Infant vocal behaviors are extremely complex. Consequently, coding these be
haviors is difficult and is typically associated with low reliability acros
s observers. Various difficulties that arise when dealing with prelinguisti
c vocalizations, especially in the first 6 months of life, are outlined her
e. A proposed database of digitized infant vocalizations that illustrates s
trategies used to deal with these difficulties is then described. These str
ategies are based on theoretical infraphonological constructs, empirical ob
servations, and information about the nature of mature phonological systems
. Furthermore, the strategies are open-ended and can be modified as new inf
ormation becomes available regarding infant vocal behaviors. At present, a
preliminary database is available on the Web that illustrates some of these
strategies. As the database is expanded, it is expected to provide a gener
al framework for observers to categorize infant vocalizations and thereby e
nhance observer reliability.