MINI RESEARCH

BY :

 DEBY YUSTIKA SARI

 ETIKA SARI
 MAY LENY
 RULI MAHYUNI
 SITI KOMARIAH

FACULTY OF TEACHERS TRAINING AND EDUCATION

UNIVERSITAS MUHAMMADIYAH SUMATRA UTARA
MEDAN
T.A 2016/2017
 CHAPTER I
BACKGROUND

A. Background of the study

Many words are semantically ambiguous, andcan refer to more than one concept. For
example, barkcan refer either to a part of a tree or to the sound made by a dog. To understand
suchwords, we must select one of these different interpretations, normally on the basis of the
context in which the word occurs.
Words can be ambiguous in different ways; aword like bark has two semantically
unrelatedmeanings, which seem to share the same writtenand spoken form purely by chance.
More common than this type of accidental ambiguity is thesystematic ambiguity between
related wordsenses. For example, the word twisthas a rangeof dictionary definitions including
to make intoa coil or spiral, to operate by turning, to alterthe shape of, to misconstrue the
meaning of, towrench or sprain, and to squirm or writhe. The meaning of this word varies
systematically according to the context in which the word is used. for example, there are
important differences between what it means to twist an anklecomparedwith to twist the truth.
However, although themeaningof the word is ambiguous betweenthese different
interpretations, the interpretations are closely related to each other both etymologically and
semantically, this is quite unlike the ambiguity for a word like bark.
The linguistic literature makes a distinction between these two types of ambiguity and refers
tothem as homonymy and polysemy (Cruse, 1986;Lyons, 1977, 1981). Homonyms, such as
the twomeanings of bark, are considered to be differentwords that, by chance, share the same
orthographic and phonological form. Specifically, homographs are different words that
sharethe sameorthographic form, while homophones share thesame phonological form.
Semantic ambiguity is very common in language, and our ability to understand
ambiguouswords is an important property of our languageprocessing system. Evidence about
how ambiguity affects human language performance can provide important constraints on
models of wordrecognition. In particular, models of word recognition have been required to
accommodate evidence that visual lexical decisions are faster forambiguous words. In this
paper, we evaluate theevidence of an ambiguity advantage in thelightof the distinction
between word meanings andword senses. In particular, we argue that while earlier studies in
the literature show that semanticambiguity can produce a processing advantage, itis not clear
whether this is caused by ambiguitybetween multiple word meanings or betweenmultiple
word senses. Despite this, all current accounts of the ambiguity advantage assume thatthe
advantage is produced by multiple, unrelated meanings; the experiments reported here
investigate whether this assumption is correct.
ambiguity advantageis the finding thatvisual lexical decisions are faster for words thatare
semantically ambiguous. Early reports of anambiguity advantagecame from
Rubenstein,Garfield, Millikan (1970) and Jastrzembski(1981), who found faster visual lexical
decisions for ambiguous words than for unambiguous words matched for overall frequency.
However, Gernsbacher (1984) discussed a possibleconfound between ambiguity and
familiarity inthese experiments; words with more than onemeaning are typically more
familiar. She foundno effect of ambiguity over andabove familiarity. Since then, however,
several papers have reported an ambiguity advantage in visual lexicaldecision experiments
using stimuli that wereontrolled for familiarity (e.g., Azuma & VanOrden, 1997; Borowsky
& Masson, 1996; Hino& Lupker, 1996; Kellas, Ferraro, & Simpson,1988; Millis& Button,
1989; Pexman & Lupker, 1999). Although these studies vary in therobustness of the effects
they report, their cumulative weight has had the effect of establishingthe ambiguity advantage
as an important constraint on theories of lexical representation and lexical access. One
interpretation of the ambiguityadvantage has been that ambiguous words benefitfrom having
multiple entries within the [Link] example, Kellas et al. (1988) assume thatwords are
represented by individual nodeswithin an inhibitory lexical network.
They suggest that while the multiple nodes of an ambiguous word do not inhibit each other,
theyboth act independently to inhibit all other competing entries, and this increased inhibition
ofcompetitors produces the faster recognitiontimes. As mentioned earlier, lexicographers
routinely distinguish between wordmeanings andword senses when they structure dictionary
entries. These dictionary entries provide a simple,yet reliable way to classify words as being
ambiguous between multiple meanings or betweenmultiple senses. We have used the entries
in TheOnline Wordsmyth English Dictionary–The-saurus (Parks, Ray, & Bland, 1998).

As we report later, the classifications made in this dictionary correspond closelyto

participants judgements about the relatedness of the meanings of ambiguous words. The
ambiguity disadvantage reported here removes a major hurdle for models in whichwords
compete to activate distributed semanticrepresentations. The ambiguity
disadvantagenaturally emerges from the semantic competition present in such models and has
been shownin a model of this type where a simple [Link] square algorithm was used to
learn themapping between distributed orthographic andsemantic representations (Rodd,
2000).
This new pattern of results can also be accommodated by those models in
whichwordscompete to activate abstract word nodes withina lexical network. Earlier, we
discussed howthese models could produce an ambiguity advantage by assuming either that
ambiguouswords are more efficient at inhibiting competi-tors, or that they benefit from
having multiplecompetitors in the race for recognition. Interestingly, these models can just as
easily accommodate a disadvantage for words with multiplemeanings. As in all experiments
of this type,the ambiguous words and unambiguous wordswere matched on total frequency.
This meansthat the frequency of each meaning of the ambiguous words is on average half
that of the unambiguous word.

This difference in the frequency of the word meanings could producefaster lexical decisions
for the unambiguouswords. Looking first at the stimuli used by Millis andButton (1989) and
Azuma and Van Orden(1997), neither study makes the direct contrastbetween ambiguous and
unambiguous words, words with many meanings are compared withwords with few
meanings. Words wereassignedto these groups by counting the number ofmeanings that
participants could provide foreach word. Crucially, both studies count highlyrelated word
senses as separate [Link] first at the stimuli used by Millis andButton (1989) and
Azuma and Van Orden(1997), neither study makes the direct contrastbetween ambiguous and
unambiguous words.
words with many meanings are compared withwords with few meanings. Words were
assignedto these groups by counting the number ofmeanings that participants could provide
foreach word. Crucially, both studies count highlyrelated word senses as separate meanings.
This can be demonstrated by example. This particular dictionary was chosen because it
reliablyseparates semantically unrelated meanings into distinct lexical entries, but unlike
some other dictionaries it does not require that senses within an entryhave the same
syntacticclass. This reflects the intuitions of participants that meanings from different
syntactic classes can be highly related(Azuma, 1996).
B. Identification of The Problem
1. The students’ do not understand about semantic ambiguity in the sentence
2. The students difficulties in identifying semantic ambiguity in sentence

C. The scope and limitation

The scope of this study is semantic and the limitation of this study are semantic
ambiguity.

D. The formulation of the problem

The problem of this research are formulated as the following:
1. Are the student able to understand about semantic ambiguity in the sentence?
2. Are the students difficulties in identifying semantic ambiguity in the sentence?

E. The objective of the study

The objectives of this study are as following:
1. To describe student’s difficulties in identifying semantic ambiguity
2. To find out the students’ difficulties in identifying semantic ambiguity

F. The significances of the study

The findings of this study are theoretically and practically:
1. Theoretically
To add up references for the next research in identifying semantic ambiguity.
2. Practically
English teacher:
To know the students’ ability in identifying semantic ambiguity.
To make student more active in learning English especially about semantic
ambiguity.
 CHAPTER II
RESEARCH METHODS

A. Research Approach.
In this journal Authors dicide a title Making Sense of Semantic Ambiguity:
Semantic Competition in Lexical Accessto prove anybenefit ambiguity sense given toward
interpretation process. Earlier survey models of this ambiguity advantage assume that it is the
presence of multiple unrelatedmeanings that produce the advantage. This journal consist of
set of three lexical decision experiments reported challenge the assumption.
The authors contrast the ambiguity seen in words like bark, which have multiple
unrelated meanings, with words that have multiple related word senses (e.g., twist). In all
three experiments te authors find that while multiple word senses do produce faster
responses, ambiguity between multiple meanings delays recognition. These results suggest
that,while competition between the multiple meanings of ambiguous words delays their
recognition, the rich semantic.

B. Determination of Population Sample.

Author journal research participant in the campus (MRC Cognition and Brain
Sciences Unit, Cambridge, United Kingdom).The participants were 25 members of the MRC
Cognition and Brain Sciences Unit subject panel. All had English as their first language, and
had normal or corrected-to-normal vision.

C. Data Collection Methods.

Data in this journal based on the result of three experiment and it is the comparation
between of some journal and proved from the opinion of several experts before. The results
of these three experiments represent an important challenge to accept views of how semantic
ambiguity affects recognition of isolated words. Previous reports of an ambiguity advantage
have been interpreted as showing that there is a processing advantage for words that have
multiple meanings. A range of models has been put forward to explain how this advantage
might arise.
 CHAPTER III
THE DISCUSSION AND RESULT

A. The description of this journal

There have been several reports in the literature of faster visual lexical decisions to words
that are semanticallyambiguous. All current models of this ambiguity advantage assume that
it is the presence of multiple unrelated meanings that produce this benefit. A set of three
lexical decision experiments reported here challenge this assumption. We contrast the
ambiguity seen in words like bark, which have multiple unrelated meanings, with words that
have multiple related word senses (e.g., twist). In all three experiments we find that while
multiple word senses do produce faster responses, ambiguity between multiple meanings
delays recognition. These results suggest that, while competition between the multiple
meanings of ambiguous words delays their recognition, the rich semanticrepresentations
associated with words with many senses facilitate their recognition.

Many words are semantically ambiguous, andcan refer to more than one concept. For
example, bark can refer either to a part of a tree or to the sound made by a dog. To
understand suchwords, we must select one of these different interpretations, normally on the
basis of the context in which the word occurs. Words can be ambiguous in different ways; a
word like bark has two semantically unrelated meanings, which seem to share the same
written and spoken form purely by chance. More common than this type of accidental
ambiguity is the systematic ambiguity between related word senses. For example, the word
twist has a rangeof dictionary definitions including to make intoa coil or spiral, to operate by
turning, to alterthe shape of, to misconstrue the meaning of, towrench or sprain, and to
squirm or writhe. The meaning of this word varies systematically according to the context in
which the word is used; for example, there are important differences between what it means
to twist an ankle comparedwith to twist the truth. However, although the meaning of the word
is ambiguous between these different interpretations, the interpretations are closely related to
each other both etymologically and semantically; this is quite unlike the ambiguity for a word
like bark.
 THE AMBIGUITY ADVANTAGE

The ambiguity advantage is the finding thatvisual lexical decisions are faster for
words that are semantically ambiguous. Early reports of an ambiguity advantage came from
Rubenstein, Garfield, Millikan (1970) and Jastrzembski (1981), who found faster visual
lexical decisions for ambiguous words than for unambiguouswords matched for overall
frequency. However, Gernsbacher (1984) discussed a possible confound between ambiguity
and familiarity inthese experiments; words with more than one meaning are typically more
familiar. She found no effect of ambiguity over and above familiarity. Since then, however,
several papers have reported an ambiguity advantage in visual lexical decision experiments
using stimuli that were controlled for familiarity (e.g., Azuma & Van Orden, 1997; Borowsky
& Masson, 1996; Hino & Lupker, 1996; Kellas, Ferraro, & Simpson,1988; Millis & Button,
1989; Pexman & Lupker, 1999). Although these studies vary in the robustness of the effects
they report, their cumulative weight has had the effect of establishing the ambiguity
advantage as an important constraint on theories of lexical representation and lexical access.
One interpretation of the ambiguity advantage has been that ambiguous words benefit from
having multiple entries within the lexicon. For example, Kellasetal. (1988) assume that words
are represented by individual nodes within an inhibitory lexical network. They suggest that
while the multiple nodes of an ambiguous word do not inhibit each other, they both act
independently to inhibit all other competing entries, and this increased inhibition of
competitors produces the faster recognition times. A related account (e.g., Jastrzembski,
1981) assumes that the benefit arises from the presence of noise or probabilistic activation;
because ambiguous words are assumed to have multiple entries, they benefit from having
more than one competitor in the race for recognition.

On average, by a particular point in time, one of these competitors is more likely to

have reached the threshold for recognition than a word that has only one entry in the [Link]
response to this inconsistency, there have been several attempts to show, with varying
degrees of success, that this class of model can show an advantage for ambiguous words.
Joordens and Besner (1994) and Borowsky and Masson (1996) both suggest that because
ambiguous words have more than one meaning, on average the randomly determined initial
state will be closer to a valid finishing state for ambiguous words, and this could reduce the
time ittakes for the network to settle. Kawamoto, Farrar, and Kello (1994) suggested that if
anerrorcorrectinglearning algorithm was used to learn the mapping from orthography to
semantics and then to compensate for the increased error produced by the ambiguous words
in the semanticunits, stronger connections are formed between the orthographic units. If
lexical decisions are made on the basis of orthographic representations, then this could
improve performance for ambiguous words.

In summary, all current accounts of the ambiguity advantage assume that it is

ambiguity between unrelated meanings that produces the ambiguity advantage. None of these
models explicitly predict what the effect of multiple word senses should be. For those models
in which the benefit arises because of the presence of multiple localist lexical entries for
ambiguous words, the presence of a benefit for words with multiple senses will depend on
whether multiple senses are represented as separate entries within the network. Those models
that involve distributed semantic representations predict that words with multiple senses may
show a processing advantage, but that this should be reduced compared with words with
multiple meanings.

WORD SENSES AND WORD MEANINGS

As mentioned earlier, lexicographers routinely distinguish between word meanings and

word senses when they structure dictionary entries. These dictionary entries provide a simple,
yet reliable way to classify words as being ambiguous between multiple meanings or
betweenmultiple senses. We have used the entries in The Online Wordsmyth English
Dictionary–Thesaurus (Parks, Ray, & Bland, 1998).1 As we report later, the classifications
made in this dictionary correspond closely to participants’ judgements about the relatedness
of the meanings of ambiguous words. Looking first at the stimuli used by Millis and Button
(1989) and Azuma and Van Orden (1997), neither study makes the direct contrast between
ambiguous and unambiguous words; words with many meanings are compared with words
with few meanings. Words were assigned to these groups by counting the number of
meanings that participants could provide for each word. Crucially, both studies count highly
related word senses as separate meanings. This can be demonstrated by example.
 B. The discussion of this journal

A. EXPERIMENT 1

In this first experiment, lexical decisions for alarge set of ambiguous and unambiguous
words are analyzed using multiple regression analysesto determine the effects of multiple
meanings,multiple senses, and meaning relatedness. Tothe extent that different effects
emerge for these factors, this would provide the basis for further investigations.
MethodParticipants.
The participants were 25 members of the MRC Cognition and Brain Sciences Unit subject
panel. All had English as their first language and had normal or corrected-to-normal vision.
Stimuli and design.
One hundred twenty-four ambiguous words were selected to be included in the
experiment. One hundred thirteen were taken from the Twilley, Dixon, Taylor, and Clark
(1994) homograph norms. While most of the selected words had only two meanings, a few
words with three meanings were included where the third meaning had a meaning probability
in the Twilley et al. (1994) norms of less than 0.1. The remaining 11 ambiguous words did
not appear in the norms, but were considered to have similar properties. For half these
ambiguous words, the two meanings correspond to separate entries in the Wordsmyth
dictionary, and are therefore ambiguous between two meanings according to the criteria used
by lexicographers. The remaining ambiguous words were judged to have more than one
meaning by Twilley et al. (1994), but these meanings corresponded to a single entry in the
Wordsmyth dictionary. For these words it ismnot clear whether their different interpretations
should be classed as meanings or senses; their inclusionwill allow us to look for effects of the
relatedness of the meanings of ambiguous words.

B. EXPERIMENT 2

Experiment 1 suggests that the two types of semantic ambiguity have very different
effects on lexical decision performance. While multiple meanings delay recognition, multiple
senses produce a processing benefit. This result is clearly controversial; all existing models of
the ambiguity advantage have assumed thatmultiple meanings produce faster visual lexical
decisions. Experiment 2 attempts to replicate the contrasting effects of ambiguity seen in the
multiple regression analysis of Experiment 1, using a factorial design to directly compare the
effects of lexical ambiguity and multiple word senses.
MethodParticipants.
The participants were 25 members of the MRC Cognition and Brain Sciences Unit subject
panel. All had English as their first language, and had normal or corrected-to-normal vision.
Stimuli and design.
The word stimuli were selectedto conform to a 2 3 2 factorial design where the two factors
were ambiguity and number of senses. Words were classed as being un- ambiguous if they
had only one entry in the Wordsmyth Dictionary (Parks et al., 1998) and as ambiguous if
they had two or more entries.

Two measures of the number of senses were used. These were the total number of word
senses listed in the Wordsmyth dictionary for all the entries for that word, and the total
number of senses given in the WordNet lexical database (Fellbaum, 1998). Thirty-two words
were selected to fill each cell of the factorial design such that the number of word meanings
was matched across each level of number of word senses, and the total number of word
senses was matched across each level of the number of word meanings. Therefore, unlike
Experiment 1, the numbers of ambiguous and unambiguous words used in this experiment
were equal. Of the words used in this experiment,16% were also used in Experiment 1.

C. EXPERIMENT 3

Experiments 1 and 2 have shown that word senses and word meanings have very different
effects on visual lexical decisions. Ambiguity between multiple meanings produces a
disadvantage, while multiple senses produce faster responses. This experiment investigates
whether these contrasting effects of ambiguity are also present in the auditory domain, using
the same factorial design as Experiment 2. We argue that the ambiguity effects seen in
Experiments 1 and 2 reflect the influence of amodal semantic representations on visual word
recognition. If this is the case, then it is of interest to see whether the same pattern emerges in
the auditory domain. It is possible that semantic information plays a similar role across the
two domains, but it is also possible that the temporalcharacteristics of speech may reduce the
role of semantic information on spoken word recognition compared with visual word
recognition. However, we do know that the semantic information relating to spoken words is
accessed early on, prior to the word becoming unique (Marslen-Wilson, 1987; Zwitserlood,
1989), and this makes it at least possible that auditory lexical decision will show an influence
of semantic ambiguity. A study by Holley-Wilcox (1977) (cited in &McCusker, Hillinger,
Bias, 1981) supports this idea that it is possible to detect effects of semantic ambiguity using
auditory lexical decision. They found that auditory lexical decisions were significantly slower
for homophones like plane and plain, which although sharing the same phonology are spelled
differently, thanfor nonhomophones. This result can be explainedby assuming that
competition between the different meanings of the homophones is slowing the recognition.
This is consistent with the ambiguity disadvantage seen the visual domain in Experiment 1
and suggests that competition between the different meanings of ambiguous words does play
a role in the auditory domain. However, a possible problem with using homophones that are
not homographs to look for semantic ambiguity effects is that there may be interference
between their different orthographi representations. Although it may seem unlikely that
interference between orthographic representations should affect an auditory task, this idea is
supported by Ziegler and Ferrand (1998). They found slower auditory lexical decisions for
words whose rimes could be spelled in more than one way (e.g., sleep). This raises the
possibility that orthographic interference may have contributed to the finding of Holley-
Wilcox (1977) and makes it preferable to avoid such items in any experiment designed to
show the effects of semantic ambiguity.
MethodParticipants.
The participants were 26 students at Cambridge University. All had English as their first
language.
Stimuli and design.
The word stimuli were selected to conform to the same 2 3 2 factorial design as in
Experiment 2. Seventy-seven percent of the words selected were also used in Experiment 2.
Twenty-three words were selected to fill each cell of the factorial design; the number of
words in each cell is smaller than that used in Experiment 2 because of the additional
phonological constraints used to match the groups.

C. The result of this journal

The results of these three experiments represent an important challenge to accepted views
of how semantic ambiguity affects recognition of isolated words. Previous reports of an
ambiguity advantage have been interpreted as showing thatthere is a processing advantage for
words that have multiple meanings. A range of models has been put forward to explain how
this advantage might arise.
 Our analyses of the stimuli used in three ofthe clearest demonstrations of this effect
suggested to us that the accepted interpretation might be incorrect and that related word
senses and not unrelated meanings might be responsiblefor this processing advantage. The
results ofthe three experiments reported here support this view. In all three experiments we
found a significant benefit for words that have many senses, compared with words with few
senses. In contrast,ambiguity between unrelated meanings consistently produced a processing
disadvantage; this ambiguity disadvantage was significant in Experiments 1 and 3. We now
consider the implications for models of word recognitionWe have already discussed how
models of word recognition have tried to explain the apparent advantage for words with
multiple meanings, but our data suggest that they must accommodate exactly the reverse
effect. This challenge is less problematic than might be expected.
The ambiguity disadvantage is a natural prediction of models in which words compete for the
activation of semantic representations (Gaskell & Marslen-Wilson, 1997; Hinton &Shallice,
1991; Joordens & Besner, 1994; Plaut, 1997; Plaut & Shallice, 1993). As discussed earlier, in
these models interference between the different meanings of ambiguous words would delay
their recognition relative to an unambiguous word. As noted by Joordens & Besner(1994), an
ambiguity advantage can be produced by these models only if an additional mechanism is
present to overcome this semantic competition. Our results suggest that no such mechanism is
required.
The ambiguity disadvantage reported here removes a major hurdle for models in
which words compete to activate distributed semantic representations. The ambiguity
disadvantage naturally emerges from the semantic competition present in such models and
has been shown in a model of this type where a simple leastmean square algorithm was used
to learn the mapping between distributed orthographic and semantic representations (Rodd,
2000). This new pattern of results can also be accommodated by those models in which
words compete to activate abstract word nodes within a lexical network. Earlier, we
discussed howthese models could produce an ambiguity advantage by assuming either that
ambiguous words are more efficient at inhibiting competitors, or that they benefit from
having multiple competitors in the race for recognition. Interestingly, these models can just as
easily accommodate a disadvantage for words with multiple meanings. As in all experiments
of this type, the ambiguous words and unambiguous words were matched on total frequency.
This means that the frequency of each meaning of the ambiguous words is on average half
that of the unambiguous word. This difference in the frequency of the word meanings could
produce faster lexical decisions for the unambiguous words. Further, if lateral inhibition were
presentbetween all word nodes within the lexical network, including the nodes
corresponding to the different meanings of an ambiguous word, this would act to slow the
recognition of ambiguous words. Therefore, it appears that both classes of models considered
here can accommodate the ambiguity disadvantage. The question that remains is whether the
ambiguity disadvantage should simply be explained in terms of an effect of frequency of
word meanings, or whether we can claim that it provides evidence of competition between
the different meanings of words. We suggested earlier that looking for an effect of meaning
relatedness might help us to determine the mechanisms by which any observed ambiguity
effects might arise.
 CHAPTER IV
CONCLUSION

The data reported here require us to reconsider how semantic ambiguity affects the
recognition of isolated words. While we do not consider these data to be inconsistent with the
existing ambiguity advantage data, they do contradict the accepted interpretation of these
data. We have found that multiple meanings delay recognition, while multiple senses produce
a processing advantage. We therefore claim that the ambiguity advantage reported in
previous studies should be interpreted as showing a benefit for words with many senses; this
is consistent both with the data reported here and our analysis of the stimuli from previous
studies. Our claim that ambiguity between multiple meanings can slow lexical decisions is
entirely new. Yet it can apparently be incorporated into most current models of word
recognition by assuming that there is competition between the different meanings of
ambiguous words or that there is an advantage for the more frequent meanings of
unambiguous words compared with ambiguous words matched on overall frequency.
However, the finding that the ambiguity disadvantage is modulated by meaning relatedness
suggests that the effect must, at least in part, be due to competition to activate a distributed
semantic representation. It is less clear how the word sense benefit should be interpreted, and
further work is required to determine the cause of this intriguing effect. Nonetheless, the
overriding message from this series of experiments is that the word recognition process is
intimately tied in with the competitive process by which the stored meanings of words are
retrieved.