In resultatives, atelic events are turned into telic events by adding a secondary predicate. An important generalisation about resultative predicates is that they must be bounded (Talmy, 1986; Barbiers, 1995; Klein, 1997; Wyngaerd, 2001).

(1) Tim danced himself {completely/almost/half/*very} tired.

The scale of resultative predicates is maximally closed as shown below (Wechsler, 2005):

(2) John hammered the metal flat.

If the secondary predicate is not a gradable adjective with a maximum scale, resultatives are not grammatical (Napoli, 1992; Goldberg, 1995).

(3) She watered the tulips {flat, *droopy}.

However, gradable adjectives without a maximum scale are available if the main predicates ‘somehow' encode an endpoint in their meaning (Ono, 2007; Mihara, 2009).

(4) I froze the ice cream hard. 

Washio (1997) hypothesises the dichotomy of strong and weak resultatives; Japanese only allows the former one, in which the affected theme is predicated by a main verb (control resultatives). On the other hand, Japanese is flexible in that open scale adjectives are available as a secondary predicate as shown below:

(5) John-ga gomu-o nagaku nobasi-ta.

John-NOM rubber-ACC long stretch-PAST

‘(Lit.) *?John stretched the rubber long.' ‘John stretched the rubber and it became long.'

I claim that the boundedness of resultative events is contained in the main predicate that encodes ‘result' in Japanese (Rappaport Hovav and Levin, 2010; Beavers and Koontz-Garboden, 2012). If both manner and result can be encoded in a single lexical item a la Beavers and Koontz-Garboden (2012), the meaning of nobasu can be written as follows:

(7) [[nobasu]] = λxλe1[long'(x, e1) ∧ ∃e2[cause'(e2, e1)] ∧ ∀e3[cause'(e3, e1) → streching'(e3)]]

The Japanese type resultative is also possible in English, whereby hard is available as a secondary predicate in (4).