Abstract
The gap between a definite and an indefinite is reduced in the case of specific indefinites, in the sense that they both have determined reference. This fact is reflected in most accounts, particularly so in the analysis of Schwarzschild where specific indefinites are argued to quantify over singleton sets. If definites and specific indefinites share the singleton domain property, how are we to separate the two? The Privacy Principle, which is cued to the information state of discourse participants, is argued to apply to definites as well as indefinites, making it even harder to draw the line between definites and specific indefinites. We look at issues surrounding the distinction between definiteness and specificity through the lens of two empirical phenomena, bare nominals and specificity markers. Bare nominals in languages with determiners, for example, English, and bare nominals in languages without determiners, for example, Hindi, highlight different aspects of the problem. English bare plurals are known to have indefinite readings but not specific indefinite readings. Hindi bare nominals are thought to have definite and indefinite readings but can be shown not to have specific indefinite readings. The second empirical phenomenon we consider are specificity markers like certain. In many languages, they require a determiner in the singular but not in the plural, raising questions about their syntactic status as determiners or modifiers. They also do not occur in certain constructions, for example, imperatives. Probing such restrictions provides a window into their semantic profile. This paper thus presents some puzzles that center around the creation of singleton sets and its relation to the Privacy Principle.