Pernilla Hallonsten Halling

Adverbial modification and secondary predication

Secondary predicates can be defined as adjuncts that predicate (Himmelmann & Schultze- Berndt 2005b: 4). Thus, they share with modifiers the characteristic of being adjuncts. Two main construction types are generally discussed as instantiating secondary predica- tion: depictives and resultatives (see section 2.5.2 for a detailed discussion), as illustrated with English examples in (3.14) and (3.15).

(3.14) I ate the fish raw.
(3.15) We painted the barn red.
depictive resultative

The question we are interested in here is whether manner adverbs should also be seen as instances of secondary predication. This is a matter that seems to divide linguists. Win- kler (1997: 11) suggest a “there-sentence test” following Milsark (1976 cited in Winkler 1997: 11) which can be applied in order to resolve the matter. Based on the assumption of Drubig (1992), who assumes that “depictive secondary predicates denote stage-level pred- icates, whereas subject-oriented manner adverbs like cleverly, stupidly, and cheerfully are predicates of properties of individuals”, Winkler (1997: 11-12) illustrates the applicability of the there-sentence test with the following examples.

(3.16) (a) John dropped his spoon drunk/*clever.
(b) John dropped his spoon cleverly.
(3.17) (a) There are some men drunk.
(b) *There are some men cleverly.
depictive manner adverb

While the manner adverb in (3.16b) cannot be used in a there-sentence as illustrated in (3.17b), the depictive in (3.16a) can, as illustrated in (3.17a). But as pointed out by Winkler, Milsark’s original there-sentence test is intended to allow predicates of states but not properties (1976, cited in Winkler 1997: 11). This is evident when applied to the depictive in (3.14): *There was some fish raw is clearly unacceptable.

Loeb-Diehl (2005) and Verkerk (2009) both define manner adverbs (as well as depictives and resultatives) as secondary predicates. Loeb-Diehl (2005) bases the classification of manner adverbs as secondary predicates on the assumption that manner adverbs fulfill the semantic function of predicating a property of an event. Since this event is denoted by a primary or main predicate, the element that denotes the property is logically classified as a secondary predicate. But Loeb-Diehl points to the difference between depictives and resultatives on the one hand, and manner adverbs on the other, that is often referred to. The former are participant-oriented and the latter are event-oriented (cf. Geuder 2000, discussed in section 2.3.2). This is obvious in example (3.14) above, where the property raw relates to the fish, at the time of being eaten, and ate is the primary predicate. Similarly, in (3.15), red relates to the barn that is being painted, and painted is the primary predicate. Both the fish and the barn are participants in the respective events. This can be compared to the manner adverb in (3.18).
(3.18) He ran slowly.
In (3.18), slowly modifies ran, which denotes an event and not a participant. To analyze slowly as a secondary predicate, it must thus be seen as separate from the primary predi- cate ran. The participant orientation that unites depictives and resultatives is important for Loeb-Diehl, but a number of similarities between depictives and manner predications, as they are termed, are also pointed to. Depictives and manner predications are argued to share the feature of expressing a property that is simultaneous with the main event.

Resultatives, however, denote a result of the main event, and are therefore consequent to it. For depictives and manner predications, the difference between event and participant orientation can be blurred in certain cases, e.g. with expressions of human propensity. For instance, enthusiastically can be interpreted as either participant- or event-oriented (cf. discussion in section 2.3.2). Loeb-Diehl comes to the conclusion that “manner en- coding and depictive encoding form domains which are semantically distinct for their prototypical ‘core’ cases, but which turn out to be non-discrete in their peripheral instan- tiations” (2005: 12). This means that the encoding in these two domains may overlap in specific languages, a fact which Loeb-Diehl also exemplifies (2005: 12–13). One lan- guage that has this pattern is Lezgian (Nakh-Daghestanian), which is also in my language sample (cf. section 4.4).

(3.19) Lezgian (Nakh-Daghestanian) (Haspelmath 1993: 196)

1. 
   Jusuf.A
   

Jusuf(erg)
ne-lAj
who-srel
xˆAjit’Ani
indf