Journal of Linguistics

Trapped morphology 1

a1 SUNY Stony Brook, Center for Advanced Study
a2 Centre for Advanced Study, University of Oslo

Article author query
harris ac   [PubMed][Google Scholar] 
faarlund jt   [PubMed][Google Scholar] 


We argue that there is a diachronic process, distinct from phonological erosion, that results in the loss of inflectional morphology that is trapped when a clitic attaches to a host, becoming an affix. This is supported with attested examples from Mainland Scandinavian, Georgian, Spanish, and Greek, as well as shallow, well-accepted reconstructions from Slavic and Georgian. It is further supported by new reconstructions from Zoque (Mixe-Zoquean) and Andi (Northeast Caucasian). For example, in Old Norse the postposed article is a clitic, and there is a case ending between the noun stem and the article: hest-s=in-s ‘the horse (gen)’. The first s is trapped morphology, and it is subsequently lost: hest-en-s. Similarly, in pre-Georgian, the postposed article traps the ergative case marker, *-n: *k'ac-n=ma-n ‘the man (erg)’; it is subsequently lost: k'ac-man. We argue that the loss of trapped morphology is not sound change or another phonological process, but a morphological process.

(Received July 25 2005)
(Revised December 12 2005)

c1 Department of Linguistics, SUNY Stony Brook, Stony Brook, NY 11794-4376, U.S.A. E-mail:
c2 Institutt for lingvistiske og nordiske studium, Postboks 1102, Blindern, NO-0317 Oslo, Norway. E-mail:


1 The research reported here was supported in part by the National Science Foundation under Grant Number BCS-0215523 (ACH) and by the Centre for Advanced Study in Oslo in 2005 (both authors). We are grateful to our colleagues at the Centre for Advanced Study for their comments, especially to Henning Andersen for help with the Slavic data. We also appreciate comments and challenges from members of the audience at the workshop Weak Words: Their Origin and Progress, held in Konstanz in April 2005, where an earlier version of this paper was presented. We would also like to thank Brian Joseph for discussion of the Greek data.