. Number and Percentage of Participants Who Had Above 80% Accuracy Rate in
Ungrammatical Sentences of Embedded Clause of Wh$Q of Xiang zhidao
Accuracy Rate
Number of Participants (Percentage)
80% or above
6 (28.57%)
Below 80%
15 (71.43%)
Table 44. Number and Percentage of Participants Who Had Above 80% Accuracy Rate in
Ungrammatical Sentences in Embedded Clause of Wh$Q of Juede
Accuracy Rate
Number of Participants (Percentage)
80% or above
10 (47.62%)
Below 80%
11 (52.38%)
Thus the ANOVA test on OP and ungrammatical sentences of GJ data yielded the
same results regarding the developmental stages of participants’ knowledge of this
grammatical feature in these different clauses. Only when number of participants who
had above 80% accuray rate in ungrammatical sentences of GJ data was used as an
approach to examine the developmental stage, participants’ knowledge of this
grammatical feature in embedded clauses of wh$questions of
of wh$questions of
is shown not on a par.
and embedded clauses
116
Are participants who acquired the [$wh] feature strength in simple wh$questions
the same group of people as those who acquired the matrix [$wh] feature strength in wh$
questions of
? Did participants who learned that wh$words are not allowed in
embedded clause$initialposition of wh$questions of
stop putting wh$words in
embedded clause$initialposition of wh$questions of
) The questions will be
answered in the next section.
4.2.2.4. Major Research Question 2.4: Connection Between
Participants’ Performance on Different Clauses Regarding
the Absence of Wh$Words in Clause$Initial Position
4.2.2.4.1. The OP data
As shown in Tables 45 that summarizes descriptive statistics of the relationship
between participants’ performance on simple wh$questions and main clauses of wh$
questions of
regarding the grammatical feature of no wh$words in clause$
initialposition, among the 21 participants participating in the study, 13 participants
acquired this grammatical feature in both of them, and 5 participants acquired neither. As
shown in Table 46, 7 participants acquired this grammatical feature in both the embedded
clause of wh$questions of
, and 9 acquired neither.
and in embedded clauses of wh$questions of
117
Table 45. Consistency between Participants’ Acquisition of Absence of Wh$Words in
Clause$initialPosition in Simple Wh$Questions and Acquisition of that in Main Clauses
of Wh$Questions of Juede in the 4Th OP Test
Did not acquire
Acquired simple Wh$Q
simple Wh$Q
Acquired MC of Wh$Q of -
13
1
2
5
Did not aquire MC of Wh$Q of -
Note: “The absence of wh$words in clause$initial position” is absent in the row and
column headings for brevity.
Table 46. Consistency between Participants’ Acquisition of Absence of Wh$Words in
Clause$initialPosition in Embedded Clauses of Wh$Q of Xiang zhidao and Acquisition of
that in Embedded Clauses of Wh$Questions of Juede in the 4Th OP Test
Acquired MC of Wh$Q of -
Acquired Wh$Q of
Did not Acquire Wh$Q
Xiang zhidao
of Xiang zhidao
7
2
3
9
Did not aquire MC of Wh$Q of -
Note: “The absence of wh$words in clause$initial position” is absent in the row and
column headings for brevity.
The tetrachoric correlation test on participants’ performance on various kinds of
clauses regarding the absence of wh$words in clause$initialposition yields that
participants’ performance on simple wh$questions has the highest correlation with the
118
main clauses of wh$questions of
regarding this grammatical feature in comparison
with participants’ performance on simple wh$questions with that in other clauses (r =
0.82, p < 0.05, see Table 47). There is a medium$high correlation between their
performance on embedded clauses of wh$questions of
clauses of wh$questions of
and embedded
(r = 0.74, p < 0.05). Surprisingly, the correlation
between their performance on embedded clauses of wh$questions of
matrix clauses of wh$questions of
and
is very high (r = 1.00, p < 0.05).
Table 47. Correlation Between Participants’ Performance on Different Clauses Regarding
the Absence of Wh$Word in Clause$Initial Position in the 4Th OP Test
Simple Wh$Q
Simple Wh$Q
EC of Wh$Q
of Xiang
zhidao
MC of Wh$Q
of Juede
EC of Wh$
Q of Juede
1.00
EC of Wh$Q of Xiang
zhidao
1.00 (0.09)
1.00
MC of Wh$Q of Juede
0.82* (0.03)
1.00* (0.04)
EC of Wh$Q of Juede
1.00 (0.10)
0.74* (0.03)
1.00* (0.05)
1
Note: Number in parenthesis is the corresponding p value. *p < 0.05
4.2.2.4.2. The GJ Data
Only participants’ judgment of ungrammatical sentences was used for the same
reason as in section 4.2.2.3.2. As shown in Tables 48 that summarizes descriptive
statistics of the relationship between participants’ performance on simple wh$questions
and main clauses of wh$questions of
regarding the grammatical feature of no wh$
words in clause$initialposition, among the 21 participants of this study, fourteen acquired
this grammatical feature in both of them, and 4 participants acquired neither. As shown in
Table 49, five participants acquired this grammatical feature in both the embedded clause
119
of wh$questions of
and in embedded clauses of wh$questions of
, and
10 acquired neither.
Table 48. Consistency between Participants’ 80% Accuracy Rate in Simple Wh$
Questions and in Main Clauses of Wh$Questions of Juede Regarding the Absence of Wh$
Words in Clause$initialPosition in the 4Th GJ Test
Had 80% accuracy rate in MC of Wh$
Q of Did not have 80% accuracy rate in MC
of Wh$Q of -
Had 80% accuracy rate
in simple Wh$Q
Did not have 80%
accuracy rate in
simple Wh$Q
14
3
0
4
Note: “The absence of wh$words in clause$initial position” is absent in the row and
column headings for brevity.
Table 49. Consistency between Participants’ 80% Accuracy Rate in Embedded Clauses
of Wh$Q of Juede and in Embedded Clauses of Wh$Q of Xiang zhidao Regarding the
Absence of Wh$Words in Clause$initialPosition in the 4Th GJ Test
Had 80% accuracy rate in EC of Wh$Q
of Did not have 80% accuracy rate in EC
of Wh$Q of -
Had 80% accuracy rate
in Wh$Q of Xiang
zhidao
Did not have 80%
accuracy rate in Wh$
Q of juede
5
5
1
10
Note: “The absence of wh$words in clause$initial position” is absent in the row and
column headings for brevity.
The tetrachoric correlation test on participants’ performance on various kinds of
clauses regarding the absence of wh$words in clause$initialposition yields that
participants’ performance on simple wh$questions and in main clauses of wh$questions of
regarding this grammatical feature has a high correlation correlation (r = 1.00, p <
0.00). Surprisingly, there is also a high correlation between participants’ performance on
120
simple wh$questions and embedded clauses of wh$questions of
regarding
this grammatical feature. There is no correlation between their performance on embedded
clauses of wh$questions of
and in embedded clauses of wh$questions of
(r = 0.70, p > 0.05).
Table 50. Correlation Between Participants’ Performance on Different Clauses Regarding
the Absence of Wh$Word in Clause$Initial Position in the Ungrammatical Sentences in
the 4Th GJ Test
EC of Wh$Q
Simple Wh$Q
MC of Wh$Q
EC of Wh$
of Juede
Q of Juede
of Xiang
zhidao
Simple Wh$Q
1.00
EC of Wh$Q of Xiang
1.00 (0.06)
1.00
MC of Wh$Q of Juede
1.00** (0.00)
1.00 (0.28)
1.00
EC of Wh$Q of Juede
1.00** (0.00)
0.70 (0.06)
1.00 (0.09)
zhidao
1.00
Note: Number in parenthesis is the corresponding p value. **p < 0.01
The correlation tests on participants’ OP and GJ data did not yield exactly the
same results. What is consistent is that there is a high correlation between participants’
performance on simple wh$questions and main clauses of wh$questions of
regarding the grammatical feature that clause$initialposition is bad for wh$words. More
121
importantly, the fact that a higher correlation is found between their performance on
simple wh$questions and main clauses of wh$questions of
than between their
performance on simple wh$questions and other clauses indicates that participants very
much likely took these structures as the same concerning this grammatical feature, and
thus used the same strategy.
Participants’ performance on embedded clauses of wh$questions of
has a medium$high correlation with embedded clauses of
in the OP data, but no
correlation was found in the GJ data. There is no very strong evidence for the hypothesis
that participants used the same strategy for the embedded clauses of the two types of wh$
questions.
122
CHAPTER 5 DISCUSSION
This dissertation consists of two studies, which respectively explore L2 English$
speaking learners’ acquisition of simple wh$questions over their first year of Chinese
learning, and their acquisition of two types of complex wh$questions by the end of the
first year. The results of the studies yield some interesting research findings. This chapter
summarizes the research findings and discusses them in relation to the hypothesis of L2
acquisition of parameters and other second language acquisition theories and findings.
The limitations of the study and directions for future research will also be presented.
5.1. Findings on the Acquisition of the [$wh] Feature
Strength in Simple Wh$questions
5.1.1. Difficulty Hierarchy of Wh$questions with Wh$
words Playing the Role of Different Grammatical
Functions
Study 1 first explored the answers to the question of whether participants’
knowledge of [$wh] feature strength is extended to all grammatical functions at the same
time, or whether there is a difficulty hierarchy of different grammatical functions
regarding this feature strength, as shown by a large number of L1 studies and a smaller
body of L2 research. If it is the latter, what is the difficulty hierarchy?
The oral production (OP) and grammaticality judgment(GJ) tests results indicate
that different grammatical functions that wh$words play in the questions did not affect
learners’ acquisition of the [$wh] feature strength substantially, and its influence was only
obvious at a very early stage as shown in the OP Test. The evidence is that learners’
performance on wh$questions with wh$words playing different grammatical functions
(wh$questions of different grammatical functions hereafter) was generally the same
123
except in OP Test 2. In this test, participants’ performance on object wh$questions was
statistically significantly better than on attributive wh$questions. Their performance on
adverbial wh$questions and object wh$questions was not significantly different from each
other, and their performance on adverbial wh$questions and attributive wh$questions was
not statistically different from each other.
With respect to the GJ data, the data on judgment of grammatical sentences from
the GJ tests does not reveal a significant difference between participants’ performance on
wh$questions of different grammatical functions in any test. The only exception is in the
data of their judgment of ungrammatical sentence from GJ Test 4, where learners’
performance on attributive wh$questions was not as good as on object and adverbial wh$
questions. The lower scores stem from an unusual drop in accuracy rate on attributive
wh$questions in Test 4. This unusual development of attributive wh$questions is most
likely due to the effect of a single test item, which is discussed below. Based on the
above observation, the difficulty hierarchy is adverbial wh$questions = object wh$
questions; object wh$questions = attributive questions; and attributive wh$questions are
more difficult than adverbial wh$questions for the learners.
5.1.1.1. Troublesome Attributive Wh$Questions
The OP and GJ tests both indicate that participants had more difficulty with
attributive wh$questions at a certain point.
The fact that there was no statistically significant difference between attributive
wh$questions and object wh$questions in OP Test 1 but there was between participants’
performance on attributive wh$questions and object wh$questions in OP Test 2 could be
due to the slower increase in accuracy rate of attributive wh$questions (63.10% to
68.57%) than that on object wh$questions from Test 1 to Test 2 (77.38% to 88.10%). In
other words, participants did not improve as fast on attributive wh$questions as on object
wh$questions. This finding can be taken as evidence for the greater difficulty that
124
attributive wh$questions pose to the learners at the beginning stage of the learning
process in comparison with object wh$questions.
The statistically significant difference in performance between attributive wh$
questions and object and adverbial wh$questions in data on participants’ judgment of the
ungrammatical sentences in Test 4 was a result of both the substantial increase in
accuracy rates of object wh$questions and adverbial wh$questions from Test 3 to Test 4
(85.71% to 95.24% for object wh$questions, and 71.43% to 88.10% for adverbial wh$
questions) and the substantial decrease in accuracy rate of attributive wh$questions from
Test 3 to Test 4 (77.38% to 64.29%). This developmental pattern of attributive wh$
questions is unusual, especially compared with its developmental pattern as revealed
through the data on participants’ judgment of grammatical sentences and through OP
data. Thus the test items of the ungrammatical attributive wh$questions and participants’
responses in GJ tests were examined again. It was found that in GJ Test 4 one of the test
items for ungrammatical attributive wh$questions, as shown in (42), does not give as
salient clue to the position of wh$word as other ungrammatical test item. In (42), there is
an adverbial
(totally) before
(how many), whereas in all of the other
ungrammatical test items, wh$words are placed at the very beginning of the sentences.
Participants could have mistaken this sentence as a sentence with an in$situ wh$word. The
accuracy rate on this test item is very low (52.38%) as compared to that of the other test
item for attributive questions (80.95%).
(42)
Ungrammatical sentence for
Yigong jige
ren
(how many)
ni kanjian$le?
Totally how many people you see$competion aspect
How many people did you see altogether?
Although this is only a possibility, participants’ low performance on attributive
wh$questions in GJ Test 4 should be interpreted with caution. There is reliable evidence
that attributive wh$questions pose more difficulty for learners than object wh$questions
125
only on the OP test, and this happens at a very early stage in the learning process, in Test
2.
Could the greater difficulty posed by the attributive wh$questions in OP Test 2 or
GJ Test 4 be due to the instructional order? It is unlikely to be so, because participants
had learned the formation rule of Chinese wh$questions along with wh$words acting as
all three different grammatical functions by the time they took Test 1. In addition, all of
the wh$words acting as attributives were introduced before they took Test 2, as shown in
Table 51. Two of them were introduced before they took Test 1 and the other three were
introduced before they took Test 2.
Table 51. The Introduction Order of Wh$Words Tested in the Four OP and GJ Tests
GF
OP Test 1
GJ Test 1
OP Test 2
GJ Test 2
O
shenme
(what)
shui
(who)
A
shenme
(what)
ji
(how many)
na
(which)
duoshao
(how much)
duoda
(how big)
Adv
nar
(where)
weishenme
(why)
genshui
(with whom)
zenme
(how)
zenmeyang
(how)
OPTest 3
GJ Test 3
OP Test 4
GJ Test 4
duojiu
(how long)
Could the difficulty hierarchy result from the amount of input that learners were
exposed to? Only the input from the textbook could be tracked, which is summarized in
126
Table 52. The number of attributive wh$questions that the learners were exposed to
before they took Test 2 and Test 4 were actually more than that of object and adverbial
wh$questions. For example, in Test 4, there were 14 attributive wh$questions in the
textbook but there were only one object wh$question and five adverbial wh$questions.
This consequently eliminates the possibility that attributive wh$questions were more
difficult for the learners because they could not see examples of this pattern in the
textbook as much as they did for object and attributive wh$questions. However, there was
no way to know the amount of input of these different kinds of wh$questions that the
learners might have obtained in classroom settings.
Table 52. Number of Wh$Questions of Different Grammatical Functions to which
Learners Were Exposed in the Textbook
Test 1
Test 2
Test 3
Test 4
O
6
3
2
1
A
4
5
4
14
Adv
2
2
5
5
5.1.1.2. Argument and Adjunct Asymmetry
Neither the GJ data nor the OP data indicate that object wh$questions and
adverbial wh$questions posed different degrees of difficulty for the learners even at the
beginning stage of learning. This finding contradicts the results reported in previous
studies, and it also does not agree with either the predictions of the theories proposed by
Stromswold (1995) or O’Grady (1997).
The possible reason could be that the previous studies were mostly either L1
research (Blount, 1972; Clancy, 1989; Ervin$Tripp, 1970; H. T. Lee, 1989; S. Lee, 2004;
127
Merkin, Wotten, & Bloom, 1980; Santelmann, 2003) or L2 research on nonmovement
language learners’ acquisition of L2 movement languages (Lee, 2004; Saunders, 1983),
while this study examined adult English$speaking learners’ L2 acquisition of in$situ wh$
questions.
One explanation of the L1 acquisitional order of wh$questions comes from the
point of view of children’s cognitive development (Clancy, 1989). Children’s
comprehension and production of wh$questions is constrained by their cognitive
maturity. When children begin to explore the world, people and things in their immediate
surrounding are what children first understand and are curious to find out. Manners and
reason of events are what they will understand later. Thus
and
wh$questions,
which can be subject or object wh$questions, were shown to be acquired earlier than
adverbial wh$questions, such as
and
wh$questions. The only adverbial wh$
questions that have been acquired earlier and acquired at the same time as
wh$questions are
and
wh$questions (Blount, 1972; Ervin$Tripp, 1970; H. T. Lee, 1989;
Merkin, Wotten, & Bloom, 1980; Santelmann, 2003). However, as H. T. Lee (1989)
reported, only questions about location in the immediate environment were
comprehended earlier (1; 7), and non$here$and$now locative questions were
comprehended later (2; 0), which once again indicates the influence of children’s
cognitive development on the acquisition order of wh$questions.
There is also an account from the perspective of syntactic structure. Merkin,
Wotten, and Bloom (1980) proposed that what, where, and who wh$questions occur
before why, how, and when wh$questions since what, where, and who are wh$
pronominals that stand for major sentence constituents and can be combined more easily
with auxiliary verbs. In contrast, why, how, and when are wh$adverbials that do not
replace major sentence constituents, and they are used with descriptive verbs primarily
and thus are acquired later. Based on the observation that children are more accurate in
auxiliary inversion in object wh$questions than in adjunct wh$questions, under the
128
Government and Binding framework, Stromswold (1990) proposed that the government
of object trace is more local than the government of adverbial trace. Stromswold further
argued that adjunct wh$words at the beginning stage can be adjoined to IP, which does
not trigger inversion, as shown in (43) b, or moved to Spec, CP in children’s grammar, as
shown in (43) a, which does trigger inversion. In contrast, argument wh$words are moved
to Spec, CP only. The reason for this difference is that adjunct trace within IP is not as
salient as object trace.
(43) a. Inverted argument /adjunct wh$questions
CP
Who/why
C’
C
IP
Spec
I’
I
VP
Aux Subj
Who is she pushing?/Why is she smiling?
b. Uninverted adjunct wh$questions
IP
C
IP
Spec
I’
I
Subj Aux
Why she is smiling?
VP
129
Unlike L1 child learners, the adult English$speaking L2 learners of Chinese under
investigation in this dissertation had passed the stage of learning the concepts and
semantics associated with argument and adjunct wh$questions. In addition, since Chinese
has in$situ wh$questions, the learners did not need to establish new CP projections for
wh$words and auxiliaries. Thus they did not need to overcome the pitfall of creating a
wrong projection for adverbial wh$words as L1 and L2 learners of movement languages
do. There are two possibilities for what is involved in English$speaking learners’
acquisition of Chinese wh$questions, depending on the initial stage of learners’ L2
grammar that is assumed. One is that the learners start from the English setting as
predicted from the Full Transfer Full Access (FTFA) Hypothesis, and thus need to move
the wh$words in Spec, CP to in$situ positions, which already exist in their grammar, as
shown in (44). It can be hypothesized that the differences in the distance between Spec,
CP to adverbial, to object and adjectival positions in canonical sentence order does not
affect their learning so much as to make one of them significantly more difficult than the
other, since these positions are already well established in their grammar. Attributive wh$
questions are more difficult since their in$situ positions are embedded the deepest, but
that stage was soon overcome by the learners. The other possibility is that learners start
from the default setting of [$wh] feature strength as predicted by Minimal Initial Syntax
(MIS) hypothesis, and thus no movement is involved in the learning at all, which leads to
equal difficulty of wh$questions of different grammatical functions.
130
(44)
CP
Spec
IP
VP
V’
Adv
V’’
V
O
Adj
O’
5.1.2. The development of the [$wh] Feature Strength, Full
Transfer and Full Access Hypothesis, and Minimal Initial
Syntax Hypothesis
The next question that study 1 explored is which hypothesis can better describe
the initial state of the [$wh] feature strength in learners’ L2 grammar and what is the
developmental path of this feature strength in their first year of Chinese language
learning. This study shows that the initial stage of learners’ L2 grammar displays very
little presence of [+wh] feature strength but there is not a successful acquisition of [$wh]
feature strength yet. The later development for the whole group can be characterized as a
slow progress in completely deleting the [+wh] feature strength from their interlanguage
and gaining the [$wh] feature strength in their interlanguage.
131
5.1.2.1. The Initial Stage: MIS vs. FTFA
Although participants’ acquisition of the [$wh] feature strength was traced as
early as the sixth week of their Chinese learning, there was no evidence for participants
starting out with the [+wh] feature strength influenced by their L1 English as predicted by
the FTFA hypothesis. For example, in OP Test 1, participants had as high as a 71.42%
accuracy rate in producing the [$wh] feature strength, and the rate of producing the [+wh]
feature strength was only 5.16%.
The early emergence of [+wh] feature strength is consistent with what has been
reported in Kim’s (1995) study and Yuan’s (2007) study. Can this be accounted for by
the MIS hypothesis, as has been done Kim and Yuan? This author does not consider MIS
a very convincing hypothesis for the following two reasons. One is that if participants
had started with the [$wh] feature strength, but not the [+wh] feature strength, then the
[+wh] feature strength in their L2 grammar should become less and less in the following
tests. On the contrary, the rate of [+wh] feature strength in the following tests was greater
than that in Test 1 and reached a peak in Test 3 (6.08% for Test 2, 15.87% for Test 3, and
7.93% for Test 4). The second reason is that the low occurrence rate of [+wh] feature
strength can be due to participants’ general low production ability at the beginning stage
of learning Chinese, which may have prevented them from producing sentences showing
their acquisition of [+wh] feature strength. The rate of participants’ zero or incomplete
production was 10.43% in Test 1, the highest among all of the tests.
The phenomenon that the initial state of participants’ interlanguage often does not
manifest obvious effects of L1 setting and reveals an early emergence of L2 properties
has been reported by other research on the L2 acquisition of parameter settings (White,
1992). For example, White (1992) examined the acquisition of three properties associated
with the verb movement parameter by L1 French learners of English. For wh$questions,
one of the surface properties associated with this parameter, beginner French$speaking
children of English produced only 0.17% of them in L1 French word order in an elicited
132
oral production task; in a preference task they wrongly accepted wh$questions in French
order only 14% of the time. For the position of negative verbs and adverbs, the other two
properties associated with this parameter, participants’ rates for wrongly accepting the
sentences mirroring the L1 French order were 15% and 77%, respectively.
The MIS hypothesis, which claims that L2 learners always assume the default
weak feature strength, cannot explain the high acceptance rate of the wrong adverbial
position in relation to the verbs mirroring the L1 situation. It also cannot explain why in
Parodi, Schwartz, and Clahsen (1997), the beginning/mid L1 romance language learners
of L2 German did not assume a weak N$feature, but produced the strong N$feature as
much as 20% of the time, as a result of L1 influence. However, if FTFA is assumed, then
what is puzzling to researchers is why some parameters reveal a very early emergence of
L2 properties in participants’ interlanguage while others do not, and that even the
properties associated with one parameter did not show the same degree of L1 influence
(White, 2003).
For this study, perhaps a similar explanation as in section 5.1.1.2 can be adopted.
Learning Chinese wh$questions by English$speaking learners involve the movement of
wh$words back to an established position in L2 learners’ grammar, which would be easier
than establishing CPs to hold the wh$words and the auxiliary verbs as done by learners
who L1s are in$situ languages when they learn English wh$questions. As a result, [$wh]
feature strength emerged very early in this study. Perhaps the initial state as proposed by
the FTFA hypothesis should also be modified as a L2 grammar showing the L1 influence
of various degrees, but not necessarily the L1 setting.
133
5.1.2.2. The Developmental Path of the [$wh] Feature
Strength
5.1.2.2.1. U$Shaped Learning Curve
What is interesting in the course of the development of [$wh] feature strength is
that the course of learning is not linear. The participants made progress during the period
between Test 1 and Test 2, then seemed to regress in Test 3, and then progressed again
from Test 3 to Test 4. For example, in the OP data, the accuracy rate increased from
71.42% in Test 1 to 80.20% in Test 2, dropped to 71.13% in Test 3, and increased to
83.33% again in Test 4. Another piece of evidence is that [+wh] feature strength, a
manifestation of the L1 English setting, did not suddenly diminish. As can be seen from
Table 22 (See in chap. 4, p. 83), it increased to its maximum to 15.87% in Test 3, and
then dropped in Test 4 to a similar level as in Test 1. The GJ data reveals a similar trend,
as can be seen from Figure 3 (See in ch. 4, p. 86).
This kind of unsteady development of a grammatical feature has been reported in
other L2 studies (Ke, 2005; Kellerman, 1983; Lightbown, 1983; Sjoholm, 1989; Wode,
Bahns, Bodey, & Frank, 1978). The learning course characterized by a sharp increase in
accuracy rate at the beginning, a dip somewhere in the middle and a rise in the end shared
by results of these studies is called U$curve. Lightbown (1983) attributed U$shaped
learning curve to a process whereby learners have mastered some forms and then
restructure the whole system when encountering some new ones. McLaughlin (1990)
pointed out that the U$shape is associated with restructuring: substitution of more
complex internal representation for less complex ones.
For example, in Lightbown’s (1983) study, in which French learners of English
were examined for their acquisition of present continuous tense, learners were reported to
correctly produce the present continuous in a picture$elicitation task in stage I, followed
by a production of simple present tense when present continuous tense should be used in
134
stage II, and then the correct present continuous tense again in stage III. Lightbown said
that after the simple present tense was introduced, learners had to redefine the limits of
the present continuous they had overgeneralized and learn to distinguish when to use
which.
Ke’s study (2005) is a cross$sectional study of English$speaking learners’
acquisition of 19 major Chinese structures over four instructional levels. He used
participants’ simulated oral proficiency interview (SOPI) data. Three acquisition patterns
were found: linear, U$shape, and plateau. The structures that display a U$shaped learning
curve are the Ba$contruction, the Bei$contruction, the aspect marker Guo, verb
duplication, resultative complements, potential complements, passives, relative clauses,
time position, and topic comment structure. Though he could not explain why there is a
U$shaped learning curve for all of the structures, Ke explained that some of them, such as
the Ba$construction, resulative complements, potential complements, Bei$construction,
and passive structures, share some semantic and syntactic similarities. Semantically, they
all express the results of actions. Syntactically, some of these structures can be embedded
in others. Thus the temporary drop in accuracy rate in the middle of the learning process
can be caused by learners’ relearning the boundaries of these similar patterns, which led
to a better command of them at a later stage.
If the tests in study 1 of this dissertation are assumed to have the same difficulty
level and participants’ lower performance in Test 3 was not the result of higher difficulty
level of that test, then what could have led to the drop of accuracy rate in Test 3 and what
restructuring process were learners going through?
It could be that the adverbial wh$questions with omitted subjects that learners
obtained from the input of the classroom gave some of the learners a wrong clue that wh$
words could go at the beginning of wh$questions, although the textbook that the learners
used did not have a single sentence like this. Unlike English, Chinese subjects are not
obligatory and can be dropped as long as they can be understood from the context.
135
Adverbial wh$words would appear at the beginning of wh$questions when subjects are
dropped, as shown in (45) a, b, and c. Learners could have misanalyzed (45) as (46),
when their knowledge of wh$words in relation to the other elements is not stable yet.
Why did sentences like (45) not affect these learners earlier? These sentences
would only appear when the communication went beyond sentence level, since subject
omission happens when its identity is established in the previous sentence. Learners at the
beginning level did not have a chance to encounter these sentences when the focus of
training was at the sentence level. However, since classroom input was not tracked, the
above explanation is only a possibility.
(45) a. (Ni) shenme shihou qu?
you when
go
When are you going?
b. (Women) zenme cai
We
how
neng xuehao zhongwen?
condition adv
can learn well Chinese
How can we learn Chinese well?
c.(Ta) weishenme mei lai?
he why not come?
Why did he not come?
(46) a. Shenme (ni) shihou qu?
you
when
go
When are you going?
b. zenme (women) cai
how we
condition adv
neng xuehao zhongwen?
can learn well Chinese
How can we learn Chinese well?
c. weishenme (ta) mei lai?
why
he not come?
Why did he not come?
136
5.1.2.2.2. Progress or Not
Despite the drop and rise in accuracy rate over the four tests, an ANOVA test
does not indicate statistically significant differences between any of the tests (See Tables
11, 18, and 20 in chap. 4, pp 76$80). Does this finding suggest that participants did not
make any progress from Test 1 to Test4? First of all, due to the small sample size of the
participants (= 21) in the study, this conclusion should be verified in future studies with a
larger sample. Furthermore, as reported in the discussion of individual results in Chapter
4, more participants were joining in the group who had above 80% accuracy rate starting
in Test 2. Thus it is safe to say that although statistically speaking participants as a whole
group were not better in Test 4 than in Test 1, there was evidence for their making
progress at the individual level.
5.1.3. Acquisition or Not: Group Results vs. Individual
Results
Now, it is important to consider the following question: Do participants succeed
in acquiring [$wh] feature strength by the end of the first year of Chinese study? If so,
when? The group results of the OP and GJ data yielded similar answers if the above$80%
accuracy rate in two consecutive tests is taken as the criterion for acquisition: the
participants as a group had not actually acquired [$wh] feature strength at the time they
took Test 2 since their accuracy rate dropped to below 80% in Test 3. Although
participants’ accuracy rate for the [$wh] feature strength rose to above 80% in the OP
Test 4 and close to 80% in the GJ Test 4, by the same line of reasoning, there is a
likelihood that they acquired the feature, but further evidence is needed.
When the individual results were examined, it was found that participants did not
all perform at around the average accuracy rate of the group as a whole, and that there
were participants who acquired the [$wh] feature strength any stage of the learning
process when 80% accuracy rate in consecutive tests up to Test 4 is used as criterion of
137
acquisition (See Figure 4 in ch.4, p. 88). For example, according to the OP data, 28.57 %
of the participants acquired it as early as in Test 1, 19.05% of them acquired it in Test 2,
and 4.76% of them acquired in Test 3. Finally, 19.05% of the participants had above 80%
accuracy rate in Test 4. Thus altogether 71.41% of the participants were very likely to
have acquired the [$wh] feature strength by the end of first$year of learning, among
whom 52.38% of the participants have definitely acquired it, and 19.05% of the
participants might have acquired it. According to the GJ data, only 28.57% of the
participants were very likely to have acquired it, among whom 23.81% definitely
succeeded in resetting the [+wh] feature strength to [$wh]. This study is another one
convincing us that only by examining group results and individual results alongside each
other can we get a full picture of the reality of the acquisition.
That there were fewer participants who acquired the [$wh] feature strength in GJ
tests than in OP tests looks puzzling at first sight, since participants usually perform
better on perception or comprehension tasks than on production tasks. This unusual result
in the study is due to the criterion of acquisition in the GJ data that was adopted: Only
when participants had 80% accuracy rate in both the grammatical sentences and
ungrammatical sentences are they regarded as having successfully acquired the [$wh]
feature strength. Thus although the accuracy rates for grammatical sentences,
ungrammatical sentences, and oral production in each time period did not vary much,
there is a big discrepancy between the number of participants who acquired the [$wh]
feature strength according to the OP data and according to the GJ data.
138
5.2. Findings on the Acquisition of the [$wh] Feature
Strength in Complex Wh$questions
5.2.1. Initial State of the [wh] Feature Strength Necessary
for Acquisition of Wh$Questions of %
questions of -
and Wh$
and Learning Results as Compared to
That of Simple Wh$Questions
In this section, findings on the acquisition of complex wh$questions, which
consist of indirect questions (i.e., wh$questions of
object complements (i.e., wh$questions of
questions of
), and wh$questions in
), are going to be presented. For wh$
, learners need to reset the embedded [+wh] feature strength to
[$wh]. It is found that the initial stage of participants’ L2 grammar showed their L1
English influence, according to both the OP data and the GJ data. For example, there was
a 21.90% occurrence rate for the embedded [+wh] feature strength in the OP data.
For wh$questions of
, all that learners need to do is to reset the matrix [+wh]
feature strength to [$wh] feature strength. It is found that participants’ L2 grammar did
not show as much L1 English influence, as indicated by a low 4.76% occurrence rate for
matrix [+wh] feature strength in the OP data. The rejection rate of the ungrammatical
sentences in the GJ test is consistent with the result of the OP test. However, the major
type of error the learners made in both the OP and GJ tests is questions with wh$word in
clause$initialposition of the embedded clause, which does not manifest either the L1
setting or the L2 setting. For example, its occurrence rate in the OP test is 13.33%.
How did learners perform on wh$questions of
and wh$questions of
as compared to their performance on simple wh$questions? The ANOVA test
indicates that in the OP tests, learners’ performance on simple wh$questions is
significantly better than on wh$questions of
and wh$questions of
. In
the data on their judgment of grammatical sentences of the GJ test, learners’ performance
139
on wh$questions of
patterns together with that on simple wh$questions, both of
which are close to the criterion for successful acquisition (76.98% and 78.10%,
respectively). In the data on their judgment of ungrammatical sentences of the GJ test,
learners’ performance on wh$questions of
manifests a pattern similar to their
performance on wh$questions of
questions of
. Overall, learners’ performance on wh$
patterns with that on wh$questions of
, both of which are
inferior to their performance on simple wh$questions at the end of their first year of
Chinese learning.
5.2.2.1. Discrepancy between Judgments of Grammatical
Sentences and Ungrammatial Sentences
For wh$questions of
, there is a big discrepancy between learners’ judgments
of grammatical and ungrammatical sentences. They accepted the grammatical sentences
of wh$questions of
, exemplified by sentence (47) a, close to full mastery (78.09%),
while they could only reject ungrammatical sentences at a chance level (50.48%), as
exemplified by sentence (47) b. This phenomenon suggests that the participants could
accept matrix [$wh] and also the zero embedded [wh] feature. They could not, however,
completely reject the embedded [+wh].
(47) a. Ni juede ta zai fangzili dai$le
?
You feel he in house$inside stay$Perf how long
How long do you think that he stayed in the house?
b. *Ni juede
ta zai fangzili dai$le?
You feel how long he in house$inside stay$Perf
How long do you think that he stayed in the house?
Interestingly, for wh$questions of
, participants accepted the
grammatical sentences, as exemplified by sentence (48) a, at a slightly higher than the
chance level (61.90%). They rejected the ungrammatical sentences, as exemplified by
140
sentence (48) b, at around chance level (45.71%). It seems to show that participants could
not accept the embedded [$wh]. Neither could they reject the embedded [+wh].
(48) a. Wo xiang zhidao ta zai fangzili dai$le
I want
?
know he in house$inside stay$Perf how long
I wonder how long he stayed in the house.
b. *Wo xiang zhidao
ta zai fangzili dai$le?
I want know how long he in house$inside stay$Perf
I wonder how long stayed in the house.
5.2.2.2. Embedded [+wh] in Wh$Questions of Juede
What looks puzzling at first sight is why there are participants who produced and
accepted a medial [+wh] feature for wh$questions of
at all, since this is not allowed
in either their L1 English or their L2 Chinese. Does this indicate that learners’ L2
grammar is not UG$constrained? Is there a language in the world that has medial [+wh]
for wide scope wh$questions so that this error that learners made is not wild but falls into
a possibility of a natural language?
It is then found that wide scope wh$questions with medial [+wh] are cross$
linguistically attested in languages such as Malay (49) and Madurese (50), and also in
languages such as German (51) and Romani (52). However, unlike Malay and Madurese,
wide scope wh$questions in German and Roman have a matrix [+wh] in addition to a
medial [+wh]. Based on these languages, L2 grammar of some of the participants in study
two of this dissertation is not wild but UG$constrained.
(49) Malay (Cole & Hermon, 1998, p. 225)
Ali memberitahu kamu tadi [CP apai (yang) [IP Fatimah baca ti]]?
Ali told
you just now what that
Fatimah read
What did Ali tell you that Fatimah was reading?
(50) Madurese (Davies, 2003, p. 240)
141
Hasan a$bala dha’ Atin [CP apai [IP se e$baca
H
AV$say to
A what
Siti ti]]?
REL ov$read S
What did Hasan tell Atin that Siti read?
(51) German (Thornton, 1990, p. 216)
Wasi glaubt Hans [[mit wemi [Jakob jetzt ti spricht]]?
What believes Hans with who Jacob now talk to
With whom does Hans believe that Jacob is now talking?
(52) Romani (Thornton, 1990, p. 216)
Soi O Demiri mislinol [CP kasi [IP i Afrifa dikhla ti]]?
What does Demir think
whom
Afrifa saw
Whom does Demir think Afrifa saw?
The immediate question is why the participants in this study produce wide$scope
wh$questions resembling Malay and Madurese but not German and Romani.
Simple wh$questions in Malay and Madurese are either in$situ or fronted14. Wide$
scope wh$questions can either resemble English wide$scope wh$questions, or have an
embedded [+wh] like (49) and (50) above. In contrast, simple wh$questions in German
and Romani are fronted. Wide$scope wh$questions in these languages can either resemble
the English setting, or have both matrix and embedded [+wh] like (51) and (52). The
hypothesis is that if the knowledge of the relationship of the parameter settings between
simple wh$questions and wide$scope wh$questions is preinstalled in learners’ L2
grammar, then when learners realize that they are learning a movement language, they
would be programmed to produce either wide$scope wh$questions like in the target
language or questions mirroring German (51) and Romani (52). The latter is exactly what
some of the participants in Wakabayashi and Okawara’s study (2003) did. In their study,
14
The fronted Madurese wh$questions are clefted.
142
which examined Japanese university students learning L2 English, 12.5% of the
participants were reported to have produced wide$scope wh$questions resembling
German and Romani wide$scope wh$questions.
On the other hand, if learners realize that a language allows in$situ and fronted
wh$questions, they would either produce wide$scope wh$questions like in the target
language or questions mirroring Malay (49) and Madurese (50). This has been borne out
by Oiry and Demirdache’s (2006) study on L1 acquisition of French wide$scope wh$
questions. French is a language that allows both in$situ and fronted simple wh$questions.
In their study, 40% of the participants were reported to have produced wide scope wh$
questions mirroring Malay and Madurese, and its occurrence rate in the production data is
8%. Whether there exists a language that has both in$situ simple wh$questions and wide$
scope wh$questions like (49) and (50) is not certain to the author. However, what is for
certain is that errors like (51) and (52) are impossible since participants in study two of
this dissertation had a close to full mastery of in$situ simple wh$questions, and thus
would not choose the German setting for wide$scope wh$questions as L2 learners of
English did in Wakabayashi and Okawara’s study (2003).
5.2.2. Acquisition or Not
The results of this study show that participants as a group did not succeed in
acquiring either wh$questions of
or wh$questions of
, since the
accuracy rate for neither of them reached 80%. For example, in the OP tests, the accuracy
rate for wh$questions of
is 59.05%, and that for wh$questions of
is
60.95%. The results of the GJ tests were consistent with those of the OP tests.
Similar to what has been found in participants’ acquisition of simple wh$
questions, the group accuracy rates are not representative of every learner in the group.
Some participants successfully acquired it and some did not. In addition, there are almost
as many who acquired wh$questions of
(47.62%) as those who acquired wh$
143
questions of
(42.86%) in Test 4.
Participants’ performance on these two types of wh$questions in judgment of the
grammatical sentences did not show an obvious difference in terms of accuracy rate.
Neither was there a marked difference in their performance on the ungrammatical
sentences. However, when only 80% accuracy rate in both grammatical sentences and
ungrammatical sentences is regarded as a successful acquisition, it was shown that there
were fewer people who successfully acquired wh$questions of
than those who acquired wh$questions of
(23.80%)
(42.86%).
5.2.3. Learners’ Competence in Different Clauses
Regarding the Absence of Wh$Words in Clause$Initial
Position
The error of medial [+wh] feature strength in wh$questions of
in both OP
and GJ data indicates that L2 learners do more than just reset the matrix [wh], by which
English and Chinese wide scope wh$questions differ. They face two tasks: One is to learn
that wh$word is not allowed in matrix clause$initial position and the other is to learn that
wh$word in not allowed in embedded clause$initial position either. What was learners’
competence in this grammatical feature as measured by accuracy rate in the following
clauses: simple wh$questions, embedded clauses of wh$questions of
clauses of wh$questions of
, embedded clauses of wh$questions of
, matrix
?
The results of this study indicate that in the OP tests their performance on simple
wh$questions and matrix clauses of wh$questions of
patterned together, which were
statistically significantly better than their performance on wh$questions of embedded
clauses of wh$questions of
and embedded clauses of wh$questions of
Since grammatical sentences of wh$questions of
.
could not indicate specifically
participants’ performance on main clause or embedded clause, ungrammatical sentences
of all the above mentioned clauses were used for comparison for their performance on the
144
GJ tests. The GJ tests revealed a similar pattern.
Individual results in the OP tests indicate that as many participants acquired the
grammatical feature mentioned above in simple wh$questions as matrix clauses of wh$
questions of
(71.43%), and there were as many participants who acquired this
grammatical feature in embedded clauses of wh$questions of
embedded clauses of wh$questions of
as in
. Individual results in the GJ tests do not
indicate such a nice fit (see Tables 41, 42, 43, and 44 in chap. 4, pp. 115$116). Matrix
clauses of wh$questions of
were acquired by the greatest number of participants
(76.19%), regarding this grammatical feature, followed by simple wh$questions
(66.67%), followed by embedded clauses of wh$questions of
questions of
(47.62%), and by wh$
(28.57%).
Generally speaking, simple wh$questions and matrix clauses of wh$questions of
regarding this grammatical feature followed the same developmental pattern.
Embedded clauses of wh$questions of
questions of
and embedded clauses of wh$
also shared similarity in their developmental path.
5.2.3.1. Developmental Path of Wh$Questions of Both the OP and GJ data in Study 2 suggests that for wh$questions of
, the
developmental path is as follows: (a) first get rid of matrix [+wh] feature strength as in
(53) a; (b) then get rid of medial [+wh] feature strength as in (53) b, and (c) finally a
successful acquisition, as in (53) c. Another piece of evidence for this order of acquisition
is that participants who correctly rejected (53) b always rejected (53) a, but not vice
versa, as pointed out in Chapter 4.
(53) Wh$questions of a. *Shenme ni juede xuedao$le ne?
What you feel learn$le particle?
What did you feel you learned?
145
b. *Ni juede shenme ni xuedao$le ne?
You feel what
ni learn$le particle?
What did you feel you learned?
c. Ni juede xuedao$le shenme ne?
You feel learn$le what particle?
What did you feel you learned?
5.2.3.2. Nature of Wh$Words in Embedded Clauses of Wh$
Questions of %
.
and of Wh$Words in Embedded
Clauses of Wh$Questions of Since the ANOVA test revealed that learners’ performance on the embedded
clauses of the two types of wh$questions is no different from each other concerning the
grammatical feature that clause$initial position is bad for Chinese wh$words, the author
turned to a consideration of the nature of these two kinds of errors.
First of all, the participants were aware that the scope of the two types of
questions was different. For example, sentences in (54) were produced according to the
English translations in the OP tests. The embedded wh$words in wh$questions of
as shown by (54) a indicates that participants still used [+wh] feature strength in
the scope marking position as a result of L1 influence. The embedded wh$words in wh$
questions of
as shown by (54) b are not in a scope$marking position. As explained
in section 5.2.2, this kind of error does not come from L1 influence, but rather from UG.
Therefore, the nature of the medial [+wh] in the two types of complex wh$questions is
very different.
The similar accuracy rates for the two types of complex wh$questions regarding
the grammatical feature of no embedded wh$words in both OP data and the data on
participants’ judgment of ungrammatical sentences makes this author wonder if the
participants were using a similar strategy for the two types of wh$questions. In other
146
words, is it possible that some participants were using an English simple wh$question in
the complements of the two types of wh$questions despite the fact that the verbs are
different?
(54) a.*Wo xiangzhidao [CP [+wh] shenme difangi ni hai canguan$le ti.]
I
wonder
what
place you also visit$le.
I wondering what other place you visited.
b. [CP [$wh] i *Ni juede [CP [+wh] shenmei ni xuedao$le ti]]?
You feel what
ni learn$le particle?
What did you feel you learned?
5.2.4. Connection between Participants’ Performance on
Different Clauses Regarding the Absence of Wh$Words in
Clause$Initial Position
The GJ and OP data reveal that participants’ performance on [$wh] feature
strength of simple wh$questions has a high correlation with the [$wh] feature strength of
main clauses of wh$questions of
in both the OP tests (r = 0.83) and the GJ tests (r =
1.00). These results suggest that participants may use a similar strategy in learning the [$
wh] feature strength in the two types of clauses because of the similar syntactic
mechanism that these two clauses share. As for the embedded clauses of wh$questions of
and of wh$questions of
, a medium$high correlation existed in the OP
test (r = 0.74) but no correlation was found in the GJ data. Thus, there is no strong
evidence that the participants were using a similar strategy in acquiring the grammatical
feature that wh$words are not allowed in clause$initial position for embedded clauses of
the two types of wh$questions. This seems to suggest that if the underlying motivation for
wh$ in$situ for the embedded clauses are different, one being not in wh$ scope as in wh$
questions of
, the other being under control of [$wh] feature strength, participants
could be using different strategies to learn it. In other words, they might not be using a
147
strategy such as verb plus complement for wh$questions of
and wh$questions of
.
On the other hand, could the above discrepancy between the GJ and OP data be
caused by the GJ data not being representative of participants’ L2 linguistic knowledge?
As explained above, grammaticality judgment task has been more and more questioned
as a valid and reliable method to elicit L2 learners’ grammar (Ellis, 1991; Goss, Zhang, &
Lantolf, 1994; Davies & Kaplan, 1998). However, it is still hard to explain why
participants were using one strategy for wh$questions of
wh$questions of
and another strategy for
in the GJ task, the influence of which should be the same
upon the participants’ judgment of the test items for the two types of sentences.
5.3. Summary of the Findings
This dissertation consists of two studies investigating L2 Chinese learners’
acquisition of wh$questions. Study 1 reports that only attributive wh$questions posed
more difficulty to the participants than object wh$questions in OP test at the beginning
stage, but this difference soon diminished. Adverbial wh$questions and object wh$
questions were not found to be significantly different to the participants as reported in
some L1 (Blount, 1972; Clancy, 1989; Ervin$Tripp, 1970; H. T. Lee, 1989; S. Lee, 2004;
Merkin, Wotten, & Bloom, 1980; Santelmann, 2003) and L2 studies (Lee, 2004;
Saunders, 1983). This author attributed this discrepancy to the different syntactic
mechanism that L2 learners of movement language and L2 learners of in$situ language go
through. The initial stage of the [wh] feature strength did not display a clear presence of
[+wh] feature strength, and the development of the [$wh] feature strength in learners’ L2
grammar was not linear, but U$shaped. The individual results indicate that 71.43% of
participants acquired the [$wh] feature strength in OP Test 4, but the success rate was
only 28.57% in GJ Test 4 due to the stringent criterion for acquisition adopted for the GJ
test.
148
Study 2 reveals that learners’ performance on indirect questions and wh$questions
in object complements was not as good as their performance on simple wh$questions to a
statistically significant degree at the end of first year of Chinese learning. The initial
stage of indirect questions showed an obvious L1 influence. For wh$questions in object
complements, participants made a kind of error that manifests neither the L1 nor the L2
setting. This kind of error was found to fall into the possibility of a natural language,
indicating that L2 grammar is not wild but UG$constrained. Participants’ performance
regarding the grammatical feature that wh$words should not occur in clause$initial
position in simple wh$questions, indirect questions, or wh$questions in complements was
examined. The participants’ performance on simple wh$questions patterned together with
their performance on matrix clauses of wh$questions in complements, which was better
than their performance on embedded clauses of the two types of complex wh$questions.
Evidence for connection between their performance on simple wh$questions and matrix
clauses of wh$questions in complements regarding the grammatical feature were
revealed, but there was no strong evidence for connection between their performance on
embedded clauses of the two types of complex wh$questions regarding this grammatical
feature. It suggests that the participants might have used similar strategy in learning
structures of the same nature but different strategies in learning structures of a
fundamentally different nature.
5.4. Limitations of the Study
While reporting the findings from these two studies, this author is fully aware of
the limitations; therefore, the findings of studies needed to be applied to the field with
caution. Mainly, there are four limitations of the studies. The first limitation is the small
sample size. There were only 21 participants involved. As the F values reported in the
studies are associated with sample size, it is possible that the reported results did not
completely reflect the reality.
149
Another problem is that the number of object, attributive, and adverbial wh$
questions in the tests was not in proportion to each other in each test, and was not in the
same proportion from test to test. In designing the tests, wh$words were chosen out of the
consideration of whether they were key words with which a coherent dialog could be
made in the OP task, and also whether they had just been learned by the participants. This
led to the very small number of wh$questions of certain grammatical functions. For
example, in Test 2 and Test 4, there were only two items for nominal wh$words, but four
items for adjectival and adverbial wh$words. This means that an error in nominal wh$
words would weigh more than an error in adjectival or adverbial wh$words, which may
have affected the results in such a way that the real accuracy rates of the participants in
these wh$questions were not revealed.
The third problem is with the method of the GJ test. When the GJ data were
analyzed, a suspicion arose that the participants’ rejection of some of the grammatical
sentences could be due to their dislike of other grammatical functions or choice of words
within the sentence, but not the [$wh] feature strength. This author had thought of asking
the participants to circle the part they believed to be incorrect and to give an explanation.
However, asking the participants to explain what is wrong has its own problems. It can
intensify their conscious use of the grammar knowledge, since L2 learners were already
reported to use their conscious knowledge of the grammar (Davies & Kaplan, 1998), and
it can also take an even longer time, thus leading to participant fatigue. As a middle
ground, in future studies, maybe the participants could be asked to indicate which part in
a test sentence they think is wrong.
Last, in Test 4, in which participants’ performance on wh$questions of
and wh$questions of
was measured, their performance on complementation
in general should be measured as well. This way, it could tell whether the low production
rate of the correct target form could be due to their unfamiliarity with complementation in
150
general. Similarly, in GJ test, it could reduce the possibility that participants wrongly
reject because of their inability to parse the test items properly.
Nonetheless, the limitations of the studies do not override the meaningful
contribution that the studies have made to the field. These are the first studies that
examined L2 English$speaking learners’ acquisition of Chinese simple wh$questions
longitudinally over a one$year learning period, and their acquisition of indirect questions
and wh$questions in complement at the end of their first year of Chinese language
learning. This author believes that the results of these studies verify existing L2
acquisition theories in general and help construct a model of acquisition of wh$question
in Chinese in specific.
151
REFERENCES
Aoun, J. (1985).
Dordrecht: Foris.
/
0!
1
Aoun, J., & Li, Y. H. A. (1993). Wh$elements in situ: Syntax or LF?
24 (2), 199$238.
.
(
,
Beck, M.$L. (1998). L2 acquisition and obligatory head movement: English$speaking
learners of German and the local impairment hypothesis.
2 (
, 20, 311$348.
Bloom, L., Merkin, S., & Wootten, J. (1982). Wh$questions: the linguistic factors that
contribute to the sequence of acquisition. +
3
$
, 53 (4), 1084$1092.
Bromberg, H.S., & Wexler, K. (1995). Null subjects in child wh$questions. " !
# $
, 26, 221$247.
Cairns, H. S., & Hsu, J. R. (1978). Who, why, when and how: A development study.
+
, 5 (3), 477$488.
Clancy, P. M. (1989). Form and function in the acquisition of Korean wh$questions.
+
, 16 (2), 323$347.
Cheung, L. L.$S. (1992). 4
$
1(
. Unpublished doctoral
dissertation, Massachusetts Institute of Technology.
Cheung, A. S., & Lee, T. H. (1993). Cantonese speaking children’s comprehension of
wh$questions. In E. Clark (Ed.), ! #
56
+
7
*
(pp. 106$117). Chicago, IL: University of Chicago Press.
Chomsky, N. (1965). 2 $
. Cambridge, MA: MIT Press.
Chomsky, N. (1975). 7
. New York: Pantheon.
Chomsky, N. (1981).
Chomsky, N. (1986a).
/
. Dordrecht: Foris.
. Cambridge, MA: MIT Press.
Chomsky, N. (1986b). 8
Praeger.
0
. New York:
Chomsky, N. (1993). A minimalist program for linguistic theory. In K. Hale and S. J.
Keyser (Eds.), !
/
590 :
/
(pp. 1$52). Cambridge, MA: The MIT Press.
Chomsky, N. (1995). "
$
. Cambridge, MA: MIT Press.
Cole, P., & Hermon, G. (1998). The typology of wh$movement: Wh$questions in Malay.
, 1 (3), 221$258.
152
Corder, S. P. (1967). The significance of learners’ errors. In J. C. Richards (Ed.), :
0# $
(
. London: Longman (pp. 18$
27). London: Longman
Corder, S. P. (1973).
Books.
$$
. Harmondsworth, UK: Penguin
Davies, W. D. (2003). Extreme locality in Madurese wh$questions.
259.
, 6 (3), 237$
Davies, W. D., & Kaplan, T. I. (1998). Native speaker vs. L2 learner grammaticality
judgement. 2$$
, 19 (2), 183$203.
De Villiers, J. (1991). Why questions? In T. L. Maxfield, & B. Plunkett (Eds.), # $
2 (
0#
;"
7
/ , (pp. 155$173).
Amherst, MA: University of Massachusetts Occasional Papers
Ervin$Tripp, S. (1970). Discourse agreement: How children answer questions. In J. Hayes
(Ed.), +
$
(pp. 79$106). New York: Wiley.
Eubank, L. (1993). Sentence matching and processing in L2 development.
7
, 9, 253$280.
Eubank, L. (1994). Optionality and the initial state in L2 development. In T. Hoekstra &
B. D. Schwartz, (Eds.),
(
(pp.
369$388). Amsterdam: John Benjamins.
Eubank, L. (1996). Negation in early German$English interlanguage: More valueless
features in the L2 initial state.
7
, 12, 73$106.
Fahn, R. L. S. (2003). Chinese$speaking children’s production of wh$questions.
+
0
:
. 29 (2), 82$117. Retrieved
from http://www.eng.ntnu.edu.tw/concentric$linguistics/PDF/Vol.29/no2924.pdf.
Gess, R., & Herschensohn, J. (2001). Shifting the DP parameter: A study of anglophone
French L1ers. In C. R. Whilshire & J. Camps. (Eds.), 7
5 (
(pp. 105$119). Amsterdam: John Benjamins.
Grinstead, J. (2001). Wh$movement in child Catalan.
(1), 5$28.
2$$
, 12
Grinstead, J. (2004). Subjects and interface delay in child Spanish and Catalan.
, 80 (1), 40$72.
Gruter, T. (2006). Another take on the l2 initial state: evidence from comprehension in
L2 German.
2 (
, 13 (4), 287$317.
Hanna, K., & Wilhelm, A. (1992). On the acquisition of wh$questions. +
# $
, 15 (fall), 89$98.
Hattori, H. (2004). The acquisition of wh$movement by Japanese advanced learners of
English. 3
2/
, 65 (03), 584$C. (UMI No. C817102).
153
Hawkins, R. (1998). The inaccessibility of formal features of functional categories in
second language acquisition. Paper presented at the Pacific Second Language
Research Forum, Tokyo.
Hawkins, R. & Chan, Y.$H. C. (1997). The partial availability of Universal Grammar in
second language acquisition: The ‘failed functional features hypothesis’.
7
, 17, 347$367.
Hawkins, R., & Hattori. H. (2006). Interpretation of English multiple wh$questions by
Japanese speakers: A missing uninterpretable feature account.
7
, 22 (3), 269$301.
Huang, C.$T. J. (1982).
+
.
Unpublished doctoral dissertation, Massachusetts Institute of Technology.
Hulk, A. (1996). Parameter setting and the acquisition of wh$questions in L2 French. In
P. Jordens & J. Lalleman (Eds.),
(
(pp 237$
259).
Kayne, R. (1984). +
/
/
. Dordrecht: Foris.
Ke, C. (2005). Patterns of acquisition of Chinese linguistic features by CFL learners.
+
!
2
, 40 (1), 1$24.
Kellerman, E. (1983). U$shaped behavior in advanced Dutch EFL learners. In S. Gass &
C. Madden (Eds.), $
(
(pp ) Rowley Mass:
Newbury House.
Kim, J. T. (1995). !
(
1(
:
doctoral dissertation, University of Hawai‘i at Manoa.
8
. Unpublished
Kim, J$T. (2001). The initial state of second language syntax: An investigation of L2 wh$
movement and null$subjects from the minimalist perspective. 3
2/
, 62 (02), 552$A$553$A. (UMI No. DA3004306).
Kim, J$T. (2003). L2 initial syntax: Wh$movement and the most economical syntactic
derivation. #
2
3
$
, 27 (1), 414$424.
Kim, S. C. (1999). The subject$object asymmetry in the acquisition of wh$questions by
korean learners of english. # 1#
2
2$$
,
3, 1$11.
Johnson, J. S., & Newport, E. L. (1989). Critical period effects on universal properties of
language: the status of Subjacency in the (
of a second language. +
#
, 21 (1), 60$99.
Lasnik, H., & Saito, M. (1984). On the nature of proper government.
235$289.
(
Lasnik, H., & Saito, M. (1992). "
Cambridge, MA: MIT Press.
$ .
$
0+
$$
Lee, H. T. (1989). Development of a mandarin$speaking child's comprehension of wh$
questions. +
( 2 4
, 18, (1), 29$62.
,
154
Lee, S. (2004). 2
(
/ :
<
1$
(
1
8
:
0* (
. Unpublished doctoral dissertation, University of
Hawai‘i.
Li, J. (1995). 3 and wh$questions in Mandarin Chinese. , 4 (4), 313$323.
:
2
Lightbown, P. M. (1985). Great expectations: Second language research and classroom
teaching. 2$$
, 6, 173$189.
Manzini, M. (1992).
02
Cambridge, MA: MIT Press.
$
Martohardjono, G. (1993).
1
1
=
doctoral dissertation, Cornell University.
(
(
.
02
. Unpublished
McLaughlin, B. (1990). Restructuring. 2$$
, 11, (2), 113$128.
Merkin, S, Wootten, J, & Bloom, L. (1980). Linguistic and pragmatic factors in the
acquisition of wh$questions. Paper presented at the 12th Stanford Child Language
Research Forum.
O’Grady, W. (1997).
$
0!
(
:
. Chicago:
University of Chicago Press. Oiry, M., & Demirdache, H. (2006). Evidence from L1
acquisition for the syntax of wh$scope marking in French. In V. Torrens& L. Escobar
(Eds.), !
(
7
(pp 289$315). Amsterdam:
John Benjamins.
Paradi, T., Schwartz, B. D., & Clahsen H. (2004). On the L2 acquisition of the
morphosyntax of German nominals.
, 42 (3), 669$705.
Park, H. (2003). Why$questions in L2 acquisition. 4
(2), 415$440.
>
<
7
, 39
Platzack, C. (1996). The initial hypothesis of syntax: A minimalist perspective on
language acquisition and attrition. In H. Clahsen (Ed.),
$ $
(
0: $
$
(pp 369$414). Amsterdam: John Benjamins.
Radford, A. (1994). The syntax of questions in child English. 21 (1), 211$236.
Rizzi, L. (1990). 7
+
. Cambridge, MA: MIT Press.
Rojina, N. (2004). The acquisition of wh$questions in Russian. ?
, 32 (1), 68$87.
Rowland, C., & Pine, J. M. (2000). Subject$auxiliary inversion errors and wh$question
acquisition: “What children do know?” +
, 27 (1), 157$181.
Santelmann, L. (1997). Wh$less questions in early Swedish: An argument for continuity
in language development. +
# $
, 15, 217$253.
,
155
Santelmann, L. (2003). The acquisition of Swedish wh$questions. In G. Josefsson, C.
Platzack, & G Hakansson, !
(
(pp. 261$307).
Amsterdam: John Benjamins.
Sarma, J. (1991). !
(
1(
dissertation, University of Conneticut.
:
Unpublished doctoral
Saunders, N. J. (1983). Will you get off this plane somewhere? Form and function in the
L2 acquisition of English by Japanese. 4
# $ 1 2$$
2
2
, 6, 66$83.
Schachter, J. (1988). Second language acquisition and its relationship to universal
grammar. 2$$
, 9 (3), 219$235.
Schwartz, B. D., & Sprouse, R. (1994). Word order and nominative case in nonnative
language acquisition: a longitudinal study of (L1 Turkish) German interlanguage. In
T. Hoekstra & B. D. Schwartz (Eds.),
(
(pp. 317$368). Amsterdam: John Benjamins.
Schwartz, B. D., & Sprouse, R. (1996). L2 cognitive states and the full tranfser/full
access model.
7
, 12 (1), 40$72.
Seidl, A., & Hollich, G. (2002). Infants’ and toddlers’ comprehension of subject and
object wh$questions. #
2
;
+
3
$
, 26 (2), 596$607.
Sjoholm, K. (1989). Characteristic features in lexical restructuring among advanced
Finnish$ and Swedish$speaking learners of English.
# $
, 8, 103$113.
Slabakova, R. (2000). L1 transfer revisited: the L2 acquisition of telicity marking in
English by Spanish and Bulgarian native speakers.
, 38, 739$770.
Stromswold, K. (1988). Linguistic representations of children’s wh$questions. # $
7 $
+
3
$
, 27, 107$114.
Stromswold, K. (1995). The acquisition of subject and object wh$questions.
2 (
, 4 (1$2), 5$48.
Thornton, R. J. (1991). 2
1
0!
(
1(
Unpublished doctoral dissertation, University of Connecticut.
$
Tsai, W.$T. D. (1994). On nominal islands and LF extraction in Chinese. ?
@
!
, 12 (1), 121$175
Tsai, W.$T. D. (1999). On lexical courtesy. 73.
:
2
, 8 (1), 39$
Tyack, D., & Ingram, D. (1977). Children’s production and comprehension of questions.
+
, 4, (2), 211$244.
156
Vainikka, A., & Young$Scholten, M. (1994). Direct access to X’$theory: Evidence from
Korean and Turkish adults learning German. In T. Hoekstra & B. D. Schwartz (Eds.),
(
(pp. 265$316). Amsterdam:
John Benjamins.
Vainikka, A., & Young$Scholten, M. (1996a). Gradual development of L2 phrase
structure.
7
, 12 (1), 7$39.
Vainikka, A., & Young$Scholten, M. (1996b). The early stages of adult L2 syntax:
Additional evidence from Romance speakers.
7
, 12 (2),
140$176.
Vainikka, A., & Young$Scholten, M. (2002). Restructuring the CP in L2 German.
#
2
;
+
3
$
26 (2), 712$722.
,
Van Valin, R. D., Jr. (2002). The development of subject$auxiliary inversion in English
wh$questions: an alternative analysis. +
, 29 (1), 161$175.
Wakabayashi, S., & Okawara, I. (2003). Japanese learners’ errors on long$distance wh$
questions. In S. Wakabayashi (Ed.),
$$
(
:
/
$
- $
. Berlin: Mouton de Gruyter.
White, L. (1985). The pro$drop parameter in adult second language acquisition.
, 35, 47$62.
White, L. (1986). Implications of parametric variation for adult second language
acquisition: An investigation of the ‘pro$drop’ parameter. In V. Cook (Ed.),
: $
$$
(
(pp. 55$72). Oxford:
Pergamon Press.
White, L. (1990). Second language acquisition and universal grammar.
2 (
, 12 (2), 121$133.
White, L. (2003).
Cambridge University Press.
(
. Cambridge:
Wode, H., Bahns, J., Bedey, H., & Frank, W. (1978). Developmental sequence: An
alternative approach to morpheme order.
, 28, (1), 175$185.
Yoshinaga, N. (1999). Wh$questions: 2
$
(
:
- $
. Unpublished doctoral dissertation, University of Hawai‘i.
Yuan, B. (1998). Interpretation of binding and orientation of the Chinese reflexive
English and Japanese speakers.
7
, 14, 324$340.
by
Yuan, B. (2001). The status of thematic verbs in the second language acquisition of
Chinese.
7
, 17 (3), 248$272.
Yuan, B. (2007). Behaviours of $words in English speakers’ L2 Chinese wh$questions:
Evidence of no variability, temporary variability and persistent variability in L2
grammars.
0
+
, 10 (3), 277$298.
157
APPENDIX
TESTS USED IN COLLECTING THE DATA
Test 1
Oral Production Task 1.1
Instructions: You are going to role$play with your instructor on the following two
topics. In both situations, you would like to obtain some information by asking questions,
and your instructor will play the role of the person you are asking questions from. This is
not a test. Your performance will not affect your grade. Please relax and do the best you
can. Thanks very much for your cooperation!
你几点到家?
你几点到家?
Topic 1:
Your friend is inviting you to his home for his birthday party. You would like to
obtain the following information by asking him questions.
!
1) Where is your home? (哪儿
2) Who did you invite? (谁
,家
, 请
)
)
3) How many people did you invite? (几个
)
,玩
)
4) What will we play? (什么
5) What dinner will we have? (什么
,饭
)
Topic 2:
At the birthday party, you chat with a friend. What questions would you ask her to
get the following information?
!
158
6) What would you like to eat? (什么
7) Who is that American? (谁
,吃
,美国人
8) Where do you often go to movies? (哪
)
)
!
, 看电影 "
9) How many movies have you seen? (几个
10) What movies has she seen this month? (什么
#
)
)
)
Grammaticality Judgment Task 1.2
Your name:
Years of Learning Chinese:
Time you started:
Time you finished:
Instructions: This is a grammar exercise. Your performance in the test will not
affect your grade. Please relax and do the best you can. Our help is greatly appreciated.
What you need to do in this exercise is to judge the goodness of each of the underlined
sentences. If it is good, please circle Y; otherwise please circle N. Please circle “DNK”, if
you do not know the answer.
1)小王: 小李,明天是我的 生日。 我 想请你到我家。
小李:好啊。 你的家在哪儿?
Y N
DNK
2)小李:这个月有很多好 电影。 几个电影你看了?
小白: 是啊, 有很多好电影。 我看了三个 电影了。
Y N
3)小李:你 也 喜欢 看 电影!什么电影你喜欢看 ?
小白:我 喜欢 看法国 电影。
DNK
159
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
4)小王: 小李, 明天是我的生日。你来 我 的 家,好吗?
小李: 好啊。你还请了谁?
5)小李:周末我们一起 看 电影 吧!
小白:好啊。我们看电影几点?
6)小王: 小李,我请了朋友明天到我家玩儿。
小李:是吗?你请了几个人?
7)小王: 小李,明天是我的生日。 我想请你到我家。
小李:好啊。在哪儿你的家 ?
8) 小王: 小李,明天是我 的 生日。
小李:是吗?我们一起看电影 吧。
9) 小李:是吗? 你请了小白吗?
小王:我 请了 小白。
10)小王: 周末到我家玩儿吧。
小李:好啊。我们在你家玩什么?
160
11) 小王: 小李,我请了朋友明天到我家玩儿。
小李:是吗? 几个人你请了 ?
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
12) 小李:好啊。我们在哪儿见?
小王: 我们见在学校。
13)小王: 小李, 周末 我 请 你 吃饭,怎么样?
小李:好啊。我们 吃 什么饭?
14)小李:小白,这儿有鸡 (chicken),鱼 (fish),还有饺子
(dumplings)。你想什么吃?
小白:我想吃点儿饺子(dumplings)。
15)小李:你也 喜欢看电影!你常常看电影在哪儿?
小白:我常常在家看电影。
16)小李:小白,这儿有鸡 (chicken),鱼 (fish),还有 饺子
(dumplings)。你想吃点儿 什么?
小白: 我想吃点儿饺子(dumplings)。
161
17)小王:你喜欢 不喜欢吃中国饭?
小李:我喜欢吃中国 饭。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
18)小王: 小李, 明天是我的生日。 你来我 的家,好吗?
小李:好啊。谁你还 请了 ?
19)小李:今天有很多人。 谁你认识 ?
小白:我 认识 你和小王。
20)小王: 小李, 明天 是 我 的生日。 你来我的家, 好吗?
小李:好啊。 我到你家七点。
21)小李:你也喜欢看电影! 哪儿你常常去看电影?
小白:我常常在家看电影。
22)小王: 周末到我家玩儿吧。
小李:好啊。什么我们在 你家玩 ?
23)小李:今天 有很多 人。你认识谁?
小白:我认识你和小王。
162
24)小李:今天有很多人吗?
小白: 是,有很多人。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
25)小李:这个月有很多好电影。 你看了很多电影吗?
小白: 是啊, 我看了很多电影。
26)小李:你也喜欢看电影!你常常去哪儿看电影?
小白:我常常在家看电影。
27)小李:这个月有很多好 电影。你看了几个电影 了?
小白: 是啊, 有很多好电影。 我看了三个电影了。
28)小王: 小李, 周末 我 请 你 吃饭,怎么样?
小李:好啊。 什么饭我们吃 ?
29)小李:小白,这儿 有 鸡 (chicken),鱼 (fish),还有 饺子 (dumplings)。
什么你想吃点儿 ?
小白: 我想吃点儿饺子(dumplings)。
30)小李:你也喜欢看电影!你喜欢看什么电影?
小白:我喜欢看法国电影。
Y N
DNK
163
Y N
DNK
Test 2
Oral Production Task 2.1
Instructions: You are going to role$play with your instructor on the following two
topics. In both situations, you would like to obtain some information by asking questions,
and your instructor will play the role of the person you are asking questions from. This is
not a test. Your performance will not affect your grade. Please relax and do the best you
can. Thanks very much for your cooperation!
你几点到家?
你几点到家?
Topic 1:
Your friend invites you to play ball with him. You ask him what ball you are
going to play. You also discuss with him who else you can invite since you need more
people. The following are the questions you ask.
!"
#
!
1) What (kind of) ball are we going to play? (what=什么,play ball=打球d2qi{)
2) How many people do we need to get? (how many=几个j&ge,
need to=得d6i, get=找zh2o)
3) Who do you want to get? (who=谁,get=找)
4) Who did you play basketball with last time? (play basketball=打篮球
d2l1nqi{,last time=上次sh3ngc*)
5) How does XiaoLi play basketball? (how=怎么样z6nmey3ng)
Topic 2:
You are accompanying your friend to buy clothes at a store where bargain is
allowed. You are supposed to help him select clothes and also to bargain. These are the
164
questions you ask while you are helping him.
!"
#
!
6) What do you want to buy? (what = 什么)
7) Where are you buying your clothes? (where = 哪儿$
)
8) Which shirt do you like? (which = 哪,shirt = 衬衫
)
9) What size of shirt do you wear? (what size = 多大, wear = 穿
)
10) How much money do you want to spend? (how much = 多少,spend
money=花钱
)
Grammaticality Judgment Task 2.2
Your name:
Years of Learning Chinese:
Time you started:
Time you finished:
Instructions: This is a grammar exercise. Your performance in the exercise will
not affect your grade. Please relax and do the best you can. For each
, decide if it sounds good in that context. If it is, then circle "Y". If not, then
circle "N", and
$
. If you don't know or can't decide, then circle "DNK". "
#
%
. If you do not know a word or
phrase, please let us know.
1) 小王: 衬衫在这儿。哪件衬衫你喜欢?
小李: 这件白色的很好。
Y N
DNK
165
2)小王: 打篮球,好啊。你还想找谁 ?
小李: 我 不知道。
Y N
DNK
Y N
DNK
3)小李:哪家商店的衣服好?
小白:我常常买衣服在Younkers。
4)小王: 这件衬衫很合适。
小李:是啊。 多少钱你想花? 我跟他们讲价 (I will bargain with them)。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
5)小王: 我不知道找谁。
小李: 你上次跟谁一起打球?
6)小王: 我请你听音乐会吧。
小李:对不起。我要去跟我的男朋友。
7)小王: 你周末常常做什么?
小李:我常常跟一个朋友打球。
8) 小王: 衬衫在这儿。你穿多大号的衬衫?
小李:大号的。
166
9) 小王: 衬衫在这儿。你喜欢哪件衬衫?
小李: 这件白色的很好。
Y N
DNK
Y N
DNK
10)小王: 小李,你有空吗? 我想请你打球。
小李:好啊。 什么球我们打?
11)小王: 这件衬衫很合适。
小李:是啊。 你想花多少钱? 我跟他们讲价 (I will bargain with
them)。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
12)小王: 你能跟我一起买衣服吗?
小李:好啊。 你想在哪买衣服?
13)小李:我们一起打球吧。
小白:好啊。我们见在明天晚上七点,怎么样?
14)小王: 我们一起打球吧。
小李:好啊。我们明天晚上见在学校。
15)小王: 我请你听音乐会吧。
167
小李:对不起。我要跟我的男朋友一起去。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
16)小王: 小李,你有空吗? 我想请你打球。
小李:好啊。 我们打什么球?
17)小李:我不知道找谁。
小白:怎么样小李打球打得?
18)小李:我们一起打球吧。
小白:好啊。我们明天晚上七点见,怎么样?
19)小王: 我们到了。什么你想买?
小李:我想买一件衬衫、一条裤子。
20) 小王: 衬衫在这儿。多大号的衬衫你穿?
小李:大号的。
Y N DNK
21)小王: 我们一起打球吧。
小李:好啊。我们明天晚上在学校见。
Y N
22)小王:打篮球,好啊。 我们得找几个人?
DNK
168
小李: 再找三个人。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
23)小王: 打篮球,好啊。谁你还想找 ?
小李:我 不知道。
24)小王: 你能跟我一起买衣服吗?
小李:好啊。 在哪你想买衣服?
25)小王: 我不知道找谁。
小李: 跟谁你上次一起打球?
26)小王: 你周末常常做什么?
小李:我常常打球跟一个朋友。
27)小王:打篮球,好啊。 几个人我们得找?
小李: 再找三个人。
28) 小李:我不知道找谁。
小白:小李打球打得怎么样?
29)小李:哪家商店的衣服好?
169
小白:我常常在Younkers买衣服。
Y N
DNK
Y N
DNK
30)小王: 我们到了。你想买什么?
小李:我想买一件衬衫、一条裤子。
Test 3
Oral Production Task 3.1
Instructions: You are going to role$play with your instructor on the following two
topics. In both situations, you would like to obtain some information by asking questions,
and your instructor will play the role of the person you are asking questions from. This is
not a test. Your performance will not affect your grade. Please relax and do the best you
can. Thanks very much for your cooperation!
你几点到家?
你几点到家?
Topic 1:
You visit the Chinese section of the library and then check out books. You ask the
following questions during this process.
!"
#
!
1) Which floor are the Chinese books at? (which = 哪
2) What ID do I need to use? (What = 什么
)
)
3) (For a book that is not immediately available)
How long do I need to wait? (how long = 多久
%
4) How many books can I check out? (how many = 几)
5) How much money do I pay if my book is past due?(how much = 多少
pay = 付
&, past due = 过期
' %
,
170
Topic 2:
You are invited to a party. You ask the following questions to find more
information about it.
!"
#
6) Where shall we meet? (where = 哪
!
, meet = 见
)
7) How should we get to your house? (how = 怎么
8) What do we need to bring? (bring = 拿
)
)
9) Who (else ) did you invite?
10) With whom did you play cards last time? (with whom = 跟谁 (
=上次
, last time
))
Grammaticality Judgment Task 3.2
Your name:
Years of Learning Chinese:
Time you started:
Time you finished:
Instructions: This is a grammar exercise. Your performance in the exercise will
not affect your grade. Please relax and do the best you can. For each
, decide if it is correct or not. If it is, then circle "Y". If not, then circle "N".
Please
. If
$
you don't know or can't decide, then circle "DNK". "
#
%
. If you do not know a word or phrase,
please ask your instructor for help.
1)图书馆员:你好。
小白: 你好。 请问, 中文书在几层楼?
171
Y N DNK
2)图书馆员:别人把这本书借走了。 你得等一段时间.(wait for some time)
小白: 多久我得等?(等=wait)
Y N DNK
3) 小李:我想请你们星期五到我家玩儿。
小王:太好了。怎么我们到你家呢?
Y N DNK
4)图书馆员:找到你要的书了吗?
小白: 是的。我得用什么证件?
Y N DNK
5)图书馆员:你好。
小白: 你好。请问, 几本书学生可以借?
Y N DNK
6) 图书馆员:你好,有什么问题吗?
小白: 是的。 如果书过期,多少钱每天我得付?
Y N DNK
7)图书馆员:你好。
小白: 你好。请问, 学生可以借几本书?
Y N DNK
8)图书馆员:我能借这本书吗?
172
小白:不可以。你只能看这本书在图书馆。
Y N DNK
9)小李:到你家去玩儿?太好了。
小白:我们见在明天晚上七点,怎么样?
Y N DNK
10) 小李:我想请你们星期五到我家玩儿。
小王:太好了。我们怎么到你家呢?
Y N DNK
11)小李: 到你家去玩儿?太好了。谢谢你。
小白:好啊。我们明天晚上见在你家。
Y N DNK
12)图书馆员:你好。
小白: 你好。 请问, 在几层楼中文书?
Y N DNK
13)小李: 你们上次玩什么了?
小白:我们上次跟小王小张打牌。
Y N DNK
14) 小李: 我们可以下课以后, 一起走。
小白:好啊。 我们在哪儿见?
Y N DNK
15) 小李:小王和小张都不会打牌。我们这次不能打牌了。
173
小白:你上次跟谁打牌?可以请他们来。
Y N DNK
16) 图书馆员:你好,有什么问题吗?
小白: 是的。 如果书过期,每天我得付多少钱?
Y N DNK
17)图书馆员:别人把这本书借走了。 你得等一段时间.(wait for some time)
小白: 我得等多久?(等=wait)
Y N DNK
18) 小李:我想请你到我家玩儿。
小白: 好啊。 我们带点儿什么?
Y N DNK
19)小白: 什么时候还?
图书馆员:你要在三月八号还书。
Y N DNK
20) 小李:我想请你星期五到我们家玩儿。
小白: 太好了。 你还请了谁?
Y N DNK
21)小李: 到你家去玩儿?太好了。谢谢你。
小白:好啊。我们明天晚上在你家见。
Y N DNK
22)图书馆员:找到你要的书了吗?
174
小白: 是的。什么证件我得用?
Y N DNK
23) 小李: 我们可以下课以后, 一起走。
小白:好啊。 在哪儿我们见?
Y N DNK
24)小李: 你们上次玩什么了?
小白:我们上次跟小王小张打牌。
Y N DNK
25)小李:到你家去玩儿?太好了。谢谢你。
小白:别客气。 我们明天晚上七点见,怎么样?
Y N DNK
26) 小李:我想请你到我家玩儿。
小白: 好啊。 什么我们带点儿?
Y N DNK
27)小白: 我能借这本书吗?
图书馆员:不可以。你只能在图书馆看这本书。
Y N DNK
28) 小李:我想请你星期五到我们家玩儿。
小白: 太好了。 谁你还请了?
Y N DNK
29) 小李:小王和小张都不会打牌。我们这次不能打牌了。
175
小白:跟谁你上次打牌?可以请他们来。
Y N DNK
30)小白: 什么时候还?
图书馆员:你要还书在三月八号。
Y N DNK
Test 4
Oral Production Task 4.1
Instructions: You are going to role$play with your instructor on the following two
topics. In both situations, you would like to obtain some information by asking questions,
and your instructor will play the role of the person you are asking questions from.
Example: the time
arrives home. 你几点到家?
This is not a test. Your performance will not affect your grade. Please relax and
do the best you can. Thanks very much for your cooperation!
Topic 1:
You are interviewing Thomas for a special TV program. Thomas is a student who
just comes back from intensive study$abroad Chinese program in China. You need to find
out the following information. You can add more questions if you want. Please make sure
to use the words in the brackets.
1) Which city did you go? (which = 哪
)
2) How long did you stay in China? (how long = 多久
)
3) I am wondering who you went with (wonder = 想知道
4) I am wondering where you visited (wonder = 想知道
5) I am wondering which city you liked the best (wonder = 想知道
)
)
)
176
6) How do you think we should study Chinese? (think/feel =
认为/觉得
)
7) What gift do you think we should give to Chinese people?
(think/feel=认为/觉得
)
8) What do you think that you have learned? (think/feel = 认为/觉得
)
Topic 2:
Last night there was a burglary (入室抢截
)taking place at
*& )
Jim’s house. Jim had been away on a trip for some time. You are interviewing Jim’s
neighbors, who were witnesses. Your questions should cover but are not limited to the
following areas. Please make sure to use the words in the brackets.
9) What was he (the suspect) wearing? (What = 什么
10) Where did he (suspect) go finally? (where = 哪儿
)
)
11) How many people did you see? (how many = 几 )
12) I am wondering how many weeks Jim has been away. (wonder =
想知道
)
13) I am wondering how you found the burglary. (wonder = 想知道
14) how do you think the suspect entered the house? (think/feel =
认为/觉得
)
15) how long do you think that he stayed in the house?(think/feel =
认为/觉得
)
Grammaticality Judgment Task 4.2
Your name:
Years of Learning Chinese:
)
177
Time you started:
Time you finished:
Instructions: This is a grammar exercise. Your performance in the exercise will
not affect your grade. Please relax and do the best you can. For each
, decide if it sounds good in that context. If it is, then circle "Y". If not, then
circle "N", and
$
. If you don't know or can't decide, then circle "DNK". "
#
%
. If you do not know a word or
phrase, please let us know.
1) A: 听说你去了中国。 你去了哪个城市?
B: 我去了北京和上海。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
2) A:他可以玩什么?
B: 他玩什么都行。
3)A: 听说你去了中国。 多久你在中国呆(stay)了?
B: 我在中国呆了两个月。
4) A: 你在中国旅行了一个月。 什么你觉得你学到了?
B: 我觉得我学到了很多中国的文化。
5) A: 听说你去了中国。 我想知道你跟谁一起去的。
178
B: 我是跟我的同学一起去的。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
6) A: 听说你的中文说得很好。 你认为怎么我们应该学中文?
B: 多说多练。
7) A: 你在中国呆了那么久,一定很了解中国。
什么礼物你觉得我们应该送给中国人?
B: 这要看(it depends on) 你去哪了。
8) A:他可以玩什么?
B: 他都玩什么行。
9) A: 听说你去了中国。 什么地方我想知道你参观(visit)了。
B: 我还参观了南京。
10)A: 听说你去了中国。 哪个城市你去了?
B: 我去了北京和上海。
11) A: 听说你去了中国。 我想知道跟谁你一起去的。
B: 我是跟我的同学一起去的。
179
12) A:我们可以在哪说中文?
B: 我们在哪儿都可以说中文。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
13) A: 听说你去了中国。 我想知道你最喜欢哪个城市。
B: 我最喜欢上海。
14)A: 听说你去了中国。你在中国呆了多久?
B: 我在中国呆了两个月。
15) A: 听说你去了中国。 哪个城市我想知道你最喜欢。
B: 我最喜欢上海。
16) A:他能吃什么?
B: 他都什么东西吃。
17) A: 听说你去了中国。 我想知道什么地方你参观(visit)了。
B: 我还参观了南京。
18)A: 听说你的中文说得很好。 你觉得我们应该怎么学中文?
B: 多说多练。
180
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
19) A:你在中国旅行了一个月。 你觉得什么你学到了呢?
B: 我觉得我学到了很多中国的文化。
20) A: 听说你的中文说得很好。 怎么你认为我们应该学中文?
B: 多说多练。
21) A:我怎么到你家?
B: 你怎么到我家都行。
22) A: 听说你去了中国。 跟谁我想知道你一起去的。
B: 我是跟我的同学一起去的。
23)A:在中国呆了那么久,一定很了解中国。你觉得我们应该送中国人什么礼物?
B: 这要看你去哪了。
Y N
DNK
Y N
DNK
24) A: 听说你去了中国。 我想知道哪个城市你最喜欢。
B: 我最喜欢上海。
25) A: 听说你去了中国。 我想知道你参观(visit)了什么地方。
B: 我还参观了南京。
181
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
26) A: 你在中国呆了那么久,一定很了解中国。
你觉得什么礼物我们应该送中国人?
B: 这要看你去哪了。
27) A: 我们可以在哪说中文?
B: 我们都在哪儿可以说中文。
28)A: 你在中国旅行了一个月。 你觉得你学到了什么呢?
B: 我觉得我学到了很多中国的文化。
29) A: 她穿什么衣服好看?
B: 她都穿什么衣服好看。
30)A: 你看见了那个人。 他穿着什么?
B: 他穿着一件黑色的衬衫。
31)A: 你看见了那个人。 他最后去哪了?
B: 这我不知道。
182
32) A: 你看见了这个入室抢劫(burglary) 。 一共几个人你看见了?
B: 只有他一个人。
Y N
DNK
33) A: 你看见了这个入室抢劫(burglary) 。你觉得多久他在房子里呆了?
B: 几十分钟。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
34) A: 她穿什么衣服好看?
B: 她穿什么衣服都好看。
35) A: 你看见了这个入室抢劫(burglary) 。你一共看见了几个人?
B: 只有他一个人。
36) A:你看见了这个入室抢劫(burglary) 。
怎么我想知道你发现的入室抢劫。
B: 我看见Jim家的灯亮了,但是Jim不在家。
37)A: 你看见了那个人。 什么他穿着?
B: 他穿着一件黑色的衬衫。
38)A: 你是Jim的邻居。 我想知道Jim离开(leave)家几个星期了。
B: 有两个星期了。
183
Y N
DNK
Y N
DNK
39) A: 我怎么到你家?
B: 你都怎么到我家行。
40 ) A: 你看见了这个入室抢劫(burglary) 。 我想知道你怎么发现的入室抢劫。
B: 我看见Jim家的灯亮了,但是Jim不在家。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
41) A: 你是Jim的邻居。我想知道几个星期Jim离开(leave)家了。
B: 有两个星期了。
42) A: 你看见了这个入室抢劫(burglary) 。怎么你觉得他进的房子?
B: 不知道。
43) A: 你看见了这个入室抢劫(burglary) 。
我想知道怎么你发现的入室抢劫(burglary) 。
B: 我看见Jim家的灯亮了,但是Jim不在家。
44)A: 你看见了这个入室抢劫(burglary) 。你觉得他怎么进的房子?
B: 不知道。
Y N
DNK
184
45) A: 你看见了那个人。 哪他最后去了?
B: 这我不知道。
Y N
DNK
Y N
DNK
Y N
DNK
Y N
DNK
46) A: 你是Jim的邻居。 几个星期我想知道Jim离开(leave)家了。
B: 有两个星期了。
47) A: 他能吃什么?
B: 他什么东西都吃。
48) A: 你看见了这个入室抢劫(burglary) 。你觉得怎么他进的房子?
B: 不知道。
49) A: 你看见了这个入室抢劫(burglary) 。你觉得他在房子里呆了多久?
B: 几十分钟。
Y N
DNK
50) A: 你看见了这个入室抢劫(burglary) 。多久你觉得他在房子里呆了?
B: 几十分钟。
Y N
&
%
!&
'
DNK