下載/瀏覽

Download Report

Transcript 下載/瀏覽

Processing English Compounds
in the First
and Second Language: The
Influence
of the Middle Morpheme
指導教授:鍾榮富教授
Reporter:NA1C0008陳慧齡
Introduction
• A compound is made up of two or more words concatenated
to form another word
– e.g., pan and cake→pancake
• Compounds usually consist of a “head” word and a
modifier
– The head word is usually the rightmost word
– The modifier word precedes the head and qualifies the sense
denoted by the head.
– pancake, the head word is cake and the modifier word is pan.
• Compounds can also have a head-complement relationship
– e.g., taxi driver, dish washer
– synthetic compounds
Introduction
• When producing English noun-noun
compounds, Native English speakers tend to
– exclude regular plural inflection
• rat-eater not rats-eater
– allowing irregular plural inflection
• mice-eater
• Exposure to the input alone is insufficient to
explain the observed dissociation between
regular and irregular plural inflection in the
production of English compounds. (e.g.,
Gordon, 1985).
Pinker and Prince’s dual-mechanism
model
• The processing of regular morphology is
mediated by classic symbolic rules of grammar
– e.g., rat + [-s] = rats
• Irregulars are stored as memorized pairs of
words (mouse-mice) in the mental lexicon.
• 6% of occasions in which regular plurals appeared.
• The possessive [-s] morpheme is a relatively
common “regular” inflectional morpheme that
appears within compounds (e.g., Adam’s apple).
– either the phoneme /s/ or /z/ (which can also function
as allomorphs of the regular plural morpheme) appears
in a noun-noun sequence, it is almost always a
possessive [-s] morpheme (Hayes, 2003).
• Murphy (2000) suggested that one reason
children might omit regular plurals from
English compounds
– the plural [-s] morpheme consistently is
found at the end rather than in the middle of
words.
Haskell et al. (2003) argued that two
input-driven constraints
• The semantic constraint works alongside their
second phonetic constraint.
• The phonetic and semantic constraints are
learned from general properties of plurals and
prenominal modifiers that children experience
in the input they receive.
– mice-chaser / big box
The aims of this study
• To explore the extent to which these more
input-based or probabilistic explanations of
how plural inflectional morphology and
compounding interact might account for L2
learner behavior.
• How nonhead nouns ending in the phoneme /s/
(or /z/) are treated in compounds
Research Questions
1. Will compounds containing possessive nouns
be processed more quickly than compounds
containing plural nouns?
2. Will the same preferences as shown by native
speakers (NSs) be manifest by nonnative
speakers (NNSs) who have had considerably
less exposure to the input?
Research Questions
3. Will compounds in which the first noun ends in
/s/ (/z/), whether it is the plural form or not, be
processed more slowly than compounds that do
not include a first noun ending in /s/ (/z/)? Will
this difference (if observed) be manifested by the
NNSs who have had significantly less exposure to
English? Assuming that the role of exposure is a
critical variable, then one might predict that the
L2 learners will not show this preference in the
LDT task.
Design
5 different categories of nonhead word
1. regular plural:regular [-s] morpheme
2. possessive:possessive [-s] morpheme
3. irregular plural
4. phonetic /s/ (/z/) :no morphological
function but was phonetically realized as
an /s/ (/z/) phoneme
5. phoneme other than /s/ (/z/)
• To preclude the order effects
– half of the participants saw the stimuli in one
random order
– half of the participants saw it in a second random
order.
Mean age:24
22 NSs
Undergraduate
students
Participants
Mean age:23
13 NNSs
EAP
Advanced-level
The frequencies of these first
nouns
M
Set 1 (regulars) 76.89
Set 2
39.17
(possessives)
Set 3
215.25
(irregulars)
Set 4 (ending 53.41
in /s/ [/z/])
Set 5 (not
140.76
ending in /s/
[/z/])
SD
75.15
49.52
means:oneway ANOVA
Tukey HSD
post hoc tests
F(4, 89) p
irregular plural
3.92
< .01 nouns used
as the first
nouns in the
compounds
260.91
had a higher
mean overall
64.26
frequency
191.81
The frequencies of these
second nouns
M
Set 1 (regulars) 7.42
Set 2
17.13
(possessives)
Set 3
215.25
(irregulars)
?
Set 4 (ending 21.52
in /s/ [/z/])
Set 5 (not
16.61
ending in /s/
[/z/])
SD
9.71
20.80
means:oneway ANOVA
Tukey HSD
post hoc tests
F(4, 67) p
the frequencies
9.53
< .01 of the second
nouns in the
irregular set
were
260.91
significantly
?
higher relative
24.29
to the other
four sets of
second nouns.
22.46
• The apostrophe was omitted from all the
possessive nouns, making it impossible to
distinguish between the plural and possessive
solely on the basis of punctuation.
Lexical Decision Task (LDT)
• Table 2
• some nonwords included in the set of stimuli.
– e.g., pent rasser
• 48-point type
• Apple iMac computer using Psyscope software
• Response times were recorded by the Psyscope
software.
Procedure
Participants were
told about the
rules
Read the
sentence out
loud
pressed the space bar ,
an asterisk to appear
on screen and the
sentence to disappear
216 test trials,
45 min
Participants pressed
one of two clearly
marked keys
pressed the space
bar again and the
compound
appeared
Accuracy Data
Repeated measures multivariate ANOVA
independentsamples factor
(Group)
NS
one relatedsamples factor
(Word Type)
NNS five levels
effect of Group
effect of Word Type
F(1, 33)
= 16.42
F(4,132) = p < .01
7.37
p < .01
No Group × Word Type interaction
Tukey’s post hoc tests: reliable difference in the number of errors made
between items with a nonhead noun ending in
phoneme /s/ (/z/) and all other word groups(and
possessive [-s]).
Accuracy Data
Accuracy Data
• Figure 1 illustrates
– the English native speakers were over 96%
accurate,
– the Chinese NNSs were over 86% accurate.
• Neither group had any difficulty correctly
distinguishing the real words in the compounds
from the nonce compounds.
Reaction Time Data
Overall (omnibus) F test
independentsamples factor
(Group)
NS
one relatedsamples factor
(Word Type)
NNS five levels
effect of Group
effect of Word Type
F(1, 33)
= 21.87
F(4,132) = p < .01
28.79
p < .01
A marginal Group ×Word Type interaction, F(4, 132) = 2.29, p = .059.
Reaction Time Data
Reaction Time Data
• Figure 2 illustrates that for all different types
of nonhead nouns, the Chinese NNSs were
slower to respond on the LDT than the English
NS.
Comparison Examination-1
• Participants respond more quickly to
compounds:
nonhead noun
possessive regular plurals
NSs
T tests, p < .05
NNs
Comparison Examination-2
• Participants responded differently to nonhead
nouns that ended in /s/ (/z/)
nonhead noun
T tests, p < .05
ended in
/s/ (/z/)
did not end in
/s/ (/z/)
NSs
• NSs processed singular nonhead nouns not
ending in /s/ (/z/) (e.g., drink) faster than they
did those singular nonhead nouns that did end
in /s/ (/z/) (e.g., grass).
• There was no difference in this comparison for
the Chinese NNSs.
Comparison Examination-3
• To compare possessive morphology on
nonhead nouns relative to irregulars
nonhead noun possessive
irregulars
morphology
NSs
T tests, p < .01
• No reliable difference for the Chinese NNSs.
• Separate comparisons revealed that the NSs
– faster to respond to compounds with possessive
morphology relative to regular plural morphology.
– processed compounds without an /s/ (/z/) on the
nonhead noun faster than those compounds with an
internal /s/ (/z/) on the nonhead noun.
– processed singular nonhead noun compounds
more quickly than other singular nonhead nouns
that ended in a phoneme /s/ (/z/).
• NS participants responded more quickly to the
possessive [-s] items
– occur relatively frequently in the input.
• NS participants responded more slowly to the
phoneme /s/ (/z/) items
– comparatively infrequent.
• RT data for NSs show that they are faster to
respond to possessive morphology
– the input is an important factor here because
irregular plural nouns are rare in naturally
occurring noun-noun compounds.
• Advanced L2 learners of English did not show
the same preferences (as measured by RTs) as
the NSs
– had not had sufficient exposure to English
– have excluded regular plurals from compounds in
production.
The importance of using different tasks
• Compound production task or preference task
– Irregular plural
• Present study
– Take longer time to process a irregular plural than
a process noun.
Limitations
• Constraints inherent in the design
– small sample sizes
– Unequal variability across groups
– limitations of the stimuli
• Need exploring include uncovering how much
input is required
• Sophisticated measures