Marc van Oostendorp
Schwa in phonological theory

Email:oostendorp@rullet.leidenuniv.nl


Wednesday

Vowel reduction in Germanic, a cross-linguistic overview of the facts; reduction as feature loss; reduction and halfway-reduction in Russian; problems and advantages of an account in terms of faithfulness to the head for Dutch.


Vowel reduction in Germanic languages

(High) German
No productive vowel reduction to schwa at all
English
Reduction is described within the SPE-tradition as relatively straightforward: all unstressed vowels get reduced to schwa. According to Burzio (1994) this unnecessarily complicates the analysis of English stress (e..g with destressing rules). Therefore, he proposes a new analysis of vowel reduction:
  1. In the structure VCX, where X is not a vowel:
    1. Reduction of V is generally inhibited, because C requires vocalic support (not provided by X): adirond[ae]ck, adjectival.
    2. Reduction of V is permitted if either (i) or (ii):
      1. C has high intrinsic sonority, being either a sonorant or s: sere[@]ndipity, apr[@]n.
      2. C requires a low sonority downstep, being t/d: connectic[@]t.
    3. Reduction of V is forced (plus or minus idiosyncracies) by ''constant transition,'' when the preceding stressed syllable is light (``Arab rule''): hamm[@]ck, rec[@]gnition.
  2. In the structure VC1C2X, where X is not a vowel:
    1. If C1 is (relatively) sonority-transparent, being a sonorant or s, then C2 is maximally licensed if it requires a low sonority downstep (t): elephant; and only minimally licensed if it requires a high sonority downstep (p, k): podunk. Minimal licensing (as with clusters (Cp, Ck) results in both of (i), (ii)/
      1. Relative infrequency: podunk
      2. Non-reduction of V (whose sonority is critical): pod[/\]nk
    2. If C1 is (relatively) sonority-opaque, being a stop, then C2 is maximally licensed only if it either requires a low sonority downstep (t): aprub[t]; or if it is intrinsically sonorous (sonorant, s): particip[l]. Clusters stop-p, stop-k are thus excluded.

      Note: The contrast between *tp, *tk and sp, sk, sonorant-p, sonorant-k indicated that t(/d) cannot be treated on a par with sonorants/s, underscoring the independence of (i), (ii) in Ib above.

Dutch
Reduction is possibly even more complicated than in English. Factors that presumably play a role:

Frequency:
vowels in more frequent words tend to get reduced more easily than those in non-frequent ones.
Differences between sociolects and dialects:
certain dialects/sociolects show more reduction than others

It is hard to see how these factors could possibly be taken into account within the present form of generative grammar; presumably, this is the reason why they are usually ignored. We will not go into these factors either. Other factors that we will take into account:

Syllable type:
vowels in open syllables are easier to reduce than those in closed syllables; in some idiolects reduction in closed syllables is only possible if the following consonant is deleted
/bEnzin@/ [b@zín@] (*[b@nzín@]) `petrol'
/kAntor/ [k@tór] (*[k@ntor]) `office'
Stress position:
vowels in so-called 'semi-weak' positions are easier to reduce than those in 'weak' positions.
/fonoloGi/ formal style [fònoloGí]
less formal style [fòn@loGí]
informal style [fòn@l@Gí]
impossible *[fònol@Gí]
Style of speech:
we find more vowel reduction in informal styles of speech than in formal styles.
Position in the word:
vowels in absolute word-initial or absolute word-final position do not reduce
/plezir/ [pl@zír] `fun'

/eGal/ [egál] (*[@Gal]) `even'
/tOfe/ [tÓfe] (*[tÓf@]) `toffee'
Vowel quality:
mid vowels are easier to reduce than high and low vowels; front vowels are easier to reduce than back vowels; rounded vowels are easier to reduce than unrounded ones.
Reduces in Weak position Semi-weak position

/e/ formal formal
/a/ semi-formal semi-formal
/o/, /i/ semi-formal informal
/u/, /y/ informal excluded
Segmental environment:
Vowels next to /h/ cannot be reduced.
/plezir/ [pl@zír] `fun'

/helas/ [helas] (*[h@las]) `alas'


Reduction as feature loss

If schwa is an empty vowel, reduction to schwa obviously means feature loss.The question is: why do vowels loose features?

Two possible answers:

  1. Schwa is better than full vowels in an unstressed position.
  2. Schwa is generally better than full vowels, but reduction is blocked in stressed position.

I advocate the former option; the second has been proposed by John Alderete and will be discussed below.

My analysis centers around the following violable constraint:

PROJECT(V,Ft): a vowel dat dominates a vocalic feature should be the head of a foot

In essence, this constraints states that only schwa is legitimate outside the head position. In this analysis, as in all others, stress is therefore taken as the crucial factor in reduction.


Faithfulness to the head

Correspondence
Given two strings S1 and S2, correspondence is a relation R from the elements of S1 to those of S2. Segments a E S1 and B E S2 are referred to as correspondents of one another when ARB.

MAX
Every segment of S1 has a correspondent in S2.

DEP
Every segment of S2 has a correspondent in S1.

IDENT(gF)
Correspondents segments agree in the value for feature F. If ARB and A is [gF] then B is [gF].

HEAD-DEP
Every segment contained in a prosodic head in S2 has a correspondent in S1. If B is contained in a prosodic head in S2, then B E Range(R).

In the Mississippi Valley Siouan language Dakota, stress regularly falls on the second syllable from the beginning of the word. Epenthesis into the second syllable, however, for the purpose of syllabifying certain root-final consonants as onsets, creates exceptions to canonical second syllable stress.

chi-kté `I kill you'
ma-yá-kte `you kill me'
wichá-kte `you kill them'
o-wícha-ya-kte `you kill them there'

/Cek/->[Céka] `stagger'
/khuS/-> [khúza] `lazy'


Reduction and halfway-reduction in Russian

In Russian, it appears that vowel reduction to low vowels, /stol-á/ -> stal-á, only occurs in the syllable directly preceding the stressed syllable (the pre-tonic syllable henceforth). All vowels reduce to [@] in unstressed, non pre-tonic syllables.

Nom. Sg. stól slóv-o
Gen. stal-á slóv-a
Dat. stal-ú slóv-u
Instr. stal-óm slóv-om
Loc. stal-é slóv-e

Nom. Pl. stal-ý slav-á
Gen. stal-óf slóv
Dat. stal-ám sl&av-ám
Instr. stal-ámi slav-ámi
Loc. stal-áx slav-áx
'table' 'word'

v[ó]d@j nom. pl.
v[a]d-á nom. sing.
v[@]davóz 'water carrier'

z[a]vót 'winding mechanism'
z[@]vadít 'to bring, wind up'

sk@v[a]rót gen. pl. 'frying pan'
sk@v[@]radá nom. sing.