Pan-Pacific Association of Applied Linguistics 13(2), 125-140

Explaining English Middle Sentences

Kabyong Park
Namseoul University

K. Park. 2009. Explaining English Middle Sentences. Journal of

Pan-Pacific Association of Applied Linguistics, 13(2), 125-140.

The current paper attempts to account for the formation of English

middle sentences. Discussing a set of previous analyses on the
construction under investigation we show, following the assumptions of
Oosten(1986) and Iwata(1999), that English middle constructions should
be divided into two types: generic middle constructions and non-generic
middle ones. The distinction is shown to be closely related with the tense
aspect of the verbs: the structural subject in the former is interpreted as
generic with the present tense, while the latter can take past tense and
progressive aspect with a non-generic subject. Other thematic roles than
Agent can be realized as the structural subjects. In the generic
interpretation, the intrinsic property of the plays the role of cause and is
primary responsible for the event denoted by middle verb. In the non-
generic interpretation, on the other hand, a specific event plays the part
of cause and thus can take the past tense. Middle verbs are lexically
derived from a set of activity or accomplishment verbs that carry [-state,
+process] aspect features. After derivations, the verbs in the generic
interpretation possess [+state, +process, +repetition] aspect feature, and
those of the non-generic one, [-state, +process, -repetition]. Another
contrast lies in the definiteness of the subjects: the generic interpretation
involves an indefinite/generic subject and the non-generic one needs a
definite/non-generic subject.

Key Words: middle construction, generic, non-generic, aspectual

features, definiteness

1 Introduction

A middle construction refers to the clause where the theme or patient of a

verb is structurally realized as the subject of a predicate in an active voice.
For example, in the sentences below, the verbs read and drive occur with the
active voice and the logical objects the book and the car appear in the
structural subject position.

(1) The book reads easily.

(2) The car drives well.

125 C 2009 PAAL 1345-8353/00

○
 Kabyong Park

This construction has been widely discussed in many works including

Keyser & Roeper (1984) and Fagan (1982). However, even native speakers
have reported inconsistency about the grammaticality of the middle
construction and a set of issues have not been resolved: the definition of the
middle construction, the conditions on external arguments and implicit agent
subjects, the semantic property of the middle verbs and the existence of
adverbials.
The current paper distinguishes two types of the middle construction,
following Oosten (1986) and Iwata (1999): the generic middle construction
and the non-generic middle construction. In the former type, an intrinsic
property of the subject takes the primary responsibility for the event
expressed in a predicate. The non-generic construction does not show such a
relationship.
This paper is organized as follows. Section 2 is devoted on the
discussion of a few distinct approaches to the conditions on the formation of
the middle construction: Hale & Key's (1987) and Fellbaum & Zribi-Hertz's
(1987) Affectedness Condition, Vendlers's (1967) Aspect Condition, Oosten
(1984) and Fellbaum's (1986) Primary Responsibility Condition, and Chung's
(1995) Causative Condition. We will show that each perspective falls short of
providing an explanatory account. The next section attempts to resolve the
limitations discussed in section 2 by discussing such issues as non-generic
interpretation, aspect features of the middle verbs, subjects of the middle
constructions and the impact of manner adverbials. We will also show that
the middle construction can occur in the progressive and past tense and can
be interpreted non-generically in certain contexts.

2 Conditions on the Formation of the Middle Construction

The researches in the early generative grammar assume that the transitive
verbs should occur in the underlying representation of the middle
construction, since those verbs carry a passive meaning. This section
discusses some previous analyses of the constraints on the construction under
investigation and point out their limitations.

2.1 Affectedness constraints

According to the Affectedness Constrains, the middle construction can be

possible only when the internal argument of a transitive verb is affected by
the event or action 1 . Fellbaum & Zribi-Hertz (1989) assume that the intrinsic

1
An argument A of a verb or predicate is AFFECTED by the action or process P
referred to by the verb if the referent of A exists prior to P and if its inherent
properties are modified by P (Fellbaum & Zribi-Hertz 1989:28).

126
 Explaining English Middle Sentences

property of the argument is affected. This hypothesis can account for the
grammaticality of the following sentences 2 .

(3) The bottle breaks easily.

(4) * The Eiffel Tower sees from my window.

It is argued that the patient the bottle is affected by the event of

breaking easily; hence the grammaticality of (3). The logical object Eiffel
Towel in (4), on the other hand, is simply seen to people but can not be
affected by the action of seeing. Thus, the contrast in (3-4) can well be
explained by the Affectedness Constraint. Moreover, the Constraint can
explain why the verbs denoting creation can not occur in the middle
construction. Consider the following examples.

(5) a. *This bridge builds easily.

b. *This poem writes easily.

We can perhaps simply say that the subjects bridge and poem in (5)
are 'created' by the action of building and writing. Thus, we might not say
that the process of creation results in the change of any intrinsic properties of
these particular arguments. Then, the Affectedness Constraint can account for
the ill-formedness of the examples in (5). Along the same line of thought, we
can explain the contrast in (6), where the same verb occurs in the middle
construction.

(6) a. This piano plays easily.

b. *This sonata plays easily.

As the same verb play is used in both sentences, the choice of subjects
is solely responsible for the contrast. In (6a), the subject this piano can be
affected by the action of playing, though this sonata can not 3 . However, let us
consider the following example in Fellbaum & Zribi-Hertz.

(7) Mary photographs well.

The grammaticality of this particular sentence forces us to say that the

subject Mary is affected by the action of photographing. This claim seems to

2
Change of state is some property of the theme held before the time with respect to
which the proposition containing the predicate is evaluated and fails to hold after that
time, or vice versa (Roberts 1985:394).
3
However, it is not easy to claim that the action of playing affects the piano in any
way.

127
 Kabyong Park

be extremely hard to maintain in any sense. Felbaum & Zribi-Hertz simply

stipulates that taking a picture transforms a person into an image and then the
image can be affected by the even of photographing, which is very hard to be
supported. It will not be easy to believe that any property or state of the
subject can be affected by being photographed.
Moreover, the Affectedness Constraint can not provide a plausible
account of the middle structure in the following sentences.

(8) This book reads easily.

(9) Greek translates easily.

In contrast to the prediction of the Affectedness Constraint, these

sentences are judged grammatical although the subjects this book and Greek
can not be considered being affected by the event of reading and translating.
This Constraint also can not effectively explain the fact that the middle
construction can also involve logical objects denoting other semantic rules
than the patient. The following data show that the arguments denoting
Instrument or Locative can be structurally realized as subjects.

(10) The knife cuts well.

(11) The aluminum pan bakes higher and browns evenly.

2.2 Aspectual features of verbs

According to Vendler (1967), verbs can be categorized into four different

subclasses on the basis of their semantic properties: activity, accomplishment,
achievement and state. These can be exemplified as in the following
sentences. The first two can appear with a progressive tense, since they can
denote a progress of an event taking time. These activity /accomplishment
verbs differ in the presence/absence of a definite time period for the relevant
event. An achievement verb, in turn, denotes an event that happens and ends
at a particular time. The event or state expressed by a state verb, on the other
hand, lasts for period of time. According to Fagan (1992) and Vendler (1967),
the crucial factor for the formation of middles is whether the verb can occur
in the present tense 4 . For instance, activity and accomplishment verbs can,
whereas state and achievements can not, as illustrated in the following
examples.

(12) The car drives easily. (activity)

4
Roberts(1986)'s agentivity test also differentiate two groups of verbs: activity and
accomplishments involve an external agent, whereas achievement and state verbs do
not.

128
 Explaining English Middle Sentences

(13) This book reads easily. (accomplishment)

(14) * This poem understands easily. (achievement)
(15) * The answer knows easily. (state)

The contrast in (12-15) leads Fagan to assume that only activity and
accomplishment verbs are allowed in the middle construction and to claim
that the aspectual features of a verb is responsible for the formation of the
middle construction.

(16) a. She is driving the car.

b. She is reading this book.
(17) a. * She is understanding this poem.
b. * She is knowing this answer.

We can clearly see the similarities in grammaticality between the

middle sentences in (12-15) and progressive sentences in (16-17). The verbs
in (16) can appear in a progressive form, whereas the achievement and state
verbs can not. However, the aspect feature constraints can not explain the
ungrammaticality of the following examples.

(18) a. *Mary invites easily.

b. They are inviting Mary.
(19) a. * That issue discusses easily.
b. They are discussing that issue.

The verbs invite and discuss should belong to activity/accomplishment

verbs, since they can appear in a progressive form as shown in (18), Thus,
they are expected to appear in a middle sentence according to Fagan's aspect
constraints. However, this prediction is not borne out as illustrated in (18a)
and (19a). In other words, not all the activity/accomplishment verbs can
always appear in a middle construction. Moreover, these aspect constraints
can not account for the contrast found in (20-21) resulting from different
types of patients and adverbials.

(20) a. * This sonata plays easily.

b. This piano plays easily.
(21) a. * The soup eats rapidly.
b. The soup eats like a meat.

We might conclude that the aspect feature constraints solely based on

different types of verbs would not provide a satisfactory account of the
formation of middle construction.

129
 Kabyong Park

2.3 Primary responsibility constraints

Van Oosten (1986) and Fellbaum (1986) claim that the subject of the middle
construction takes the primary responsibility for the event expressed by a
verb. In other words, only the patient taking the primary responsibility can be
allowed as the subject of the middle construction. Consider the following
examples.

(22) a. The lawn mower handles easily.

b. Sweatshirts wash in the machine.

According to Fellbaum(1986), the structural subject/the logical object

the lawn mower has the property of being handled easily by anybody in (22a)
and sweatshirts in (22b) has a certain characteristic that allows them to be
washed in the machine. In other words, a certain property of the structural
subject in the middle construction is primarily responsible for the action or
event expressed by a verb 5 .
However, this primary responsibility constraint also has a limitation.
First, it can not explain the role of adverbials in the formation of the middle
construction.

(23) a. * The ball hits easily.

b. The ball hits like a dream.

In this pair of sentences, patients are the same. Likewise, the primary
responsibility should be the same for hitting or being hit. Thus, Oosten's
primary responsibility constraint may not be able to explain the contrast in
(23). Moreover, this model does not capture the fact that the primary
responsibility could vary depending on the patients, thought the same verb is
employed, as illustrated below.

(24) a. This piano plays easily.

b. * This sonata plays easily.

Note that in (24) the only difference between the two is the patient and
still the acceptability is different. The contrast might show that the middle
formation can not be due to the primary responsibility constraint only.

5
The logical object is called 'quasi-agent' in van Oosten(1986: 460-461) and
'constructional agent role' in Hale and Keyser(1986).

130
 Explaining English Middle Sentences

2.4 Causative constraints

Chung (1995) observes that all the middles are allowed with a transitive verb
but not all transitive verbs are allowed in the middle construction. He argues
that a transitive verb can appear in the middle construction only when the
internal argument carries the feature of causer.

(25) Causative Condition 6

Unless there is a cause for an event, the caused event can not
take place (Chung 1995: 276)

Based on the condition, he assumes that all the middles carry a

causative property in terms of semantics. According to this causative
condition, all the noun phrases carrying patient, theme, instrument or even
place roles can appear as the structural subject of a middle sentence, as
shown in (26).

(26) a. The glasses break easily. (Patient)

b. The books read easily. (Theme)
c. The marks hit easily. (Goal)
d. The piano plays easily. (Instrument)
e. The dogs frighten easily. (Experiencer)

In (26a), a property of the glasses is the cause of the event of being

broken. Along the same line, a certain property of the subjects plays the role
of cause of the event in other examples; for example, a property of the books
in (b) is the cause of the event. However, the causativity constraint does not
overcome the limitations of the previous analyses. Moreover, it has its own
problem. First, according to Chung (1995), state verbs can not form a middle
construction. As a middle construction involves a cause-effect relationship, it
follows that the verb involved should denote an event. But, if we follow
Chung (1995), we can not explain why a middle construction expresses a
certain non-event property or state feature at least at the surface.
Chung (1995) attributes the contrast in (27-28) between the
achievement verb break and the activity verb hit to their lexical
idiosyncrasies.

(27) This bottle breaks easily.

(28) * This wall hits easily.

6
Chung(1995)'s concept of the causer seems to be the same as Van Oosten(1986)'s
responsibility of the subject.

131
 Kabyong Park

The cause of all the middle construction is a certain property of the

subject, not the subject itself. The sentence (27) is grammatical, as the
intrinsic property of the bottle causes the event of breaking easily. And (28)
is ruled out, since it is hard to believe that the intrinsic property of the wall
causes the event of hitting easily. If we follow Chung and assume that the
grammaticality results from the lexical idiosyncrasies of the verbs, all the
achievements should be allowed in the middles, while all the activity verbs
should not. His causative constraint can not completely substitute the aspect
feature constraints 7 .
To summarize, we have discussed four previous perspectives on the
formation of middles and shown that all the four conditions have limitations
and can not successfully account for the contrast under discussion. In the next
section we make a new proposal that can deal with the problems mentioned
above.

(29) a. Affectedness Condition

b. Aspect Features Condition
c. Primary Responsibility Condition
d. Causative Condition

3 A More Explanatory Approach

We follow Grimshaw (1990) and Chun (2003) regarding the argument

structure of middle verbs in that the internal argument of a transitive verb can
project in the structural subject position when it carries the role of a cause.
According to Grimshaw (1990), the argument structure of a lexical item is
arranged according to the thematic hierarchy and the aspectual hierarchy. The
external argument is projected at the subject position and the internal
arguments are placed inside the projection of a verb. However, some internal
arguments of a certain verb can appear at the subject position; for example,
psychological verbs. Grimshaw argues that the placement of an argument at
the structural position depends on the aspectual hierarchy. Thus, the internal
theme argument can appear at the subject position if it carries the cause role
of the aspectual hierarchy. Therefore, it might mean that the subjects of the
middle construction can appear at the surface position with any thematic role
except Agent if they carry the cause role of the aspectual hierarchy.
We have seen that most of the examples of the middle construction
involve a generic interpretation. Oosten (1986) and Iwata (1999), however,
shows another type of the middle construction: non-generic middle sentences.

7
Chung (1995:270) claim that the Goal role can not appear as the subject of a middle
does not hold for the following sentence.
(i) The target hits easily.

132
 Explaining English Middle Sentences

Both types involve an implicit agent and the verb appears in the active voice
with a passive meaning. The difference lies in the crucial observation that the
non-generic middle can take a progressive tense and a past tense. And the
subject is non-generic, as the name indicates, and is not primarily responsible
for the event expressed by the predicate.
In this section we attempt to make a new proposal that can account for
the middle formation. More specifically, we assume that there exist two types
of middle construction: generic and non-generic middle construction. And we
aim to show that a set of aspectual features of the middle verbs are
responsible for the middle formation. We also argue that the definiteness of
the subject of a middle sentence is closely related with the aspectual features
of the verb.

3.1 Non-generic middle construction

According to Iwata (1999), the existence of an implicit agent subject is

crucial for the definition of a middle sentence. However, some middle
sentences lack genericity and modality of possibility which were assumed to
typically characterize middle sentences. And the middle verb in those
sentences can take both progressive and past tense, as shown below in (30 b-
c). The implicit agent subject carries the feature of specificity.

(30) a. This car handles smoothly.

b. This car is handling smoothly.
c. This car handled smoothly.

All the middle sentences in (30) are judged grammatical. The

difference, however, lies in the genericity of the implicit agent: (a) involves a
generic person, while (30b-c) a specific speaker. Thus, we can predict that
the sentences (30b-c) should be ruled out if the event is independent of the
speaker. And the prediction is borne out, as shown in (31-32).

(31) *This car was handling smoothly while I was sleeping in the
backseat.
(32) *This car handled smoothly while I was sleeping in the
backseat.

It is clear that the implicit agent is not a generic subject but the
speaker. Thus, while the speaker is sleeping, this car can not be handled.
When a middle sentence involves a specific event, the interpretation of the
subject plays a key role.
Iwata (1999) raises a question about the judgment on the sentences in
(33), which are judged ungrammatical in Keyser and Roeper (1984). If we

133
 Kabyong Park

assume that these are examples of the non-generic middle, these sentences
can be given a legitimate interpretation.

(33) a. ?*Yesterday, the mayor bribed easily, according to the

newspaper.
b. ??At yesterday's house party, the kitchen wall painted
well.
c. Grandpa went out to kill a chicken for dinner, but the
chicken he selected didn't kill easily.
d. If it hadn't been for the wet weather, my kitchen floor
would have waxed easily.
e. The wall is painting easily.
f. The floor is waxing easily.

As seen in (30), a typical generic middle sentence carries a different

interpretation from a non-generic one in the progressive and past tense. In the
generic middle construction a generic agent can be interpreted to carry out an
action due to the intrinsic property of a middle subject, whereas in the non-
generic middle a certain event is being or was performed by an implied
specific agent regardless of the intrinsic feature of the subject.
If we posit the non-generic middle interpretation, all the conditions in
(29) pose a problem. The conditions in (29a), (c) and (d) will be discussed in
3.2 and (29b) in Section 3.3.
The claim that the middle construction must involve a non-generic
interpretation has a thread of connection with the argument that a middle verb
can not be presented in the progressive and/or past tense. As pointed out in
Iwata(1999), however, a middle sentence can appear in the progressive and
past tense. Otherwise, the grammaticality of many middles can not be
accounted for.

(34) a. The boat sank all by itself.

b. * This book reads easily all by itself.

One thing worth mentioning here is that if the progressive or past

tense appears with an intransitive verb, it might not be easy to tell whether
the sentence is a middle structure or an ergative one. The apparent
equivocalness can be cleared up with the addition of adverbial phrases like
all by itself, as illustrated in (34).
According to Keyser and Roeper(1984:405), an ergative verb can co-
occur with the adverbial all by itself that carries the meaning 'without any
outside help'. This coexistence might be explained with the assumption that
the ergative verb does not involve an agent. The ill-formedness of the middle
sentence in (34b), in contrast, can be attributed to the existence of an implied
agent, since the adverbial phrase literally negates it. The same contrast

134
 Explaining English Middle Sentences

between the middle and the ergative can be found in (35) and (36). The
typical middle in (35) can not be saved unless with the addition of an
adverbial 8 .

(35) a. * Bureaucrats bribe.

b. * The wall paints.
(36) a. This branch broke.
b. The cheese molded.

3.2 Aspectual features constraints

According to the aspectual constraint in (29b), both activity and

accomplishment verbs can appear in the middle sentence. They both describe
a process during a certain time and they differ in terms of time period.
Adopting J. Seo (1991) 's introduction of the features [stativity] and
[process] for the classification of verbs, we might assume that activity and
accomplishment verbs carry the features [-state] and [+process], achievement
verbs [-state] and [-process] and state verbs [+state] and [+process]. Still,
some accomplishment verbs are not always allowed in the middle structure.
Consider the following examples.

(37) a. * This bridge builds easily.

b. * This poem writes easily.

The verbs write and build could be categorized as accomplishment

verbs, since the event of building and writing takes a certain period of time
until it is finished up. And once the event of creating a certain project is
accomplished, the same event can not continuously or repeatedly happen.
Thus these verbs may not appear in the middle. Now let us assume that a
middle verb should carry the features [+state, +process]. Then the
grammaticality of (38-9) can be easily accounted for.

(38) This pen writes well.

(39) Love letters write easily. (Chun 2003:145)

The action of writing in (38-9), in contrast to (37b), can be repeated

for a certain period of time. It might be argued that the contrast in
grammaticality between (37b) and (38-9) can be attributed to the aspectual

8
Fellbaum(1986:6) points out that an ergative verb can be modified by an adverbial in
certain contexts.
(i) a. The door closes easily; you just have to press down. (middle)
b. The door closes easily; it only takes a gust. (ergative)

135
 Kabyong Park

features; the middle verb write carries the features [+state, +process,
+repetition].
Now let us turn our attention to the non-generic interpretation in
(30b,c) and (33). The non-generic middle verb can take the progressive and
past tense and hence is assumed to carry [-state, -repetition]. Thus, the
assumption above that a middle verb should carry [+state, +process,
+repetition] does not hold for the non-generic interpretation. A certain
modification of the hypothesis is in order.
M. Lee (2001) claims that the middle construction can possibly be
formed only if recursiveness based on regular repetition and definiteness
inside the predicate is presented 9 . However, she does not investigate the non-
generic interpretation. To account for the existence/contrast of both generic
and non-generic middle sentences, we must hypothesize that a transitive verb
with the features [-state, +process] can also appear in the non-generic middle
with the features [+state, +process, +repetition] as well as in the generic
middle with the features [-state, +process, -repetition]. It seems to be a
theory-internal choice whether we posit a lexical device 'shifting' features in
the lexicon or we assume more than one subentries for the same verb; a
typical transitive, a generic middle and/or a non-generic middle.
Vendler (1967) also mentions that it is not easy to make a clear cut
classification of verbs, which means that a verb might exhibit semantic and
aspectual variety and that shifts of aspectual features should be allowed 10 .
For example, such verbs as think, know, understand, see, hear can not easily
belong to a specific subcategory. Consider the following examples.

(40) a. Oh, I am quite tall, I saw him all the time he was in the
courtroom. I was watching him.
b. At that moment I saw him.

9
(i) a. * The suicide website terminates easily.
b. * That natural disaster prevents easily.
She attributes the ungrammaticality of the examples above to the
nonrecursiveness of the event. It does not happen repeatedly.
10
Brinton(1988) also points out that the same verb can belong to all the four different
categories. Some examples are found below.
(i) a. activity : The child is touching the breakable glassware.
b. accomplishment : Hannah touched all the buttons in the elevator to make them
light up.
c. achievement : Just then he touched the buzzer.
d. state : The wainscoting touches the floorboard at a right angle all along the
southern wall.

136
 Explaining English Middle Sentences

The same verb see carries the meaning of state in (40a) perhaps due to
the adverbial all the time in (40a), whereas might be classified as an
achievement verb in (b) where it refers to a specific time. Thus we might
conclude that the subcategorization should depend on the contexts.

3.3 The subjects of middle sentences

The constraints discussed in section 2 show that the middle construction does
not describe a specific event but expresses an intrinsic property or general
state. For example, Fellbaum & Zribi-Hertz(1989) presents the following
examples to support this generalization.

(41) The clothes wash with no trouble because

a. they're machine-washable.
b. * I have lots of time.

A property of the subject the clothes in (a) can appear in the

subordinate clause after the middle sentence, whereas any predication of the
agent I is not permitted as shown in (b). The contrast might lead us to
conclude that the middle sentence refers to an intrinsic property of the
structural subject. Chung (1995: 276), on the other hand, claims in (25) that
unless there is a cause for an event, the caused event can not take place.
However, we can see here that the causative condition in (25) may not
account for the contrast in (41). If we take the intrinsic property of the subject
as the cause, it does not cause any new event to occur but simply maintains a
general state of the subject. There is not really a cause-causee relationship in
the middle sentences. Thus the condition in (29d) might be reformulated as in
the following.

(42) Constraint on the Middle Construction X contains a certain

property or state due to its intrinsic nature.
(X may carry all thematic roles except Agent)

Next let us discuss the definiteness of the middle subjects. The

contrast in (43-4) is closely related to the definiteness of the middle subjects:
the ones in (44) with the determiner are permitted.

(43) a. ?* Bureaucrats are bribing easily.

b. * Chickens are killing easily.
(44) a. These bureaucrats are bribing easily.
b. These chickens are killing easily.

We can clearly see from the above examples that the specificity of the
non-generic middle verb with progressive and past tense is closely related

137
 Kabyong Park

with the definiteness of the subjects. In this case an intrinsic property of the
subject is not really responsible for the specific event. Rather, a specific event
is performed by a specific agent. Thus we might conclude that the primary
responsibility constraint can not appeal to the interpretation of the non-
generic middle sentences. Consider the following examples.

(37) a. * This bridge builds easily.

b. * This poem writes easily.
(38) This pen writes well.
(39) Love letters write easily. (Chun 2003:145)

The same line of thought can apply to the repeated examples above.
The sentences in (37) have definite subject, and thus can not exhibit a generic
interpretation, whereas in (38-39) the indefinite subjects this pen(meaning a
pen of this kind) and love letters can carry a generic interpretation. Again the
definiteness of the subject is closely related with the generic/non-generic
interpretation. In fact, many native speakers have reported that the following
examples are acceptable in certain contexts.

(37a') ? This bridge is building easily.

(37b') ? The poem is writing easily.

3.4 Adverbials

Another question worth exploring is whether an adverbial is a necessary

condition for the formation of a middle sentence. To paraphrase the question,
must an adverbial occur to express an intrinsic property of the subject?
Consider the following ungrammatical sentences.

(45) a. *That idea communicates.

b. *That passage translates.
c. *The home wrapping center stores.
d. *The document photocopies. (Iwata, 1999:535)

It seems apparent that we can not dispense with an adverbial for the
formation of a middle sentence. However, many examples are found to show
that other elements including verbs themselves can perform adverbial effects,
as shown below.

(46) This dress buttons.

(47) This umbrella folds up.
(48) a. * The meat cuts.
b. This meat doesn't cut. (Fellbaum, 1986:9)
(49) a. ?? This car drives.

138
 Explaining English Middle Sentences

b. I thought we were out of gas, but the car DRIVES!

(Fellbaum, 1986:9)
(50) These red sports models do sell, don't they?

4 Concluding Remarks

We have shown that none of the previous analyses are successful in

accounting for the formation of the middle sentences in English. To
summarize, the cause of the middle sentence is an intrinsic property of the
subject, an incidental event or state. Thus, we assume two types of middles
sentences: generic and non-generic. The subject in the former is primarily
responsible for the event, whereas the subject in the latter is not. The
structural subjects in the middle sentence do not carry their own will and
other thematic roles than Agent are realized as the subject. Mainly activity
and accomplishment verbs may be derived as middle verbs and all the middle
verbs share the feature [+process]. These typical transitive verbs contain the
features [-state, +process] and then the features are shifted in the middle
construction after a certain lexical derivation to [α state, +process, α
repetition]. If α is assigned a positive value, it results in a generic
interpretation and if negative a non-generic one. Thus, the generic
interpretation must involve the feature [+state, +process, +repetition], while
the non-generic one has [-state, +process, -repetition]. We also show that the
definiteness of the subject of the middle is closely related with the
interpretation: the generic one with [-definite] subject and the non-generic
with [+definite] one, which enables the middle verb to take the progressive
and past tense for a specific event.

References

Chun, B. (2003). Middle formation revisited. The Linguistic Association of

Korea Journal, 11(1), 135-158.
Chung, T. (1995). A Semantic condition on English middles: A causative
approach. Korean Journal of Linguistics, 20(4), 271-288.
Fagan, S. (1992). The syntax and semantics of middle constructions: A study
with special reference to German. Cambridge: Cambridge
University Press.
Fellbaum, C. (1986). On the middle construction in English. Bloomington,
IN: Indiana University Linguistic Club.
Fellbaum, C., & Zribi-Hertz, A. (1989). The middle construction in French
and English. Bloomington, IN: Indiana University Linguistic Club.
Grimshaw, J. (1990). Argument structure. Linguistic Inquiry Monograph, 18,
Cambridge, MA: MIT Press.

139
 Kabyong Park

Hale, K., & Keyser, S. (1987). A view for the middle, lexicon project working
papers, 10, Cambridge, MA: Center for Cognitive Science, MIT.
Lee, M. (2001). Recursiveness condition on English middles. Eoneo, 26(4),
735-753. (written in Korean)
Keyser, S., & Roeper, T. (1984). On the middle and ergative constructions in
English. Linguistic Inquiry, 15, 381-416.
Iwata, S. (1999). On the status of an implicit arguments in middles. Journals
of Linguistics, 35, 527-553.
Massam, D. (1992). Null objects and non-thematic subjects. Journal of
Linguistics, 28, 115-137.
Roberts, I. (1987). The representation of implicit and dethematized subjects,
Dordrecht: Foris Publications.
Seo, J. (1991). Research on Korean grammar. Hanshin Publishers, Seoul,
Korea. (written in Korean)
Striok, T. (1992). Middles and movement. Linguistic Inquiry, 23, 127-137.
Van Oosten, J. (1986). The nature of subjects, topics, and agents: A cognitive
explanation. Bloomington, IN: Indiana University Linguistic Club.
Vendler, Z. (1967). Linguistic in philosophy. Ithaca, NY: Cornell University
Press.

Kabyong Park
English Department
Namseoul University
21 Maeju-Ri, Seonhwan-Eup
Chonan-shi, KOREA
82-41-580-2162(Office)
82041-580-2921(Fax)
E-mail: kpark@nsu.ac.kr

Received: January 27, 2009

Revised: May 22, 2009
Accepted: June 5, 2009

140