What do unstressed syllables
and empty nuclei have in common?
Harry van der Hulst & Grazyna Rowicka
In ongoing work we examine parallels in the distribution of unstressed
syllables and empty nuclei. We will discuss some of the parallels we think
exist and then suggest that the structural principles (among which are
government relations) which license empty nuclei can be conceptualized
as 'intralexical prosodic structure'. We will also suggest that post-lexically
this intralexical structure ceases to play a role and that at that level
a 'new' extralexical prosodic structure is formed (cf. Anderson & Ewen
1987:122ff.).
Korean vowel harmony
Duck-Young Lee & Shohei Yoshida
This paper discusses Vowel Harmony in Korean (VH) within Element Theory.
Traditionally, VH has been viewed as involving a semantic contrast of "darkness"
and "lightness". It is assumed that vowels are either dark or
light and all the non-initial vowels within a word agree with the initial
vowel in darkness/lightness. In the previous feature-based analyses, VH
is expressed as rightward spreading of either [+low] or [+ATR]. However,
these analyses are flawed by numerous exceptions. In this paper, we argue
that once VH is seen as a head-alignment of the A-element, VH turns out
to be a purely phonological process with no exceptions.
In Hamburg bin ich über
einen spitzen Stein gestolpert
Jean Lowenstamm & Tobias Scheer
In standard German, underlying s becomes sch before a
consonant word-initially (except before k). We provide a government based
account of this puzzling distributional restriction. In the course of our
presentation, we will motivate a specific view of word-initial configurations,
and of the structure of "sch...".
Pasiego Vowel Harmony, or
How OT Can Solve a Problem for Unary Feature Theories
Krisztina Polgardi
The Pasiego dialect of Spanish has been a challenge for unary feature
theories, since it involves raising harmony. Harris (1990) analysed raising
as "reduction harmony", i.e. as delinking of a dependent A in
unstressed positions, unless a governing A licenses it. The problem with
this analysis is that the condition on delinking refers to the absence
of the element A, a possibility crucially denied by unary feature theorists.
In this talk, I will argue that the problem can be solved by combining
the Government Phonology approach with Optimality Theory. Since OT is non-procedural,
the only thing that has to be stated is constraints on the output. More
precisely, the constraint expressing that combination of elements in a
governed position needs to be licensed by a governing A has to be ranked
above PARSE (A). Thus there is no need to define when a certain process
is required to apply, and reference to the absence of elements can be dispensed
with.
Hungarian
Péter Rebrus, Péter Siptár, Péter Szigetvári
and Miklos Törkenczy
Hungarian provides an array of phenomena that current GP literature
caters for on a piecemeal basis. These include vowel harmony (general backness
and parasitic rounding harmony; e.g., házhoz 'to a house',
kerthez 'to a garden', tzhöz 'to a fire'), vowel transparency
and non-harmonizing suffixes (moziba 'to a cinema', házig
'until a house'), vowel--zero alternations (e.g., sarok / sarkot
'corner NOM, ACC'; mozog / mozgok 'move 3SG, 1SG'), vowel length
alternations (e.g., nyár / nyarak 'summer SG, PL'; tél
/ telek 'winter SG, PL' vs. gyár / gyárak 'factory
SG, PL', szél / szélek 'edge SG, PL'; tiszta /
tisz-ta-ság 'clean, cleanness' vs. tiszta / tisztában
'clean, in the clean one'), lowering properties (dependent on stem types,
word categories and also other factors, e.g., hal / halak 'fish
SG, PL' vs. dal / dalok 'song SG, PL', fül / fülek
'ear SG, PL' vs. sül / sülök 'porcupine'; csinál-om
'make PRES 1SG' vs. csinál-t-am 'make PAST 1SG').
There are implicational relations between these phenomena, e.g., stems
exhibiting vowel length alternation are all lowering stems (nyár
/ nyarak), but the opposite does not hold (gyár / gyárak;
forms involving lowering or vowel--zero alternation are all harmonizing
domains, but not all harmonizing suffixes induce lowering (nyár
/ nyárság 'summer, summerness').
The above have been explained by various devices including floating
and ambient elements, empty categories, onset-to-onset government and some
ad-hoc stipulations. The fact that there are implications among these phenomena
provides an opportunity to study the interactions between these devices.
In addition, distinguishing synthetical and analytical concatenation seems
to be unavoidable in order to account for the alternations listed. Analyticity
and syntheticity can be supported by independent evidence provided by the
phonotactic constraints on Hungarian morphemes. This does not only mean
constraints on possible consonant clusters but also apply to the types
of vowels appearing word-finally (e.g., Hungarian words never end in short
[o] and [ö], verb stems always end in a consonant).
In the workshop offered we would show some weak points of the analyses
provided so far and propose alternative approaches which account for the
phenomena outlined above using the devices of GP in a more efficient and
less stipulative way. Accepting the principle that phonological processes
apply as their conditions are met, the interactions between the processes
are well observable. As a conclusion it will turn out that the application
of the devices is limited by the implicational relations, and the devices
themselves have such relations.
Mòoré vowels revisited
John Rennison
The analysis of the vowel system and vowel harmony of "standard
Mòoré" (=the dialect of Ouagadougou) which I previously
proposed (1988, 1992) is called into question by data from northern dialects
presented by R. Kabore (1994). Here, suffix vowels which in Ouagadougou
show the alternations [i]~[e]
and [u]~[o]
have a third variant, giving [i]~[e]~[E]
and [u]~[o]~[O].
In addition, there exist suffixes with the vowel [I]
which never changes (i.e. does not harmonize for ATR or A), but which causes
a mid vowel in the stem to change from [e]
to [E] or from [o]
to [O]. Superficially, this looks like
RTR harmony. I propose that my 1988 analysis is still correct for Ouagadougou,
but that the northern dialects have different parameters for A-umlaut (a
process which also manifests itself in other ways). Therefore there is
no RTR harmony.
|