Vous êtes sur la page 1sur 25

THE PREFERENCES FOR AMBIGUOUS RELATIVE CLAUSES IN ENGLISH

DEPARTMENT STUDENTS BATCH 2007 OF FKIP UNLAM BANJARMASIN


Akhmad Alfian Noor
Program Studi Pendidikan Bahasa Inggris
FKIP Universitas Lambung Mangkurat
Abstract: This research uses descriptive method. English department students of FKIP Unlam
batch 2007 seem less aware and understand when they encounter ambiguity especially in the
structurally or syntactically ambiguous sentences that related to syntactic analysis based on the
interview with Syntax teacher who taught them at that time. This study is about their relative
clause (RC) attachment preferences for ambiguous relative clauses between noun phrase 1
(NP1) attachment or high attachment and noun phrase 2 (NP2) attachment or low attachment.
To collect the data, grammatical judgment test and off-line questionnaire are used. A
grammaticality judgment test are carried out in order to verify that the participants are
sensitive to the structure in off-line questionnaire, that of a RC with complex antecedent. In
analyzing the data, the steps are done as follows: (1) calculating the grammaticality judgment
test scores (in percentages), (2) calculating the off-line questionnaire to see whether the number
of either NP1 or NP2 responses is significant above chance level or not for each condition
namely of (NP-V-[NP1 of-NP2]-RC) and with (NP-V-[NP1-with-NP2]-RC) condition.
This research found that the participants showed a strong preference for NP2
disambiguation for NPs linked by the preposition with (with condition), but none showed any
significance attachment preferences for sentences containing complex genitive antecedents (of
condition).

Keywords: Preferences, Ambiguous Relative Clauses, High Attachment vs. Low
Attachment.



Background
The centrality of the study on
ambiguity especially structural or
syntactic ambiguity focuses on the
concept that dealing with sentences with
multiple readings or interpretations
would be a necessary component of
sentence processing. Structural or
syntactic ambiguity is the phenomenon
in which the same sequence of words or
the same sentence has two or more
meanings that is accounted for by
different phrase structure analyses
(Fromkin et al, 2011).
The sentence Someone shot the
maid of the actress who was on the
balcony is syntactically ambiguous
because in one phrase structure analysis
the relative clause (RC) who was on
the balcony can be attached to the
noun phrase (NP) the maid while in
the other phrase structure analysis the
RC can be attached to the other NP the
actress. This kind of syntactically
ambiguous sentence is called relative
clause attachment ambiguity.
Although English department
students of FKIP Unlam Banjarmasin
batch 2007 are mostly at their last
semester and had studied many braches
of linguistics from basic to advanced
level, based on the interview with
Syntax teacher who taught them at that
time, they seem less aware and
understood when they encounter
ambiguity especially in the structurally
or syntactically ambiguous sentences
which related to the concept of syntactic
analysis.
So it is interesting to study about
the preferences for ambiguous relative
clauses in English department students
batch 2007 of FKIP Unlam Banjarmasin
because individually the readers have
their own respective tendency towards
the meaning or interpretation they
prefer in structurally ambiguous
sentences. And moreover it is suspected
that there is a possibility the students
would have a tendency to choose the
meaning or interpretation with less
processing load for their preferences
in this study.

The Nature of Ambiguity
A word, phrase, or sentence is
ambiguous if it has more than one
possible meaning (Bach, 1998:6).
Meanwhile, Payne (2006:180) states
that some linguistic utterances are
ambiguous because they may be
interpreted in more than one way.
Levine (1988:20) states that no
linguist disputes the point about
ambiguity as an inherent property of all
natural languages, and adds that
ambiguous modes of expression are
rooted in the very nature of language
and thought. Hopper and Traugott in
Schutze (1997:4) explains that
according to historical linguists
hypotheses, people create new
meanings to enhance the expressivity
and in-formativeness of words.

Types of Ambiguity
In general, the types of
ambiguity can be divided between
lexical ambiguity and structural or
syntactic ambiguity. Lexical ambiguity
is the most commonly known form of
ambiguity (Payne, 2006:180).
Fromkin et al. (2011) defines
lexical ambiguity as sentences that have
multiple meanings due to the presence
of words that have multiple meanings
within them. For example, the sentence
This will make you smart, off course
without any additional information or
context, is ambiguous because of the
two meanings of the word smart:
clever or burning sensation. The
other example of the words are nouns
like chip, pen and suit, verbs like
call, draw and run, and adjectives
like deep, dry and hard.
On the other hand, structural or
syntactic ambiguity is the phenomenon
in which the same sequence of words or
the same sentence has two or more
meanings that is accounted for by
different phrase structure analyses
(Fromkin et al., 2011). If this ambiguity
occurs in a sentence, then it is called a
structurally or syntactically ambiguous
sentence.
The sentence The girl saw the
man with the telescope is structurally
ambiguous. It is because in one phrase
structure analysis the prepositional
phrase (PP) with the telescope can be
attached to the noun man which
means the girl saw a man who had a
telescope, while in the other phrase
structure analysis the PP can be attached
to the verb saw which means the girl
used a telescope to see the man.
Another well-know syntactic
ambiguity is the relative clause (RC)
attachment ambiguity. When a noun
phrase is made up of a noun-head, a
prepositional phrase and relative clause,
sometimes it is rather confusing as to
whether the relative clause modifies the
noun-head or the head of the noun
phrase of the prepositional phrase. For
example, the complex noun phrase The
secretary of the president who sacked
the typist, whereas the secretary here
is as noun-head, of the president as
prepositional phrase, and who sacked
the typist as relative clause, is
structurally or syntactically ambiguous
because it could be the secretary who
sacked the typist, or it could be the
president the one who sacked the
typist. This kind of ambiguity is used
in this study.
Another example of this kind of
ambiguity in a sentence is Someone
shot the maid of the actress who was on
the balcony. This sentence is
structurally or syntactically ambiguous
between two interpretations or
meanings, one in which the maid is on
the balcony, the other in which the
actress is on the balcony.

Relative Clause (RC) and RC
Attachment Ambiguity
Relative clauses (RCs) are
typically found after a noun phrase and
provide some information about the
person or thing indicated by that noun
phrase, and also they are sometimes
called adjective clauses because, like
many adjectives, they often describe
and help to identify the person or thing
being talked about (Yule, 1998:240).
For example, the relative clause who is
wearing blue jeans as in the sentence
The girl who is wearing blue jeans is
my sister, gives information about the
noun phrase in that the girl being
talked about in the main clause as the
sister is the one who is wearing blue
jeans.
There are some structurally or
syntactically ambiguous sentences
containing a relative clause (RC) that
can refer to either of the two preceding
noun phrases (NPs). This is called
ambiguous relative clause or usually
called relative clause attachment
ambiguity because the relative clause
could be attached to either noun phrase
1 or NP1 (high attachment) or noun
phrase 2 or NP2 (low attachment).
This ambiguity may occur
because of the confusion head within
complex noun phrase or complex NP.
As stated before, when a complex noun
phrase is made up of a noun-head, a
prepositional phrase and relative clause,
sometimes it is rather confusing as to
whether the relative clause modifies the
noun-head or the head of the noun
phrase of the prepositional phrase.
For example, the complex noun
phrase, The secretary of the president
who sacked the typist, whereas the
secretary here is as noun-head, of the
president as prepositional phrase, and
who sacked the typist as relative
clause, is structurally or syntactically
ambiguous because it could be the
secretary who sacked the typist, or it
could be the president the one who
sacked the typist.
By studying relative clause
attachment ambiguity, it can be known
more about a parametric function of
syntactic analysis in sentence
processing of human language. This in
fact has become the focus of many
studies such as the attachment
preferences for ambiguous relative
clauses (e.g., Felser et al., 2003 and
Bertenshaw, 2003). Another example of
such ambiguity is given below as used
by both studies.

Someone shot the maid of the actress
who was on the balcony.
a. The maid was on the balcony.
b. The actress was on the balcony.

In this example the relative
clause (RC) which is introduced by the
relative pronoun who could be attached
to the first noun the maid, as in (a) or to
the second noun the actress, as in (b). If
the parser attaches the RC to noun
phrase 1 (NP1), this is called NP1
attachment, early closure or high
attachment, and if the parser attaches
RC to noun phrase 2 (NP2), it is called
NP2 attachment, late closure or low
attachment. The parser here means the
one who does the syntactic analysis of
the relative clause ambiguity in the
sentence.
In general theoretical
considerations between high and low
attachment is represented by different
tree diagram as show below.

High Attachment (HA):




N1` Det
N1 PP
P`
NP1
NP1 CP (RC)
P NP2
Det N2`
N2
Low Attachment (LA):













The relative clause (RC)
attachment ambiguity is especially
intriguing because attachment
preferences were found to vary cross-
linguistically. Frazier and Clifton
(2006) reported that native speakers of
English were found to prefer attachment
to the second noun (NP2). The
Japanese-speaking learners of English
were also found to prefer for NP2
attachment (Bertenshaw, 2003). While
Papadopoulou (2006) reported that
native speakers of Greek were found to
prefer attachment to the first noun
(NP1).
















Sentence Processing and Parsing
Sentence processing is the
process in which our brains undergo in
order to comprehend an utterance.
Garrett (1990) in Papadopoulou
(2006:1) defines sentence processing as
a rapid and automatic process, one
which is closely linked in time to the
input. This means that sentence
processing is described to proceed in an
incremental fashion that is once each
word is encountered, it is integrated to
the sentence analysis.
There are some sentence
processing models which related to the
relative clause attachment ambiguity.
N1 PP
P`
P NP2
NP2 CP (RC)
Det N2`
N2
NP1
Det N1`
The Garden Path Model, Construal
Theory, Recency/Predicate Proximity,
and Tuning Hypothesis are some of the
examples. Below it will be discussed
about these models.

The Garden Path Model
The Garden Path model (Frazier,
1978) in Papadopoulou (2006:11)
assumes that parsers processing
resources are limited, therefore during
the initial processing of a sentence only
one syntactic analysis is pursued rather
than pursuing multiple syntactic
analyses. This is guided by the
principles of Minimal Attachment and
Late Closure.
Minimal Attachment requires
that new incoming material be attached
in a way that the fewest necessary
phrase structure nodes are used in
accordance with the well-formedness of
language rules. On the other hand, Late
Closure requires that if grammatically
permissible, attach new items into the
clause or phrase currently being
processed (i.e., the clause or phrase
postulated most recently).
Only Late Closure principle will
be discussed here, since it could predict
the preference for relative clause
attachment ambiguity. This study
specifically talks about relative clauses
(RCs) with complex noun phrase (NP)
antecedents. Therefore, it is relevant to
see what the Garden Path Model
especially related to the Late Closure
principle will predict for a sentence of
this type, which illustrated below.

Someone shot the servant of the actress
who was on the balcony.

On encountering the RC who
was on the balcony, according to Late
Closure principle, the parser is predicted
to attach it to the most recently
processed phrase, which in this case is
the second NP the actress. This
predication in fact has been found to be
the preference of native English
speakers in several studies (e.g. Frazier
and Clifton, 1996 and Traxler et al.,
1998) and the preference of Japanese
Speaking learners of English (e.g.
Bertenshaw, 2003).

Construal Theory
Proposed by Frazier and Clifton
(1996, 1997) (cited in Papadopoulou,
2006), the Construal Theory is a refined
of the earlier Garden Path Model. The
main difference is that in this theory,
syntactic constructions are divided into
two sets: primary and non-primary
phrases. Primary relations or primary
phrases include the subject and main
predicate of any finite clause as well as
the complements and obligatory
constituents of primary phrases (Frazier
and Clifton, 1996:41).
Relative clause constructions are
considered as non-primary phrases.
Therefore, according to this theory,
instead being attached to a NP, relative
clause constructions are merely
associated with it. This relationship is
determined by the Relative Clause
Construal Hypothesis. The hypothesis
states that when a RC can be attached
to two heads, it will not be immediately
attached to the most recent constituent
(the second NP), as Late Closure would
require, but rather it will be associated
to the extended maximal projection of
the last theta-role assigner, however
when more than one potential head is
available within the current thematic
domain, interpretative principles
determine the attachment preferences
for the RC.
For example, in the structurally
or syntactically ambiguous sentence,
Someone shot the maid of the actress
who was on the balcony, the relative
clause (RC) will be associated with the
entire NP the maid of the actress, as this
is the maximal projection of the theta
domain. By using interpretative
principles which mean that they
introduce entities in the thematic
domain, in this case because theta-
assigning preposition of creates the
thematic domain headed by the maid,
the prediction, then, is for NP
high

attachment to the RC. In other words,
the maid would be chosen as the
preferred reading.

Recency/Predicate Proximity
The Recency/Predicate
Proximity model is proposed by Gibson
et al. (1996a), and cited in
Papadopoulou (2006). This model
suggests that parsing choices are the
result of the competition between two
factors, namely Recency and Predicate
Proximity principles. Recency which
favouring low-attachment (LA) requires
new elements to be attached to the most
recently built structures. Meanwhile,
Predicate Proximity principle which
favouring high-attachment (LA) states
that incoming material should be
attached as close as possible to the head
of a predicate phrase.
The strength of these two factors
is determined by the processing load
and working memory involved when
processing, for example relative clause
attachment ambiguity. The factor with
the smallest load and working memory
involved would be preferred. For
example, native English speaker tend to
prefer NP
low
attachment because it is
thought to be related with the stricter
word order in English. Therefore NP2 is
chosen to reduce the processing load
involved.

Tuning Hypothesis
Tuning theory, proposed by
Brysbaert and Mirtchell (1996) as cited
in Papadopoulou (2006), asserts that the
parser as the processor keeps statistical
records of the way structural
ambiguities are most frequently
resolved in the language, and the initial
analysis of the sentence is based on
frequencies at a structural basis.
According to the Tuning
Hypothesis, structural decisions are
based on statistical records of the way
the structural ambiguity is most
frequently resolved in the language. In
the case of the relative clause (RC)
attachment ambiguity, this hypothesis
predicts that the RC attachment
preferences are mostly or more
frequently based on disambiguations
which so far encountered in different
languages or solved based on cross-
linguistic differences. This means that if
the relative clause ambiguity is resolved
more frequently towards low-
attachment, the speakers of these
languages are influenced by this
experience and prefer a low-attachment
disambiguation and vice versa.

Relative Clause Processing by L2
Learners of English
One of the examples of this
study can be found in Felser et al.
(2003) and Berthensaw (2003). Both
studies consisted of two parts: an off-
line questionnaire and a self paced
reading experiment, but here it will only
discuss about the off-line questionnaire
since its directly related to the present
study.
The off-line questionnaire
investigated relative clause (RC)
attachment preferences of German and
Greek learners of English in Felser et al.
(2003) and of Japanese-speaking
learners of English in Berthensaw
(2003). The procedures and instruments
used in both studies were identical to
each other and to the present study. The
sentences in the off-line questionnaire
were presented in two conditions. In
one condition the RC complex was
joined by the preposition of which is
called the of condition and in the other
by with which is called the with
condition. These two conditions are
illustrated in the structurally or
syntactically ambiguous sentences
below.

The dean liked the secretary of the
professor who was reading a letter (of
condition).

The dean liked the professor with the
secretary who was reading a letter (with
condition).

For this off-line questionnaire
task, the participants were asked to
indicate which noun phrase (NP) they
thought modified the RC (e.g. the
secretary or the professor in the
examples above). For both the German
and Greek groups, the participants
showed no clear preference for either
NP in the of condition, but the
participants from both groups clearly
showed a significant preference for
NP
Low
in the with condition, and also
there were significantly more NP
low
responses for the with condition than
that of the of condition which indicated
that either type of sentence (of or with
condition) is treated any differently
from the other.
Within the Japanese-speaking
learners of English, there was a
significant preference for NP2
attachment in the both of and with
conditions, and also there were slightly
more NP
low
responses for the with
condition than that of the of condition,
but this difference was not significant.
This indicated that neither type of
sentence (of or with condition) is treated
any differently from the other.
The conclusion drawn by Felser
et al. (2003) was that L2 learners in this
case neither transfer their L1 attachment
nor adopt English preference when
processing a sentence in their L2.
Meanwhile, Bertenshaw (2003)
suggested that the fact that the learners
performed in a similar way to the native
English speakers in similar studies
might suggest that they have attained a
native-like competence.

Methodology of Research
Research Method
A descriptive method which is
designed to gather information about
present conditions is used in this study.
Gay (1976) in Sevilla et al. (1992)
defines the descriptive research as
involving collection of data in order to
test hypotheses or to answer questions
concerning the current status of the
subject of the study.
More specifically the descriptive
quantitative research is used to collect
and analyze the data and information
about the attachment preferences for
ambiguous relative clauses in English
department students batch 2007 of
FKIP Unlam Banjarmasin. Aliaga and
Gunderson (2002) in Muijs (2004)
defines quantitative research as
explaining phenomena by collecting
numerical data that are analysed using
mathematically based methods (in
particular statistics). This means that the
quantitative data in this study are in the
form of numbers, and descriptive
statistics is used in presenting and
analyzing the data which states the facts
as they are.
Ambiguous relative clauses
(RCs) here are those that with two
possible noun phrases or hosts, namely
noun phrase 1 (NP1) or NP
HIGH
which is
called NP1 or high attachment and noun
phrase 2 (NP2) or NP
LOW
which is
called NP2 or low attachment for the
RCs interpretations or meanings. This
research is intended to describe about
the relative clause (RC) attachment
preferences of English department
students batch 2007 of FKIP Unlam
Banjarmasin, or more specifically to
describe whether there is a significant
preference for either NP1 or NP2
attachment in the sentences that contain
structurally or syntactically ambiguous
relative clauses with the formula (NP-V-
[NP1-of-NP2]-RC) and with the
formula (NP-V-[NP1-with-NP2]-RC) or
not. The latter of the two formulas is
called with condition whereas the
former is called of condition.
One of the important elements in
descriptive quantitative study is the
existence of the hypotheses (if there is
one). A hypothesis is a tentative
explanation that accounts for a set of
facts and can be tested by further
investigation (Muijs, 2004:8).
The number of either NP1
attachment (high attachment) responses
or NP2 attachment (low attachment)
responses for each condition (of or with
condition) is significant if the null
hypothesis (H
0
) is rejected while the
alternative hypothesis (Ha) is accepted
within one sample t-test under one-
tailed. The alternative hypothesis is the
one we want to be true whereas the null
hypothesis is the opposite. The null
hypothesis (H
0
) is rejected, and the
alternative hypothesis (Ha) is accepted
if the mean response (in percentage) of
either NP1 or NP2 responses for each
condition is more than 50% within one
sample t-test under one-tailed (H
0
:
50%, and Ha: > 50%).

Sampling Technique
Population
Fraenkel and Wallen (2006:93)
defines the population as the group of
interest to study about, the group to
whom the result to be generalized is
based on. Meanwhile, Suharsimi
Arikunto (2006:130) defines a
population as a set (or collection) of all
elements processing one or more
attributes of interest.
According to the definition
above, the population of this research is
all English department students batch
2007 academic year 2011/2012 at the
odd semester. According to the data
from English department of FKIP
Unlam Banjarmasin, there are definitely
more than 100 students or at around 120
students for both Reguler A and B
students at odd semester academic year
2011/2012 in January, 2012 when this
study is conducted.

Sample
A sample in a research study is
the group on which information is
obtained (Fraenkel and Wallen,
2006:92). This research uses random
sampling as the sampling technique
because there are a quiet of numbers of
the students of English department of
FKIP Unlam Banjarmasin batch 2007
academic year 2011/2012 at odd
semester in January, 2012 when this
study is conducted.
After a careful calculation, it is
decided that the sample for this study is
around 37% of the population. So the
sample for the study is around 44
students as the participants. The sample,
44 students, will be taken from 22
students of Reguler A and 22 students
of Reguler B, which will be divided into
each respective group namely, group
Reg. A and group Reg. B participants.

Data Collection
According to Gray (2004:400),
instrument refers to the tools used to
gather data as part of a research project.
The instruments used in this descriptive
quantitative research to collect the data
about the preferences for ambiguous
relative clauses in English department
students batch 2007 of FKIP Unlam
Banjarmasin are both the
grammaticality judgment test and the
off-line questionnaire (Felser et al.,
2003 & Bertenshaw, 2003).

Grammaticality Judgment Test
Grammaticality judgment test in
this study is intended to verify that the
students are sensitive to the structure
used in the off-line questionnaire, that
of a relative clause (RC) with a complex
noun phrase (NP) antecedent. This test
consists of 20 sentences, 10
grammatical and 10 ungrammatical
which are all taken from Felser et al.
(2003). They include 10 sentences
relevant to the study, with complex
antecedents of the form NP1-of-NP2-
RC (n=6) and NP1-with-NP2-RC (n=4).
These are divided equally between
grammatical and ungrammatical
sentences. The examples of 10
sentences relevant to the study are given
below.

The chairman phone the boss of the
clerks who were working on the new
project (grammatical).

The customer spoke to the assistant with
the pharmacist who were preparing the
medicine (ungrammatical).
The distracters or fillers are
consisted of 10 sentences with various
relative clause constructions. The
sentences are arranged in such a way
that 10 sentences which relevant to the
study, that with complex antecedents of
the form NP1-of-NP2-RC and NP1-
with-NP2-RC either grammatical or
ungrammatical sentences, would be put
in even numbers and would be preceded
or followed by one distracter either
grammatical or ungrammatical one. The
sentences are laid out with a five point
scale following each one.
There are twenty sentences for
this task. The participants from both
group Reg. A and group Reg. B are
asked to indicate their judgment on each
one by circling or crossing one of the
five possible responses namely A-E.
They are told to rely on their knowledge
of English grammar. They can also
circle or cross the Dont know response
to any question. The score is given in
every students answer that is, for the
grammatical items: definitely correct (3
points), probably correct (2 points),
dont know option (excluded), probably
incorrect (1 point), and definitely
incorrect (0 points). This system is
reversed for the ungrammatical items.
This yields a possible 30 points for each
of the two conditions (10 grammatical
items and 10 ungrammatical items).

Off-Line Questionnaire
Off-line questionnaire in this
study is used to probe the attachment
preference of the ambiguous relative
clauses in the sentences. All
experimental sentences (henceforth, the
items or sentences used to probe the
attachment preference of the ambiguous
relative clauses in this study are called
experimental items/sentences) are
structurally or syntactically ambiguous
with the formula NP-V-[NP1-P-NP2]-
RC, where V is the matrix verb, P is
preposition, and RC is relative clause.
The relative clauses (RCs) used
in the study are those that introduced by
the relative pronoun who, and are
syntactically ambiguous because each
one of the RC can be attached to either
noun phrase 1 or NP1 (high attachment)
or noun phrase 2 or NP2 (low
attachment) in the relative clause
attachment ambiguity. As stated above,
all experimental sentences are
syntactically ambiguous in that either
noun phrase (NP1 or NP2) is a possible
host for the relative clause (RC). There
are a total of 20 experimental sentences
in this study. More specifically, there
are two versions of the experimental
sentences for the off-line questionnaire,
either of the type NP-V-[NP1-of-NP2]-
RC (of condition) or NP-V-[NP1-with-
NP2]-RC (with condition).
Moreover, there are two
versions of each sentence, with the two
noun phrases (NPs) namely noun phrase
1 (NP1) and noun phrase 2 (NP2) in the
noun phrase (NP) complex being joined
either by genitive or occupational of
or by accompaniment or attributive
with, as illustrated by the example
sentences respectively below. This is
done by reversing the relative ordering
of NP1 and NP2 in the with conditions,
as the examples showed below so that
the so called experimental sentences
sound equally natural in both the of and
the with conditions. Both noun phrases
(NPs) and the auxiliary in the relative
clause are in the singular forms. And
also all nouns in the sentences are of the
type [+human]. All so called
experimental sentences here are taken
from Felser et al. (2003).

The clerk asked for the consultant of the
economist who was reading the paper
(of condition).

The clerk asked for the economist with
the consultant who was reading the
paper (with condition).

Twenty filler sentences are also
formed. These are made up of
unambiguous relative clause
constructions different from the
experiment items. Almost all of the
filler sentences here were taken from
Bertenshaw (2003).
After both experimental and
filler items are available, there would be
question about the sentence with two
possible answers. For the so called
experimental sentences these questions
off course are intended to probe the
attachment preference of the participant.
The example is given below.

The clerk asked for the consultant of the
economist who was reading the paper.
Who never smiled?
the consultant the economist

For this question is intended to
probe the attachment preference of the
ambiguous relative clause in the
sentence, that means if the participant
prefers to attach to NP
High
the consultant
will be chosen. This is called NP1
attachment or high attachment. On the
other hand, if the preference is for
NP
Low
then the participant will choose
the economist. This is called NP2
attachment or low attachment.
The answer that indicated high
attachment would come first for half of
the sentences and second for the other
half. It is done to prevent a participant
to come up with any strategy for
answering the question. The 20
experimental items then are intermixed
with the 20 fillers and pseudo-
randomised in such a way that the two
experimental sentences, always from
two different conditions (of or with
conditions) would be always intervened
by one filler sentence. The experimental
sentences would be always in even
numbers, and filler sentences would be
always in odd numbers on the off-line
questionnaire paper.
There would be also two
different questionnaires with the same
filler sentences so that each participant
would only read one version of each
experimental sentence. Furthermore,
there would be also an equal number of
sentences of each condition (of and with
conditions) for both questionnaires.
For this task the participants
from both group Reg. A and group Reg.
B are asked to read the sentence and
then answer the question that followed
it. Especially for the even numbers, that
where the experimental sentences are,
they are instructed not to rely on their
grammatical knowledge of English, but
rather to use their initiation. They are
also instructed to make their choices as
spontaneously as possible. The interest
here lies in their initial reaction in
answering each question to probe the
attachment preference of the ambiguous
relative clause in the sentence. They are
also asked not to go back and change
their answers. However, it is most likely
impossible to assure that the
participants are not over-thinking of
each sentence.

Data Analysis
Fraenkel and Wallen (2006:190)
states that quantitative are reported in
term of scores. One sample t-test under
one-tailed with null hypothesis (H
0
:
50%) and alternative hypothesis Ha: >
50%) is used to see whether the number
of either NP1 or NP2 responses is
significant above chance level or not for
each condition (of and with conditions).
This also means that the participants
may not show any reliable preference
for either NP1 attachment (high
attachment) or NP2 attachment (low
attachment) for each condition. After
that, insert the t-statistic value into
probability (p) value t-test calculator to
get p-value under one tailed. This is
done to get the actual significance level
. The formula for one sample t-test is
given below.
n
s
x
t

=

Where:
t
: t-value
x
: the sample mean

: a specified value to be tested (

=
50%)
s
: the sample standard deviation
n
: the size of the sample (n = 22 for
each group)

A paired t-test under two-tailed
as suggested by Bertenshaw (2003) is
used to see whether either/neither type
of sentence (of or with condition) is
treated any differently from the other.
For this test, it could be done by using
NP1 responses for of condition vs.
NP1 responses for with condition, or
NP2 responses for of condition vs.
NP2 responses for with condition.



Interpretation of the Research Result
Grammaticality Judgment Test
The result demonstrates that
both group Reg. A and Reg. B
participants were able to handle
different constructions of English
relative clause, including those that
were used in the off-line questionnaire.
In summary, the idea of the
grammaticality judgment test was to
ensure that the participants were fully
sensitive to relative clauses with
complex antecedents because these
kinds of construction were also used in
the off-line questionnaire. From these
results from both group Reg. A and
Reg. B participants, it seems to be the
case. And also for both groups, as
suspected by Bertenshaw (2003) there is
a possibility that the lower score for the
ungrammatical sentences might be
because the learners of a language are
often not easily to totally reject an
ungrammatical sentence than to accept a
grammatical sentence. This could
happen because of the lack of grammar
knowledge.

Off-line Questionnaire
Group Reg. A Results
First of all, for RCs following
the preposition of (of condition), group
Reg. A participants preferred to attach
high (NP1) by as much as 57.3%
(SD=38.9), and they preferred to attach
low (NP2) by as much as 42.7%
(SD=38.9). Based on these data, it is
known that group Reg. A participants
showed no reliable preference for either
NP1 or NP2 disambiguation for RCs
following the preposition of (of
condition). This had been proved by one
sample t-test under one-tailed, t
(21)=0.876, p>0.15, with null
hypothesis (H
0
: 50%) and
alternative hypothesis Ha: > 50%).
Because the confidence level obtained
from this test was less than 70%, then
H
0
that the mean response (in
percentage) of either NP1 or NP2
responses for this condition is equal to
or less than 50% has been accepted
which meant the responses were not
significant.
Meanwhile for RCs following
the preposition with (with condition) the
participants preferred to attach high
(NP1) by as much as 26.8% (SD=28.8),
and they preferred to attach low (NP2)
by as much as 73.2% (SD=28.8). The
participants in this group only showed a
reliable preference for RCs following
the preposition with (with condition),
and that was the preference for NP2
attachment or low attachment. It had
been proved by one sample t-test under
one-tailed, t (21)=3.769, p=0.0005, with
null hypothesis (H
0
: 50%) and
alternative hypothesis Ha: > 50%).
Because the confidence level obtained
from this test was at about 99.9%, then
Ha that the mean response (in
percentage) of NP2 responses for this
condition is more than 50% has been
accepted which meant the responses
were significant. However, as stated
before the group showed no reliable
preference for either NP1 or NP2
disambiguation for complex genitive
NPs (of condition). And also there were
significantly more NP
low
responses for
the with condition than that of the of
condition. A paired t-test under two-
tailed showed that this difference was
significant (t(21)=3.088, p<0.1), which
means that either type of sentence was
treated any differently from the other.

Group Reg. B Results
First of all, for the of condition,
group Reg. B participants preferred high
attachment (NP1) by as much as 43.2%
(SD=35.6), and they preferred low
attachment (NP2) by as much as 56.8%
(SD=35.6). From these results, for the
of condition, the participants showed no
clear preference for either NP1 or NP2
disambiguation. One sample t-test under
one-tailed, t (21)=0.898, p>0.15, with
null hypothesis (H
0
: 50%) and
alternative hypothesis Ha: > 50%) had
proven this insignificance. Because the
confidence level obtained from this test
was less than 70%, then Ha that the
mean response (in percentage) of either
NP1 or NP2 responses for the of
condition is more than 50% has been
rejected which meant the responses
were not significant.
On the other hand, for the with
condition, the participants preferred
high attachment (NP1) by as much as
23.6% (SD=27.9), and they preferred
low attachment (NP2) by as much as
76.4% (SD=27.9). From these results,
like within group Reg. A participants, it
is known that the participants in this
group only showed a clear preference
for the with condition, and that was the
preference for NP2 attachment or low
attachment. One sample t-test under
one-tailed, t (21)=4.438, p<0.0005, with
null hypothesis (H
0
: 50%) and
alternative hypothesis Ha: > 50%) had
proven this clear reliable preference for
NP2 attachment for the with condition
where the number of NP2 responses
was significant above chance level for
this condition. Because the confidence
level obtained from this test was more
than 99.9%, then H
0
that the mean
response (in percentage) of NP2
responses for the with condition is equal
to or less than 50% has been rejected
which meant the responses were
significant. Moreover, either type of
sentence was treated any differently
from the other. This is because there
were significantly more NP
low
responses
for the with condition than that of the of
condition, and a paired t-test under two-
tailed showed that this difference was
significant (t(21)=3.123, p<0.1).

Analysis and Interpretation of the
Research
Neither of the two groups
showed any significance relative clause
attachment (RC) preferences for either
noun phrase 1 (NP1) attachment (high
attachment) or noun phrase 2 (NP2)
attachment (low attachment) for the of
condition (this had been tested with one
sample t-test). In the with-condition,
like native English speakers all groups
preferred NP2 attachment or low
attachment. This finding is also in line
with previous studies on L2 (English)
RC attachment preferences (Felser et
al., 2003 & Bertenshaw, 2003).
Since a cross-linguistic low
attachment preference is generally
reported for this condition, the learners
low attachment preferences in the
present study could be related to the
possible employment of native-like
processing strategies. None of the
participant groups showed a native-like
low attachment preference in the of-
condition though, which indicates that
the participants are sensitive to the
distinction between the theta-assigning
preposition with and the non-theta
assigning preposition of.
There is no evidence from the
results, on the other hand, that English
foreign language learners in English
department of FKIP Unlam
Banjarmasin batch 2007 apply either of
the two phrase-structure based locality
principles (Recency or Predicate
Proximity) when processing ambiguous
sentences containing complex NPs
joined by of (of condition). As for
sentences containing complex with
antecedents (with condition), the
application of Recency and Late
Closure principles could be clearly seen
as the participants from both group Reg.
A and group Reg. B were found to be in
favor for NP2 or low attachment.
Therefore as English has stricter word
order the participants preference for
NP2 or low attachment in this case
(with condition) is related to the
Recency principle in order to reduce the
processing load involved when
processing sentences containing
complex with antecedents.
In line with Felser (2003), it is
unlikely that the participants non-
native like performance in the study
should have anything to do with
insufficient grammatical knowledge of
the construction under investigation
because the participants in both group
Reg. A and group Reg. B had scored
well on the grammaticality judgment
test itself. For example, if the
participants had wrongly analyzed the
of-PPs as independent thematic
domains, they should have shown an
NP2 preference for complex genitive
antecedents (of condition), too. But
from the results of the study, they
didnt. So it means that it wasnt the
case because of insufficient
grammatical knowledge of the
construction under investigation that
there was no clear preference for NP2
for complex genitive antecedents (of
condition).


Conclusion
Based on the data findings, the main
results can summarized as follows:
1. In the grammaticality judgment
test, the two groups of EFL
learners (group Reg. A and Reg.
B participants from English
department of FKIP Unlam
Banjarmasin batch 2007) were
able to handle different types of
English relative clause
construction, including those of
the kind that were used in off-
line questionnaire of the main
study, and that they were
sensitive to number agreement
violations.
2. In the off-line questionnaire, the
two groups of EFL learners
(group Reg. A and Reg. B
participants from English
Department of FKIP Unlam
Banjarmasin Batch 2007)
pattern with native English
speakers in that they showed a
strong preference for noun
phrase 2 (NP2) disambiguation
for NPs linked by the
preposition with (this had been
tested with one sample t-test).
3. Contrary to native English
speakers, however, neither
group Reg. A nor group Reg. B
of EFL learners showed any
significance attachment
preferences for either noun
phrase 1 (NP1) attachment (high
attachment) or noun phrase 2
(NP2) attachment (low
attachment) for sentences
containing complex genitive
antecedents (of condition) in the
off-line questionnaire, which
had been tested with one sample
t-test.
4. At a certain degree, it could be
implied that there is a possibility
they have a tendency to choose a
meaning that related to the noun
phrase which is closest to the
modifier in the relative clause
(RC) attachment ambiguity of
syntactic ambiguity with two
possible hosts for the RC
according to Late Closure and
Recency principles in order to
reduce the processing load
involved when reading
sentences that contain these kind
of ambiguous sentences
especially those with relative
clause attachment ambiguities in
this study. This case can be
clearly seen in the case of the
preference for the with
condition, but not for the of
condition.

Suggestion
Based on the conclusion above, several
suggestions are addressed:
1. This study only used off-line
questionnaire instrument
proposed by Felser et al. (2003)
and Bertenshaw (2003), while
there is also on-line study or
self-paced reading task. For
further research on this topic, it
is better if this task is included to
achieve comprehensive
understanding on relative clause
(RC) attachment preferences.
2. In English department of FKIP
Unlam Banjarmasin, the topic on
ambiguity should be taught more
comprehensively as ambiguity is
the most specific feature of
natural languages which
pervades languages at all levels
such as Phonology, Syntax and
Semantics, etc.
3. Specifically, the topic on
syntactic ambiguity should be
taught more comprehensively in
Syntax course, so that the
students would be able to
distinguish syntactically
ambiguous sentences from
syntactically unambiguous ones
by using appropriate syntactic
analysis or parsing, or at least
they would be able to become
more aware when encountering
or reading such ambiguous
sentences.




































REFERENCES

Bach, Kent. 1998. "Ambiguity". Routledge Encyclopedia of Philosophy: Logic &
Mathematics, 6-9.
Bertenshaw, Nicholas. 2003. The processing of ambiguous relative clauses by
Japanese-speaking learners of English. Unpublished MA dissertation. UK:
University of Essex.
Brysbaert, M. and Mitchell, D. 1996. Modifier attachment in sentence parsing:
Evidence from Dutch. Quarterly Journal of Experimental Psychology, 49:
664695.
Celce-Murcia, M. & Larsen-Freeman, D. 1999. The grammar book: An ESL/EFL
teachers course, 2
nd
Ed. Boston: Heinle & Heinle.
Felser, C., Roberts, L., Gross, R.&. Marinis, T. 2003. The processing of
ambiguous sentences by first and second language learners of English.
Applied Psycholinguistics 24, 453- 489.
Fraenkel, Jack R. and Wallen, Norman E. 2006. How to Design and Evaluate
Research in Education. New York: McGrew-Hill Companies.
Frazier, L. and Clifton, C. 1996. Construal. Cambridge. MA: MIT Press.
Frazier, L. and Clifton, C. 1997. Construal: Overview, motivation and some new
evidence. Journal of Psycholinguistic Research, 26(3): 277296.
Fromkin, Victoria, Robert Rodman, Nina Hyams. 2011. An Introduction to
Language, International Student Edition. Canada: Nelson Education, Ltd.
Gibson, E., Pearlmutter, N., Canseco-Gonzalez, E. and Hickock, G. 1996a.
Recency preferences in the human sentence processing mechanism.
Cognition, 59: 2359.
Gray, David E. 2004. Doing Research in the Real World. UK: SAGE Publications
Inc.
Levine, Donald N. 1988. The Flight from Ambiguity: Essays in Social and
Cultural Theory. Chicago: University of Chicago Press.
Muijs, Daniel. 2004. Doing Quantitative Research in Education with SPSS.
London: SAGE Publications Ltd.
Papadopoulou, Despoina. 2006. Cross-Linguistic Variation in Sentence
Processing: Evidence from RC Attachment Preferences in Greek.
Dordrecht: Springer.


Payne, Thomas E. 2006. Exploring Language Structure: A Students Guide. New
York: Cambridge University Press.
Pickering, M. 1999. Sentence comprehension. In S. Garrod, and M. Pickering
(eds.), Language Processing (pp. 123154). Hove: Psychology Press.
Program Studi Pendidikan Bahasa Inggris. 2011. Pedoman Penulisan Skripsi.
Program Studi Pendidikan Bahasa Inggris, Fakultas Keguruan dan Ilmu
Pendidikan, Universitas Lambung Mangkurat: Banjarmasin.
Quirk, Randolph and Greenbaum, Sidney and Leech, Geoffrey and Svartvik, Jan.
1985. A Comprehensive Grammar of the English Language. London and
New York: Longman.
Schutze, Hinrich. 1997. Ambiguity Resolution in Language Learning:
Computational and Cognitive Models. Leland Stanford Junior University:
CSLI Publications.
Sevilla, Consuelo G, Jesus A. Ochave , Twila G. Punzalan, Bella P. Regalla,
Gabriel G. Uriarte. 1992. Research Methods, Revised Edition. Quezon
City: Rex Printing Company Inc.
Suharsimi, Arikunto. 2006. Prosedur Penelitian (Suatu Pendekatan Praktek)
Edisi Revisi VI. Rineka Cipta: Jakarta.
Traxler, M.J., Pickering, M.J. and Clifton, C. 1998. Adjunct attachment is not a
form of lexical ambiguity resolution. Journal of Memory and Language,
39: 558592.
Yule, George. 1998. Explaining English Grammar. Oxford: Oxford University
Press.
Yule, George. 2006. The Study of Language. New York: Cambridge University
Press.
Nordquist, Richard. ____. Relative Clause. (Online),
(http://grammar.about.com/od/rs/g/relativeclterm.htm, retrieved on July
20, 2012).
Shier, Rosie. 2004. Paired t-tests. (Online),
(http://mlsc.lboro.ac.uk/resources/statistics/Pairedttest.pdf).
http://easycalculation.com/statistics/p-value-t-test.php.
http://en.wikipedia.org/wiki/P-value. (Online). accessed on July 20
th
, 2012.
http://en.wikipedia.org/wiki/Parsing. (Online). accessed on July 20
th
, 2012.
http://wikieducator.org/images/a/a8/LESSON_THIRTEEN.pdf. (Online).
accessed on July 20
th
, 2012.


http://www.cliffsnotes.com/study_guide/One-Sample-t-test.topicArticleId-
267532,articleId-267511.html. (Online).
https://statistics.laerd.com/calculators/standard-deviation-sample-population-
calculator.php.

Vous aimerez peut-être aussi