LINGUIST List 9.825

Thu Jun 4 1998

Disc: The Term "Morph"

Editor for this issue: Martin Jacobsen <>


  1. Larry Trask, Second sum: the term `morph'

Message 1: Second sum: the term `morph'

Date: Wed, 3 Jun 1998 10:05:14 +0100 (BST)
From: Larry Trask <>
Subject: Second sum: the term `morph'

My earlier summary on this topic has attracted some further replies.
Those replies were quite various, and they represented much the same
division of opinion as before: one or two more people endorsed my
proposed usage as normal, one or two didn't know it but liked it, and
several took exception to it. There were a couple more proposals for
a term: `substring' and `meta-morph'. And, by the way, I
inadvertently omitted one earlier suggestion from my first summary:
`(still unanalyzed) sequence of inflectional morphs'.

However, I can now reveal the particular reason I got interested in
this. I am acting as linguistics consultant for the third edition of
the OED, and I was commenting on the entry for `morph'. Crucially, we
have now located a couple more published uses of `morph' in my sense,
which means that my sense will now be included in the entry, along
with the narrower sense originally put forward by Hockett. Amusingly,
one of those published uses occurs in the second edition of the OED,
though not under `morph' itself.

For immediate purposes, then, we have adequate documentation of
`morph' in the sense I pointed to: roughly, `a stretch of phonological
material which is morphologically unanalyzed', or perhaps `a stretch
of phonological material representing zero or more morphemes'. It is
not my purpose to defend that usage here, though I really do think it
is both appropriate and maximally convenient, as well as required by
such familiar usages as `portmanteau morph', `empty morph' and

Oh, a couple of people got interested in my example of Basque
<mendietan> `in the mountains', the locative plural of <mendi>
`mountain', or more accurately of <mendia> `the mountain'. Here the
morph <-etan> can be analyzed as containing the ordinary locative
case-ending <-n>, and perhaps also as containing the ordinary oblique
plural marker <-e-> (though not all my colleagues are happy with this
last). That leaves <-ta-> unassigned as an empty morph, obligatory
but representing no morpheme at all: this morph is required in all
local case-forms of all non-singular NPs. Diachronically, <-etan>
probably consists of an old collective suffix <-eta>, probably
borrowed from Latin, plus locative <-n>, but that collective suffix
now has no independent existence in the language: it is fossilized in
a handful of items like <mendieta> `mountain range, sierra', from
<mendi> `mountain', and in a number of place names and surnames.
You've all seen Basque surnames like Zuloeta, Elordieta and
Zubizarreta, and I imagine that Maria Luisa Zubizarreta gets tired of
being taken for an Italian called "Zubizaretta". But <-eta> is no
longer productive, except in local case-forms.

My further thanks to Waruno Mahdi, Robert Hoberman, Tapani Salminen,
Sean Jensen, Chet Creider, Pier Marco Bertinetto, Michael Beard, Sam
Martin, Grover Hudson and Karl Teeter.

Larry Trask
University of Sussex
Brighton BN1 9QH

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue