end of an ongoing continuous process. Sleeping is a continuous process, sountilbounds it by simple composition.
However, a verbthatdenotesa temporallybounded actioncannotbefurther bounded byuntil:consider*Dave died until
the bell rang. A jump also has an inherent ending: when one lands. But unlike dying, jumping can be repeated, and
repeated jumpingis potentiallyan ongoing process. By construing jumping as repeated, it can compose in a well-formed
fashion withuntil.
What is the source of the construed repetition? One might propose thatjumpis lexically polysemous, meaning either
‘jump once’or‘jump repeatedly.’However, such a proposal fails on two grounds. First,everyverb that describes a
repeatable temporally bounded action is understood as repeated in the context ofuntil. Thus, as in reference transfer,
we are dealing with a regular process that should not have to be listed in the lexicon with every verb. Second, and
worse,thesense ofrepetitioncandependonthefreecompositionoftheverbphraseand onvariouspragmaticfactors.
Consider (21).
(21)
a. Sue slept all night. [Continuous sleeping, one night]
b. Sue slept all night until she started drinking too much coffee.
[Multiple acts of sleeping, multiple
nights]
c. Sa msang until the bell rang. [Continuous singing]
d. Sa msang the scale until the bell rang.
[Multiple performances of the scale]In (21a),sleephas beenbounded byall night, making it intoa closed action.Hence when it combines withuntilin (21b),
repetition is generated again. In (21c),singis a continous ongoing action, anduntiltells us when Sam's singing ceased.
But in(21d),singing a scaleisan action thatendswhenonegetstothetop(orgetstothetopand returns tothebottom).
Hence the only way to construe it withuntilis that Sa msang the scale over and over.
We conclude that there is no way to code every individual verb as polysemous between singular and repeated action;
rather construed repetition is a consequence arising fro mthe co mbination of all the words in the verb phrase. Hence
thesense of repetition must be“coerced”into thesentence: it is constructedonline in workingmemory, withno overt
phonologicalor syntacticevidence. Itsrole inthesentence's meaningis similar tothat ofadverbialslikerepeatedlyorover
and over, so fro mthe point of view of conceptual structure it is nothing special. All that is unusual about it is its lack of
syntacticand phonologicalform. As withreferencetransfers, I suggestwecanthink ofitas a conventionalized element
of meaning that speakers can use to“abbreviate”their utterances, trusting that hearers can reconstruct this element
when it is