Although bijna (almost) and net (just) could be seen as internal negations in the sense that bijna (almost) is equivalent to net niet (nearly) and bijna niet (almost not) to net (just), the modifiers bijna and net niet do not exhibit identical distribution patterns with nominal quantification:

(1) bijna geen/niemand (almost no/nobody) *net niet geen/niemand (*nearly ...)

(2) bijna alle/iedereen (almost all/everybody) net niet alle/iedereen (nearly ...)

(3) bijna vijftig/de helft (almost fifty/half ..) net geen vijftig/net niet de helft (nearly ...)

(4) bijna evenveel (almost as many/as much) net niet evenveel (nearly as many/much)

The semantics of bijna consists of two components, namely negation and small distance: bijna means "not, but very close": bijna alle/bijna evenveel is "not all/not as many, but very close to it". The same two components characterize net niet but its overt negative marker renders it incompatible with the negative forms in (1). Bijna and net (niet) also interact differently with the aspectual continuity adverbials nog niet (not yet), niet meer (no longer) and nog (still):

(5) a. De gezichten zijn bijna te herkennen. b. De gezichten zijn nog net niet te herk.

(‘The faces are almost recognizable’). (‘The faces are just not yet recogn.’)

(6) a. nog bijna niet (‘almost not yet’) b. (nog maar) net (‘only just/barely’)

(7) a. bijna niet meer (‘almost no longer’) b. nog net (‘still just’)

(8) a. ?? nog bijna (‘still almost’) b. net niet meer (‘just no longer’)

Both with the aspectual transition from minus to plus recognizable in (6a) and that from plus to minus recognizable in (7a) bijna reverses the original polarity of nog niet and niet meer from negative to positive: a reference point is located inside the positive area of recognizability but very close to the respective transition points. The addition of net in (5b) and (8b), by contrast, only adds the idea of small distance but does not affect the negation. A graphical representation format is proposed for (5-8) which captures these two semantic dimensions. The decreased acceptability of nog bijna in (8a) can then be accounted for by the greater complexity of the original representation for nog as opposed to those of nog niet and niet meer. Similar constraints can be observed with the interaction between bijna and the aspectual focus adverbials which describe discrepancies between the actual course of events and the expected alternatives:

(9) a. al bijna b. nog altijd niet c. ? eindelijk bijna

‘already almost’ ‘still not’ ‘finally almost’

(10) a. al b. ? nog altijd bijna niet c. eindelijk

‘already’ ‘still almost not’ ‘finally’

(11) a. al bijna niet meer b. nog altijd c. ?? eindelijk bijna niet meer

‘already almost no longer’ ‘still’ ‘finally almost no longer’

(12) a. al niet meer b. ?? nog altijd bijna c. eindelijk niet meer

‘already no longer’ ‘still almost’ ‘finally no longer’

Again a graphical representation format is proposed which captures the natural compatibility between the semantics of bijna and that of the positive focus adverbials al (niet meer) in (9a-12a), where the actual course of events is evaluated as faster than expected. The semantic conflict between bijna and the negative focus adverbials nog altijd (niet) in (9b-12b) and eindelijk (niet meer) in (9c-12c), where the actual course of events is evaluated as slower than expected, receives an equally straightforward account in these terms.

K.U.Leuven

Departement Linguïstiek, Blijde-Inkomststraat 21, 3000 Leuven, België, e-mail: Hans.Smessaert@arts.kuleuven.ac.be