This study is concerned with the mental representation of the entities
denoted by plural expressions (e.g., "the houses", "most orphans",
"they"). One possible assumption is that the denotation of a plural
expression is represented by a number of distinct tokens standing for
the atomic individuals of the set referred to ("atomic-tokens view").
However, it is conceivable that plurality is mentally construed at a
level of granularity that is coarser than the one for individual
things. An utterance with a plural expression may be used for and
understood as conveying information about an assemblage of individuals
conceived of as a single whole. Accordingly, it could be assumed that
the denotation of a plural expression is represented by a single token
("assemblage-token view"). Linguistic data suggest that neither of the
two views is valid in general. Rather, the two views seem to hold for
different types of plural expressions. More specifically, the
atomic-tokens view seems to be appropriate for "partitioning" plural
expressions, i.e., partitive plural NPs (e.g., "most of the orphans",
"both of the cars") and functionally similar expressions (e.g., "most
orphans" "both"). The assemblage-token view, in contrast, seems to
hold for unmarked plural expressions (e.g., "the cars", "they"). This
hypothesis was investigated in two experiments with German
participants, contrasting sentences that contained either the
partitioning pronoun "beide" ("both") or the non-partitioning pronoun
"sie" ("they") as external argument. Experiment 1 investigated how the
predicate of such sentences are interpreted. We hypothesized that with
a sentence containing a partitioning expression as external argument
(e.g., "both brought a gift"), the predicate would be interpreted as
applying to each atomic individual (i.e., two gifts are brought),
whereas with a sentence containing a non-partitioning expression as
external argument (e.g., "they brought a gift"), the predicate would
be interpreted as applying to the whole assemblage (i.e., one gift was
brought). The aim of Experiment 2 was to obtain information as to
whether anaphoric reference resolution of "they", but not of "both"
involves grouping the individuals referred to. To this end, reading
times of sentences with either "they" or "both" as external argument
were measured, and ease of grouping was varied by manipulating the
degree of commonality among the individuals referred to. The results
of the experiments support the hypothesis that the denotation of a
partitioning plural expression is represented by a number of distinct
tokens whereas the denotation of a non-partitioning plural expression
is represented by an assemblage token.