Adjective ordering restrictions revisited


         Adjective ordering restrictions have been widely discussed (Sproat and Shih 1991, Svenonius
1993, Bernstein 1993, Cinque 1994, Bouchard 1998 among many others), but they are still not very well
understood. For example, there is still no explanation for why in some languages SHAPE adjectives have
to precede COLOR ones, but not vice-versa (1). Is it the syntactic component that determines these
orderings or the semantic one? This paper contributes to the debate by arguing that semantics does play a
role in the grammar of adjective orderings: it doesn't determine what the adjective ordering restrictions
are, but it determines when these apply. More specifically, I propose that adjective ordering restrictions
apply only to sequences where different orders yield different semantic interpretations.
         According to Sproat and Shih 1991 not all modifying adjectives are subject to ordering
restrictions: there are some principled exceptions. For instance, adjectives that can be derived from
relative clauses are freely ordered, since the grammar contains no principle for ordering (relative) clauses
with respect to each other function of the predicate inside. I would like to add to this discussion by
observing that operator adjectives like former or alleged (i.e. of type <>,>) represent another
principled exception to the ordering requirement. Specifically, in modification structures involving at
least one operator adjective, ordering restrictions do not apply. In (2) and (3) both the (a) and (b) orders
are possible. In contrast, in (4), where there is no operator adjective, only one order is acceptable, namely
the one in (4a). In the absence of special intonation, all the other word orders are ungrammatical (4b-d).
The question is why do adjective ordering restrictions apply in one case but not the other?
         A possible account is based on a refinement of Cinque 1994 and starts from the idea that operator
adjectives are semantically and syntactically special. For Cinque adjective word order is the result of
adjectives being base generated in dedicated Specifier positions. If we slightly modified this proposal by
stipulating that operator adjectives like former or alleged are adjuncts we could then derive the free word
order effect seen in (2) and (3). However, I argue that this is not the right approach.
         Proposal: I would like to propose an account in which the syntactic component imposes ordering
restrictions only on semantically equivalent structures. It may not be that all such structures are subject to
ordering constraints, but it is only these that can be. Under this view, if two sequences A1 A2 N and A2 A1
N have different denotations, the syntax will always allow both orders. This is the case in examples (2)
and (3) where the (a) and (b) sequences contain one or more operator adjectives and have thus different
interpretations, as shown by their respective paraphrases. Reversely, if only one ordering is possible the
prediction is that the sequences A1 A2 N and A2 A1 N are semantically equivalent and that the ordering
attested is the one imposed by the syntax, as in example (4).
         Interesting evidence for this approach comes from the so-called non-definite superlatives
discussed in Herdan and Stateva 2005 (5). In the absence of an operator, adjectives like short and Italian
are strictly ordered (6a vs 6b). Remarkably however, when a degree operator is present, such as the ­est
morpheme, the same two adjectives are no longer subject to ordering restrictions (7a vs 7b). The contrast
between (6) and (7) is clearly not due to the lexical heads themselves but rather to the presence/absence of
an operator. I suggest that in (7) the presence of ­est triggers different interpretations and as such, the
adjectives are allowed to be freely ordered with respect to each other.
         Conclusion: This paper provides further illustration of the fact that principles of syntactic well-
formedness can be sensitive to semantic interpretation (Fox 2000). Adjective ordering can be seen as
another domain where semantics restricts the application of syntactic operations.

(1)    Quality > Size > Shape > Color > Provenance (Sproat and Shih 1991)

(2)    sequences with 1 operator adjective:
       a. a famous former actor (#who is now forgotten)
               MEANING: someone famous who is no longer an actor
       b. a former famous actor (who is now forgotten)
               MEANING: someone who is no longer famous or no longer an actor

(3)    sequences with 2 operator adjectives:
       a. an alleged former thief
                MEANING: someone who is alleged to have formerly been a thief
       b. a former alleged thief
                MEANING: someone who is no longer alleged to be a thief

(4)    sequences without an operator adjective:
       a. a beautiful small black purse
       b. #a beautiful black small purse
       c. #a small beautiful black purse
       d. #a small black beautiful purse
       etc.

(5)    This class has a shortest student. ("non-definite superlative", Herdan and Stateva 2005)
               MEANING: there is a unique student in this class who is shorter than
                          all the other students in the class

(6)    a. a short Italian student
       b. #an Italian short student

(7)    Every class in this school has a shortest student.
       a. My class has [a shortest Italian student].
             MEANING: an Italian student who is shortest among the Italian students in my class
       b. My class has [an Italian shortest student].
             MEANING: the shortest student in this class is Italian



Selected references:
Cinque, G. 1994. "On the evidence for partial N movement in the Romance DP". In G. Cinque,
       J. Koster, J.-Y. Pollock, L. Rizzi and R. Zanuttini (eds.), Paths Toward Universal
       Grammar, 85­110. Georgetown: Georgetown University Press.
Fox, D. 2000. Economy and Semantic Interpretation. Cambridge, MA: MIT Press
Herdan, S. and Sharvit, Y. 2005. Definite and Non-definite Superlatives and NPI Licensing. Ms.
       University of Connecticut
Sproat R. and C. Shih. 1991. "The Cross-linguistic Distribution of adjective ordering
       restrictions". In C. Georgopoulos and R. Ishihara (eds.), Interdisciplinary Approaches to
       Language. Essays in Honor of S.-Y. Kuroda, 565-593. Dordrecht: Kluwer Academic
       Publishers