University of Colorado, Boulder CU Scholar Undergraduate Honors Theses Honors Program Spring 2017 Verb Bias and Plausibility in English Sentence Processing Kendra Buck [email protected] Follow this and additional works at: http://scholar.colorado.edu/honr_theses Part of the First and Second Language Acquisition Commons, Other Linguistics Commons, Psycholinguistics and Neurolinguistics Commons, and the Syntax Commons Recommended Citation Buck, Kendra, "Verb Bias and Plausibility in English Sentence Processing" (2017). Undergraduate Honors Theses. 1304. http://scholar.colorado.edu/honr_theses/1304 This Thesis is brought to you for free and open access by Honors Program at CU Scholar. It has been accepted for inclusion in Undergraduate Honors Theses by an authorized administrator of CU Scholar. For more information, please contact [email protected]. Verb bias and Plausibility in English Sentence Processing By Kendra Buck Linguistics Department Defense Date: April 10, 2017 Thesis Advisor Professor Barbara Fox Linguistics Department Honors Council Representative Professor Kira Hall Linguistics Department Committee Member Professor Bhuvana Narasimhan Linguistics Department Committee Member Doctor Zhiying Qian Asian Languages & Civilizations Department Buck 1 INTRODUCTION Linguists have long begged the question of what really goes on in the brain when one reads a sentence and processes its meaning. This study aims to discover more about the linguistic processing strategies used by native English speakers when reading a temporarily ambiguous sentence. Ambiguity can be found in written English regularly, but native speakers may not always realize how ambiguous a sentence is because they have learned through practice how the sentence should be interpreted. We will examine how long it takes readers to read each part of a sentence that contains a reduced relative clause in order to analyze where readers encounter difficulty in processing the sentence. This will tell us whether readers are parsing out the structure of the sentence all at once or piece by piece, and in what way they do this by comparing two primary models of language processing. Native Speaker Language Processing Contending models of language processing strategies argue that we may read sentences one part at a time and add to the overall meaning as we go along while others contend that we assume the outcome of the sentence as soon as we start and simply revise our initial interpretation when contrary information comes along. Among studies concerning the methods native speakers use to process sentence structures and information, two primary theoretical houses emerge. These two competing theories are the garden-path model and the constraintbased model. The garden-path model essentially likens the processing of a sentence to walking down a winding path through a garden and choosing between certain splits in the path and continuing down a subsidiary path until reaching the end, and similarly reaching the end of the sentence (Frazier 1978). According to this model, when a reader encounters a phrase, for example, like Buck 2 The man accused [. . .] The reader will likely immediately choose between the possible structural outcomes of the sentence before reading the next words. Based on the frequency of a verb’s use in past participle sentence structures, readers will often choose the more frequent structural scenario before the less frequent usage. For a verb like “accused”, which is frequently used in past-participle, or High-PP, structures, the reader will likely assume upon reading the verb that the man will be the subject of a reduced relative clause (Trueswell 1996). In another case, consider a sentence that begins with The explorer searched [. . .] In this case, because “searched” is a Low-PP verb (Trueswell 1996) and is more frequently used acting upon a direct object rather than as a verb within a relative clause, readers are more likely to choose the direct-object path. They would begin the sentence assuming that the explorer has searched something, not that the explorer is being searched by something or someone else. In both of the above cases, readers are more likely to choose the sentence structure that occurs most frequently with the verb as soon as they read it because they are the most intuitive based on a lifetime of speaking the language and encountering verbs in certain sentence structures and contexts. Another principle that guides native speakers’ sentence processing tactics through the garden-path model is the notion of minimal attachment. The principle of minimal attachment posits that a reader will choose the garden path with the fewest syntactic branches before a more complex one (Frazier 1978). For example, one could read a sentence beginning with The explorer searched [. . .] Buck 3 and immediately assume that the explorer is the subject of a reduced relative clause and is being search by someone or for a certain reason. However, it would be much simpler syntactically to have the explorer simply searching a cave. See the diagrams below for a comparison of two possible syntactic arrangements. Figure 1 – Main clause construction Figure 2 – Reduced relative construction As illustrated in the two syntactic trees above, the main clause construction that could occur after the verb is much simpler than the structure required for a reduced relative construction. When readers choose a sentence structure as they begin a sentence, not only are they likely to consider the more frequently used structure for the type of verb, whether it is High-PP and biased towards reduced relative clause contexts, or whether it is Low-PP and biased towards a main clause outcome, but they are also influenced by the simplicity of the possible structures. They are most likely to choose a main clause outcome if the verb is Low-PP because the main clause structure Buck 4 for a sentence beginning such as the example above is less complex than the reduced relative structure. Similarly, the late closure principle influences readers to choose certain sentential outcomes based on the syntactic links between words and phrases within the sentence. Readers are more likely to attach new words to the most recent previous phrase and keep adding to the same phrase rather than start a new phrase and thus a new syntactic structure. For example, consider the following sentence The boy said that she was going to sing loudly. While reading the sentence above, the reader starts with a noun phrase ‘the boy’ and a verb phrase ‘said that’. If the sentence stops at this point, it remains simple and with minimal attachment. However, when the reader encounters ‘she’, the sentence must be changed into a noun phrase and a verb phrase which includes a sentential complement. The reader then attaches ‘she’ to the verb ‘said’ as an object rather than the agent of a new line of action and continues adding to the same verb phrase. When the reader attaches ‘going to sing loudly’ to the same verb phrase, it keeps the sentence structure relatively simple. The last word of the sentence, ‘loudly’ offers the reader a choice. The adverb can either be a modifier of ‘sing’ or of ‘said’. The late closure principle dictates that the reader will keep adding to the same phrase describing ‘she was going to sing’ rather than close that syntactic branch prematurely and instead jump back up in the syntactic tree to ‘the boy said’. This tendency of late closure along with minimal attachment are key influencers in Frazier’s garden-path model for sentence processing, all combining to support the theory that readers immediately choose one possible outcome of the sentence with the most simple and most frequent syntactic structure. After deciding which outcome the sentence should have, as the reader continues and come across contradictory information, the brain must Buck 5 backtrack to change the interpretation again and again until the sentence is finished, like a person walking down a garden path with twists and forks, backtracking whenever the path leads to a dead end and finding a new path. In opposition to Frazier’s garden-path model is the constraint-based cognition model. The constraint-based theory can be likened to a sort of funnel. As a reader begins a sentence, he keeps all possible structures and outcomes available until reaching a point in the sentences that offers new information and eliminates one or more of the possible outcomes and so on until the end of the sentence (Trueswell 1996; Garnsey et al. 1997; McDonald et al. 1994). Consider the following temporarily ambiguous sentence: The pilot considered by the airline was unavailable for the flight. Under the constraint-based model, when a reader begins the above sentence and reads the verb ‘considered’, he simultaneously considers both the direct object structure and the past participle structure as possible outcomes of the sentence. Because the verb ends in –ed in this sentence, its past participle form and its past tense form are homophonous, and because a pilot could either consider something or be considered, both outcomes at this point are equally possible. This is known as main clause/reduced relative clause ambiguity because until the reader sees more information, the sentence’s outcome is ambiguous between these two structures. Under the constraint-based model, the fact that both outcomes are still possible outweighs the fact that they might not be equally plausible. As the reader keeps these two outcomes in mind and continues reading, when the next portion appears as ‘by the airline’, the reader then can rule out the possibility of a main clause outcome. Only after the disambiguating phrase ‘by the airline’ is read does the reader choose the reduced relative structure as the likely outcome. If more information further along in the sentence, such as ‘was unavailable’ supports the same outcome, Buck 6 the reader continues on, sifting out impossible outcomes and interpretations until the end of the sentence. When native English speakers see a temporarily ambiguous sentence such as “The scientists released by the institution were no longer helpful to the research,” the verb offers information beyond the syntactic information on the surface. Syntactically, this sentence is ambiguous until the reader reaches ‘by the institution’ because of the main clause/reduced relative clause ambiguity in the verb form. However, the verb itself is a High-PP verb and occurs frequently in past participle, reduced relative sentence structures (Trueswell 1996). This kind of frequency information is learned by native speakers only through practice. It is not taught in classrooms when non-native speakers learn English, because learning the PP frequency of every verb in the language would be difficult and require the memorization of arbitrary categorizations. It is, however, this type of intuitive frequency knowledge that allows native speakers to predict that the sentence will have a reduced relative clause despite the equal possibility of a main clause structure simply based on the initial noun phrase and the verb ‘released’, then upon reading ‘by the scientist’, will further confirm the reduced relative clause sentence interpretation. The same type of parsing occurs even before the verb comes into the reader’s mind, when reading the initial noun phrase. Native speakers can also use plausibility cues to predict a sentence’s outcome before finishing it (Garnsey et al., 1997; Trueswell, 1996). Plausibility is semantic information found in the noun phrase of a sentence that tells the reader how common a particular subject is to occur as an agent or a patient for a given verb. Consider the following two sentences. a) The leopard hunted by the explorer proved to be elusive. b) The gem hunted by the explorer proved to be elusive. Buck 7 The two sentences a) and b) above differ only in their subjects. One sentence has a leopard as the object of ‘hunted’ while the other has a gem as the object of the verb. Because leopards are very likely to hunt other animals, it is plausible as the agent of the verb. A gem, however, is very unlikely to hunt something else and is therefore implausible as the agent of the verb. It is more likely to be a patient. Lexical information carried within the Low-PP verb ‘hunted’ that would lead a reader to believe that the sentence will end with a direct object in a main clause. Regardless of this verb bias, native speakers will recognize the plausibility of the subject and begin to interpret and predict the sentence outcome before reading the verb. Even when the sentence contains an implausible noun like ‘the gem’, despite the bias towards a main clause sentence outcome from the verb, the reader will likely entertain a reduced relative outcome because a gem is highly implausible as an agent of the verb ‘hunted’ that could take a direct object. Previous Studies One related study is Trueswell (1996). Trueswell’s study concerns the computational methods likely used to determine whether a verb is in the past tense or past participle form in an ambiguous context within a sentence. In other words, Trueswell tests parsing strategies L1 speakers use to overcome main clause and relative clause ambiguity. He explains that the parallel constraint cognitive model predicts that readers will consider both options at once until context or further lexical information proves it to be one or the other. Frequency information most often points to a main clause structure in general, but certain verbs have a higher frequency for participle form rather than past tense form. He also mentions that the Lexicalist approach links these two types of frequency together to predict that general structure frequency information will Buck 8 most effectively constrain interpretation when verbs have a high participle frequency. The likelihood of the preceding noun to be an agent or a patient of the verb might also play a part in processing, but Trueswell’s paper tests in two experiments whether participle frequency plays a part regardless of syntactic constraints. The first experiment, through self-paced reading tests, showed that subjects indeed had more difficulty with processing sentences with low participle frequency verbs, despite semantic information in the initial noun that would point to a relative clause. The second experiment instead tests subjects using nouns that tend to pair with main clause structures. This experiment’s results indicated that verb participle frequency does not help readers interpret sentences on its own, but depends in part on the quality of the preceding noun as a “good” or “bad” agent. However, high participle frequency verbs were shown to cause less difficulty with “main clause” nouns than low participle frequency verbs, proving that verb frequency plays a role in any of the conditions. Verb transitivity may also play a part, as Trueswell explains, but studies have yet to be conclusive on this particularly. Trueswell determines that this type of pattern might lead to a cognitive model involving ambiguity effects within a parallel framework rather than the previously touted “garden-path” revision effect. Overall, Trueswell’s results support the constraint-based lexicalist approach to sentence processing, which predicts that resolution of syntactic ambiguity mirrors the processes of lexical ambiguity resolution. To conclude, the current study combines the Trueswell’s test sentences and methods with our test sentences and combines Trueswell’s two experiments to manipulate variables more efficiently. This study will also test native English speakers. Our primary research question concerns the relationship between verb bias and plausibility, if one exists, in native language sentence processing. Only Trueswell’s single previous study has examined this. His study, Buck 9 however, manipulated variables one-by-two in the sense that he controlled for plausibility in two separate experiments while manipulating verb bias within each experiment. Our study, however, will manipulate both verb bias and plausibility simultaneously to improve upon Trueswell’s design and pave the way for future studies in this field. We predict for this study that native speakers will combine verb bias and plausibility cues to accurately predict the outcome of sentences most reliably when verb bias and plausibility cues are in agreement between the main clause and reduced relative outcomes. We also predict that, similar to Trueswell’s study, plausibility will have the strongest effect when verb bias information is contradictory in terms of indicating either a main clause or a reduced relative clause structure. METHOD Subjects Thirty-seven native English speakers participated in the experiment. This includes 21 females and 16 males between the ages of 18 and 35 years. These participants are all native English speakers. In all but one of the participants, the first language learned was American English or one of the first languages learned for simultaneous bilingual participants, the exception being one monolingual native speaker of British English. Participants had normal or corrected-to-normal vision and each received $10 as payment for their time in the experiment. Participants were recruited mostly from the University of Colorado Boulder, all but one participant being a current student at the undergraduate or graduate level. Buck 10 Materials Sentences were constructed with 10 High-PP verbs and 10 Low-PP verbs, based on the sentences and verbs used in Trueswell’s (1996) study. 64 items, items 1-20 in Appendix A, from Trueswell’s experiment were duplicated and used along with 44 additional item sets constructed from the same 20 verbs but changing other elements of the sentence. Each of the 20 verb items were put into four different sentences, two containing a plausible subject, and the other two containing an implausible subject. One of each of these two included a relativizing phrase such as “that was” or “who was”. This was done two more times for ten of the verbs and three more times for the remaining ten verbs, resulting in a total of 160 sentences, half using a High-PP verbs and half using Low-PP verbs, half containing a relativizing phrase, and half using plausible nouns and the other half using implausible nouns as subjects. This made half of the sentences at least partially ambiguous without the relativizer and the other half unambiguous including the relativizer. Sentences were normed using Amazon Mechanic Turk and Qualtrics to test for equal plausibility between the two subjects used for each set of sentences. None of the participants in the norming study also participated in the self-paced sentence reading study. Participants in the norming study were asked to rate scenarios on a scale of 1-7 based on their likelihood. A scenario would be rated 7 if it was highly likely and 1 if it was highly unlikely. An example scenario given to norming participants was “How common is it for a lion to hunt something?” or “How common is it for a lion to be hunted by something?”. These two questions and their mean ratings would tell us whether the subject in each of the four sentences we created for every verb item such as “The lion hunted by the tiger escaped into the bushes” was plausible or implausible as an agent and as a direct object of the verb. The norming study results confirmed that all of our Buck 11 plausible subjects were equally as plausible as our implausible subjects were implausible. This minimized the effects of imbalanced plausibility in the read times for the noun phrase and embedded verb phrase of our sentences during the self-paced reading experiments later on. The norming study was done to ensure that any disparity in a subject’s read time would be caused most likely by the temporary ambiguity of the sentences due to the reduced relative clause rather than by noun phrases and subjects that would have been incongruous in the context of the verb and the rest of the sentence. A table of the norming results is below. Implausible Nouns Plausible Nouns Agent Patient Agent Patient High-PP verbs 2.1 5.4 5.7 4.8 Low-PP verbs 2.0 5.3 6.0 4.8 Procedure The experiment was done on the University of Colorado Boulder campus in a quiet, darkened office to minimize distractions for participants. Participants were tested individually using a laptop computer with self-paced moving window reading software. Participants were instructed to use one hand on the spacebar of the laptop to progress through the sentences on screen while using their other hand on the computer’s mouse to answer comprehension questions. Sentences were shown to the participant one word at a time in a self-paced moving window fashion using white text on an otherwise black screen. After each sentence, participants were asked to answer a short yes-or-no comprehension question about the sentence. Half of the answers to comprehension questions for the entire study were ‘yes’ and the other half were ‘no’. Buck 12 The comprehension questions for each of the four sentences in each verb item were all either ‘yes’ or ‘no’. Each of the four sentences for every verb item were sorted into four stimuli lists along with filler sentences randomly so that each list contained the same number of critical and filler sentences as well as High-PP and Low-PP verbs and sentences containing or omitting the relativizer that/who. Participants were given one of the four stimuli lists so that no participant saw more than one condition for each item. For example, any contestant that read the implausible, ambiguous condition “The rabbit hunted by the tiger. . .” would not have also seen either of the plausible conditions for the same item nor the unambiguous version of the same sentence. Stimuli lists were then broken into four blocks and critical and filler sentences were sorted into blocks randomly but maintaining an equal ratio of filler and critical sentences between all four blocks of 40 sentences and their respective questions. The experiment took between 30 and 60 minutes for each participant depending on individual response and read times. RESULTS Preliminary results for the first 28 native speakers tested during this study show a disparate pattern between the ambiguity effect in plausible and implausible sentence conditions. The graphs below show read times over separate regions of the sentence in plausible sentences and in implausible sentences for both High-PP and Low-PP verbs. Read times were measured in milliseconds from the time between presses of the spacebar. The ambiguity effect was computed by subtracting the read times for the unambiguous condition from the read times for ambiguous condition of the same item’s plausible and implausible variations. An ambiguity effect measurement of more than 0 means that there was an ambiguity effect present causing slower Buck 13 read times due to the structural ambiguity between main clause and reduced relative clause outcomes. Ambiguity Effect (ms) Implausible Subject Plausible Subject 20 ● ● ● ● ● ● ● 0 ● ● ● ● ● ● High−PP Low−PP ● ● ● −20 The audience /manager entertained by the comedian left in The audience /manager entertained by the comedian left in Noun Phrase Data were analyzed using mixed-effect modeling, with t>2 being interpreted as significant. Prior to data analysis, reading times that were faster than 100 milliseconds or slower than 2000 milliseconds were deleted. The noun phrase in each of our critical sentences, illustrated here by the ‘The audience/manager’ column in both graphs, shows rather different ambiguity effects for the plausible and implausible conditions. There was a significant interaction between verb type and plausibility (β=18.63, SE=9.00, t=2.1). In implausible sentences, there was no ambiguity effect found in read times for the noun phrase region. An ambiguity effect only appears in plausible conditions and only with Low-PP verbs. This is likely to be a spurious effect, since the ambiguous and unambiguous conditions at this region were comprised of the same words, there should not be any reading time difference. This pattern is likely to disappear after more data are collected. Verb Phrase Buck 14 In the VP region, illustrated in the graphs above by the ‘entertained’ columns, shows slightly more varied ambiguity effect measurements than the noun phrase region. There was a numeric trend that in the implausible conditions, there was an ambiguity effect in the sentences with High-PP verbs. There was also a numeric trend that in plausible conditions, there was an ambiguity effect for Low-PP verbs. However, statistical analysis showed that these numeric patterns were not significant. There were no main effects of ambiguity, plausibility or verb type at this region (ts<2), nor was there any three-way or two-interactions among these three predictors (ts<1). This pattern is expected, because at the verb region, readers are more likely to interpret the verb as main clause verb in ambiguous conditions and as the verb in the reduced relatives in unambiguous conditions. In both cases, readers should not slow down in reading because no ambiguity is encountered. Disambiguating Region The disambiguating region is represented in the graphs above as ‘by the comedian’. At this region, there was a marginally significant main effect of ambiguity (β=6.56, SE=3.49, t=1.88) and a marginally significant interaction between plausibility and ambiguity (β=-12.21, SE=6.84, t=-1.79), which showed that when the subject was implausible as the agent of the following verb, readers were not garden-pathed in either the Low-PP or High-PP conditions. When the subject was plausible as the agent of the subsequent verb, however, there was an ambiguity effect for both Low-PP and High-PP verbs. This shows that when plausibility and verb bias cues are both available for predicting the upcoming sentence structure, native speakers rely more on the plausibility cue than the verb bias cue, such that implausible subjects alone are sufficient for guiding readers away from garden-pathing even when the verbs are more Buck 15 frequently used in main clause structures. Similarly, when the subject is plausible as the agent of the following noun, readers predict an upcoming main clause structure regardless of the biases of the verb, thus experiencing garden-path effect at the disambiguation. There was a numeric trend that in plausible conditions, such as ‘The lion hunted by the wolves. . .’ or ‘the person described by the lady. . .’ verb bias weighs heavily on the ambiguity effect of the disambiguating region. The verb ‘hunted’ is Low-PP, while the verb ‘described’ is High-PP. This numeric trend shows that the ambiguity effect was higher on the disambiguating region in sentences with Low-PP verbs. This tells us that when readers see ‘the lion hunted’, because both plausibility and verb bias point towards a main clause sentence structure, the disambiguating region which points towards a reduced relative outcome cause the reader to pause and take more time to read this region as he or she reinterprets the sentence. The same process happens for High-PP verbs like ‘described’ though the effect is less severe because the verb bias begins pointing the reader toward the reduced relative interpretation before the disambiguation solidifies the reduced relative interpretation despite the noun’s plausibility in a main clause construction. Post-Disambiguating Region At this region, there was a marginally significant three-way interaction between verb bias, plausibility and ambiguity (β=28.49, SE=17.07, t=1.67). The ambiguity effect in the region of the sentence after the disambiguating phrase, represented in the graphs above as ‘left in’ is only apparent in plausible, Low-PP conditions. This may be due to the main clause bias of the verb and its agreement with the plausibility of the noun being directly opposed to the eventual reduced relative outcome indicated by the disambiguating region. Because the ambiguity effect is strongest in all regions of plausible, Low-PP sentence conditions, the ambiguity effect seen Buck 16 after the disambiguation and through the end of the sentence could simply be residual slow-down from the disambiguating region. Conversely, in plausible conditions with High-PP verbs, the verb bias seems to be strong enough to overcome the main-clause-indicative noun plausibility and eliminate ambiguity effects after the disambiguation. Implausible conditions showed no ambiguity effects for either High-PP or Low-PP verbs, suggesting that the plausibility cues alone were enough to overcome any ambiguity effects during the disambiguating region, particularly for Low-PP verbs. This result indicates that while plausibility trumps verb bias in making predictions about the likely upcoming structure in this type of sentences, verb bias helps readers recover from garden-pathing. Readers recover faster from garden-pathing after reading High-PP verbs than Low-PP verbs. DISCUSSION We have not yet analyzed the remaining native speaker participants and combined their results with the 28 participants discussed above. However, the results we have seen from native speakers so far support the findings of previous studies. Our native speaker participants displayed the highest ambiguity effects overall in Low-PP, plausible sentence conditions. When both the verb bias and the plausibility cues pointed toward a main clause outcome, the disambiguation and the eventual reduced relative clause outcome came unexpectedly to readers. This supports the garden-path theory, as readers had to reinterpret an initial main clause assumption as they encountered contrary information. When readers read sentences that led them initially down the main clause path via the verb bias of a Low-PP verb, regardless of the plausibility cues in most cases, the reader seemed to start down a main clause garden path until the disambiguating region caused them to reinterpret the sentence as a reduced relative clause. Buck 17 Our data does support the constraint based theory, however, in that in the implausible condition and for High-PP verbs in the plausible condition, there was no ambiguity effect after the disambiguation. This could be explained by the constraint-based model if the disambiguating region eliminated all other possible sentence outcomes and constrained the reader to only the reduced relative interpretation. This would allow readers to continue on after the disambiguating region constrains their available outcome options without difficulty, as the rest of the sentence fits within the new parameters set by the disambiguation within a reduced relative clause interpretation. Similarly to the results of Trueswell’s (1996) study, our results show an improvement in processing ability in the region following the disambiguation in plausible, HighPP conditions. This confirms the assertion that although plausible nouns cause ambiguity effects during the disambiguating region for High-PP verb sentences, the contradictory bias of the HighPP verb allows for fast recovery from the initial garden-path taken by the reader after the disambiguation. Contrary to Trueswell’s study, however, is the importance found in our results of plausibility. Plausibility had a greater effect on ambiguity and sentence processing difficulty than verb bias in our results, while Trueswell’s results showed more processing difficulty due to verb bias than due to plausibility. This difference in results may be because our subjects were more polarized between highly plausible and highly implausible. This would cause more of a noticeable effect than it would have had the subjects of our sentences all been similarly plausible or implausible as they were in Trueswell’s design. The difference between ours and Trueswell’s results may also be because we manipulated verb bias and plausibility simultaneously within one experiment while Trueswell tested only for verb bias effects while controlling plausibility between two experiments. Because we tested for two variables at once, we are better able to analyze the relationship and interaction between the two factors that we are concerned with at Buck 18 play in ambiguity processing. By directly comparing plausibility effects and verb bias effects on the same sentences and test items, we have improved upon Trueswell’s design for the purpose of examining the interaction between verb bias and plausibility working in native English speakers as they parse out temporarily ambiguous reduced relative clause sentences. Moving forward, we plan to analyze more of the data we have collected from native speakers using these sentences and tests, then to gather and analyze data from L2 English speakers whose L1 is Mandarin or Japanese. We predict that L2 learners will exhibit similar parsing tendencies as those we have seen in our results so far, but that lower proficiency learners will have slower read times and higher ambiguity effects when verb bias and plausibility cues are in agreement toward a main clause structure. This is because this pattern can be found in native speakers, but low proficiency non-native speakers will even more so rely on strong verb bias cues and plausibility cues to help predict the outcome of the sentence, but will not be able to use these two cues interactively until they reach higher proficiency levels. Similar predictions have been confirmed in studies such as Qian et al. (2016) and Lee et al. (2013). This is why the current study will continue to test the same predictions in order to confirm the link proposed by Qian et al. (2016) that L1 word order plays a role in L2 sentence parsing. Studies have tested how Korean speakers interpret ambiguity in L2 English sentences, but Korean has a mandatory complementizer, unlike Japanese, though both have an SOV word order as opposed to the SVO word order of English and Mandarin. We predict that, when native Japanese speakers are tested using the same stimuli presented here, they will show similar parsing patterns to the Korean speakers tested in previous studies due to the difference between their L1 and L2 word orders unless the effect obligatory complementizer in Korean that was found to affect sentence parsing Buck 19 in previous studies is not also found in the results of native Japanese speakers concerning the presence of a relativizer. L2 Language Processing Models Language cognition models differ slightly from the Garden-path and Constraint-based models discussed above when considering how a second-language speaker of English might interpret temporarily ambiguous sentences. One model that has been tested and supported in studies like this one is the Shallow Structure Hypothesis (Clahsen & Felser 2006; Papadopoulou & Clahsen 2003). The Shallow Structure Hypothesis contends that when someone learns a second language later in life, not young enough to be considered a simultaneous bilingual speaker, that person has trouble learning some of the finer nuances of the second language. Second language learners of English, according to the Shallow Structure Hypothesis, must process sentences like the following differently. a) The leopard hunted by the explorer proved to be elusive. b) The gem hunted by the explorer proved to be elusive. L2 speakers of English might only pick up on the shallow information in each word of the sentence and often miss the frequency bias and plausibility information that English speakers learn after extensive use. Both of the sentences above will be ambiguous to an L2 speaker in this case because the surface syntactic information that allows ‘the leopard’ and ‘the gem’ to be agents or patients of the verb ‘hunted’ from a purely grammatical standpoint, so an L2 speaker might entertain both of these options simultaneously or neither of them until reading more information that will clarify such as the disambiguating region ‘by the explorer’. It is only when L2 learners of a language reach higher proficiency levels and approach native-like fluency that they will be more likely to exhibit native-like sentence parsing techniques (Lee, Lu, & Garnsey Buck 20 2013). The Shallow Structure Hypothesis remains at the forefront of L2 language processing theories. A previous study done by Lee, Lu, and Garnsey (2013) tests L2 English speakers’ sentence processing strategies using native Korean speakers particularly. Korean is an SOV language, meaning it places the verb after the subject and object, at the end of the sentence. This makes the verb essentially useless in predicting the outcome of a sentence before the end, so Korean speakers should logically rely on information from other, earlier parts of the sentence to resolve ambiguity and predict the sentence ending. In addition, the Korean language contains an obligatory complementizer equivalent to ‘that’ in English in a sentence like ‘The man read that the article was published last week’. This would make all sentences in Korean that contain a similar sentential complement unambiguous, except the complementizer in Korean appears after the verb in the sentence, again rendering it unhelpful for eliminating ambiguity before the end of the sentence. Lee et al. studied the reading times of L1 Korean speakers reading English sentences as an L2 learner to compare how long it took L2 learners to read each part of a temporarily ambiguous sentence as compared to an L1 native English speaker. They found that native speakers can use verb bias information and complementizer cues independently whether one or both are present in a sentence to predict the outcome of a sentence more or less accurately. Korean speakers, however, only showed similar results at high levels of proficiency. Lower proficiency learners relied on verb bias cues to predict the outcome of the sentence, but had trouble combining these cues with the presence of a complementizer. When the complementizer and the verb bias both pointed towards an SC interpretation, lower proficiency L2 learners had shorter read times than sentences with both a complementizer and a DO-bias verb. This suggests Buck 21 that the DO-bias of the verb threw lower proficiency L2 learners off of the SC interpretation despite the presence of the complementizer. This study supports the previous findings of Garnsey et al. (1997) concerning L2 learners of English approaching native-like parsing strategies as proficiency increases. Two other previous studies (Qian, Lee, & Garnsey (under revision); Qian, Lee, Lu, & Garnsey, 2016) compare native English speakers to L2 learners. These studies involve native Mandarin speakers as well as native Korean speakers. Their studies concern the use of verb bias and plausibility to help readers avoid garden-pathing through ambiguous phrases. As previous studies have shown, Korean speakers rely on information other than verb bias early in the sentence to process it because the verb appears later and they have a mandatory complementizer that aids in the correct interpretation of embedded clauses. The prediction of Qian et al.’s studies is that because Mandarin, like English, has an SVO word order and places the verb before embedded clauses and additionally has a rarely used optional complementizer, speakers are more likely to rely on verb bias as English speakers have in previous studies. They tested 70 Mandarin speakers in English through moving window self-paced reading trials. Reading times were then analyzed for the disambiguating region and the ambiguous noun region. The study found that ambiguous sentences were read more slowly at the disambiguating region than unambiguous sentences but only after DO-bias verbs, and plausibility had no effect on reading times. Overall L2 learners performed similarly to native speakers in pattern, though learners who were more proficient performed better than those who were less proficient, and all had the least difficulty with SC-bias sentences including the complementizer. Reading times for the ambiguous noun revealed a pattern involving verb bias only for SC-bias verbs, with no effects involving the speakers’ proficiency or the noun’s plausibility. These results are seemingly contrary to the claim Buck 22 that L2 learners rely on plausibility more than other kinds of cues to process ambiguous sentences. Congruent with previous studies, though, is the result that L2 learners relied heavily on the complementizer and the verb’s bias, though not as effectively as native speakers. This is likely because Mandarin speakers do not exactly have a complementizer like the ‘that’ phrase that English speakers use. Qian et al.’s study suggests that L1 word order influences how quickly L2 learners learn word-order-dependent cues about structures in their L2 language. If the L1 word order is similar to that of the L2, learners will rely on cues more similarly to native speakers of the L2 than learners whose L1 has a different word order. The results of our study are inconclusive concerning L2 learners of English at this stage, as we have not yet been able to test native Mandarin and Japanese speakers with these sentences. We plan to test native Japanese speakers using the same methods of the current study because Japanese, like Korean, is an SOV language, but does not have an obligatory clarifying complementizer like the one found in Korean nor does it have an obligatory relativizer. As such, we predict that Japanese speakers will rely on verb bias to a degree just as Korean speakers in previous studies have been shown to do, but will also rely on plausibility information when verb bias points to the DO construction. L2 English speakers whose native languages are Mandarin and Japanese should exhibit similar read times and parsing strategies to the Mandarin and Korean speakers from previous studies, confirming the findings of previous studies that as proficiency increases, language processing of an L2 approaches native-speaker-like tendencies but is more strongly presented in speakers of an SVO L1 which shares the same word order as the L2, English. Buck 23 REFERENCES 1. Clahsen, H., & Felser, C. (2006a). Continuity and shallow structures in language processing. Applied Psycholinguistics, 27, 107-126. 2. Clahsen, H., & Felser, C. (2006b). Grammatical processing in language learners. Applied Psycholinguistics, 27, 3-42. 3. Felser, C., & Roberts, L. (2007). Processing wh-dependencies in a second language: a cross-modal priming study. Second Language Research, 23, 9-36. 4. Ferreira, F., & Clifton Jr, C. (1986). The independence of syntactic processing. Journal of Memory and Language, 25, 348-368. 5. Fodor, J. (1983). The Modularity of mind: an essay on faculty psychology: The MIT Press. 6. Frazier, L. (2002). Sentence processing. Psycholinguistics: critical concepts in psychology. Vol. 3. 7. Garnsey, S., Pearlmutter, N. J., Myers, E., & Lotocky, M. A. (1997). The contributions of verb bias and plausibility to the comprehension of temporarily ambiguous sentences. Journal of Memory and Language, 37, 58-93. 8. Jackson, C. (2008). Proficiency level and the interaction of lexical and morphosyntactic information during L2 sentence processing. Language Learning, 58, 875-909. 9. Jackson, C. N., & Dussias, P. E. (2009). Cross-linguistic differences and their impact on L2 sentence processing. Bilingualism: Language and Cognition, 12, 65-82. 10. Lee, E-K., Lu, H-Y., & Garnsey, S. (2013). L1 word order and sensitivity to verb bias in L2 processing. Bilingualism: Language and Cognition, 16, 761-775. 11. MacDonald, M. C., Pearlmutter, N. J., & Seidenberg, M. S. (1994). The lexical nature of syntactic ambiguity resolution. Psychological review, 101, 676-703. 12. Marinis, T., Roberts, L., Felser, C., & Clahsen, H. (2005). Gaps in second Language sentence processing. Studies in Second Language Acquisition, 27, 53-78. 13. McRae, K., Spivey-Knowlton, M. J., & Tanenhaus, M. K. (1998). Modeling the influence of thematic fit (and other constraints) in on-line sentence comprehension. Journal of Memory and Language, 38, 283-312. 14. Papadopoulou, D. (2005). Reading-Time Studies of Second Language Ambiguity Resolution. Second Language Research, 21, 98-120. 15. Papadopoulou, D., & Clahsen, H. (2003). Parsing strategies in L1 and L2 sentence processing. Studies in Second Language Acquisition, 25, 501-528. 16. Qian (2015). The reanalysis and interpretation of garden-path sentences by native speakers and second language learners (Dissertation), University of Illinois at Urbana-Champaign. 17. Qian & Garnsey (in press). Verb bias and plausiblity in the resolution of temporarily ambiguous sentences in English and Mandarin. Proceedings of the 28th North American Conference on Chinese Linguistics. 18. Qian, Lee, & Garnsey (under revision). Native and non-native speakers of English differ in online use of verbbased cues about sentence structure. Bilingualism: Language and Cognition. Buck 24 19. Qian, Lee, Lu, & Garnsey (2016a). Contributions of verb bias and plausibility in L2 processing. Proceedings of the 40th Boston University Conference on Language Development. Someville, MA: Cascadilla Press. 20. Qian, Lee, Lu, & Garnsey (2016b). The comprehension of English garden-path sentences by Mandarin and Korean learners of L2 English. Proceedings of the 38th Annual Conference of the Cognitive Science Society (pp. 608-613). Austin, TX: Cognitive Science Society. 21. Rayner, K., Carlson, M., & Frazier, L. (1983). The interaction of syntax and semantics during sentence processing: eye movements in the analysis of semantically biased sentences. Journal of Verbal Learning and Verbal Behavior, 22, 358-374. 22. Roberts, L., & Felser, C. (2011). Plausibility and recovery from garden paths in second language sentence processing. Applied Psycholinguistics, 32, 299-331. 23. Tremblay, A. (2011). Proficiency assessment standards in second language acquisition research. Studies in Second Language Acquisition, 33, 339-372. 24. Trueswell, J. C. (1996). The Role of Lexical Frequency in Syntactic Ambiguity Resolution. Journal of Memory and Language, 35(4), 566-585. 25. Trueswell, J. C., Tanenhaus, M. K., & Garnsey, S. M. (1994). Semantic Influences On Parsing: Use of Thematic Role Information in Syntactic Ambiguity Resolution. Journal of Memory and Language, 33, 285318. 26. Wilson, M. P., & Garnsey, S. M. (2009). Making simple sentences hard: Verb bias effects in simple direct object sentences. Journal of Memory and Language, 60, 368-392. Buck 25 APPENDIX A – STIMULI; CRITICAL SENTENCES Item 1 1 1 1 2 2 Vtype LowPP LowPP LowPP LowPP LowPP LowPP Plausibility pl pl impl impl pl pl 2 2 LowPP LowPP impl impl 3 3 3 3 4 4 4 4 LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl impl impl 5 5 LowPP LowPP pl pl 5 5 LowPP LowPP impl impl 6 6 6 6 7 7 7 7 8 8 8 8 9 LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl impl impl pl pl impl impl pl Sentence The manager entertained by the comedian left in high spirits. The manager who was entertained by the comedian left in high spirits. The audience entertained by the comedian left in high spirits. The audience that was entertained by the comedian left in high spirits. The host entertained by the pianist was pleased with her choice. The host who was entertained by the pianist was pleased with her choice. The guest entertained by the pianist was pleased with her choice. The guest who was entertained by the pianist was pleased with her choice. The man helped by the fireman was saved from serious injury. The man who was helped by the fireman was saved from serious injury. The victim helped by the fireman was saved from serious injury. The victim who was helped by the fireman was saved from serious injury. The doctor helped by the nurse was confident in the procedure. The doctor who was helped by the nurse was confident the procedure. The patient helped by the nurse was confident in the procedure. The patient who was helped by the nurse was confident in the procedure. The woman helped by the assistant was sent to the biggest office. The woman who was helped by the assistant was sent to the biggest office. The presentation helped by the assistant was sent to the biggest office. The presentation that was helped by the assistant was sent to the biggest office. The lion hunted by the wolves escaped into the bushes. The lion that was hunted by the wolves escaped into the bushes. The rabbit hunted by the wolves escaped into the bushes. The rabbit that was hunted by the wolves escaped into the bushes. The leopard hunted by the explorer proved to be elusive. The leopard that was hunted by the explorer proved to be elusive. The gem hunted by the explorer proved to be elusive. The gem that was hunted by the explorer proved to be elusive. The tiger hunted by the poachers was very difficult to find. The tiger that was hunted by the poachers was very difficult to find. The fur hunted by the poachers was very difficult to find. The fur that was hunted by the poachers was very difficult to find. The man hunted by the woman was all she needed to fix the problem. Buck 26 9 LowPP pl 9 9 LowPP LowPP impl impl 10 10 10 10 11 11 11 11 LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl impl impl 12 12 LowPP LowPP pl pl 12 12 LowPP LowPP impl impl 13 13 LowPP LowPP pl pl 13 13 LowPP LowPP impl impl 14 14 LowPP LowPP pl pl 14 14 LowPP LowPP impl impl 15 15 15 15 LowPP LowPP LowPP LowPP pl pl impl impl 16 16 LowPP LowPP pl pl 16 16 LowPP LowPP impl impl The man who was hunted by the woman was all she needed to fix the problem. The tool hunted by the woman was all she needed to fix the problem. The tool that was hunted by the woman was all she needed to fix the problem. The workers lifted by the crane were deposited on the roof. The workers who were lifted by the crane were deposited on the roof. The bricks lifted by the crane were deposited on the roof. The bricks that were lifted by the crane were deposited on the roof. The man lifted by the platform was able to reach the ceiling. The man who was lifted by the platform was able to reach the ceiling. The chandelier lifted by the platform was able to reach the ceiling. The chandelier that was lifted by the platform was able to reach the ceiling. The woman lifted by the man was light and could be moved easily. The woman who was lifted by the man was light and could be moved easily. The furniture lifted by the man was light and could be moved easily. The furniture that was lifted by the man was light and could be moved easily. The teacher loved by the students was informative and insightful. The teacher who was loved by the students was informative and insightful. The textbook loved by the students was informative and insightful. The textbook that was loved by the students was informative and insightful. The woman loved by the man was more beautiful than imagined. The woman who was loved by the man was more beautiful than imagined. The scenery loved by the man was more beautiful than imagined. The scenery that was loved by the man was more beautiful than imagined. The baby loved by the family was eternally precious to them. The baby who was loved by the family was eternally precious to them. The heirloom loved by the family was eternally precious to them. The heirloom that was loved by the family was eternally precious to them. The specialist requested by the hospital finally arrived yesterday. The specialist who was requested by the hospital finally arrived yesterday. The equipment requested by the hospital finally arrived yesterday. The equipment that was requested by the hospital finally arrived yesterday. Buck 27 17 17 LowPP LowPP pl pl 17 17 LowPP LowPP impl impl 18 18 18 18 19 19 19 19 20 20 20 20 21 21 21 21 22 22 22 22 23 23 23 23 LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl 24 24 24 24 25 25 LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl 25 25 LowPP LowPP impl impl The musician requested by the audience was too loud and out of tune. The musician who was requested by the audience was too loud and out of tune. The song requested by the audience was too loud and out of tune. The song that was requested by the audience was too loud and out of tune. The witness requested by the judge helped the case. The witness who was requested by the judge helped the case. The evidence requested by the judge helped the case. The evidence that was requested by the judge helped the case. The woman scratched by the cat was seriously hurt. The woman who was scratched by the cat was seriously hurt. The sofa scratched by the cat was badly damaged. The sofa that was scratched by the cat was badly damaged. The cat scratched by the fence was no longer perfect. The cat that was scratched by the fence was no longer perfect. The car scratched by the fence was no longer perfect. The car that was scratched by the fence was no longer perfect. The child scratched by the knife had a mark afterwards. The child who was scratched by the knife had a mark afterwards. The plate scratched by the knife had a mark afterwards. The plate that was scratched by the knife had a mark afterwards. The knife scratched by the dishwasher lost its shiny coating. The knife that was scratched by the dishwasher lost its shiny coating. The bowl scratched by the dishwasher lost its shiny coating. The bowl that was scratched by the dishwasher lost its shiny coating. The thief searched by the police had the missing weapon. The thief who was searched by the police had the missing weapon. The room searched by the police contained the missing weapon. The room that was searched by the police contained the missing weapon. The man searched by the detective had the evidence. The man who was searched by the detective had the evidence. The file searched by the detective had the evidence. The file that was searched by the detective had the evidence. The explorer searched by the natives was suspiciously unassuming. The explorer who was searched by the natives was suspiciously unassuming. The device searched by the natives was suspiciously unassuming. The device that was searched by the natives was suspiciously unassuming. Buck 28 26 26 26 26 27 27 27 27 28 28 28 28 29 29 29 29 30 30 30 30 31 31 LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP LowPP pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl 31 31 LowPP LowPP impl impl 32 32 32 32 33 33 33 33 34 34 LowPP LowPP LowPP LowPP HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl impl impl pl pl 34 34 HighPP HighPP impl impl 35 HighPP pl The student searched by the principal had the missing books. The student who was searched by the principal had the missing books. The desk searched by the principal had the missing books. The desk that was searched by the principal had the missing books. The client wanted by the agency was worth a lot of money. The client who was wanted by the agency was worth a lot of money. The account wanted by the agency was worth a lot of money. The account that was wanted by the agency was worth a lot of money. The criminal wanted by the FBI was never found. The criminal who was wanted by the FBI was never found. The proof wanted by the FBI was never found. The proof that was wanted by the FBI was never found. The judge wanted by the CIA remained hidden for years. The judge who was wanted by the CIA remained hidden for years. The artifact wanted by the CIA remained hidden for years. The artifact that was wanted by the CIA remained hidden for years. The president wanted by the cabinet was nowhere to be found. The president who was wanted by the cabinet was nowhere to be found. The clipboard wanted by the cabinet was nowhere to be found. The clipboard that was wanted by the cabinet was nowhere to be found. The director watched by the detective was in a bad part of town. The director who was watched by the detective was in a bad part of town. The building watched by the detective was in a bad part of town. The building that was watched by the detective was in a bad part of town. The critic watched by the police was boring and uninsightful. The critic who was watched by the police was boring and uninsightful. The scene watched by the police was boring and uninsightful. The scene that was watched by the police was boring and uninsightful. The friend accepted by the man was very impressed. The friend who was accepted by the man was very impressed. The award accepted by the man was very impressive. The award that was accepted by the man was very impressive. The woman accepted by the manager was sent to headquarters. The woman who was accepted by the manager was sent to headquarters. The application accepted by the manager was sent to headquarters. The application that was accepted by the manager was sent to headquarters. The student accepted by the school was very impressive. Buck 29 35 35 35 36 36 36 36 37 37 37 37 38 38 38 38 39 39 39 39 40 40 HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl 40 40 HighPP HighPP impl impl 41 41 41 41 42 42 HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl 42 42 HighPP HighPP impl impl 43 43 HighPP HighPP pl pl 43 43 HighPP HighPP impl impl The student who was accepted by the school was very impressive. The transcript accepted by the school was very impressive. The transcript that was accepted by the school was very impressive. The witness accused by the investigator had no real alibi. The witness who was accused by the investigator had no real alibi. The suspect accused by the investigator had no real alibi. The suspect who was accused by the investigator had no real alibi. The lawyer accused by the man was innocent after all. The lawyer who was accused by the man was innocent after all. The dog accused by the man was innocent after all. The dog that was accused by the man was innocent after all. The person adopted by the couple was happy to have a home. The person who was adopted by the couple was happy to have a home. The child adopted by the couple was happy to have a home. The child who was adopted by the couple was happy to have a home. The lady adopted by the millionaire was eternally grateful. The lady who was adopted by the millionaire was eternally grateful. The hamster adopted by the millionaire was eternally grateful. The hamster that was adopted by the millionaire was eternally grateful. The secretary considered by the committee had some limitations. The secretary who was considered by the committee had some limitations. The alternatives considered by the committee had some limitations. The alternatives that were considered by the committee had some limitations. The caterer considered by the bride was the best choice. The caterer who was considered by the bride was the best choice. The dress considered by the bride was the best choice. The dress that was considered by the bride was the best choice. The manager considered by the company was all wrong for the job. The manager who was considered by the company was all wrong for the job. The strategy considered by the company was all wrong for the job. The strategy that was considered by the company was all wrong for the job. The pilot considered by the airline was unavailable for the flight. The pilot who was considered by the airline was unavailable for the flight. The plane considered by the airline was unavailable for the flight. The plane that was considered by the airline was unavailable for the flight. Buck 30 44 44 44 44 45 45 45 45 46 46 HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl impl impl pl pl 46 46 HighPP HighPP impl impl 47 47 47 47 48 48 48 48 49 49 49 49 50 50 50 50 51 51 HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl impl impl pl pl impl impl pl pl impl impl pl pl 51 51 HighPP HighPP impl impl 52 52 HighPP HighPP pl pl 52 52 HighPP HighPP impl impl The person described by the lady was quite beautiful. The person who was described by the lady was quite beautiful. The necklace described by the lady was quite beautiful. The necklace that was described by the lady was quite beautiful. The journalist described by the witness was vague yet alarming. The journalist who was described by the witness was vague yet alarming. The situation described by the witness was vague yet alarming. The situation that was described by the witness was vague yet alarming. The child described by the woman was smaller than she expected. The child who was described by the woman was smaller than she expected. The house described by the woman was smaller than she expected. The house that was described by the woman was smaller than she expected. The guy described by the victim matched the police profile. The guy who was described by the victim matched the police profile. The gun described by the victim matched the police profile. The gun that was described by the victim matched the police profile. The mailman expected by the doctor arrived much too late. The mailman who was expected by the doctor arrived much too late. The package expected by the doctor arrived much too late. The package that was expected by the doctor arrived much too late. The performer expected by the venue manager arrived on time. The performer who was expected by the venue manager arrived on time. The stereo expected by the venue manager arrived on time. The stereo that was expected by the venue manager arrived on time. The girl expected by the couple was two weeks early. The girl who was expected by the couple was two weeks early. The letter expected by the couple was two weeks early. The letter that was expected by the couple was two weeks early. The professor expected by the board had been sent to the wrong room. The professor who was expected by the board had been sent to the wrong room. The reports expected by the board had been sent to the wrong room. The reports that were expected by the board had been sent to the wrong room. The speaker proposed by the group works perfectly for the program. The speaker who was proposed by the group works perfectly for the program. The solution proposed by the group works perfectly for the program. The solution that was proposed by the group works perfectly for the Buck 31 53 53 53 53 HighPP HighPP HighPP HighPP pl pl impl impl 54 54 54 54 55 55 HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl 55 55 HighPP HighPP impl impl 56 56 56 56 HighPP HighPP HighPP HighPP pl pl impl impl 57 57 57 57 58 58 58 58 59 59 59 59 60 HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP pl pl impl impl pl pl impl impl pl pl impl impl pl 60 HighPP pl 60 HighPP impl 60 HighPP impl 61 HighPP pl program. The artist proposed by the planner was appropriate for the job. The artist who was proposed by the planner was appropriate for the job. The budget proposed by the planner was appropriate for the job. The budget that was proposed by the planner was appropriate for the job. The leader proposed by the CEO ended up failing later on. The leader who was proposed by the CEO ended up failing later on. The plan proposed by the CEO ended up failing later on. The plan that was proposed by the CEO ended up failing later on. The engineer proposed by the advisor helped the project immensely. The engineer who was proposed by the advisor helped the project immensely. The database proposed by the advisor helped the project immensely. The database that was proposed by the advisor helped the project immensely. The writer recorded by the secretary could not be understood. The writer who was recorded by the secretary could not be understood. The message recorded by the secretary could not be understood. The message that was recorded by the secretary could not be understood. The woman recorded by the notary was surprisingly truthful. The woman who was recorded by the notary was surprisingly truthful. The testimony recorded by the notary was surprisingly truthful. The testimony that was recorded by the notary was surprisingly truthful. The witness recorded by the jury revealed the real murderer. The witness who was recorded by the jury revealed the real murderer. The story recorded by the jury revealed the real murderer. The story that was recorded by the jury revealed the real murderer. The delegate released by the terrorist had been unharmed. The delegate who was released by the terrorist had been unharmed. The hostage released by the terrorist had been unharmed. The hostage who was released by the terrorist had been unharmed. The scientists released by the institution were no longer helpful to the research. The scientists who were released by the institution were no longer helpful to the research. The findings released by the scientist were no longer helpful to the research. The findings that were released by the scientist were no longer helpful to the research. The gourmet selected by the judges did not deserve to win. Buck 32 61 61 61 62 62 HighPP HighPP HighPP HighPP HighPP pl impl impl pl pl 62 62 63 63 63 63 64 64 HighPP HighPP HighPP HighPP HighPP HighPP HighPP HighPP impl impl pl pl impl impl pl pl 64 64 HighPP HighPP impl impl The gourmet who was selected by the judges did not deserve to win. The recipe selected by the judges did not deserve to win. The recipe that was selected by the judges did not deserve to win. The candidate selected by the firm worked perfectly for the task. The candidate who was selected by the firm worked perfectly for the task. The plan selected by the firm worked perfectly for the task. The plan that was selected by the firm worked perfectly for the task. The woman selected by the contest won an expensive prize. The woman who was selected by the contest won an expensive prize. The act selected by the contest won an expensive prize. The act that was selected by the contest won an expensive prize. The actor selected by the producer was too expensive for the budget. The actor who was selected by the producer was too expensive for the budget. The camera selected by the producer was too expensive for the budget. The camera that was selected by the producer was too expensive for the budget.
© Copyright 2026 Paperzz