Íslenskt mál og almenn málfræði - 01.01.2013, Page 216

214
Michael Schulte
marized and graded as follows (starting with those which probably will have a
major research impact):
1. the existence of tonemes in Old and Middle Icelandic
2. the development of the reflexive or medio-passive sk-forms
3. vowel epenthesis, i.e. the intrusion of svarabhaktis
4. internal rhyme between /a/ and umlauted /9/
5. different alliteration matrices, particularly alliteration of velar and palatal
consonants, the alliteration between semi-vowels and vowels, and Ai-allit-
eration (h — hl— hn— hr)
6. metre and rhythm in late medieval poetry (rímur)
7. textual criticism of medieval rímur, particularly Ormars rímur.
The worthiest research results pertain to the first three issues and the most
rewarding part of the work is probably the study of tonemes in conjunction with
epenthetic vowels. The author himself remarks in the English summary that the
section on epenthetic vowels or stodhljóð “is the longest and most data-heavy
chapter” (pp. 355—373). The detailed discussion of tonal distinctions in Icelandic
is likely to be the most solid and durable part of the dissertation. Apart from
these rich and multi-faceted research-modules, Haukur’s textual criticism of
Ormars rímur as well as the evaluation of other passages in dróttkv&tt verse and
in the medieval rímur deserve particular mention. In what follows, the opponent
will present an evaluation of the dissertation’s major pros and cons.
2. The notion of deep structure — dated research anyway?
Haukur convincingly argues that surface structure rather than deep structure
plays the decisive role in metrics. The dissertation thus marks a clear turn away
from abstract generative and structuralist analyses (see §10.1 “Baklægar gerðir”,
pp. 233—234, and §10.2 “Formgerðarkenningar”, pp. 235—236). In scrutinizing
earlier scholarship, Haukur is able to amend and even reject several earlier works,
not least generativist studies. Two cases in point are Edmund Gussmann’s gen-
erative study on velar palatalization from 1984 and Stephen R. Anderson’s work
on umlaut from 1973, also in a generative vein. Thus Anderson’s proposal involv-
ing “identity [of /a/ and the umlaut product /9/; M.S.] at a certain abstract level
of representation”, is safely dismissed. Apart from a series of counter-arguments,
Haukur points out a misinterpretation of the data on Anderson’s side.
However, from the viewpoint of 2ist-century linguistics, Anderson’s and
Gussmann’s studies from the 197OS and 1980S are dated in their theoretical
groundwork and the opponent wonders why Haukur has given them so much
attention.8
8 Haukur’s criticism of Anderson’s theory is based on Eiríkur Rögnvaldsson (1981:
45) and Kristján Árnason (2000:16—17).