During the development of Binding Theory in the 1980s, picture NP reflexives (1) have been considered as highly problematic since they resist a treatment in terms of Principle A and thus preclude a simple theory of reflexive distribution.

(1) John_i told Mary_j that pictures of themselves_i+j were on sale.

With the advent of an argument structure based theory of reflexive binding in Pollard and Sag (1992, 1994) -- and the quite similar theory of Reinhart and Reuland (1993) -- a rather different picture emerged: Picture NP reflexives were now considered as non-anaphoric and hence fall outside the scope of Principle A. Consequently, they were terms as exempt reflexives.

Although this theory works nicely for English, a comparative survey reveals that not all picture NP reflexives follow this pattern: In many languages, picture NP reflexives have to be considered as anaphoric. Moreover, it remains somewhat unclear under which circumstances binding theory turns a reflexive into an exempt one.

My talk focusses on a prominent subclass of picture NP reflexives, viz. picture NP reflexives in the subject of so-called object-experiencer psych verbs (2).

(2) These nude pictures of himself_i annoyed John_i.

Considering data from a variety of languages, I will entertain the followingassumptions:

1) Picture NP reflexives are not per se exempt, but may become exempt if Principle A in a language receives a particular, conditional formulation.
2) Reflexives in the subject of object-experiencer psych verbs will only emerge if a language allows exempt reflexives, i.e. there is no 'inverse binding' in cases like (2).

I will illustrate the claim by considering data from German, English, Italian, Korean, and Portuguese. The theory developed is a generalization of the theory of reflexive binding proposed for English in Pollard and Sag (1992, 1994), which allows the determination of individual, language-specific principles A by fixing a small class of parameters which constrain reflexive binding on argument structure.