| AUTHOR: Allison Wetterlin
TITLE: Tonal Accents in Norwegian
SUBTITLE: Phonology, Morphology and Lexical Specification
SERIES TITLE: Linguistische Arbeiten 535
PUBLISHER: Walter de Gruyter
Stanimir Rakić, Belgrade University
In Scandinavian, scholars usually distinguish two kinds of tonal accents, dubbed
Accent 1 (Acc1) and Accent 2 (Acc2). In this book, which represents a revised
version of her dissertation, W challenges ''the firmly established view that in
Scandinavian tonal dialects Accent 2 is the lexically specified word accent''.
She proposes a contrary hypothesis that Acc1 is a lexical accent, while Acc2 is
governed by rules. This approach has already been introduced in papers
co-authored by W (Lahiri, Wetterlin, Jönsson-Steiner 2005a, b, 2006). The book
provides a comprehensive study of the morphophonology of Standard East Norwegian
(SEN) with some excursions into Central Swedish (CS) and Danish. At the same
time, W offers readers an overview of ''facts and arguments of competing
approaches'', with the ultimate aim of providing ''an account for the distribution
of word accent in all dialects of North Germanic''.
The book consists of eight chapters of unequal length, two appendices and a list
of abbreviations. Chapter 1 introduces the subject matter and outlines the
argument. Chapter 2 presents basic facts about Scandinavian tonal accents, and
points to its characteristics, both those which are agreed on and objects of
controversy. The majority of Norwegian and Swedish dialects belong to the small
set of Germanic varieties which have a tonal opposition in polysyllabic words.
These opposing melodies are usually called Acc1 and Acc2, although their
properties differ from one dialect to another.
The lack of tonal opposition in monosyllabic words is usually accounted for by
the fact that Acc2 is phonetically more complex than Acc1 and requires a
disyllabic trochee for its realization. This implies that all monosyllables can
have only Acc1. Kristoffersen (2000) detects an additional high tone (H) in the
contour of Acc2 which is lacking in Acc1 words. This additional H is assumed to
be a lexical tone distinguishing Acc1 and Acc2 words. Both Acc1 and Acc2 are
linked to primary stress, and in this respect Scandinavian tonal dialects differ
from Tokyo Japanese, which has no stress.
In the first part of Chapter 3, W shows how different affixes affect accent
distribution in SEN. She separately presents Germanic and non-Germanic affixes,
starting with prefixes. The great majority of words with non-Germanic prefixes
have initial stress and bear Acc1, while the effect of Germanic prefixes depends
on the fact whether they are stressed or not. The effects of suffixal
derivations depend not only on the stress, but also on the syllabic structure of
suffixes. The analysis of inflection singles out only the forms in the
superlative, in the umlauted plural and in some present tense verbs having Acc1.
All other inflectional syllabic suffixes when attached to simplex words
regularly induce Acc2.
In the second part, W analyses the contributions of Withgott & Halvorsen (1984,
1988), Kristoffersen (2000) and Riad (1998, 2003), which all assume that Acc2 is
lexically specified accent. W&H frame their analysis in autosegmental and
lexical phonology, but W objects that W&H's theory is unnecessarily complicated
in that they differentiate between weakly and strongly dominant classes of
affixes. A novelty in their approach is that monosyllabic words also can be
specified for H, which becomes visible when some suffix is added.
Kristoffersen's (2000) also assumes that both stems and affixes are specified
for bearing a high tone (Acc2) or no tone at all (Acc1). After examining the
inventory of words specified for Acc2, K attempts to capture the distribution of
tones with three constraints. W finds that K's proposals fit better into the
synchronic grammar of SEN than W&H's.
Riad's (1998) paper is devoted to the analysis of the accent distribution in
Central Swedish. Accent assignment in CS is very similar to East Norwegian, the
major difference being that all compounds, and in general words consisting of
two prosodic words, have Acc2. As in SEN, Acc2 contour in CS has an additional H
which Riad also treats as a lexical tone. Like Kristoffersen (2000), Riad (2003)
proposes some constraints banning the assignment of high tones in some
circumstances. W remarks that some general constraint preventing the
post-lexical prosodic derivation of tone is missing in Riad (1998), when a
lexical tone is left stranded on the lexical level.
Chapter 4 provides W's own view on Scandinavian tones. Most scholars have
observed that loanwords usually have Acc1, yet they generally view Acc1 as the
default. W considers such an approach unnecessary complex and counterintuitive
as it assumes that material from foreign languages follows rules, while familiar
native words have to be stored in the lexicon. Taking a reverse stance, W views
Acc1 as the special accent ''that stands out from the rest and does not follow
rules'', while Acc2 is ''the default'' (p.49). According this view, Acc1 can be
lexically specified on words, suffixes, stress-bearing prefixes, or as ''post
accenting'' on unstressed prefixes. Its basic characteristic is pertinaciousness
– it always dominates in complex words. If no lexical accent is specified,
polysyllabic words with a disyllabic trochee are assigned Acc2, and words with
final stress are assigned Acc1. For example, ''kirke'' 'church' gets Acc2 because
it has a disyllabic trochee, and does not contain any lexically specified
morpheme. The plural ''kirker'' 'churches' keeps Acc2 because the plural suffix
-er is not lexically specified for Acc1. The rules determining the accent W dubs
as Disyllabic Trochee Rule (DTR).
The analysis of verbal and adjectival inflection proceeds along the same lines,
applying DTR in derivations. Lexical specification is used most economically in
the analysis of inflection – only three inflectional suffixes are lexically
specified for accent: the indefinite umlauting plural -er, the comparative
umlauting suffix -re and the indefinite superlative suffix -st. W encounters
more problems in her analysis of derivational affixes in SEN. A special problem
is presented by stressed prefixes which occur with Acc1 in verbal derivations,
and with Acc2 in nominal derivations. W solves this problem by assuming two
allomorphs of these prefixes. The verbal prefixes are specified for Acc1, and
nominal prefixes are unspecified – the nouns with these prefixes get default Acc2.
W carefully analyses the impact of suffixal derivations on accent assignment.
First, W examines how each suffix affects the accent of monosyllabic and
polysyllabic stems. The suffixes -bar, -dom, -else, -er, -ig, -ing, -lig, -løs
and -som pass both these tests and are classified as lexically unspecified for
accent. For some problematic suffixes other options have to be considered. For
example, for the suffix -(i)sk, which appears both in Acc1 and Acc2 derivation,
W assumes two different forms – one lexically specified for Acc1, and the other
Chapter 5 is devoted to compounds. W first reviews relevant literature. W&H
(1984, 1988) observed that the first constituent always determines the accent of
the whole compound, irrespective of the kind of accent it possesses.
Kristofferen's (2000) approach to tone assignment in compounds is marked by his
assumption that every lexical item has a special compound stem with an accent
which may differ from the accent of the stem itself. W finds it odd that a
common Germanic word such as ''land'' has to be stored in the lexicon with a
number of stems. To avoid such complications, W assumes that all words, not only
monosyllabic, can be lexically specified for Acc1. The problematic cases are
compounds with linking -s. For example, '''landgang'' 'landing' and '''landscap''
‘landscape’ have Acc2, but '''landsman'' 'compatriot' has Acc1. W suggests that in
combination with -s in '''landsman'', ''land'' is specified at word level for Acc1.
Aware that this solution is not satisfactory, W suggests that future research on
the history of linking elements and foot structure of compounding will shed more
light on this problem.
In Chapter 6, W tries to account for the introduction of Acc1 words by
diachronic evidence. From the 13th to 16th centuries a great number of complex
words from Middle Low German (MLG) entered Scandinavian and were gradually
analyzed as consisting of stems, suffixes and prefixes. Proto-Norse syncope had
eliminated most unstressed vowels, and with them most unstressed prefixes, so
that only few verbal prefixes containing heavy syllables survived. W assumes
that before syncope, nouns were stressed on prefixes, and verbs stressed on
stems because this is still “prevailing pattern in many Germanic languages
today” (p. 129). Therefore, the surviving prefixes either contained heavy
syllables, or they were nominal prefixes. According this view, the nouns
borrowed with stressed prefixes were assigned Acc2 because there was a native
pattern for them to fit in. For the verbs with unstressed prefixes there was no
pattern for them to follow, and they were assigned Acc1, which already existed
as the accent of some ‘special’ category of words. The independently borrowed
verbal bases, which in most cases were disyllabic, have usually been endowed
with Acc2. The difference in the accent was consequently ascribed to prefixes
which are understood as lexically specified for Acc1.
In Chapter 7, W reviews previous acoustic investigations of East Norwegian
dialects and reports the results of her own investigation of the accents in the
Trondheim dialect. In previous work (Fintoft 1970, Kristoffersen 2000, 2007),
SEN dialects show Acc1 with lower F0 at onset of the first vowel and earlier F0
minimum, while Acc2 has higher F0 at onset of first vowel and later F0 minimum.
W's own work is consistent with these findings, but she also tries to answer
some specific questions about the contours of stressed syllables in prefixed
verbs and stems. In particular, she detects acoustic clues regarding accents
already in the first stressed syllable – in Acc1 the fall of F0 is steeper than
in Acc2. Statistical analysis with ANOVA reveals more variation in stressed
vowels in Acc2 words than in Acc1 words. W claims that this supports her
assumption that Acc2 is default accent, ''since traditionally it has been
maintained that phonemic differences are categorical whereas post-lexical
phonology is more gradient'' (p. 163).
Chapter 8, which is barely more than a page, reaffirms the conclusions of the
preceding chapters and points to the research which could further expand our
understanding of Scandinavian accents.
The book contains a detailed overview of distribution of accents in SEN, a
critical analysis of major studies of Scandinavian accents, and, importantly, a
number of fine morphophonological analyses of accent in SEN. In this respect,
the book will be a useful addition to the libraries of all linguists interested
in Scandinavian accent. The book contains several misprints which in most cases
do not interfere with intelligibility.
The main thesis of the book is, as noted, that Acc1 is a lexically specified
accent “which does not follow rules”. It may seem problematic in view of the
evidence provided in the book. In the course of the book, W has given several
rules describing the distribution of both Acc1 and Acc2. Here are some of the
rules pertaining to the distribution of Acc1:
1) the words stressed on the final syllable get Acc1;
2) the words ending in vowels other than schwa tend to have Acc1;
3) the polysyllabic words of foreign origin tend to get Acc1;
4) the words with unstressed prefixes have regularly Acc1;
5) the verbs with prefixes regularly have Acc1;
6) the umlauted plural in nouns and comparative in adjectives are specified for
7) Acc1 is very pertinacious – it always dominates in complex words.
The qualification “lexically specified” cannot cover all observations in 1) – 7)
without contradiction, especially those in 1) – 3), which are essentially
phonological in nature. Additionally, the property 7) is not lexical in
character, but rather a consequence of the general prosodic organization of SEN.
The book however does not offer a systematic overview of the syllabic and
metrical structure of SEN, although it is highly relevant for the distribution
of tone. It is clear that Acc2 cannot be assigned to the words with final accent
1), because no disyllabic trochee is available, but no synchronic explanation is
given for the absence of Acc2 in the word-classes 2) – 6). In fact, Acc1 is
morphologically or phonologically specified for these word-classes, while Acc2
is specified for the nouns with stressed prefixes. The domain where Acc2 is
“default” seems to be restricted to the set of disyllabic or trisyllabic words
which do not contain unstressed prefixes. Therefore, Acc2 is “default” only in a
subset of the lexicon, while in the rest it is morphologically or lexically
specified. Felder, Jönsson-Steiner, Eulitz and Lahiri (2009) have shown in an
experiment that Acc1 is identified faster than Acc2 in Swedish, but this result,
which suggests that Acc1 is lexically specified, has been supported only for the
set of disyllabic monomorphemic nouns.
In her comment of the property 3), W notes that monosyllabic loanwords “have a
form that fits better into the Scandinavian phonological templates than
polysyllabic words. Indeed, the disyllabic trochee – a form which monosyllables
have when suffixed – is a very common template for a Scandinatian word” (p.
143). This statement clearly implies that Acc2 is probably a default accent only
in a part of the lexicon of SEN, or even Scandinavian in general.
W’s claim expressed in chapter 7 that Acc2 is a post-lexical rule is difficult
to accept regarding the fact that Acc2 is lexically and morphologically
restricted in many ways. In terms of lexical phonology, Acc2 can only be a
lexical rule as it allows a huge number of exceptions which comprise beside
particular examples, whole morphological classes as shown in 1) – 6).
To sum up, we can conclude that the book offers a useful overview of the
distribution of Acc1 and Acc2 in SEN, and especially some fine analyses of the
role of accents in inflection and derivation in SEN. However, the general thesis
of the book does not follow from the evidence provided and cannot be
unconditionally accepted in its present form.
Felder, V., E. Jönsson-Steiner, C. Eulitz, A. Lahiri. 2009. Asymmetric
processing of tonal contrast in Swedish. In: Attention, Perception &
Psychophysics 71(8), 1890–1899.
Fintoft, K. 1970. Acoustical Analysis of and Perception of Tonemes in some
Norwegian Dialects. Oslo: Universitetsvorlaget.
Kristoffersen, G. 2000. The Phonology of Norwegian. Oxford: Oxford University Press.
Kristoffersen, G. 2007. Dialect variation in East Norwegian tone. In: T. Read,
C. Gussenhoven (eds.). Tones and Tunes: Studies in Word and Sentence Prosody.
Berlin: Mouton de Gruyter. 91–111.
Lahiri, A., A. Wetterlin, E. Jönsson-Steiner. 2005a. Lexical specification of
tone in North Germanic. Nordic Journal of Linguistics 28(1). 61–96.
Lahiri, A., A. Wetterlin, E. Jönsson-Steiner. 2005b. Sounds definite-ly clitic:
Evidence from Scandinavian tone. Lingue e Linguaggio IV. 243–262.
Lahiri, A., A. Wetterlin, E. Jönson-Steiner. 2006. Scandinavian lexical tone:
Prefixes and compounds. In: G. Bruce, M. Horne (eds.). Nordic Prosody IX. Lund:
Riad, T. 1998. The origin of Scandinavian tonal accents. Diachronica 15. 63–98.
Riad, T. 2003. Distribution of tonal accents in Scandinavian morphology. Paper
presented at the First International Workshop on Franconian Tone Accent. Leiden,
13-14 June, 2003.
Withgott, M., P.K. Halvorsen.1984. Morphological constraints on Scandinavian
tone accent. Stanford: CSLI Report No. 84–11.
Withgott, M., P.K. Halvorsen. 1988. Phonetic and phonological considerations
bearing on the representations of East Norwegian accent. In: H. van der Hulst,
N. Smith (eds.). Autosegmental Studies on Pitch Accent. Dordrecht: Foris
ABOUT THE REVIEWER