Document Sample
Hermans_Meertens_Amsterdam Powered By Docstoc
					   The representation of the ‘tonal’ accents of Limburg and Ripuarian
           Ben Hermans; Meertens Institute;
Many Limburg and Ripuarian dialects are characterized by the fact that they have two
contrastive tonal accents, usually called Accent1 and Accent2. A few representative
examples from the dialect of Roermond are: kal2 ‘nonsense’ - kal1 ‘to talk’; bei2n ‘leg,
sg.’ - bei1n ‘leg, pl.’; sjri:2f ‘write, imp.’ - sjri:1f ‘write, 1st. p. sg.’.
         In the modern generative tradition there is (almost) a consensus that the
phonological representation of the two accents is basically tonal. Although analysts
differ in the details, they agree that the tonal approach is correct.
         In my talk I argue that the tonal approach is fundamentally wrong, because it
cannot capture the basic properties of the accents. I illustrate this with one of these
core properties. This concerns the fact that in many dialects the two accents change
the quality of the vowels that support them. Three changes are particularly frequent.
1) High vowels with Accent1 have a tendency to diphthongize, whereas high vowels
with Accent2 do not have this tendency. This is shown by Maastricht. In this dialect
the two forms for ‘write’ mentioned above are realized, resp. as sjrei1f and sjri:2f.
2) the members of homorganic diphthongs with Accent1 tend to dissimilate ([ei]
becomes [ai]), whereas under Accent2 homorganic diphthongs are not changed at all
in this way. This happens in many dialects of Belgian Limburg. In dialects of this type
alternations like the following are attested: bei2n ‘leg, sg.’ - bai1n ‘leg, pl.’ (compare
this with the examples in the first paragraph).
3) Diphthongs are monophthongized under Accent2, whereas they are not changed in
this way under Accent1. This phenomenon is also active in Maastricht. This dialect
therefore has alternations of the type be:2n - bei1n.
         It is difficult to see how these relations can be accounted for in a tonal
approach. In ‘real’ tone languages segments are not changed by tones, so, when seen
as direct tone-segment relations, the three phenomena create a typological oddity.
         Instead of a tonal approach I propose a prosodic approach. The essence of the
proposal is that the Accent2 contains a branching nucleus, whereas the Accent1 is a
sequence of two non-branching nuclei. Schematically, the difference between
Accent1 and Accent2 is as in (1). The important part of these representations are the
two capitalized timing units. In Accent1 this position occupies the dependent position
of the foot. However, at the syllable level it occupies a head position. It is therefore a
relatively strong weak position. In Accent2, on the other hand, the big X occupies a
dependent position both at the foot and the syllable level. It is therefore utterly weak.
         Viewed from this perspective one would expect that the big X in Accent2
exhibits reduction effects; the big X in Accent1, on the other hand, being relatively
strong, should exhibit these tendencies to a lesser extent. This, in fact, is exactly what
happens. I will show that our three phenomena can be explained along these lines.
         Why, for instance, is diphthongization avoided under Accent2? This process
creates an independent segment in the utterly weak position. Being very weak, this
position cannot license an independent segment. This explains why diphthongization
is blocked in a case like Maastricht sjri:2f. The analysis is summarized in (2). The
utterly weak status of the big X in Accent2 can also explain monophthongization
under Accent2 and dissimilation under Accent1, as I will show in my talk.
         In this approach, the effect the two accents have on segmental quality does not
create a typological oddity. Weakening occurring in weak positions is a widely
attested phenomenon, of course. Here we come to the basis point of the talk again: the
tonal approach is wrong, and the prosodic approach is right.
(1)       Accent1                            Accent2
           F                                  F             foot structure

            σ σ σ                              σ     σ      syllables
           /| /| /|                           /| /|
          ONONON                             ON ON          Onsets-Nuclei
          | | | | | |                        | |\ | |
          x xx x x x                         xxxxx          timing units
           | | | |                           | | | |
          be in                              be in          segments

          Accent1: two segments              Accent2: two segments in a
          in two different nuclei            single, branching nucleus

          Accent1: X is dependent            Accent2: X is dependent
          (in the foot) but also head        (in the foot) and also
          (in the syllable); so it is not    dependent in the syllable,
          utterly weak                       so it is utterly weak

(2)   sjri:2f (no diphthongization):OK! sjrei2f (diphthongization):NOT OK!
              F                                    F

            σ       σ                           σ     σ
           / | /|                               /| /|
          O N ON                             O N ON
          |\ |\ | | |                         |\ |\ | |
          xxxx x x                           x x xx x x
           ||\/ |                             | | | | |
            r i v                               rei v

Shared By: