Mann:
<> == Noun_C
<phn syl1 onset> == m
<phn syl1 peak-> == a
<phn syl1 tail-> == n.
From this entry and the axioms presented in the previous sections, we can derive the following theorems about Mann (a complete proof of the equation for the dative plural of Mann is provided):
Mann:
<mor word sing nom> = m ^ a n
<mor word sing acc> = m ^ a n
<mor word sing gen> = m ^ a n s
<mor word sing dat> = m ^ a n
<mor word plur nom> = m ^ E n @ r
<mor word plur acc> = m ^ E n @ r
<mor word plur gen> = m ^ E n @ r
<mor word plur dat> = m ^ E n @ r n.
If one examines a representative sample of lexical entries for nouns, one can see that those that depart from the minimalist Mann ideal do so for one or more of three reasons: (i) they are polysyllabic, (ii) their focal syllable is nonfinal, or (iii) they embody sui generis irregularity.
Case (i) arises because we have chosen to treat words as monosyllabic by default. If a root is polysyllabic then this is stipulated. This leads to a redundancy in the individual lexical entries: if, say, a peak is given for syl2 then the root clearly cannot be a monosyllable. We could have avoided this redundancy with some clever DATR code that checked the lexical entry for the highest numbered defined root syllable segment (to discover how many syllables were present) and then routed the <phn root form> path to Monosyllable, Disyllable, Trisyllable, etc., as appropriate. We could have written such code but you would not have wanted to read it. From our perspective it would also have been a conceptually perverse thing to have done since it would imply that syllable structure was epiphenomenal, merely contingent upon the segments that happened to be present. We have thus chosen to live with this particular redundancy in our lexical entries. Zwicky has pointed out that the typical morphological lexicon ``will usually admit of multiple minimizations, thanks to the fact of mutual predictability (as when it is true both that nouns in a particular declension class are by default of the [MASC] gender and that nouns of the [MASC] gender are by default in that declension class)'' (1992, 337). And, with specific reference to German noun inflection, Köpcke remarks that ``while gender does not predict the plural morpheme, it does limit the choice'' (1988, 307).
Case (ii) arises because polysyllabic roots may have nonfinal focal syllables. We have assumed that this is not systematic and must therefore be stipulated when it occurs. But, if it were the case, say, that disyllabic roots have the first syllable focussed by default then this could be captured in our analysis with some minor surgery to the way the <phn root focus> path is handled at the Word and Tri/Di/Syllable nodes.
Case (iii) just happens. Even in German. We consider some examples below:
Gott:
<> == Noun_C
<phn syl1 onset> == g
<phn syl1 peak-> == O
<phn syl1 tail-> == t
<mor suffix sing gen> == "Suffix_es:<phn root form sing gen>".
If /t/ was a sibilant then Gott would be a (sub)regular German noun in the Noun_C declension. But it isn't and Gott has a pattern of inflection that is unique in the German lexicon:
Gott:
<mor word sing nom> = g ^ O t
<mor word sing acc> = g ^ O t
<mor word sing gen> = g ^ O t @ s
<mor word sing dat> = g ^ O t
<mor word plur nom> = g ^ / t @ r
<mor word plur acc> = g ^ / t @ r
<mor word plur gen> = g ^ / t @ r
<mor word plur dat> = g ^ / t @ r n.
Although the /@s/ alternant is available as a stylistic variant for most nouns, Gott is alone in requiring it, and is singled out as unusual in German grammar books (e.g. Hammer, 1977, 20). Carstairs also notes that the suffix -e can appear on masculine and neuter singular dative nouns. This is again a stylistic variant. Hammer (1977, 20) says ``Except in certain fixed phrases, the -e is never essential and is in fact dying out''.
Similarly, Planet is almost a (sub)regular class Noun_J
lexeme. It would be completely (sub)regular if the nominative
singular form was /plan^e:t@/ not /plan^e:t/.
Planet:
<> == Noun_J
<phn root form> == Disyllable
<phn syl2 onset> == p l
<phn syl2 peak-> == a
<phn syl1 onset> == n
<phn syl1 peak-> == e:
<phn syl1 tail-> == t
<mor suffix sing nom> == Null.
Planet:
<mor word sing nom> = p l a n ^ e: t
<mor word sing acc> = p l a n ^ e: t @ n
<mor word sing gen> = p l a n ^ e: t @ n
<mor word sing dat> = p l a n ^ e: t @ n
<mor word plur nom> = p l a n ^ e: t @ n
<mor word plur acc> = p l a n ^ e: t @ n
<mor word plur gen> = p l a n ^ e: t @ n
<mor word plur dat> = p l a n ^ e: t @ n.
Phosphor is also unique in that it combines the singular/plural focal syllable difference and the vowel lengthening in plurals that are the defining characteristics of Noun_L class lexemes with the plural suffixes associated with class Noun_A:
Phosphor:
<> == Noun_L
<phn root form> == Disyllable
<phn syl2 onset> == f
<phn syl2 peak-> == O
<phn syl2 tail> == s
<phn syl1 onset> == f
<phn syl1 peak-> == O
<phn syl1 tail-> == r
<mor suffix> == Noun_A.
Phosphor:
<mor word sing nom> = f ^ O s f O r
<mor word sing acc> = f ^ O s f O r
<mor word sing gen> = f ^ O s f O r s
<mor word sing dat> = f ^ O s f O r
<mor word plur nom> = f O s f ^ o: r @
<mor word plur acc> = f O s f ^ o: r @
<mor word plur gen> = f O s f ^ o: r @
<mor word plur dat> = f O s f ^ o: r @ n.
Herz is a deviant noun, having the forms:
Herz:
<mor word sing nom> = h ^ E r t s
<mor word sing acc> = h ^ E r t s
<mor word sing gen> = h ^ E r t s @ n s
<mor word sing dat> = h ^ E r t s @ n
<mor word plur nom> = h ^ E r t s @ n
<mor word plur acc> = h ^ E r t s @ n
<mor word plur gen> = h ^ E r t s @ n
<mor word plur dat> = h ^ E r t s @ n.
This is unusual in having two forms which are degenerate - either the nominative and accusative singular, if we consider it a Noun_K, or the genitive and dative singular if we consider it a Noun_D as Wurzel (1990, 210) does. We choose the former approach for two reasons. First, it seems preferable to have the much more common uninflected form as the unexpected forms, particularly as these two forms correspond to the forms for the default noun class in our account. Secondly, although we need two statements to define this in the present analysis, our account of adjectives and determiners (Cahill and Gazdar, 1997a) captures the fact that nominative and accusative forms are commonly linked, and a combined account could make use of this connection. We therefore give the entry for Herz as follows:
Herz:
<> == Noun_K
<syn gender> == neut
<phn syl1 onset> == h
<phn syl1 peak-> == E
<phn syl1 body> == r
<phn syl1 tail-> == t s
<mor suffix sing nom> == Null
<mor suffix sing acc> == <mor suffix sing nom>.
