國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER ONE
INTRODUCTION
1.1 Background and Motivation
Overuse of passivization on unaccusatives has been commonly seen in second
language learning. Research about the issue has been made in terms of different aspects (Balcom, 1997; Ju, 2000; Montrul, 1999; Oshita, 1998, 2001; Perlmutter, 1978; Yip, 1990, 1995; Zobl, 1989). Kellerman (1978) discussed the issue in terms of different L2 English proficiency of L1 Dutch speakers and presented a U-shaped curve in language learning. However, only the verb break is involved in the study; in addition, there exists typological difference between Dutch and Chinese. It is worth noticing to examine the performances of L2 Chinese speakers learning English in learning of non-alternating unaccusatives.
Previous studies have claimed that L2 learners of English usually overextend passive voice even if learners belong to different L1 (Balcom, 1997; Ju, 2000;
Montrul, 1999; Oshita, 1998; Yip, 1990; Zobl, 1989). These errors are often found in the unaccusatives as follows.
(1) a. *The World War Ⅲ will be happened in the future. (Chinese: Yip, 1990)
b. *The most memorable experience of my life was happened 15 years ago.
(Arbic: Zobl, 1989)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2
The phenomenon of overusing passivization is known as overpassivization. The verb happen in (1) is an intransitive-unaccusative verb, rather than transitive verb;
they cannot be passivized. According to Zobl (1989), both passives and unaccusatives have identical characteristic of internal argument, without external argument, and both internal NPs were moved to the subject slot. He also argued that there is a lexical rule of unaccusatives before learners acquire passives. The lexical rules of unaccusatives are displayed as follows.
(2) [ ___ [V NP]] (i.e. [open the door])
The lexical rule would be subsumed when L2 learners acquire passives since passive rules are the core rules (Zobl, 1989). On the other hand, the assumption seems to imply the fact that the knowledge of unaccusatives did exist before learners acquire passives. Moreover, under the viewpoints, U-shaped learning of unaccusatives
conducted by Kellerman (1978) also supported the assumption. For example, in his study, the transferability rate from L1 Dutch Het kopje brak into L2 English The cup broke is about 100% in the group of low proficiency while the transferability rate goes
down to 64% in the group of higher proficiency; finally the transferability rate goes up to the 90% in the group of advanced proficiency.
However in Kellerman’s study, there are some problems that we would like to discuss further in order to make the U-shape learning of unaccusatives more
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
convincible. First, just only the verb break was involved in the study. We need more unaccusatives to support the assumption of U-shaped learning.
(3) a. The sun melted the ice.
b. The ice melted.
(4) a. *The truck happened a car accident.
b. A car accident happened.
In the above example, both happen and melt are categorized as unaccusatives.
However, melt has a transitive counterpart while happen does not. Those
unaccusatives with transitive counterparts are known as alternating unaccusatives whereas those without transitive counterparts are noted as non-alternating
unaccusatives. In Kellerman’s study just involved the alternating unaccusative verb break. We would like to examine if the language learning of non-alternating
unaccusatives still presents a U-shaped curve as well.
Second, typological difference might influence the learning of unaccusatives.
Chinese, as a topic-prominent language, allows a thematic patient/theme in subject slot.
(5). Wen-jian ji LE Document send LE “Documents were sent”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4
In the above example, the preverbal NP wen-jian ‘document’ is the topic
followed by a comment ji le ‘sent,’ which is known as topic-comment structure. Most importantly, topic-comment structure in Chinese contains a thematic patient/theme subject with a semantic passive in active voice; similarly the structure is like English unaccusatives as below.
(6) a. The window broke b. A problem emerged.
In the example of (6), both verbs break and emerge are referred to as
unaccusatives. As we could see, both topic-comment structure and unaccusatives share the NP-V word order (i.e. an NP followed by a verb) with semantic passive in active voice. Therefore, for Chinese native speakers, it is possible that L2 Chinese learners of English might get accustomed to using NP-V word order with passive meaning in active voice. That is, topic-comment structure might facilitate the learning of unaccusatives. Thus, the typological issue of topic-comment structure is totally different from the Dutch case provided by Kellerman (1978). It is worth examining if the learning of unaccusatives still presents a U-shaped curve.
Studies also revealed that subject animacy might influence the voice form in the sentence (Croft, 1995; Ferreira, 1994). Voice form in a sentence might be determined by animacy hierarchy: human>animate>inanimate>abstract entities; human entities as
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
subjects are unmarked while inanimate entities as subjects are marked (Croft, 1995).
Unaccusatives usually contain inanimate subject, which might be the reason for L2 learners to use passive voice. Thus, the factor of animacy effect needs to be
considered in learning of unaccusatives.
1.2 Purpose of the Study
Based on the study by Kellerman (1978), there are some limitations that we would like to figure them out. There are three purposes in the study to fulfill the limitations by Kellerman (1978).
(1) To examine if the learning of non-alternating unaccusatives for L2 Chinese learners of English with different proficiency represent a U-shaped curve as the study presented by Kellerman (1978).
(2) To examine how is the influence of the alternating unaccusatives and passives in the U-shaped learning of non-alternating unaccusatives.
(3) To examine if the factor of animacy involves the usage of voice forms when L2 learners are using non-alternating unaccusatives.
The study would shed light on L2 Chinese speakers learning English
unaccusatives. The issue would unveil the problems of the learning of unaccusatives in the hopes of facilitating the language learning. We expected that this finding could contribute to the development of L2 English learning.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
6
1.3 Organization of the Paper
In Chapter two, unaccusatives hypothesis will be introduced first to disclosure
the differences of syntactic configuration, and a discussion about alternating and non-alternating unaccusatives will also be demonstrated. U-shaped learning and previous study are also included. Chapter Three will deal with the methodology of experiment designs and procedures. In Chapter Four, we will report the results and discussion based on the data collected in the experiment. Also, implication for L2 language teaching will be mentioned. Finally, In Chapter Five, we will make a conclusion to summarize the ideas of the study.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER TWO
LITERATURE REVIEW
There is some discussion about why alternating unaccusatives and passives behave similarly with non-alternating unaccusatives. U-shaped learning from Kellerman (1978) will be discussed. Finally we will move to the discussion of animacy effect.
2.1 The Unaccusative Hypothesis
Unaccusatives hypothesis proposed by Permultter (1978) and revised by Burzio (1986) claimed that intransitive verbs could be categorized into two subclasses:
unergaive (e.g. fly, paint, run, dance, play, etc.) and unaccusative verbs (e.g. melt, happen, appear, vanish, break, etc). Despite the fact that both belong to intransitive verbs, they share distinct properties in syntactic and semantic features. Unaccusative verbs, which usually lack of volition and associated with change of state, would take an internal NP, while unergative verbs, whose subjects usually get volition, would take an external NP, syntactic configuration schematized as follows.
(7) a. Unergative verbs: NP [VP V]
b. Unaccusative verbs: _____ [VP V NP] (Levin & Hovav, 1995) In generative grammar, the above structures show that in deep structures, verbs in (7a) require an external argument without internal one and its NP is usually a
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
8
thematic agent. As for the verbs in (7b), it requires an internal argument as a role of thematic patient without external one. However, Burzio (1986) claimed that the object of unaccusative verbs does not get any case underlyingly; it must move to the subject position to get a structural case in favor of case filter. Therefore, unaccusative verbs and unergative verbs share the same subject-intransitive word order but stem from distinct syntactic configurations.
(8) a. Unaccusative: the guesti [VP arrived ti]
b. Unergative: the boy [VP jumped] (Oshita, 2001) Apparently, both (8a) and (8b) look identical. However, the preverbal NP the guest goes from the internal verb phrase position and leaves a trace behind the verb
arrived while the boy in (8b) originates in the preverbal position without movement.
Both constructions process differently in syntactic configuration although they belong to intransitive verbs. Semantically, unaccusative verbs usually correlate with a
thematic patient while unergative verbs fit with a thematic agent (Dowty, 1991), as shown in (8). Since agent always refers to human entities, while patient correlate with objects the thematic differences might lead to the voice forms differences (Croft, 1995). Yip (1990) also suggested that it is possible for L2 English learners to mark the passive morphology when an object appears in subject slot. Further, previous
researchers (Yip, 1990; Zobl, 1989) claimed that learners overextend passives even if
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
they belong to different mother tongue, as exemplified below.
(9) a. *I do not think that such abusive action should be happened to a
twelve-year old child. (Chinese: Yip 1990) b. *The most memorable experience of my life was happened 15 years ago.
(Arbic: Zobl, 1989) The overuse of passivization on unaccusatives as above is known as
overpassivization. This phenomenon has been evidenced in learning English as a second language by different L1 backgrounds (Balcom, 1997; Ju, 2000; Montrul, 1999; Oshita, 1998, 2001; Yip, 1990, 1995; Zobl, 1989)
2.2 The Confusion of Unaccusatives and Passives
Unaccusatives share plenty of characteristics with passives. Studies (Montrul, 1999; Oshita, 1998; Yip, 1990; Zobl, 1989) showed that L2 learners would
overgeneralize passivization, even with different L1 backgrounds when acquiring passives. Zobl (1989) also claimed that as soon as L2 learners acquire passive voice, unaccusative verbs would be subsumed; passive rules become the core rule. In other words, L2 learners are prone to regard the unaccusative verbs as passive verbs. Yip (1995) further pointed out the idea that L2 Chinese learners of English considered unaccusative verbs as transitive ones when recognizing ungrammatical transitives as grammatical and rejecting the correct unaccusative verbs.
‧
The reasons to accept ungrammatical passivized unaccusatives and reject correct unaccusatives might be the similarities between passives and unaccusatives. The similarities between unaccusative verbs and passive verbs could be listed as follows.
Firstly, both unaccusatives and passives consist of one internal argument and their arguments are all moved to the subjection (Perlmutter, 1978) as shown below.
(10) a. The star [VP appeared ____ ] b. The house was [VP painted ____ ]
The internal argument of the unaccusative verb in (10a) moves to the preverbal position while the internal one in (10b) goes forward to the sentence-initial position as well. In addition, the internal arguments always belong to thematic patient or theme.
Thus, these similarities between unaccusatives and passives might confuse L2 learners. Secondly, both (10a) and (10b) remain agentless, in which (10a) does not need any agent in unaccusative verbs whereas the agent in (10b) was suppressed (Burzio, 1986). In contrast, some differences occurred between unaccusatives and passives. Firstly, unaccusative verbs usually lack volition such as appear in (10a) and are associated with change of state verbs (Perlmutter, 1978) while passive verbs carry volition, such as paint in (10b) associated with transitive verb taking an underlying agent with volition1. Additionally, some researchers further came up with a hypothesis
1 Usually passives derived from causative sentences; that is, there must be an agent cause something to happen. Therefore, underlyingly there is supposed to be an agent even though passives are agentless in the surface structures.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
classifying the degrees of unaccusative verbs based on the semantic difference, which is known as Unaccusative Hierarchy Hypothesis as shown below (Sorace, 1993a, 1993b; Sorace & Shomura, 2001).
(11) The Unaccusative Hierarchy
Change of Location (Unaccusative Core) Change of State
Appearance
Continuation of a Pre-existing State
Existence Uncontrolled Process
[Emission]
[Involuntary Reaction]
Controlled Motion Process
Controlled Non-motion Process (Unergative Core)
(Sorace & Shomura, 2001) In the hierarchy, the verbs to the core would be more unaccusative-like or ergative-like than the verbs to the peripheral (Sorace & Shomura, 2001). Most
importantly, the semantic subtypes of the hierarchy from the peripheral unaccusatives to the core ones show the characteristics of non-volition, except for the unergative
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
12
core. As for passive verbs, they could be either volition or non-volition.
(12) a. We know Taipei 101 building.
b. Taipei 101 building is known to us.
(13) a. We bought some books.
b. Some books were bought by us.
In (12), know is the verb classified as condition of state as like unaccusative verbs, but it has transitive counterpart; therefore the verb know could be passivized as in (12b). However, in (13), bought is the verb referred to as a controlled motion verb which could be passivized as well. In other words, passive verbs allow both volition and non-volition. The trait of unaccusatives independent of volition is departed from passive verbs. Secondly, unaccusative verbs are intransitive verbs whereas passive verbs are always transitive verbs. These distinctive features between unaccusatives and passives do not keep L2 learners from making errors in unaccusatives but it seems that their similarities confuse L2 learners. However, the confusion influences the way that L2 learners view unaccusatives as passives while it was never found in the opposite way.
(14) a. Stars appeared.
b. *Stars were appeared.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(15) a. *The fish devoured.
b. The fish was devoured.
The error of regarding unaccusatives as passives as in (14) happened in L2 speakers learning English (Balcom, 1997; Ju, 2000; Oshita, 1998, 2001; Yip, 1990, 1995; Yuan, 1999; Zobl, 1989). On the contrary, the error of incorrectly regarding passives as unaccusatives was never found as in (15). The relationship between transitivization unaccusatives and intransitivization passivization as showed below.
Figure 1: Transitivization unaccusatives and intransitivization passives a. The car accident happened b. *The letter sent
*The car accident was happened The letter was sent.
The phenomenon as above figure is widely used in L2 English at the time when L2 learners are reluctant to accept NP-V word order with respect to unaccusatives (Oshita, 2001). Zobl (1989) also supported that once L2 learnres acquire passives, they would view passives as core rules and subsume unaccusatives. It seems that unaccusatives exist prior to passives for L2 learners. On the other hand,
intransitiization passives are never found in L2 English. It seems that compared to unaccusatives, passives are likely to be more salient in L2 learners’ mind. That might
Transitivization unaccusatives Intransitivization passives
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
14
be the reason that L2 learners of English overpassivize unaccusatives and do not intransitivize passives.
2.2.1 Alternating Unaccusative and Non-Alternating Unaccusative Verbs
Within unaccusative verbs, they could be characterized by alternating unaccusative verbs (e.g. break, melt, sink, etc.) and non-alternating unaccusative verbs (e.g. happen, appear, bloom, etc.). Alternating unaccusative verbs contain causative transitive counterparts but non-alternating unaccusative verbs do not.
Examples are as follows.
(16) a. I broke the window.
b. The window broke.
In (16a), the NP the window is an object of the causative-transitive verb broke while it is the subject of the inchoative-intransitive verb broke in (16b) as an NP-V order. That is, the unaccusative verbs could add a causer as an agent to become causative sentences as in (16a), or represent as the inchoative forms without the agent as in (16b). The behavior of the verb break in (16b) acts like the passive verbs just with the difference on passive morphology. Unlike non-alternating unaccusative verbs, alternating unaccusatives ones have nothing to do with verbs of existence or
appearance. Studies showed that all unaccusative verbs are basically causative verbs (Chierchia 1989, Reinhart 1999), and later some researchers (Levin & Hovav, 1995)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
assumed that alternating unaccusative verbs belong to causative verbs.
Non-alternating unaccusative verbs refer to those verbs that are unaccusatives but, in contrast, were not derived from the causative verbs. In other words,
non-alternating unaccusatives do not have transitive counterparts. Often these verbs represent the characteristics of verbs of existence and appearance. The example is shown below.
(17) a. A star appeared in the sky.
b. *The darkness appeared a star in the sky. (Levin & Hovav, 1995) In (17a), the subject a star could not be in the object position of the causative
sentence in (17b). Also, the darkness does not belong to the internal argument of the verb appear. Within unaccusatives, the difference with respect to the transitives or intransitives leads to the diverse result in passive voice.
(18) a. The window was broken.
b. *A star was appeared.
Even though both break and appear belong to unaccusatives, the one, break, with transitive counterpart in passive voice is grammatical while the other one, appear, ungrammatical. For the phenomenon of passivized unaccustives, Yip (1990)
claimed that learners somehow underlyingly viewed unaccusative verbs as transitive verbs. That is, even though some syntactic configuration exists between alternating
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
16
unaccusatives and non-alternating unaccusatives, L2 English learners could not distinguish them apart. Permultter (1978) also suggested that both alternating and non-alternating unaccusatives fall into identical category for L2 English learners.
Owing to the reason, L2 learners might accidentally correctly use passivized
unaccusatives such as The window was broken, but were not reluctant to accept NP-V structures on unaccusatives such as The window broke. Therefore, even though
alternating and non-alternating unaccusatives share similar syntactic configurations, it is possible that the processes of learning alternating and non-alternating unaccusatives might be different.
2.3 U-shaped Learning
Throughout language development, U-shaped learning plays a key role in the
development of first and second language (Kellerman, 1978; Lightbown, 1983). The phenomenon not only implicates the fact the language learning is rule-based, but also is considered to be the process of language development.
2.3.1 U-shaped in Language Development
The research of sensitivity of alternating and non-alternating unaccusative verbs has been explored by Kellerman (1978). In the study, L1 Dutch subjects are divided into eight groups in accordance with different L2 English proficiency. Subjects are observed whether they would transfer L1 Dutch Het kopje brak as L2 English The cup
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
broke. The results in the experiment showed that the transferability rate is 100% in the
group of age 14 to 17. Then the transferability rate decrease to 64% among age twenties; and finally up to 90% for the advanced group. The transferability rate is represented as a U-shaped in the language development of break.
The U-shaped learning with respect to break unveils three stages in the
development, in which L2 learners accept alternating unaccusatives break with NP-V order at first followed by a rejection as proficiency increases; then finally L2 learners accept the structure when reaching advanced level. Similarly, the phenomenon seems to appear in early child language acquisition. Young children correctly use went referring to past tense of go. However, they incorrectly use goed as its past tense as linguistic competence progressed. Finally they achieve the usage went in the end. The phenomenon has been evidenced that U-shaped learning such as children’s learning with respect to go plays a part through language development.
The appearance of U-shaped development could also be observed in SLA. Three groups of L1 French with different proficiency learning English were examined in their use of English progressive tense – ing (Lightbown, 1983). The performances of correctness concerning the use of -ing are as follows.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
18
Figure 2: The use of –ing in U-shaped learning
Stage 1: He is taking a cake Stage 3: He is taking a cake.
Stage 2: He take a cake
Time
Learners at first use progressive tense –ing both in simple and progressive tense.
As they acquire simple tense, they overgeneralize simple tense and made errors on progressive forms. Finally, once learners could correctly use progressive, they could also use simple tense in correct contexts. Both U-shaped learning in first and second language study showed the sequences of language learning. It is believed that U-shaped learning is a key role in learning of language development.
2.3.2 Previous Study on U-shaped Learning
Unaccusative Trap Hypothesis refers to the fact there are three stages as L2 learners acquiring unaccusatives (Oshita, 1998, 2001). At the first stage,
unaccusatives are misanalyzed as unergatives, and learners paradoxically correctly use NP-V order on unaccusatives. At the second stage, learners are reluctant to accept NP-V order on unaccusatives but substitute passive voice for unaccusatives. At final
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
stage, learners achieve the use of unaccusatives. This hypothesis predicts the
U-shaped learning in the acquisition of English break by L2 Dutch (Kellerman, 1978, 1979) and the learning of progressive tense –ing (Lightbown, 1983).
U-shaped learning in the acquisition of English break by L2 Dutch (Kellerman, 1978, 1979) and the learning of progressive tense –ing (Lightbown, 1983).