3.3 Level of modification and adjective ordering
3.3.2 Intersective adjectives
Truswell’s (2004) generalisation for the nongradability of intersective adjectives is, at first glance, obvious with some classes, but not with others. For instance, it is indeed true that material adjectives sound awkward when modified by very (e.g. #a very wooden spoon). The acceptability of a very Greek dinner, on the other hand, suggests that nationality adjectives allow modification by a degree adverb. What is noteworthy, however, is that ‘a very Greek dinner’ does not imply that the dinner originated from Greece. It only means that the dinner had several features of a Greek dinner, even if it did not include foods of Greek origin like tzatziki and feta. For example, someone was playing the bouzouki and people were singing Greek songs. It seems, then, that the nationality adjective in this case expresses a quality rather than simple origin. Scott (2002) and Svenonius (2008) argue that adjectives are flexible to merge in any position that makes sense for their interpretation. What I assume happens in a very Greek dinner is that the nationality adjective is merged higher than the nationality/origin position, to a position where quality adjectives are merged:
(38) DP
D0 FPquality
AP very Greek
F0 . . .
FPnationality
AP
F0nationality NP dinner
The distinction between the two positions becomes even more obvious when two nationality adjectives are used in the same phrase. Consider the contrast between a very Cypriot Cypriot man and *a Cypriot very Cypriot man. It appears that the Cypriot that is closest to the noun is the one associated with origin. The
gradable Cypriot, which is the one found further away from the noun, is the one associated with the ‘Cypriotness’ of the individual, in other words the features of his appearance or character that make him ‘very Cypriot’ (e.g. has olive skin, plays backgammon, eats meat every Sunday, and watches football).
One language where it is evident that the modified nationality adjective not only has a different interpretation, but also surfaces in a different syntactic po-sition, is Italian. Giorgi and Longobardi (1991:127–128, (23) & (24)) show that nationality adjectives in Italian cannot appear prenominally or be modified:
(39) a. un’
‘a very quick/Italian invasion of Albania’
b. un’
‘an invasion more quick/Italian than the previous one’
If, however, the nationality adjective changes from an origin reading to a manner interpretation as in (41), then the adjective is acceptable in a prenominal position:
(41) quel
‘that very German behaviour of his’
(Giorgi and Longobardi 1991:128, (27)) The interpretational distinction between modified vs. bare adjectives is less strik-ing with other intersective adjectives, but there still seems to be a contrast. For example, a very red sweater gets a subsective/relative reading where the sweater is very red as far as sweaters go. Interestingly, it does not necessarily mean that the whole sweater is red. We can think of a sweater that has some red patterns, which are, nevertheless, quite bright and draw one’s attention to them. In this context, describing the sweater as being very red seems to be felicitous. In con-trast, a red sweater must mean that the colour of the sweater is predominantly red. It cannot be used in a context where there are only a few red patterns.
As for shape adjectives, gradability again seems to be associated with a sub-sective/relative reading of the adjective. If a table has four equal sides and four right angles, then it would be odd to describe it as a very square table, as it is obvious that it is simply a square table. On the other hand, a very square face is acceptable, as faces do not typically have four equal sides and four right angles, and, as a result, some faces might be closer to a square shape than others. The interpretation in this case, therefore, is ‘very square as far as faces go’.
This characteristic of intersective adjectives to shift to a relative reading when modified, is also observed when they take a complement. This is illustrated in the Greek example in (42), which is interpreted as ‘blue for a typical London sky’. The colour of the sky in (42) might still not be as blue as a typical blue sky in Cyprus.
(42) o the
(asiniTista) unusually
ble blue
Ja for
to the
lonDino London
uranos sky
‘the (unusually) blue sky as for London standards’
In general, what seems to be happening with colour and shape adjectives is that when they are bare they can either have a definitional or a prototypical reading, but when they appear with a modifier, or when they take a complement, they are restricted to a prototypical reading. For instance, a very square face cannot have the definitional reading of square which is ‘four equal sides and four right angles.’ I take this observation and the origin vs. manner distinction of nation-ality adjectives to suggest that genuine intersective adjectives do not have any phrasal characteristics. When they do, they are interpreted similarly to subsec-tive adjecsubsec-tives, whose meaning is relasubsec-tive to the noun they modify. I consider this to be an indication that modified intersective adjectives are merged higher in the extended nominal projection than bare intersective adjectives. Supporting evidence for this comes from the contrast in the acceptability of the (a) and (b) examples in (43)–(45). The modified adjective can only appear to the left of the bare adjective, which corroborates the idea that phrasal adjectives are merged higher than head adjectives.
(43) Nationality
a. an extremely Cypriot Cypriot man b. *a Cypriot extremely Cypriot man
(44) Colour
a. a barely red red apple b. *a red barely red apple (45) Shape
a. an incredibly square square face b. *a square incredibly square face
To conclude, it seems possible that intersective adjectives modify the noun as phrases when they have a prototypical interpretation, and as heads when they have access to both a definitional and prototypical reading. The basis for as-suming two distinct syntactic positions is that the adjectives come with different interpretational properties depending on whether they are bare or complex. We now turn our attention to subsective adjectives to see how these behave under modification.