15th MFM, Manchester 24–26 May 2007

WHY CZECH CASE MARKERS SOMETIMES GET LOST Markéta Ziková & Tobias Scheer University of Brno & University of Nice, CNRS 6039 ([email protected]) ([email protected])

1. Goals •

provide an explanation of the misbehavior of a Czech declensional paradigm.

•

provide independent morphological evidence for empty Nuclei and floating vowels.

2. Czech nominal declension Czech nominal declension is traditionally organised into 14 paradigms, each of which is conceived of as a set of case markers which combine with gender-specified stems. The neuter paradigm dělání "making" is outstanding: it shows only consonant-initial markers: -m in [InsSg], [DatPl], -ch (= digraph for [x]) in [LocPl], -mi in [InsPl] (1) Paradigm dělání SG PL Nom dělání dělání Acc dělání dělání Gen dělání dělání Dat dělání dělání-m Loc dělání dělání-ch Ins dělání-m dělání-mi In all other paradigms, case markers which contain consonants are always compound: the consonant is preceded by a vowel, either short or long. Furthermore, each of all other paradigms shows no less than three different markers consisting of single vowels.

1

(2) Neuter paradigms

Nom/Acc Gen Dat Loc Ins

město "city" [Neu] SG PL měst-o měst-a měst-a měst-Ø měst-u měst-ům měst-u/ě měst-ech měst-em měst-y

moře "sea" [Neu] SG PL moř-e moř-e moř-e moř-í moř-i moř-ím moř-i moř-ích moř-em moř-i

Why are initial vowels absent from case markers in the dělání paradigm?

3. Why the suffix -í cannot be a case marker The -í which appears in the dělání paradigm is traditionally assumed to be a case marker: (3) Paradigm dělání: simple marker -í + compound markers SG Nom dělán-í Acc dělán-í Gen dělán-í Dat dělán-í Loc dělán-í Ins dělán-ím

PL dělán-í dělán-í dělán-í dělán-ím dělán-ích dělán-ími

Arguments against this analysis: Argument 1 Provided that the -í is a case marker, the paradigm shows massive syncretism which is unprecedented in other paradigms → 12 paradigm slots (6 cases x 2 numbers) receive only 3 different markers. Argument 2 Also, the syncretism predicted by (3) is highly suspicious. (4) Types of syncretism predicted by (3) LocSg = NomSg: not attested anywhere in the declension GenPl = NomPl: not attested anywhere in the declension

2

(5) LocSg ≠ NomSg, GenPl ≠ NomPl "castle" [Masc] NomSg hrad-Ø LocSg hrad-u/ě NomPl hrad-y GenPl hrad-ů

"machine" [Masc] stroj-Ø stroj-i stroj-e stroj-ů

"woman" [Fem] žen-a žen-ě žen-y žen-Ø

"rose" [Fem] růž-e růž-i růž-e růž-í

"city" [Neu] měst-o měst-u/ě měst-a měst-Ø

"sea" [Neu] moř-e moř-i moř-e moř-í

"young, adj." [Fem] mlad-á mlad-é mlad-é mlad-ých

4. What kind of animal is the suffix -í? In the dělání paradigm, 3 main classes of stems can be identified: (6)

PP-nominalizations pří-moř-í "seaside" by-√sea-noun

V-nominalizations děl-á-n-í "making" √make-th-part-noun

collectives lan-ov-í "ropes" √rope-adj.-noun

What is the contribution of the sufix -í to the morphosyntactic structure? The -í has nothing to do with Case features. It has to be somehow linked to Genderand mass/count features.

5. Zero morphemes? Does the fact that -í is part of the stem mean that the dělání paradigm has the structure in (7)? (7) Paradigm dělání: zero markers + consonant-initial markers SG PL Nom dělání-Ø dělání-Ø Acc dělání-Ø dělání-Ø Gen dělání-Ø dělání-Ø Dat dělání-Ø dělání-m Loc dělání-Ø dělání-ch Ins dělání-m dělání-mi If 8 out of 12 paradigm cells are occupied by zero morphemes, the strange types of the syncretism mentioned under (4) are produced.

3

6. The dělání paradigm: morphological and phonological structure Proposal how to solve the syncretism problem: • • • •

the dĕlání paradigm has the same morphology as all other paradigms. it is derived by the same phonology as all other paradigms. what makes it special is the fact that its stems are vowel-final. All other paradigms are built on consonant-final stems. regular phonology prevents the vowels of vowel-initial case markers from surfacing.

(8) Paradigm dělání: vowel-initial case markers are underlyingly present SG PL Nom dělání-V dělání-V Acc dělání-V dělání-V Gen dělání-V dělání-V Dat dělání-V dělání-Vm Loc dělání-V dělání-Vch Ins dělání-Vm dělání-Vmi

Having no positive (phonetic) evidence how vowels in (8) look like, we use the symbol V for marker-initial vowels. (The identity of these vowels can be probably identified on morphological grounds. Also, in comparison with the structure of other paradigms, zero markers may be assumed for Nom/AccSg or GenPl.)

7. The best way from dělání-Vm to [
laim ] runs through CVCV Mapping between the syllable structure (CVCV) and the segmental level of representation need not be one to one: empty Nuclei, floating melody, pieces of melody linked to multiple constituents. (9) Representation of klání "tournament" C V C V C V C V C V | | | k l a n i

4

1. CV sequences are the minimal syllabic units (the existence of C implies the existence of V, and vice versa). 2. hence the syllable structure of morphemes starts with an Onset (empty or full) and ends in a Nucleus (empty or full). 3. syllable structure is recorded in the lexicon, then projected into the derivation. (10) Lexical representation of √OBIL "corn" C V C V C V | | | | o b i l Only segments linked to syllabic constituents are phonetically realized. Czech (as other Slavic languages) features vowel-zero alternations. Analysis in CVCV: alternating vowels are lexically floating melodies; the alternation is controlled by Government: → if the following Nucleus is empty, the floating melody attaches to its Nucleus: empty Nuclei cannot govern: (11)c → if the following Nucleus is contentful, the floating segment remains floating: full Nuclei govern: (11)a,b (11) Derivation of e ~ ø alternants: √KOTøL "boiler" a. kotøl-e GenSg

b. kotøl-ů GenPl

C V C V2 C V1 | | | | |

C V C V3 C V2 C V1 | | | |

k o

k o

t

e

l

e

t

e

l

u

c. kotel-Ø NomSg C V C V2 C V1 | | | | k o

t

e

l

In (11)a,b, the Nucleus following the alternation site is occupied by the vowel of the case marker → the initial vowel of the case marker associates to the final empty Nucleus of the stem

5

Analysis enforced by the properties of CVCV and the fact that any vowel-initial case marker, short and long alike, triggers zero alternants 1. short marker-initial vowels in the lexicon, marker-initial short vowels are floating segments that lack any syllabic support. In order to be pronounced, they need an empty Nucleus onto which they can link. 2. long vowels marker-initial long vowels are lexically associated to a Nucleus and specified to spread to their left. In order to do so, they need an empty Nucleus to their left. 3. consequence the vowel of vowel-initial case markers can only be realized if it is concatenated to a stem that ends in an empty Nucleus. (12) Lexical representation of marker-initial short and marker-initial long vowels a. -em [InsSg]

b. -ím [DatPl] C V C V | | i m

C V | e m (13) Derivation of case forms: vowels are audible a. moř-em √SEA-[InsSg] C V C V | | | m o ř root

b. moř-ím √SEA-[DatPl] -

C V C V | | | m o ř root

C V | e m InsSg

-

C V C V | i m DatPl

In the dělání paradigm, affix-initial vowels cannot be pronounced because the stem is vowel-final: it ends in the suffix -í. No empty Nucleus it is available that could receive case-marking melody. Hence affixes containing only vowels are not pronounced at all. Compound affixes with both lexically floating and associated melody can realize only the latter.

6

(14) Floating vowels fail to be pronounced: dělání-Vm [InsSg] C V C V C V C V C V | | | | d ě l a n i

-

C V | V m

Prediction: there is no empty Nucleus after the -í, therefore long marker-initial vowels cannot branch and should realize as short. However, the dělání paradigm shows no short vowels. All markers in the dělání paradigm begin with floating vowels.

7.1 Allomorphy Is the proposal that all markers in the dělání paradigm begin with floating vowels plausible if in the DatPl, other neuter paradigms show long vowels? měst-ům, moř-ím Yes, because: • •

there exists a DatPl allomorph which begins with the short vowel, -em, and which appears in the feminine paradigm kost "bone" moreover, the fact that the InsPl marker is -mi indicates that the dělání paradigm is a mixture of neuter and feminine markers: the -mi marks InsPl in all feminine paradigms as well

Furthermore, in Caha & Ziková (2005), we show that in fact only a minority of case markers is gender-bound.

8. The rodeo story Prediction: vowel-final stems should be unable to realize these case markers which begin with short vowels. (15) Rodeo and město: two neuters with the same markers SG

PL

Nom/Acc rode-o měst-o rode-a měst-a Gen rode-a měst-a rode-í měst-Ø Dat rode-u měst-u rode-ům měst-ům Loc rode-u měst-u rode-ech měst-ech Ins rode-em měst-em rode-y měst-y

7

1. Problem: neuters like duo, rodeo, video, or stereo do display the vowel of vowel-initial markers even though their stem ends in the vowel o. 2. They take on the same markers which appear in the main neuter paradigm město, with one exception in the GenPl: město takes Ø, while rodeo realizes -í. We submit the following explanation: The stem-final o is interpreted as the NomSg marker -o. As a consequence, a final empty Nucleus is "created". Therefore all floating vowels of the markers can be normally realized. (16) √RODEO: the root-final o is reinterpreted C V C V C V | | | | | r o d e o This analysis also explains the GenPl allomorphy: mĕsto = zero, rodeo = -í recall that both paradigms are otherwise identical. Hence the question: why do we get měst-Ø, but not *rode-Ø ? Answer: because unlike the former, the latter bears a final empty CV unit. However, a word may not end in an empty CV. (17) GenPl *rode C V C V C V | | | | r o d e Critical difference between dĕlání and rodeo: the stem-final vowel is (rodeo) or is not (dĕlání) interpreted as a case marker. Hence the question: what decides whether root vowels are interpreted as case markers? Only those stem-final vowels are interpreted as case markers which serve as the NomSg markers in the default paradigms for all genders:

8

(18) Reinterpretation of root vowels a, o

other vowels

impresário → impresári-o duo → du-o idea → ide-a

emu → *em-u šodó → *šod-ó brandy → *brand-y

To avoid deriving case forms like [DatPl] *emu-m, or [LocPl] *šodó-ch, those stems whose final vowels cannot be interpreted as case markers and which don’t belong to the dělání paradigm must be lexically specified as indeclinable, i.e. unable to combine with any case markers. To be indeclinable is a lexical property.

9. Conclusion Three independent pieces of evidence for the lexically floating vowels • e ~ ø alternation • the dělání paradigm • the rodeo-type nouns

References Caha, Pavel & Markéta Ziková. The Czech Declension: Syncretism Principle and Morphological Noise. To appear in the proceedings of the FDSL-6, Potsdam, 1–2 December, 2005. Lowenstamm, Jean. 1996. CV as the only syllable type. In Current trends in Phonology. Models and Methods. Durand, J. & B. Laks (ed.), 419–441. Salford, Manchester: ESRI. Scheer, Tobias. 2004. A Lateral Theory of Phonology. Vol.1: What is CVCV, and why should it be? Berlin: Mouton de Gruyter. Scheer, Tobias & Markéta Ziková. (Slavic) vowel-zero alternations and phase theory. Paper presented at OCP 4, Rhodes, 19-21 January 2007. (hdt: www.unice.fr/dsl/tobias.htm) Ziková, Markéta. Cyclicity in phonology: Case markers in the Czech nominal declension. To appear in the proceedings of the FDSL-6.5, Nova Gorica, 1–3 December, 2006.

9