You've already noticed the size of this post, but please don't get discouraged by it OR by the amount of quoted text! I've decided to give it in this form because it reads well and I couldn't explain it better myself.
I've found some information on ergatives and "middle constructions"/ pseudo-intransitives in
English grammar: a university course (2006) by Angela Downing and Philip Locke and in an article by Maarten Lemmens titled
Lexical constraints on constructional flexibility: English ‘Middable’ Verbs (1998)***.
Downing and Locke put ergatives and pseudo-intransitives (what was referred to earlier as 'middle constructions') under the
Causative processes heading, so that gives us an idea of where we're standing.
Ergatives
"The following examples illustrate a
transitive-causative structure (a controlling, purposeful, responsible Agent directs its energy towards something or someone (the Affected), so that this undergoes the action named by the verb, with a consequent change of state):
Paul opened the door.
Pat boiled the water.
I rang the bell.
From this perspective, the action of boiling, ringing, etc. is initiated by a controlling Agent or a Force participant:
The sun melted the ice.
The Affected is, however, the essential participant, the one primarily involved in the action. It is the door that opens, the water that boils and the bell that rings. If we conceptualise the situation from a different angle, in which no Agent initiator is present, we encode the process as ‘happening’ of its own accord. An Agent can’t be added. This is the
anti-causative structure.
The door opened.
The water boiled.
The bell rang.
When the Affected object of a transitive clause (e.g. the bell) is the same as the Affected subject of an intransitive clause, we have an ergative alternation or
ergative pair, as in
I rang the bell (transitive) and
the bell rang (intransitive). [...]
The test for recognising an ergative pair is that the causative-transitive, twoparticipant structure must always allow for the corresponding one-participant, anticausative structure. Compare the previous examples (e.g.
he opened the door/the door opened) with the following, in which the first, although transitive, is not causative. There is no intransitive counterpart, and consequently, no ergative pair:
Pelé kicked the ball. *The ball kicked."
Pesudo-intransitives
"A further type of Affected Subject occurs with certain processes (
break, read, translate, wash, tan, fasten, lock) which are intrinsically transitive, but in this construction are construed as intransitive, with an Affected subject.
Glass breaks easily.
This box doesn’t shut/close/lock/fasten properly.
Colloquial language translates badly.
Some synthetic fibres won’t wash. Usually they dry-clean.
Fair skin doesn’t tan quickly, it turns red.
Pseudo-intransitives differ from other intransitives in the following ways:
• They express a general property or propensity of the entity to undergo (or not undergo) the process in question. Compare glass breaks easily with the glass broke, which refers to a specific event.
• Pseudo-intransitives tend to occur in the present tense.
• The verb is accompanied by negation, or a modal (often
will/won’t), or an adverb such as
easily, well, any of which specify the propensity or otherwise of the thing to undergo the process.
• A cause is implied but an Agent can’t be added in a by-phrase.
• There is no corresponding transitive construction, either active or passive, that exactly expresses the same meaning as these intransitives. To say, for instance,
Colloquial language is translated badly is to make a statement about translators’ supposed lack of skill, rather than about a property of colloquial language. The difficulty of even paraphrasing this pattern shows how specific and useful it is.
Ed broke the glass. (active)
The glass was broken. (by Ed) (be-passive)
The glass broke. (anti-causative)
Glass breaks easily. (pseudo-intransitive)"
Lemmens on middle constructions in reference to ergatives:
"Keyser & Roeper (1984) argue that
it [the middle construction] differs from one-participant constructions with ergative verbs, such as
The ball bounces or
The boat sinks. They observe that middle constructions “state propositions that are held to be generally true […] they do not describe particular events in time” (1984: 384).
Similarly, Fagan notes that “middles […] are not used to report events, but to attribute a specific property to some object” (1988: 200). Consequently, Keyser & Roeper say, middle constructions, e.g.
Bureaucrats bribe easily or
Greek translates easily,
do not allow the imperative (*
Translate, Greek!), whereas ‘true’ ergative verbs do, e.g.
Sink, boat! or
Bounce, ball!
Further, the middle construction necessarily implies an agent: they “state the doability of an action as it can be, or is, performed by a non-specific agent, i.e., anybody” (Fellbaum 1985: 29). The notion of feasibility and that of the implied agent have been observed in the literature, e.g. Fiengo (1980) notes that middles retain an implicit agent whereas ergatives do not, Levin (1993: 5) talks about “an understood but unexppressed agent”. It is a point also made by Langacker who says about the following sentences
The window opened only with great difficulty. (Langacker 1991)
The dried mud scraped off effortlessly. (ibid.)
that “while the ease or difficulty of carrying out the action is attributed to inherent properties of the subject, it can only be assessed as easy or hard in relation to the capacity of an actual or potential agent” (1991: 334).
"
_________________________________________________________________
***In case the link is no longer available "the analysis of the English middle construction presented in the paper has been incorporated as a separate section in the author's more elaborate analysis of English lexical causatives: Lexical Perspectives on Transitivity and Ergativity. Causative Constructions in English. Amsterdam & Philadelphia: John Benjamins (1998), Ch. 4, 4.2, pp. 71-85. [ISBN: 90 272 3671 2; 268p.]."