This chaining, unlike the chaining ofcardinal, is semantically transparent, in that the original concept is contained in
both the derivative concepts. This sort of transparent chaining is a standard case of polysemy.
One might try to claim (with Deane 1996 and Ruhl 1989) that these three uses are not separate senses that must be
learned individually, but rather for ma single sense that can be extended fro mthe si mple core prototype (6a). This
treatment runs into problems, though, when we start dealing with long branching chains. An example is the word
smoke. The core is the nounsmoke, the wispy substance that comes out offires. Fro mthis we get at least the senses in
(7).
(7) a. smoke 1 =‘wispy substance’
b. X smokes 2 =‘X gives off smoke 1 ’
c. Y smokes 3 X=‘Y causes X to smoke 2 , where X is a pipe, cigar, etc., by putting in the mouth and puffing,
taking smoke 1 into the mouth, etc’
d. Y smokes 4 =‘Y smokes 3 something’
e. Y smokes 5 X=‘Y causes smoke 1 to go intoX, whereX is a ham, herring, etc.,by hanging X overafire in an
enclosure’Although all the steps in (7) are transparent, I would be reluctant to say that thesefive uses together for ma single
concept for which smoke 1 is the prototype. In particular, there is really no relation betweensmoke 4 andsmoke 5 other
than the presence ofsmoke 1 as a character in each action. One reason for this is that each step of the chain adds
idiosyncratic information. For instance, it could only be a joke to speak of smoking 3 a herring by settingit onfire and
putting it in your mouth and puffing, or of smoking 5 a cigar by putting it in a smoky room.
Because ofthese multiple branches and idiosyncrasies, itmakes moresensetosay there arefive senses arranged in two
transparent chains branching outward fromsmoke 1 (1– 2 – 3 – 4 and 1–5). These senses are not any more closely related
to each other than they are to, say,smokyandsmoker(the latter meaning either‘a person who smokes 4 ’or‘a vessel in
which one smokes 5 things’). And in a morphologically richer language than English they might be morphologically
distinct; for instance German hasRanchforsmoke 1 ,rauchenforsmoke 2 – 4 , andräuchernforsmoke 5. In other words, even if
all these senses are related, the language user must learn them individually.
A check on the validity of such putativerelationships is the extent to which they occur in other lexical items. Here are
the linkages forsmoke, listing other words that show the same relationships among their senses:
(8) smoke 1 →smoke 2 (V =‘give off N’):
steam, sweat, smell,flower