234 María de los Ángeles Gómez-González
819). It is no less true, however, that we often cram more than one clause or
concept into intonational frames, or stretch out a single clause into more
than one frame.- Unless otherwise stated, data are taken from the Lancaster IBM Spoken
English corpus (LIBMSEC). For further details on this corpus see Taylor
and Knowles (1988) and Gómez-González (2001: 192–206). - Langacker (2001a) proposes the following list of natural paths, or cogni-
tively natural ordering of the elements of a complex structure:- speech time (word order, expression order in general)
- locational path of access (nested or chained)
- temporal sequence of events
- successive points along any scale
- chain of reference point of relationships
- empathy (speaker > hearer > other; human > animate > inanimate;
etc.) - chain of cause-effect relationships
- clausal organization (main clause > subordinate clause)
- viewpoint chain
- partonomy (hierarchy of whole-part relations)
- grammatical relations (subject > object > other)
He contends that, ceteris paribus, there is a tendency for natural paths to
coalign, and for the sequences in which this applies to be more basic and
easier to process (i.e. unmarked), although there are also cases in which this
tendency is overridden by other factors.- Here ‘markedness’ refers to probabilities of choice within a linguistic sys-
tem. Linguistic choices normally skew, in that one or some of their terms
are marked or less probable, in contrast with unmarked, or more prototypi-
cal or recurrent items. Further, with Kies (1988: 74), we shall take the view
that marked options have: (a) comparatively lower frequency of occurrence,
(b) comparatively higher structural complexity; and (c) more restricted dis-
tribution in definable environments (Givón’s 1993: 178 condition of
‘discourse distribution’). However, it should be borne in mind that marked
patterns are not always less frequent – less expected from the addressee’s
point of view – than their unmarked counterparts, for they are affected by
such factors as text type (Siewierska 1988: 12), or by matters of position
and context (Mithun 1987: 313). A case in point is discourse-initial sen-
tences, which belong to highly marked situations and therefore give rise to
the so-called markedness reversal, i.e. to a local affinity with marked con-
structions (Fox 1987; Pu and Prideaux 1994).