Middle Verbs and Their Acquisition

The ultimate goal of this paper is to provide an in-depth analysis of the L2 learners’ acquisition of middle constructions. One of the properties of middle constructions is that the subject of middle verbs must be a definite NP. Middle verbs can occur by revealing an intrinsic characteristic or property of the definite NP. In addition, middle verbs can only occur along with short adverbs such as well and easily. In this paper, we conduct a survey and evaluate the L2 learners’ responses to middle verbs. More specifically, we have examined how the L2 learners acquired the abstract constraint, the middle construction condition, the definite DP condition, and the adverb constraint. It is worth noting that 45% of the adult subjects acquired the definite condition, 32.5% of the L2 learners acquired the middle construction condition, 50% of the adult subjects acquired the abstract noun constraint, and 20% of the L2 learners acquired the adverb constraint. This in turn indicates that that the abstract noun constraint was first acquired by the adult subjects, followed by the definite DP condition, the middle construction condition, and the adverb constraint, in that order. With regard to middle verbs, it is significant to note that English are subject to the abstract noun constraint, the middle construction condition, the definite DP condition, and the adverb constraint, whereas Korean is not subject to the abstract noun constraint, the middle construction condition, and the definite DP condition. This in turn suggests that Korean is a superset language, compared to English. Thus, learning difficulty arises. Finally, this paper argues that unlike Chomsky’s UG hypothesis (Chomsky 1981, 1982, 2019a, 2019b), the L2 learners looked for similarities between L1 and L2 and thus relied on their L1.


Introduction
The main goal of this paper is to provide a detailed analysis of college students' acquisition of middle constructions. We aim to examine how the Korean learners of English acquired middle verbs. One of the properties of middle verbs is that the subject of middle verbs must be a definite NP. Middle verbs can appear by revealing an intrinsic characteristic or property of the definite NP. In addition, middle verbs can only occur along with short adverbs such as well and easily. In this paper, we conduct a survey and evaluate the L2 learners' responses. The organization of this paper is as follows. In section 2, we argue that the internal arguments of middle verbs must be a definite DP. We further argue that an intransitive verb can occur in the middle construction only when a definite DP has an event that is an intrinsic characteristic or property of the definite DP. Additionally, we show that short adverbs such as easily and well in the middle construction are necessary with restrictions on the use of agent adverbs. In section 4 and section 5, we examine how the L2 learners acquired the abstract constraint, the middle construction condition, the definite DP condition, and the adverb constraint. It is worthwhile noting that 45% of the adult subjects acquired the definite condition, 32.5% of the L2 learners acquired the middle construction condition, 50% of the adult subjects acquired the abstract noun constraint, and 20% of the L2 learners acquired the adverb constraint. This in turn indicates that that the abstract noun constraint was first acquired by the adult subjects, followed by the definite DP condition, the middle construction condition, and the adverb constraint, in that order. With respect to middle verbs, it is significant to note that English are subject to the abstract noun constraint, the middle construction condition, the definite DP condition, and the adverb constraint, whereas Korean is not subject to the abstract noun constraint, the middle construction condition, and the definite DP condition. This in turn suggests that Korean is a superset language, compared to English. Thus, learning difficulty arises. Finally, we contend that unlike Chomsky's UG theory (Chomsky 1981(Chomsky , 1982(Chomsky , 2019a(Chomsky , 2019b, the L2 learners looked for similarities between L1 and L2 and thus relied on their L1.

The Internal Arguments of Middle Verbs
Kang (2019) proposes the definite DP condition in which the subject of middle verbs must be a definite DP: (1) Definite DP Condition: "The internal arguments of middle verbs must be a definite DP". "Definite DP: the NP, this NP, that NP, these NP, adjective NP, genitive NP, no NP, proper nouns, NP with semantic definiteness" (Kang, 2019, p. 35146) The following sentences provide confirmation that the subject of middle verbs must be a definite DP: (2) a. The glasses break easily.
b. Mary photographs well. e. *Chickens are killing easily.
One of the reasons why (2a) is grammatical is that the subject of the middle verb break is a definite NP (namely, the glasses) in accordance with the definite NP condition. Likewise, (2b) is grammatical since the subject of the middle verb photograph is Mary. In this example, the subject of photograph is the proper noun Mary, which is accordance with the definite DP condition, hence the grammaticality of (2b). (2c) is grammatical since the subject of the middle verb sell is his novel. In this example, the subject his novel meets the definite DP condition. That is, the adjective NP his novel has the definiteness, which results in the grammaticality of (2c). When it comes to (2d), things are different.
Love letters do not seem to carry the definiteness, but they are the subcategory of a letter, thus carrying the semantic definiteness, which results in the grammaticality of (2d). (2e) is ungrammatical since the subject of the middle verb kill (chickens) is not a definite DP, which leads to the ungrammaticality of (2e). Note, however, that the definite DP these chickens instead of chickens make (2e) grammatical, which meets the definite DP condition. We thus conclude that the subject of middle verbs must carry the definiteness.

The Middle Construction Condition
In the following, we aim to examine how a middle verb can occur in the middle construction. As a condition of middle construction formation, we adopt (3): (3) Middle Construction Condition "An intransitive verb can occur in the middle construction only when a definite DP has an event which is an intrinsic characteristic or property of the definite DP." (Kang, 2019, p. 35146) In order to examine whether (3) works for middle constructions or not, let us consider the following sentence: (4) The steak we ate yesterday cuts like butter.
In this example, the subject of the middle verb cut is the steak, which is a definite DP, thus meeting the definite DP condition. In addition, the definite DP the steak has the event of cutting like butter as its intrinsic characteristic or property, hence the grammaticality of (4). Thus, it is reasonable to assume that the middle verb cut can occur in the middle construction. Now let us observe the following sentence: (5) The cats frighten easily.
The subject of the middle verb frighten is the cats, thus observing the definite DP condition.
Additionally, the definite DP the cats have the event of frightening easily as their intrinsic characteristic or property, which leads to the grammaticality of (5). The reason why (5) is grammatical is that (5) meets the definite DP condition and the middle construction condition. Now let us observe the following sentence: (6) *The Eiffel Tower sees from my window.
In (6), the definite DP condition is satisfied since the subject of see is the definite DP the Eiffel Tower.
In (6), the definite DP condition is satisfied, but (6) violates the middle construction condition. That is to say, the definite DP the Eiffel Tower has the event of seeing from my window, which is not the intrinsic characteristic or property of the Eiffel Tower, hence the ungrammaticality of (6). Thus, it is reasonable to assume that "an intransitive verb can appear in the middle construction only when a definite DP has an event which is an intrinsic characteristic or property of the definite DP" (Kang, 2019, p. 32548).

Adverbs
In what follows, we aim to show that short adverbs in the middle construction are necessary with restrictions on the use of adverbs. Bassac and Bouillon (2002) point out that as illustrated in (7), there are restrictions on the use of adverbs in the middle construction: (7) a. Neutrogena rinses away completely. As indicated in (7), there are restrictions on the use of adverbs in the middle construction. More specifically, agent adverbs cannot occur in the middle construction.
The following examples provide confirmation that there are restrictions on adverbs that modify agent subjects: (8) a. *This little flashlight plugs in expertly.
b. *Red wine spots wash out carefully.
c. *Cotton irons cautiously. (Kang, 2019) An important question that naturally arises is "why do we need short adverbs in the middle construction?" Let us observe the following sentences: (9) a. The car drives well.
b. The car drives easily.
As indicated in (9a) and (9b), the middle verb drive requires short adverbs such as well and easily. The reason why we need short adverbs such as well and easily is that these adverbs indicate the intrinsic characteristics or properties of subjects along with middle verbs. In (9), the short adverbs well and easily indicate intrinsic characteristics and properties of the definite DP the car. This is why short adverbs along with middle verbs in the middle construction are necessary. In addition, this fact backs up the claim that both (1) and (3) are necessary in the middle construction. We thus conclude that short adverbs such as easily and well in the middle construction are necessary with restrictions on agent adverbs.

The Goals of Experiments
We aim to provide answers to five main questions. The main goal of this paper is to provide a detailed analysis of the L2 learners' acquisition of middle constructions. We aim to answer the following main questions: Do the Korean learners of English have the knowledge of middle verbs? Does learning take place through transfer? Do the L2 learners have the knowledge of the definite NP condition, the middle construction condition, the abstract DP constraint, and the adverb constraint? Do the Korean learners of English reflect the acquisition order with respect to these conditions and constraints? What do they show with respect to markedness?

Subjects
Forty Korean EFL college students participated in our experiment. These subjects are undergraduate students who are attending my English grammar class (3 credits). We conducted our experiment by online chat of Zoom. We asked forty students whether eight Korean sentences and eight target English sentences are grammatical or not. We did not give the L2 learners feedback during our experiment.

Results
First, we included (10a) and (10b) in our experiment to evaluate the L2 learners' general knowledge of middle verbs: (10) a. Ku piano-nun swipke yencwutoynta.
The piano-TOP easily play (The piano plays easily.) b. The piano plays easily.
In (10a) and (10b), in order to acquire middle verbs, the L2 learners must know that the internal argument of the so-called middle verb must be a definite DP. In addition, they must know that the definite DP the piano has the event of playing easily as its intrinsic characteristic or property. more than half of the adult subjects did not acquire middle verbs. That the L2 learners' incorrect responses were 55% suggests that more than half of the L2 learners did not acquire the definite DP condition and the middle construction condition. That the Korean learners of English showed the similar pattern in (10b) corresponding to (10a) suggests that the L2 learners' acquisition may have taken place through positive transfer even though L1 and L2 are not identical. This in turn implies that our transfer hypothesis does not back up "Chomsky's hypothesis that learners do not need to rely on their L1 (a zero transfer position)" (Ellis, 2015).
We included (12a) and (12b) to evaluate the abstract noun constraint that abstract nouns cannot be the subject of middle verbs: (12) a. Ku sonata-nun swipke yencwutoynta.
The sonata-TOP easily play (The sonata plays easily.) b. *The sonata plays easily.
As pointed out by Kang (2019), the abstract noun constraint can come from the definite DP condition in a broad sense since the subject of middle verbs must be a definite DP: (13) Abstract Noun Constraint "An abstract noun cannot be the subject of a middle verb". (Kang, 2019) Now let us consider the following examples: (14) a. The piano plays easily.
b. *The sonata plays easily. The reason why (14a) is grammatical is that (14a) meets the definite DP condition. In addition, (14a) does not violate the abstract noun constraint since the subject of the middle verb play is the piano, which is not an abstract noun. Another reason why (14a) is grammatical is that the definite DP the piano has the event of playing easily, which is the intrinsic characteristic or property of the piano.
However, (14b) is ungrammatical since it violates the abstract noun constraint that "an abstract noun cannot be the subject of a middle verb" (Kang, 2019). Interestingly, 39 adult subjects thought of (12a) as grammatical, whereas 20 adult subjects thought of (12b) as grammatical. More specifically, the L2 learners' correct responses to (12b) were 50%, whereas their incorrect responses to (12b) were 50%.
This in turn suggests that half of the Korean learners of English acquired the abstract noun constraint. It is important to note that in the Korean sentence (12a), the abstract noun sonata allows the middle verb yencwuhata "play". On the other hand, in the English sentence (12b), the abstract noun sonata does not allow the middle verb play, which in turn indicates that Korean is a superset language, as compared to English. Interestingly, the same applies to (15a) and (15b) The reason why (16a) is grammatical is that the subject of the middle verb break is a definite DP and that the definite DP has the event of being broken easily, which the intrinsic characteristic or property of this bottle, thus satisfying two conditions (the definite DP condition and the middle construction condition. Likewise, (16b) is grammatical since (16b) meets the two conditions. More than anything else, the subject of the middle verb break has the event of being broken easily, hence the grammaticality of (16b). When it comes to (17a), it is grammatical even though it violates the middle construction condition. More specifically, i-tali "this bridge" has the event of building easily, which is not the intrinsic characteristic or property of i-tali "this bridge". Simply put, (17a) is grammatical, despite the fact that (17a) violates the middle construction condition. On the other hand, (17b) is ungrammatical since it violates the middle construction condition. That is, the definite DP this bridge has the event of building easily, which is not the intrinsic characteristic or property of this bridge, thus resulting in the ungrammaticality of (17b). Interestingly, the L2 learners' correct responses to (16b) were 70%, whereas their incorrect responses to (16b) were 30%. On the other hand, the adult subjects' correct responses to (17b) were 32.5%, whereas their incorrect responses to (17b) were 67.5%. The reason why the L2 learners showed the different responses to (16b) and (17b) may be that in (16b), a property of this bottle is the cause of the event of being broken, whereas a property of this bridge is not the cause of the event of building easily. In this respect, (17b) is more marked than (16b). One of many hypotheses relating to markedness is that "learners acquire less marked structures before more marked ones" (Ellis, 2015). Thus, the L2 learners may have acquired less marked structures (16b) before more marked structures (17b). Again, the reason why (17a) is grammatical, despite the fact that it violates the middle construction condition may be that Korean is a superset language, compared to English. Ellis (2015) maintains that "where L1 and L2 are identical, learning can happen easily through positive transfer, but they are different, learning difficulty arises" (Ellis, 2015). Thus, the reason why the L2 learners' incorrect responses to (17b) were 67.5% may be that L1 and L2 are not identical. Note that the L2 learners' incorrect responses to (17a) were 45%.
We included (18a) and (18b) to assess the L2 learners' knowledge of the definite DP condition.
English is subject to the definite DP condition, whereas Korean is not. More specifically, in Korean, the subject of middle verbs need not be a definite DP. Thus, (18a) is grammatical, despite the fact that the subject of kill is not a definite DP. Conversely, English requires the definite DP condition that the subject of a middle verb must be a definite DP. As the status of (18b) suggests, it is ungrammatical since the subject of the middle verb kill is not a definite DP. Note, however, that the use of these chickens instead of chickens makes (18b) grammatical, as illustrated in (19) (19) These chickens are killing easily. (Park, 2009) Interestingly, the L2 learners' correct responses to (18a) were 80%, whereas their incorrect responses to (18a) were 20%. On the other hand, their correct responses to (18b) were 45%, whereas their incorrect responses to (18b) were 55%. This in turn indicates that 45% of the L2 learners acquired the knowledge of the definite DP condition. Again, it is worth pointing out that Korean is not subject to the definite DP condition, hence a superset language, compared to English. More importantly, the results from the L2 leaners' responses suggest that the Korean learners of English do not entertain "the UG hypothesis (Chomsky, 1981(Chomsky, , 1986(Chomsky, , 2019a(Chomsky, , 2019b) that learners have complete access to Universal Grammar" (Ellis, 2015). In fact, the adult subjects showed different responses to (18a) and (18b). That is to say, their correct responses to (18a) were 80%, but their correct responses to (18b) were 45%, which in turn implies that the L2 learners did not complete access to Universal Grammar.
red wine spot-TOP carefully wash out (Red wine spots wash out carefully.) b. *Red wine spots wash out carefully.) As observed earlier, there are restrictions on the use of adverbs in the middle construction. More specifically, agent adverbs cannot appear in the middle construction. The reason why (20a) and (20b) are ungrammatical is that the agent adverb carefully appears in the middle construction. Interestingly, the L2 learners' correct responses to (20a) were 72.5%, whereas their incorrect responses to (20a) were 27.5%. On the other hand, their correct responses to (20b) were 20%, whereas their incorrect responses to (20b) were 80%. This in turn indicates that one fourths of the adult subjects acquired the adverb constraint. It is worth noting that both Korean and English are subject to the adverb constraint. Again, the results from the Korean learners of English do not back up the UG theory. As observed earlier, the adult subjects' correct responses to (20a) were 72.5%, whereas their correct responses to (20b) were 20%.

Discussion
In what follows, our discussion centers on defining some characteristics of the L2 learners' acquisition of middle constructions. Let us consider the following sentences: (21) a. Ku sonata-nun swipke yencwutoynta.
As observed earlier, 45% of the adult subjects acquired the definite condition, 32.5% of the L2 learners acquired the middle construction condition, 50% of the adult subjects acquired the abstract noun constraint, and 20% of the L2 learners acquired the adverb constraint. The following graph shows the order of the adult subjects' acquisition of middle constructions. This graph in turn implies that the abstract noun constraint was first acquired by the adult subjects, followed by the definite DP condition, the middle construction condition, and the adverb constraint, in that order.

Conclusion
To sum up, we have provided a detailed analysis of the L2 learners' acquisition of middle constructions.
In section 2, we have argued that the internal arguments of middle verbs must be a definite DP. We have further argued that an intransitive verb can occur in the middle construction only when a definite DP has an event that is an intrinsic characteristic or property of the definite DP. Additionally, we have shown that short adverbs such as easily and well in the middle construction are necessary with restrictions on the use of agent adverbs. In section 4 and section 5, we have examined how the L2 learners acquired the abstract constraint, the middle construction condition, the definite DP condition, and the adverb constraint. It is worth noting that 45% of the adult subjects acquired the definite condition, 32.5% of the L2 learners acquired the middle construction condition, 50% of the adult subjects acquired the abstract noun constraint, and 20% of the L2 learners acquired the adverb constraint. This in turn implies that that the abstract noun constraint was first acquired by the adult subjects, followed by the definite DP condition, the middle construction condition, and the adverb constraint, in that order. With respect to middle verbs, it is noteworthy that English are subject to the abstract noun constraint, the middle construction condition, the definite DP condition, and the adverb constraint, whereas Korean is not subject to the abstract noun constraint, the middle construction condition, and the definite DP condition. This in turn indicates that Korean is a superset language, compared to English. Thus, learning difficulty arises. Finally, we have maintained that unlike Chomsky's UG theory (Chomsky, 1981(Chomsky, , 1982(Chomsky, , 2019a(Chomsky, , 2019b, the L2 learners looked for similarities between L1 and L2 and thus relied on their L1.