Introduction

Successful second language comprehension and production involve building grammatical structures rapidly from incoming linguistic input and then producing appropriate responses. Within milliseconds, interlocutors need to decode incoming phonemic and morphological strings, construct syntactic representations, integrate semantic information and finally respond appropriately to their interlocutors. Sentence comprehension, however, is not always easy and smooth. Sometimes if utterances involve complex grammatical structures, comprehenders need to spend more time analyzing these structures. Sometimes, when the meaning of utterances is not straightforward and involves additional meaning, comprehenders need to be capable of being aware of it so that they can “read between the lines.” These seemingly daunting and complex tasks are nevertheless conducted in an extremely rapid manner for first language speakers, who rarely encounter any difficulty during sentence comprehension. However, any of these tasks could potentially be a huge challenge for second language learners and could lead to comprehension or production breakdown. Many studies have shown that adult second language speakers, although equipped with superior linguistic knowledge and proficient second language skills, still display non-native-like proficiencies in many subtle aspects. For example, the English article system poses a difficult challenge even for advanced English learners [1, 2, 3, 4]. In addition, many different theories have been proposed to account for the fact that second language learners generally fail to attain a native-like proficiency level [5, 6, 7].

In recent years, advances in computers and imaging technology have provided the driving force for online examination and understanding of the cognitive mechanism of second language learners. Such cutting-edge technology also provides researchers the ability to examine the validity of different theories. These online techniques include reaction times, event-related potential (EPRs), functional Magnetic Resonance Imaging (fMRI), transcranial magnetic stimulation (TMS) and positron emission tomography (PET) [8, 9, 10, 11]. The advancement of modern techniques has enabled researchers to form research questions about first and second language acquisition and comprehension that could not be answered previously. For example, researchers have been using fMRI to see whether the neural substrates are the same between first and second language speakers and how other factors such as age and proficiency could influence cerebral organization among second language learners. Despite the discrepancies found among different studies using different techniques [8, 9, 12], Perani et al. [13] did find that proficiency level and age in a second language play an important role in second language processing.

Most of the studies investigating second language learning have focused on syntax [14], phonology [15], morphology [16], pedagogy [17] and so on. Few have looked at second language learning or processing in pragmatics using online methods to address these issues. The current paper attempts to bridge this gap by using the online reaction time technique to examine scalar inferences in second language learners. To be more specific, the author investigated Mandarin learners’ processes of English scalar inferences. Before I delve into our research questions and design, I will first briefly introduce Grice’s Cooperative Principles and scalar inferences in pragmatics.

Theoretical Background

From a linguistic point of view, human communication involves decoding and encoding incoming and outgoing strings of utterances. However, language communication conveys much more than what is encoded superficially. For example, when one interlocutor says, “Some of my sons are doctors,” one of the possible assumptions that is not encoded in this utterance is “Not all of my sons are doctors.” Thus during a conversation, a listener has to go beyond the literal linguistic expressions, or read between the lines, to get the implicit meaning, or implicatures, from the speaker. The processes of deriving these inferences among interlocutors were first characterized by a philosopher of language, Paul Grice [18], who systematically studied the meanings and inferences between interlocutors and laid the foundations of the modern study of pragmatics. He assumed that a successful communication involves cooperation between interlocutors and thus proposed the Cooperative Principle and four maxims to illustrate how people follow these rules to make conversation go smoothly.

1. Cooperative Principle: Contribute what is required by the accepted purpose of the conversation.
2. Maxim of Quality: Make your contribution true; do not convey what you believe to be false or unjustified.
3. Maxim of Quantity: Be as informative as required.
4. Maxim of Relation: Be relevant.
5. Maxim of Manner: Be perspicuous; avoid obscurity and ambiguity, and strive for brevity and order.

A typical implicature can be found in the following short conversation:

1. (1)Speak A: Did you do your laundry?
2. Speaker B: I was busy

In this conversation, it is quite obvious that B didn’t address A’s question directly. On the contrary, B replied with “I was busy” to imply that he did not do his laundry. According to Grice’s Cooperative Principle, if B is being cooperative, his contribution, or implicature, would be helpful in aiding his interlocutor to understand that he did not do the laundry. Grice has characterized the implicature like the example above as a particularized implicature since it occurs out of specific context, instead of arising from the linguistic or literal meaning of the words in the conversation. The latter one is known as a generalized implicature, which does not depend on a particular context to derive the meaning and can be illustrated by the following classic example:

1. (2) a. He is an Englishman; he is, therefore, brave.
2. b. His being an Englishman implies that he is brave.

In example (2), when (2a) is expressed, it entails (2b). This implication does not arise out of a specific context, as in (1), and is purely based on the linguistic or literal meaning of the words. In addition to context that distinguishes conversational implicatures from conventional implicatures, two other features also characterize conversational implicatures: cancellable and reinforceable. For example, what B said in (1) can be interpreted or inferred differently under a totally different context. In other words, the implicature that B didn’t do the laundry would be cancelled without the context of A’s utterance. These two features do not hold for conventional implicatures since whenever (2a) is expressed, it will always entail (2b).

Scalar Inferences

Another form of conversational implicatures concerns scalar inferences [19, 20, 21], which involve using quantifiers like all or some in conversation. Horn [19] stated that logical words like <or, and> or quantifiers like <some, all> actually form information scales. They form a scale because statements such as “A and B” are subsets of statements like “A or B.” Thus the logical words can be ordered in the subset/superset order. This subset/superset relationship can be applied to quantifiers such as <some, all>, with some being the lower ranking term and all being the higher ranking term. For example, assuming that a speaker complies with Grice’s maxims and says as much as s/he can and says what is relevant to the conversation, when s/he says, “I ate some of the cake,” it is possible that the speaker implicates "I did not eat all of the cake," using some as a less informative or a lower ranking term than all. The main idea behind scalar inferences is that the use of a lower ranking term by the speaker prompts the hearer to realize the speaker’s reluctance to use the higher ranking one, i.e., all. In the conversation when the speaker chooses to select “some,” it is possible that the speaker rejects the use of the higher-ranking and stronger sense “all,” which might logically be not true or not warranted. <some, all> is not the only set of scalar inferences. Other possibilities include the use of numerals, modals, adverbs, lexical contrast and partial contextual orderings. Here are some examples of them.

1. (3) Numerals:
2. John: Do you have three chairs?
3. Mary: I have two.
4. Implicature: Mary does not have three chairs.
5. (4) Lexical contrast:
6. Bryan: Did you succeed in getting on the tennis team?
7. Judie: I tried.
8. Implicature: Judie did not succeed in getting on the tennis team.
9. (5) Partial Contextual Ordering
10. Daniel: Did you go to work today?
11. Jennifer: I went shopping.
12. Implicature: Jennifer did not go to work.

Utterances like “some X are Y” can be further illustrated by two inferences: “some and possibly all” and “some but not all.” These two interpretations of “some” can also be viewed as either being encoded logically or pragmatically. Logically speaking, “some” can be interpreted as “some and possibly all,” and is encoded by semantic rules. Pragmatically speaking, “some” is interpreted as “some but not all,” as required by real-world scenarios. Note that logically, there are four possibilities to account for “some X are Y.” These four possibilities include “X is a subset of Y,” “Y is a subset of X,” “X overlaps with Y” and finally “X coincides with Y.” The interpretation “some and possibly all” is considered less informative since it is compatible with all four possibilities. Compared with “some and possibly all,” “some but not all” reduces the four possibilities to only two possibilities and thus “some but not all” is considered to be more informative.

The derivation of implicatures in the original Gricean account has been described as complex and onerous since Grice originally considered that in addition to what has been said, interlocutors also have to evaluate what has not been said in a context. However, it does not seem plausible and efficient from cognitive development and sentence processing stages. If the implicature derivations are too costly, they would not benefit human cognitive development or mental processing. Since the Gricean theory was introduced, two influential accounts have been proposed to explain scalar inferences: the Default Inference and Relevance Theory. In Default Inference, Neo-Gricean researchers [20, 21, 22, 23, 24, 25] like Levinson [20] pointed out scalar inferences resulting from Q-heuristic, which states that “What isn’t said isn’t the case.” When a speaker uses a weaker form, s/he intends not to use the stronger form since the stronger form is not warranted or called for. This has been considered the default inference among interlocutors. Under this assumption, scalar inferences belong to generalized conversational implicatures, which should be derived automatically without considering the context and can speed up the cognitive processes. For Levinson, since the generation of scalar inferences is automatic or by default, it can only be cancelled when the context requires it. In the case of “some but not all” versus “some and possibly all”, the Default Inference predicts that interlocuters would automatically assume the “some but not all” interpretation, which should be least onerous during cognitive computation.

On the other hand, pragmatists who are in favor of contextual accounts are against the default view. According to them, a listener either narrows down or broadens the speaker’s linguistic expressions. Consider the example of “some X are Y” in the above cases. Interpreting some as “some and possibly all” is an example of broadening the scope of some while interpreting some as “some but not all” is an example of narrowing down its meaning. If a listener does not find a scenario that matches the use of some and possibly all, he would realize a narrowing use of some must be applied instead. Furthermore, some pragmatists argue that inferences that the listeners draw depend on how relevant the inferences are. For example, Relevance Theory, which offers a comprehensive account on inferences [26], postulates that when hearing an utterance containing a weaker form like some, the interpretation “some and possibly all” suffices since a less informative interpretation might be enough for listeners rather than “Some but not all,” which is more informative and requires listeners to apply Q-heuristic. Unless the speaker wants the listener to apply Q-heuristic, more cognitive effort is associated with generating a more informative “some but not all” interpretation. In this way, Relevance theory predicts the opposite from the Default Inference theory in the scalar inferences. In the example of “some,” the Relevance theory assumes that the listener would tend to go with the logical interpretation. In other words, when hearing “some X are Y”, interlocutors tend to interpret it as “some and possibly all.” If there is a need for the listener to go further, interlocutors would adopt the “some but not all” pragmatic interpretation.

Previous Second Language Studies

Previously, pragmatics has relied heavily on theorists’ knowledge in theoretical linguistics and intuitions about the language. Experiments on scalar inferences in either first language or second language studies have not relied on experimental techniques until very recently. Studies on first language scalar inferences have begun to use various online techniques to investigate this issue [27, 28, 29, 30]. As shown in the previous sections, Gricean-, Neo-Gricean or Relevance theories have been trying to offer their accounts of pragmatics but none of them has been very explicit about the derivational process in the human cognitive system. In addition, despite many efforts trying to unravel implicatures, surprisingly few studies have been devoted to the psychological reality of scalar inferences. In two of the earlier studies, Bouton [31, 32] compared native speakers’ and non-native speakers’ interpretation of implicature in an attempt to investigate whether it was necessary to teach implicature to second language learners. Bouton found that L2 learners of English comprehended the implicature only 75% of the time. In a follow-up study, Bouton (1992) [32] used different methods, including structure, cloze, dictation, etc., to examine thirty L2 English learners’ comprehension of implicature and again found that even though his L2 subjects made improvement on scalar interpretations over time, their interpretation performances were still significantly different than those of native speakers.

Several downsides could be found in Bouton’s study. First, the study was a continuation of an even earlier study and given that the author decided to use the same group of subjects, it turned out that not many subjects wanted to take the same test again. The loss of many of the original subjects in the second study made the comparison between the two studies difficult. Second, Bouton did not explain the rationale for using four different methods in his study and did not describe whether there was a significant difference among them. Whether each method yielded the same pattern of implicature comprehension could not be compared across four methods. Third, Bouton’s conclusion was mainly drawn from looking at the results in proportions among groups of subjects. More adequate statistical analysis and interpretation would have been needed before any conclusions could be drawn are expected before making any conclusions.

In a more recent study examining French continuous and discontinuous quantifiers, Dekydtspotter and Hathorn [33] used the Truth Value Judgment Task to investigate whether English learners of French could correctly observe the subtle differences of a target sentence like “Something remarkable was observed by each of the researchers.” A continuous or a discontinuous French quantifier leads to subtle differences in interpretations. For a continuous quantifier, the interpretation can be either “all the researchers observed a different remarkable object” or “all the researchers observed the same object.” For a discontinuous quantifier, the interpretation can only be “the researchers did not observe the same object.” Dekydtspotter and Hathorn found that language learners with high language proficiency displayed native-like implicature interpretation while low-intermediate language learners did not show such a tendency. In a study investigating the use of contexts in deriving scalar implicatures among English and Korean first and second language speakers, Slabakova [34] found that Korean learners of English used more pragmatic interpretation in their second language than in their native language, with or without the context, and concluded that pragmatic interpretations were of no problem to second language learners. Given that no beginning learners of either English or Korean were involved in this study, it is unknown whether the same conclusion can be drawn for beginning learners.

Considerable disagreement has arisen concerning the cognitive mechanism underlying the processing of scalar inferences. In addition, how second language learners process scalar inferences is not fully understood yet. How do second language learners’ performances differ from those of native speakers? Would they be influenced by the same factors that affect L2 scalar inference processing? Or would they be less influenced by these L2 factors? What are other factors that could possibly affect cognitive computation? Previously, it has been known that working memory capacity, such as reading span, is highly interconnected with people’s processing of sentences. Would this working memory capacity affect language processing or scalar inferences among second language learners? More detailed discussion of working memory capacity and reading span task will be presented in the third experiment. Given that not many online experimental studies have been done to investigate scalar inferences among second language learners and the few existing studies still rely heavily on Truth Value Judgment Tasks or other offline methods, it is of great interest and importance to apply online techniques to elucidate scalar inferences and psychological reality among L2 language learners.

I now turn to the three experiments in this paper. In Experiment 1, I used the Truth Value Judgment Task and reaction times to examine the two interpretations of “some.” First, which theory, Default Inference or Relevance Theory, can be used to account for second language learners’ performances in scalar interpretation? How long do second language learners spend on processing scalar inferences? Both the offline Truth Value Judgment Task and the online reaction performances provide us a more in-depth window into the cognitive mechanism of scalar inferences among second language learners. If the Default Inference theory is correct, we should be able to observe more and faster acceptance of “some but not all” interpretation. In other words, there should be reliable differences in the acceptance of “some but not all” interpretation than in the acceptance of “some and possibly all”. On the contrary, if the Relevance Theory is correct, we would observe the opposite pattern in participants’ performances. That is, we should observe faster acceptance of “some and possibly all” interpretation. Experiment 2 used the same paradigm as Experiment 1 except that subjects were not given limitless time to respond. Instead, subjects were given a certain period of time to respond. This was done in order to further verify that people’s choice of interpretation would not change under the pressure of time. The second research question would thus be: Would scalar inferences remain the same under time pressure for second language learners? The predictions for the second experiment remain the same basically. If, under time constraint, comprehenders’ choose “some but not all” interpretation more and faster than “some and possibly all”, we have further evidence that Default Inference makes the right inference regarding comprehenders’ cognitive processing. On the contrary, if comprehenders choose “some and possibly all” more and faster than “some but not all”, we have evidence that the Relevance Theory is correct. Experiment 3 extended the findings from Experiment 1 and Experiment 2 and examined one important cognitive factor that contributes to human language processing: working memory capacity. Working memory capacity has been shown to affect sentence comprehension and students’ performance in scholarly aptitude tests. However, much less is known about the role of working memory capacity in second language research. In Experiment 3, I investigate this capacity during scalar inferences. Several research questions are of interest in this paper: The research question for the third experiment is: How would people with a high reading span differ from those with a low reading span? Do they show different patterns of scalar inference processing? We address these questions in the following three experiments.

Furthermore, since the current study aimed to use Mandarin learners of English as participants, two possible concerns should be addressed before running the following experiments. First of all, would Mandarin learners of English be able to read two interpretations of “some” in English? That is, would they be able to read the logical and pragmatic interpretations, just like English native speakers? To examine this issue, we conducted an experiment asking participants who did not participate in the following three experiments to evaluate both English and Mandarin stimuli sentences by using either “some but not all” interpretation or “some and possibly all” interpretation. For example, twenty participants were given instruction about using “some and possibly all” scenario to evaluate stimuli sentences based on 1–5 Likert Scale, with “1” meaning “very unlikely” and “5” meaning “very likely”. Another twenty participants were given instruction about using “some but not all” interpretation to evaluate stimuli sentences based on 1–5 Likert Scale. Then we compared whether there was a difference in their rating between these two interpretations. The average for “some but not all” interpretation was 4.38 while the average for “some and possibly all” interpretation was 4.12. Statistical analysis showed that there was also no difference in participants’ rating (F(1, 38) = 3.97, p > .05). In addition to this, we also checked whether there was a difference between English and Mandarin stimuli. Statistically, there was also no difference between English and Mandarin stimuli (F(1, 38) = 2.13, p > .05).

Experiment 1

The first experiment set out to examine the first research question: Which theory, Default Inference or Relevance Theory, can be used to account for second language learners’ performances in scalar interpretation? Instead of forcing subjects to judge stimuli sentences either logically or pragmatically or to apply real world knowledge during the experiment, Mandarin learners of English were introduced to a context that depicted a condition that could support either a logical or a pragmatic interpretation. For example, a typical context could be: “John has many dictionaries. Some of the dictionaries are used.” In this context, it is possible to interpret some as either some but not all or some and possibly all. After reading the context, subjects were then presented with one of the following target sentences: “Some and possibly all of the dictionaries are used” or “Some but not all of the dictionaries are used.” A total of 40 sets of target sentences (see Appendix I) were used in the Experiment 1 with 20 of them using “some and possibly all” and 20 using “some but not all” in the questions. These target stimuli were then randomized with 40 control sentences (see Appendix II) so that subjects did not detect the manipulation of the experiment. Controls were designed not to include scalar inferences so that they could serve as the basis in the experiment. Typical control sentences would read, “Jennifer has four borrowed books. They were from the library.” The following target sentence could read, “Jennifer has four borrowed books from the library.”

After reading the target sentence, subjects need to respond by clicking either “True” or “False.” Two kinds of responses were recorded: subjects’ reaction times and their responses to the target sentence. Reaction times were measured as the time that subjects took to respond to the question as either true or false. If the Default Inference is correct, the percentage of true responses should be greater in “some but not all” interpretation than that in the “some and possibly all” interpretation. In addition, we can observe less time spent on interpreting some as some but not all since it is the default interpretation. On the other hand, percentage of false responses should be lower in “some but not all” interpretation than that in “some and possibly all” interpretation and the reaction time for false responses in the “some but not all” interpretation should be longer than that in the “some and possibly all” interpretation.

Experiment 2

The results of Experiment 1 indicated that Mandarin learners of English prefer to interpret some as some but not all, as reflected by the results from both reaction time data and the Truth Value Judgment Task. However, since subjects were not required to make their responses within a certain period of time, it is arguably true that they might be able to come up with scenarios that were compatible with either some but not all or some and possibly all. Since the purpose of the experiment was not to let subjects brainstorm during experiments so that they could come up with a possible scenario to fit the interpretation, it was therefore necessary to know if subjects could still make the same responses when they were pressed for time. Experiment 2 set out to examine whether subjects interpreted some as some but not all or some and possibly all under the pressure of time. It was hypothesized that when people are under pressure to respond within a short range of time, they would not be able to brainstorm and possibly find a scenario to fit the other interpretation.

The results of the first experiment showed that people tended to spend around 2500 milliseconds when they were presented with some as some and possibly all and around 1300 milliseconds when presented with some as some but not all. It seemed that when presented with some and possibly all, subjects needed another 1200 milliseconds to evaluate the possibility of this interpretation of some. Experiment 2 thus gave subjects 1500 milliseconds to respond to the stimuli sentences. This was to make sure that subjects could still process the target sentence but couldn’t use real world knowledge or other techniques to come up with a possible scenario. Again, if the Default Inference is correct, the percentage of true responses, under time pressure, should still be greater in “some but not all” interpretation than that in the “some and possibly all” interpretation. In addition, we should still observe less time spent on interpreting “some” as “some but not all” since it is the default interpretation. On the other hand, percentage of false responses should be lower in “some but not all” interpretation than that in “some and possibly all” interpretation and the reaction time for false responses in the “some but not all” interpretation should be longer than that in the “some and possibly all” interpretation.

Experiment 3

Experiment 2 showed that subjects’ derivation of scalar inferences were seriously influenced by the pressure of time. Since the time it takes to process language also correlates with cognitive capabilities, another interesting question may be posed: whether the capacity of cognitive system plays a role in scalar inferences. To be more specific, we want to know if there is a difference in people’s derivation of scalar inferences with regard to high versus low working memory capacity. Before delineating how to approach this research question, I will first briefly introduce working memory and its function with regard to language processing.

Overall Discussion

The three experiments conducted in the current study tried to investigate scalar inferences among Mandarin learners of English by using online experiments. To be more specific, the current three experiments tried to examine which of the two theories, Default Inference or Relevance Theory, is correct in predicting Mandarin learners’ processing of scalar inferences. According to Default Inference, it predicts that interpreting “some” as “some but not all” is the default interpretation that learners will come up with and thus it will be cognitively less onerous for interlocutors. On the contrary, Relevance Theory predicts that the listener would tend to go with the logical interpretation, thus interpreting “some” as “some and possibly all.” If there is a need for the listener to go further, s/he would adopt the “some but not all” pragmatic interpretation.

So far, most of the studies on scalar inferences have focused on native speakers’ offline judgments of scalar inferences or on children’s interpretations of sentences containing scalars. None of them have used online measures to examine second language learners’ processing of scalars. Considerable disagreement has arisen out of these offline studies. The current study thus aims to bridge this gap by using reaction times and true/false judgment tasks to investigate this issue.

Across the three experiments, it has been found that Mandarin learners’ acceptance rate of some but not all was higher than their acceptance rate of some and possibly all. This evidence was also backed up by faster reaction times that subjects demonstrated to some but not all. It indicates that subjects did not hesitate to interpret some as some but not all. The overall reaction time for some but not all was about half as fast as that for some and possibly all. Another interesting issue that can be observed in the current study is that scalar inferences did consume more time than simple structures. Results showed that even though subjects’ reaction times in some but not all were faster than some and possibly all, they were both slower than the reaction times in control stimuli.

Faster reaction times and higher acceptance rate of some but not all could be used as a piece of evidence to support the Default Inference, which hypothesized faster responses for the some but not all interpretation. Given that the reaction times in some but not all were faster than some and possibly all and were more similar to the reaction times for control stimuli, the evidence can be used to indicate that generating a some but not all interpretation was less effortful and could be the default interpretation. In addition, the extra reaction times in some and possibly all conditions can be used to explain that when the default interpretation is not correct, people take time to reject such an interpretation.

Further evidence in support of a Default Inference account comes from the results of Experiment 2, which limited subjects’ responses to only 1500 milliseconds. In this situation, subjects’ rejection rate of some and possibly all was higher than the rejection rate in Experiment 1. This indicates that when people do not have enough time to try to come up with a suitable scenario to fit some and possibly all, they tend to reject it.

Besides finding compelling evidence in support of the Default Inference account, this paper also addressed the role of working memory capacity in scalar inferences. The results in Experiment 3 showed that learners with a high reading span are more likely to process scalar inferences faster and also more likely to generate possible scenarios to fit some and possibly all, as indicated by its higher acceptance rate. However, for subjects with a low reading span, they were slower and possibly more prone to making pragmatic mistakes. Taken all together, the three experiments in this paper demonstrate that experimental procedures help tease apart competing theories raised and contribute to our knowledge of current pragmatic theories.

Appendix I: Experimental Stimuli

1. Mary had many sisters. Some of the sisters were married.

2. Peter owns many keys. Some of the keys are lost.

3. Dennis had a lot of pencils. Some of the pencils were used.

4. May has a lot of flowers. Some of the flowers are yellow.

5. Daniel had many books. Some of the books were dirty.

6. Martha has many brothers. Some of her brothers are really tall.

7. Jessica got a lot of assignments. Some of her assignments were difficult.

8. The movie star has many fans. Some of the fans were teenagers.

9. The teacher has many students. Some of the students are diligent.

10. The man had many uncles. Some of his uncles were rich.

11. The assistant mailed many letters. Some of the letters were long.

12. The manager has many employees. Some of the employees were old.

13. The clerk has got many complaints. Some of the complaints are about his attitude.

14. The cowboy has many horses. Some of the horses are sick.

15. The book has many authors. Some of the authors are female.

16. The computer has found many viruses. Some of the viruses are removable.

17. This city has many old buildings. Some of the old buildings are collapsing.

18. There are many movies on this website. Some of the movies are classic.

19. There are many people in that house. Some of the people are really old.

20. That tall building has many windows. Some of the windows are broken.

21. The patient has got many children. Some of the children are still young.

22. There are many girls in this club. Some of the girls are good-looking.

23. The apartment has many rooms. Some of the rooms are small.

24. There are many photos in that book. Some of the photos are beautiful.

25. The old man has many dogs. Some of the dogs are cute.

26. The store had many customers. Some of the customers were rude.

27. This campus has many trees. Some of the trees are tall.

28. Tom has many gardeners. Some of the gardeners are from the south.

29. This company has many managers. Some of the managers have just retired.

30. That old doctor has a lot of patients. Some of the patients hate him.

31. The farmer has grown a lot of apples. Some of the apples are green.

32. The student has many notebooks. Some of the notebooks are dirty.

33. The king has a lot of servants. Some of the servants are proud.

34. Jasmine went to many different countries. Some of the countries were beautiful.

35. Nicole has bought many dolls. Some of her dolls are really delicate.

36. Johnny prepared many dishes. Some of the dishes tasted great.

37. The policeman has caught many bad guys. Some of the bad guys are really evil.

38. The little boy has many pets. Some of his pets are dogs.

39. Tom’s parents have many bought apartments. Some of the apartments are huge.

40. The professor is working on many projects. Some of the projects are complicated.

Appendix II: Control Sentences

1. Natalie has got many rejections. She is very sad.

2. The old painter lost his salary again. His wife is mad.

3. The businessman loves to travel. He has been to many different countries.

4. Mr. Johnson fired one of his employees. The employee stole his money.

5. The manager is mad. One of his workers got into a fight.

6. The student did not finish his report. He failed the class.

7. The mayor is sick. He needs to see a doctor.

8. Tim doesn’t like his teacher. He thinks she is too strict.

9. Carolyn loves sports. She plans to join the school team.

10. Amanda wants to study abroad. She is trying to earn more money.

11. The king has many knights. They are all very brave.

12. That single mom has two sons. They both go to high school.

13. Kate likes horror movies. She enjoys watching them at night.

14. Amy’s mom needs help. She has problems cleaning a big house.

15. The vending machine has many different products. All of them are cheap.

16. Nurses in that hospital are very busy. They need to take care of many patients.

17. A police officer is talking to a young man. He might have stolen something.

18. Richard has got many presents for his birthday. One of them is a toy car.

19. Many students find this course difficult. They can’t figure out the difficult content.

20. Martha wants to have a baby sister. But her mom wants a baby boy.

21. Many of the queen’s maids are young. They love to work in the palace.

22. John can’t wait to see his father, who has been away for one year.

23. Nancy’s mother went to see a doctor. She has been ill for a long time.

24. Vivian is lucky. She married her high school sweet heart.

25. The driver got a ticket because he went through a red light.

26. The store is not open today because the clerk overslept.

27. The babysitter is playing with the kid, who is only four years old.

28. The salesman is good at selling his products. He made lots of money.

29. The swimmer didn’t see the sign which says, “No Swimming.”

30. The singer has a lot of fans. She constantly meets with them to have fun.

31. The girl wants to be a popular singer but she doesn’t have a beautiful voice.

32. The actor is memorizing his lines, which are quite hard.

33. Mandy loved an expensive watch. She finally asked her husband to buy it for her.

34. The concert has attracted many people. All of them love classical music.

35. Brandon’s grandfather fought in the war. He is proud of him.

36. The writer has to take a part-time job because his novel didn’t sell well.

37. Diane’s parents are rich. They paid to let her go to the best school.

38. The party was fun. Many people didn’t want to leave.

39. Both of her children like to play basketball. They are professional players now.

40. Ron has been studying very hard for the exams. He finally got excellent grades.

Appendix III: Vocabulary Used in the Experiment

Instruction: Below is a list of the vocabulary used in the experiment. Please read all of them and check if there are any words you don’t understand. If there are any words you don’t understand, feel free to ask the experimenter for their meanings.

1. sisters

2. married

3. keys

4. old

5. pencils

6. flowers

7. yellow

8. books

9. tall

10. used

11. brothers

12. assignments

13. difficult

14. movie star

15. fans

16. teenagers

17. hardworking

18. uncles

19. rich

20. assistants

21. letters

22. lost

23. employees

24. manager

25. clerk

26. complaints

27. cowboy

28. sick

29. horses

30. authors

31. female

32. computers

33. viruses

34. removable

35. buildings

36. collapsing

37. movies

38. classic

39. citizens

40. building

41. broken

42. window

43. patient

44. single

45. club

46. children

47. apartment

48. room

49. photos

50. beautiful

51. huge

52. customers

53. rude

54. gardeners

55. Tainan

56. retire

57. farmer

58. notebooks

59. servant

60. proud

61. country

62. beautiful

63. dolls

64. delicate

65. policeman

66. dish

67. evil

68. apartments

69. pets

70. dogs

71. professors

72. projects

73. complicated

74. owns

75. cute

76. rejections

77. painter

78. salary

79. mad

80. businessman

81. travel

82. different

83. countries

84. employees

85. fight

86. finish

87. report

88. fail

89. mayor

90. strict

91. abroad

92. earn

93. knights

94. brave

95. single

96. vending machine

97. products

98. cheap

99. police officer

100. content

101. maids

102. stole

103. palace

104. clerk

105. overslept

106. swimmer

107. fans

108. memorize

109. expensive

110. husband

111. concert

112. attract

113. classical

114. fought

115. professional