Rit (Vísindafélag Íslendinga) - 01.06.1970, Page 476

474
units as either stops or continuants (i.e., as /d, g/ or /ð, y/).
It is assumed, however, that there is an underlying difference
of voicedness between /t, k/ and /D, G/. According to this
interpretation the stems offede, mode, brede, bred all end in /D/,
and those of klog, koge both end in /G/.
It seems immediately reasonable to assume that the forms
with stem final /D/ all have an underlying long vowel.6 If
this is true the difference of vowel length between (neut.) bredt
and modt on the one hand and (past part.) bredt and fodt on
the other must apparently be conditioned by the suffixes.
The neuter suffix is /t/; there is no question about that. But
the verb suffixes may have either /t/ or /D/, as may be seen
from the inflexion of verbs with no stem-final consonant:
[sge:’] ‘happen’—[sge:do]—[sge:’t] vs. [do:’] ‘die’—[d0:ða]—
[do:’ð] ([doð’]). The suffix consonant of the former verb
reflects /t/ (unless the suffix is added to a stem augmented
with /D/, in which case the forms skete, sket correspond to
bredte, bredt,fodte,f0dt). The suffix consonant of the latter verb
unquestionably reflects /D/.
If verbs like mode are supposed to take suffixes in /t/, and
verbs like fode, brede are supposed to take suffixes in /D/ (this
happens also to agree with historical evidence), we get the
following combinations among members of the set /t, D/:
Underlying /t/ + /t/, cp. past part. [sat] from [seda] ‘put’,
» /D/ + /t/> cp. [mot], [brst],
„ /D/ + /D/, cp. [fo:’t], [bre:’t];
also [bre:’da] ‘width’.
It is seen that combinations among members of this set all
reduce to one segment, which is identical with the surface
representation ofsingle /t/, i.e., [d] or [t]. Exceptions are some
peripheral neuter forms with /D/ + /t/ manifested as [ð] plus
stop (e.g. perfid-t, led-t\ see §3, end).
•This would immediately seem most natural since the vowel is long in mono-
syllabic forms (without a stop suffix).