Abstract

In Ancient Greek a single indefinite enclitic pronoun tis/ti was used indifferently in both negative/affective environments in the manner of a negative polarity item (henceforth NPI , cf. anyone/anything), and in positive ones in the manner of a positive polarity item (henceforth PPI , cf. someone/something). At the same time the negative pronoun oudeís/oudén was used both as a true negative quantifier (henceforth NQ, cf. no one/nothing) and as an emphatic NPI with negative concord (cf. substandard I didn’t see nothing).

The functions of each pronominal class – i.e. PPI -like vs NPI -like for indefinites, NQ vs NPI for negatives – were determined by syntactic distribution in relation to the presence or absence of a sentential negative marker (henceforth NM). As expected, in positive, non-modal statements indefinites and negatives functioned in all positions as PPIs and NQs respectively. By contrast, an indefinite before a NM functioned like a PPI (cf. something didn’t fit the evidence) and after a NM like an NPI (cf. the evidence didn’t amount to anything), while a negative pronoun before a NM was a true NQ (i.e. with double negation, cf. no one didn’t see the problem) and after a NM a ‘strong’ NPI (i.e. without double negation, cf. again substandard I didn’t see nothing).

This pattern was increasingly at odds with a clause structure in which focal constituents were often contrastively stressed and fronted to the left periphery: neither indefinite nor negative pronouns could be routinely focalised in this way because of the various prosodic and/or semantic restrictions on their distribution. The deficiency was remedied in the early middle ages by (i) formal/prosodic recharacterisation of the indefinites (distinct sets of PPIs and NPIs eventually emerged, cf. Modern Greek ‘kapjos ‘someone’ vs. ka’nis ‘anyone’), (ii) the loss of NQs and (iii) the generalisation of NPIs in negative sentences to all syntactic positions available to other DPs, including the focus position, a process that entailed their reinterpretation as involving universal quantification over negation (‘for any X it is not the case that…’) rather than, as before, existential quantification under negation (‘it is not the case that there is an X such that…’).

The final outcome is typologically to be expected in so far as NQs are clearly redundant in a system in which NPIs can appear both before and after NMs. It is suggested that there are only three logically possible systems for the interaction of sentence negation with negative indefinites (i.e NQs and/or NPIs), and that there is a characteristic cycle of diachronic development – even though no particular stage in the cycle is more inherently ‘unstable’ than any other.