This article examines the semantic functions of locative space-builder
s in Finnish existential structures ('there is an X in Y', formally in
X is Y), especially in contrast with their function in corresponding
locative structures (X is in Y). Both structures consist of a locative
modifier, a (semantically colorless) verb, and a subject, and their f
unction in both cases is to establish a (concrete or metaphorical) con
tainment relation between the space and the referent of the subject (e
.g. The baby is in the house vs. In the house there is a baby). Existe
ntials, however, have the additional function of establishing the subj
ect as a new element in the space, whereas the locative structure typi
cally predicates the location of a definite subject that has already b
een established in the discourse. The seeming similarity of the basic
containment relations expressed in existential and locative structures
is not, however, always straightforward In the following discussion I
argue that there ave certain semantic constraints on locative space-b
uilders in existential structures that do not hold true for the space-
builders of the corresponding SVX locative structures. The most fundam
ental constraint concerns the conceptual status of the space-builder:
in the existential structure the space has to be conceptually independ
ent of the element in the space (i.e. the subject), whereas in the loc
ative structure space-builders are allowed that are conceptually depen
dent on the subject. As an example, space-builders indicating nonspati
al domains such as internal mental states (which are represented metap
horically as locations) are allowed in locative structures such as The
man is in a coma but not in existential structures (In the coma ther
e is a man). This is because the man's coma exists only with respect t
o the man, and the man cannot be introduced as a new entity into a dom
ain that already presupposes his existence. The space (or domain) type
s that are conceptually dependent on other entities include metaphoric
al locations, relational terms (such as X's left/right side; X's dream
s, etc.), activity domains, and certain other types. In the following
discussion these are referred to as bound spaces, since they are conce
ptually dependent on other entities and receive their semantic interpr
etation in relation to those entities. This feature distinguishes them
from prototypical autonomous or unbound spaces.