A&HL 4008 – Interlanguage Analysis

Spring 2000

J.D. Purdy

Some elements of a second language are more difficult to learn than others. For learners of English, constructions using unaccusative verbs are particularly difficult. The purpose of this investigation is to collect and analyze data regarding judgments made by international graduate students on the verb forms preferred with two different types of unaccusative verbs. It is hoped that the data will reveal patterns in the use of unaccusative verbs, and lead to strategies to help ESL learners correctly use these verbs.

In a construction containing an unaccusative verb, the logical object, or theme of the sentence, is in the position of grammatical subject. Another characteristic of this type of construction is that the subject does not exercise volitional control over the action taking place. In addition, the proposition does not imply the existence of an agent. There are two types of unaccusative verbs. The first contains verbs with a transitive counterpart (unaccusative (+T)). The transitive counterpart can take the active or the passive voice. An example of this type of unaccusative verb is break:

1. The boiling water broke the glass.
2. The glass was broken.
3. The glass broke.

In sentence (1), with a transitive verb in the active voice, the grammatical subject is the agent and the grammatical object is the theme. This is a canonical English sentence. Sentence (2), another canonical sentence in English, is in the passive voice. Here, the grammatical object, or theme, is moved to the position of grammatical subject. In addition, the auxiliary be is used with the past participle of break to signal the use of the passive voice. Although there is no explicit agent in this sentence, an agent is implied. Of course, the agent could be made explicit by adding it in a phrase beginning with the preposition by. The verb in sentence (3) is unaccusative. Like the passive, the theme of the sentence is moved from the grammatical object position to the grammatical subject position. Unlike the passive, however, the movement of the theme to the subject position is not accompanied by a change in the verb form from active to passive. The verb form used is the same as the form that would be used if the grammatical subject were the agent, or logical subject, of the sentence. Zobl (1989), following Marantz (1984), claims that neither the passive nor the unaccusative form is derived from the other. This is said to explain the one-way direction of generalization. Learners generalize the passivization rule to unaccusative verbs, but do not take passive sentences and make them unaccusative. This is "in accord with the status of the passive as a core rule of English." Another feature of unaccusative constructions such as (3) is that they contain neither an explicit nor implicit agent. Celce-Murcia & Larsen-Freeman (1999) list five situations in which agentless sentences are preferred: 1) when the agent is irrelevant; 2) when the objective is to create an aura of mystery or suspense; 3) when the theme of the subject is so fragile or unstable that the change-of-state can take place without the intervention of an agent; 4) when "it is natural to expect change to occur," and; 5) when there are many possible causes for the change-of-state. It is expected that the data collected for this study will show that students prefer to explicitly mark the movement of the theme to the grammatical subject position by using passive verb forms with unaccusative (+T) verbs. The implication of an agent will be a secondary consideration of the students.

The second type of unaccusative verb has no transitive counterpart (unaccusative (–T)). These verbs are differentiated from active intransitive verbs like work and study in that the grammatical subject does not exercise volitional control over the action indicated by the verb. Like the unaccusative (+T) verbs, the grammatical subject is actually the theme of the sentence, and is moved to the grammatical subject position. Also like the (+T) unaccusatives, the verb appears in the active voice, and no agent is implied, either explicitly or implicitly, for example:

(4) A tragic event occurred during the parade.

An important difference between unaccusative (+T) and unaccusative (–T) verbs is that learners do not receive input suggesting that unaccusative (-T) verbs can be passivized. While it is not clear whether students are aware of the semantic difference between the passive and unaccusative in unaccusative (+T) verbs, passivization of unaccusative (-T) verbs clearly shows that the students are marking the movement of the theme to the grammatical subject position by using the be + past participle construction. The data collected for this study are expected to show that despite receiving no input to support it, students will passivize constructions containing unaccusative (-T) verbs.

The subjects in this study were all graduate engineering students at Stevens Institute of Technology in Hoboken. There were six subjects, all male. Four of the students spoke Chinese, one spoke Korean, and one spoke Spanish. These students were classmates in a class designed to improve their English language proficiency, especially in academic writing and speaking. The class was made up of the least proficient graduate students entering Stevens that semester. Although the students had good reading comprehension skills, and decent listening comprehension skills, their production skills were poor. During the semester, the students’ writing had indicated difficulty in handling structures containing unaccusative verbs:

(5) Also the birth rates is decreasing from 62 birth in 1991 to 51 birth in 1998

1. [M]oreover the girls are stopped having babies due to they are interesting

in a better future.

2. The article said that the bill has been lasted for a long time, because….
3. At the remaining time of today I was stayed in my dormitory.

In (5) and (6), the use of the passive is incorrect with the unaccusative (+T) verbs because the students improperly implied the presence of an agent in the propositions. This interpretation assumes that is in (5) is the passive auxiliary be and not the present progressive auxiliary be. This assumption is made because the action clearly took place in the past. Sentences (7) and (8), containing unaccusative (-T) verbs, appear to have been improperly passivized. Thus, it appears that unaccusative constructions involving both types of unaccusative verbs are problematic for this group of ESL learners.

Data Collection Procedures

Data was collected by means of a cloze exercise. The students were given a one-page text in which 28 verbs had been replaced with a blank (appendix A). Of these 28 verbs, eight were unaccusative (-T) verbs, 12 were unaccusative (+T) verbs, six were transitive verbs that had no related unaccusative from, one was a transitive verb with a related unaccusative form, and the last one was an active intransitive verb. Next to each blank was a number, and for each number the subjects were provided with four possible answers (appendix B). Only one of the four choices was correct for the given sentence. The four choices for each blank included at least one version of the verb in the passive voice, and at least one in the active voice. Since the text was in the past tense, the selection of the past progressive was not interpreted as indicating passivization. In these instances, it was not possible to determine whether the auxiliary be was the passive auxiliary or past progressive auxiliary. The students were instructed to choose the best of the four possible choices and write the chosen verb in the blank. The students were given as much time as they needed to complete the exercise.

Analysis

As expected, the students had considerable difficulty with constructions containing unaccusative verbs. Of the 48 opportunities the students had to select the correct form of an unaccusative (-T) verb (eight verbs times six students), the students selected the passive form 11 times, or 23 percent of the time:

4. First, I had shined my shoes last night, and had left them by the door, but between last night and this morning, they were disappeared.
5. My friend was supposed to come for lunch, but she was arrived early.
6. I found out that my uncle was died last week.
7. When the storm was over, only the foundation of the house was remained.
8. Another incident was occurred later in the afternoon.

See appendix C. With the unaccusative (+T) verbs, the students inappropriately used the passive construction at about the same rate. Of the 72 opportunities the subjects had to determine the correct form of the unaccusative verb (12 verbs times six students), the students selected the passive 18 times, or 25 percent of the time:

9. Before I could answer, the door was opened.
10. Finally, as the water was emptied, I noticed a pool of water on the floor – the dishwasher was leaking!
11. I tried to look outside, but moisture was condensed on the windows and I couldn’t see outside.
12. They were shrunk and were so tight that they hurt my feet.

In addition, six transitive verbs that do not have unaccusative forms were included in the exercise. In the one case where the passive voice was required, all six students selected the passive:

(18) His home was destroyed by a tornado.

However, some students used the passive voice when the active was appropriate:

(19) I heard a noise outside and looked out the window to see what was caused

1. My dog must be taken them and carried them out

These sentences differ from sentences containing unaccusative verbs in that the grammatical subject of each verb is the agent, and not the theme. Another verb encountered by the students was the transitive verb hurt. This verb has a related unaccusative form, and the correct form in the given context was the active voice. However, two of the six students chose to passivize the clause:

2. They had shrunk and were so tight that they were hurt my feet.

The last verb included on the cloze exercise was an active intransitive verb. This phrasal verb, look out, was not passivized by any of the students.

As expected, the students had difficulty with sentences containing unaccusative (+T) verbs. It appears that the students found the distinction between the meaning of a sentence such as:

1. The glass was broken.

and

2. The glass broke.

somewhat elusive. In both sentences, the logical object is in the grammatical subject position. This is marked by the passive participle be in (2). In (3), however, this signal is missing. On the other hand, the absence of an implicit or explicit agent in the proposition expressed in (3) is an indication that the sentence belongs in the active voice. In dealing with this conflict, the learners appear to lean towards marking the movement of the object to the subject position, disregarding the requirement for a passive construction to have an agent in either the deep or the surface structure. Perhaps the syntactic rule concerning the construction of a passive structure is more salient than the rule relating to the absence of an agent.

The difficulty learners have with unaccusative (+T) constructions is understandable, since the learner’s input does include sentences in which the logical object precedes the verb in the passive voice. What is less obvious is why learners have about the same level of difficulty with constructions containing unaccusative (-T) verbs. In this study, the frequency of inappropriate passivization was about the same for both types of verbs (23% for (-T) verbs and 25% for (+T) verbs). However, passive voice verb forms are never used with these verbs by native English speakers, and therefore the use of the passive voice with unaccusative (-T) verbs does not occur in the learners’ input. It is possible that a different mechanism leads the student to passivize these sentences. Yip (1995) suggests that learners may consider these unaccusatives to be "underlyingly transitive." She also contends that learners may have the idea that nothing happens without a reason, so an agent must exist, even if it is implicit. According to Yip, Roeper (1987) found in a study of three-year-old children that the children "tended to look for an agent or supply one if it is not present." In this view, the bias for an agent is not due to the operation of a grammatical principle, but to an extralinguistic concept. The passive construction supplies this agent. Another possible factor is that the semantic meaning of verbs of this type may be unclear, so that the learner is not able to say whether an agent plays any role in the action. If a learner does in fact assume that an unaccusative has an underlying transitive argument structure, then the verb must have an explicit or implied agent, and unaccusative (-T) verbs are therefore treated like unaccusative (+T) verbs. In addition, mismatches in the role of agent between semantically similar verbs in the students’ L1 and in English may contribute to this problem.

Another interesting phenomenon was found in the data. Not only unaccusative verbs were passivized by the students, but several structures containing transitive verbs in the active voice were passivized. One of these has a related unaccusative form:

1. They had shrunk and were so tight that they were hurt my feet.

But the other two do not:

(19) I heard a noise outside and looked out the window to see what was caused

1. My dog must be taken them and carried them out

On the other hand, all of the students correctly selected the passive voice when it was required:

1. His home was destroyed by a tornado.

All of these results indicate that the students do not make clear distinctions between unaccusative (+T) verbs, unaccusative (-T) verbs, transitive verbs in the active voice, and transitive verbs in the passive voice.

Yip addresses the learnability problems students have with the unaccusative. She suggests that the Uniqueness Principle may aid the learners in avoiding incorrect verb forms. If learners consider that the meaning of one linguistic form is distinct from the meaning of a second form, and, in the case of unaccusative (-T) verbs, the passive voice form does not appear in the learner’s input, the learner may be led to select the correct form. Unfortunately, the learnability of unaccusative (+T) verbs is still problematic, since both unaccusative and passive forms appear in the input, and the semantic difference between the two is subtle. Yip argues that the unaccusative and passive forms "are collapsed as one undifferentiated meaning, which is mapped onto a single IL form, namely the passive." Thus, acquisition of unaccusative (-T) verb forms depends on the learner’s ability to distinguish between the semantic meanings of unaccusative verb forms and passive verb forms.

Perhaps an avenue for future investigation of the learnability of unaccusative verb forms is to examine students’ perceptions of the meaning of verbs where the meanings differ in whether they imply volitional action, or whether they imply the action of an agent. Making learners aware of the degree of volitional control involved in the action, or the presence or absence of an agent, may alter the students’ cognitive processes when dealing with these types of verbs, and lead the students to produce more native-like English. Verbs that might be good candidates for such studies include stay/remain, continue/last, die/kill, start/begin/initiate, and others that are similar semantically, but differ in the amount of volitional control involved or the role of an agent.

Another approach to the problem of learnability might be to relate the unaccusative as it is expressed in English to the way in which agentless actions are expressed in the learner’s native language. This may help illuminate the mismatches between the meanings of verbs in the learner’s L1 and the corresponding verbs in the target language.

Celce-Murcia, M. & Larson-Freeman, D. (1999). The grammar book: An ESL/EFL

teacher’s course. Boston: Heinle & Heinle

Yip, V. (1995). Interlanguage and learnability: From Chinese to English. Philadelphia:

J. Benjamins.

Zobl, H. (1989). Canonical typological structures and ergativity in English L2