GRAMMATICALIZATION AND ANCESTRAL GRAMMAR
 
                       Bernard H. Bichakjian
                      University of Nijmegen
                    B.Bichakjian@mailbox.kun.nl
 
  Ever since Bybee, Perkins and Pagliuca's Evolution of Grammar (1994),
grammaticalization has been a popular topic.  The popularity is
understandable.  Since it occurs practically under the observer's eyes,
the process is both indisputable and easy to describe.  Moreover, since
it is unidirectional, it is used to provide an illustration of the birth
of grammar. 
 
  Indeed no one can dispute that items such as Engl.  have (< a cognate
of Lat.  capio 'catch') Sp.  haber (< Lat.  habeo 'to hold, keep') Pg. 
ter (< Lat.  teneo 'to hold, keep') have at least in part shed their
original lexical meaning and become free grammatical morphemes.  It
could even be argued that forms such as Fr.  (je) chanterai '(I) shall
sing' (< cantare habeo) exem plify the change of a lexeme becoming a
bound grammatical morpheme, though it probably was a two-step process. 
First the lexical item became a modal auxiliary gradually changing into
a temporal one, and then the ensuing free morpheme was reduced to an
inflectional marker. 
 
  Whatever the number of steps, the case of lexical items being
occasionally pressed into grammatical service and losing their semantic
content is well documented.  Yet, as the process is labeled
"grammaticalization" and used to illustrate the birth of grammar, one is
prompted to ask two essential questions:
 
    1.  Is "grammaticalization"a process that creates novel grammatical
functions or simply novel grammatical markers.  In other words, does
"grammaticalization" produce an expansion or an enrichment of the
grammar of the host language, or does it simply consist of replacing an
eroding or otherwise less satisfactory marker with a preferred one?
 
    2.  Can "grammaticalization" account for all grammatical markers in
natural languages?
 
  This paper will argue that the historical record suggests that the
total or partial reduction of lexical items to grammatical morphemes is
a process that renews the stock of existing markers, but more often than
not does not create new grammatical functions.  The explanation of the
creation of new grammatical functions must therefore be sought
elsewhere. 

  Addressing the second question, this paper will further argue, that an
important fraction of the grammatical markers of natural languages
cannot be accounted for by lexical erosion.  Indeed a number of "living
fossils," such as vowel alternation, consonantal gemination, syllable
reduplication, etc.  suggest that the reduction of lexical items was
neither the only process for the creation of grammatical markers, nor
indeed the most ancient one.  "Grammaticalization" is therefore one
among several ways of creating grammatical markers and by no means the
oldest.