Grebo Verbal Morphology PDF

Preview

Summary

t Grebo region Grebo Verbal MorpholoGy Hanno T. Beck May 1983 0. Introduction Grebo is a West African language spoken in southeastern Liberia by an unknown number of speakers. Grebo is a tonal language, and also regularly exhibits vowel harmony and nasality harmony. These three effects combine to yi...

Description

t

Grebo region

Grebo Verbal MorpholoGy Hanno T. Beck May 1983

0.

Introduction

Grebo is a West African language spoken in southeastern Liberia by an unknown number of speakers. Grebo is a tonal language, and also regularly exhibits vowel harmony and nasality harmony. These three effects combine to yield a rich set of allomorphs in the language’s morphology. This paper will explore and describe the behavior of all 17 processes connected with Grebo verbal morphology. Section One introduces the language’s phonology and typical behavior, and demonstrates how some of the simpler suffixes may be analyzed. Section Two will address problems in the description of complicated affixes and reduplicative forms. Section Three summarizes the findings. An appendix offers some speculations concerning the diachrony of Grebo verbal roots and its relationship to synchronic morphological processes. 1.

The Grebo Sound System

In this examination of the phonology/morphology of Grebo, a general framework of autosegmental phonology will be assumed. Specifically, segments will be treated as sets of distinctive feature specifications; each feature defines a tier, and each element on a tier, in order to receive phonetic realization, must become associated with an element of the [syllabic] tier (a distinguished tier that will also be referred to as the C-V skeleton). Association between specifications on two tiers in an expression takes place, normally from left to right, after all morphological processes have applied. The well-formedness condition on autosegmental representations will be assumed here roughly in the form adopted by Goldsmith (1976), insofar as lines of association between elements on different tiers may not cross. Grebo has the following consonant system (when pairs of consonants are given together, the leftmost is [–voice]):

–cont: +ant –cor –back

+ant –cor +back

+ant +cor –back

–ant –cor –back

–ant –cor +back

–nasal

p,

kp, gb

t,

d

ky, gy

k,

+nasal

hm, m

hn, n

ny

b

ŋm

g ŋ

+cont: –nasal

hl,

+nasal

l̃

+stri

s

–stri

l

h

hw, w

h̃

h̃w̃ , w̃

f

and /y/ is [+cont, –nasal, –ant, –cor, –back, +high] In his description of Grebo consonants, Innes (1966) does not specifically discuss the sound he transcribes as /f/, but he does say that sounds not mentioned “have the values which they commonly have in West African languages.” This leaves no clue to whether /f/ transcribes a voiceless labiodental fricative or a voiceless bilabial fricative. Neither Ladefoged (1968) nor Westermann and Ward (1933) provide any descriptions of the fricatives of Grebo or its neighbors Gweabo and Kru, and since those authors’ studies show that both the bilabial and labiodental fricatives are fairly common in other West African languages, there is no clear answer. Because of its tendency to pattern with the bilabials in phonological processes, as we shall see below, /f/ is here assumed to describe the voiceless bilabial fricative [ɸ], and thus to be [+round].1 Two other things worth noting about the consonant system of Grebo are the large number of nasals, both voiced and voiceless, and that some consonants have a rather limited distribution, presumably for phonotactic reasons; /hl/, /hn/, /h̃w̃ /, and /n/ never appear word-internally, and /hm/, /ny/, and /ŋw/ do so only exceptionally. Grebo has the following vowel system: –round –back

+round +back

+high

i, ĩ

u, ũ

–high, –low

e, ẽ, ė

o, õ, ȯ

+low

ɛ, ɛ̃

a, ã

ɔ, ɔ̃

Hayes (2009:86, footnote 5) adopts a [labiodental] feature, but notes that [strident] and [distributed] have also been used. For the purposes of this paper, the precise characterization is not crucial and we will go with [–stri, +round]. 1

2

The system of vowel harmony, by which vowels /e/ and /ė/, and /o/ and /ȯ/ come to be distinguished, will be explained below. Note that all vowels except /ė/ and /ȯ/ have nasal correlates. The syllabic structure of Grebo is extremely simple—all syllables are of form (C)V. Thus there are no consonant clusters whatever.2 Verb roots in Grebo conform to the template given in (1): (1)

(CV)1(V)

Root morphemes of types CV, CVV, CVCV, CVCVV and CVCVCV have been observed. Grebo has four levels of tone. Innes (1960) assigns numbers to the tones as follows: high = 1; high-mid = 2; low-mid = 3; low = 4. More than one tone may appear on a single [+syallbic] segment,3 subject to certain constraints. While Innes says that three tones rarely associate with a single vowel, his grammar (1966) and dictionary (1967) give no examples of such triply-toned segments whatever. We may assume, then, that triple tones, to the extent that they appear at all, are highly exceptional and peripheral. Therefore we shall adopt the constraint represented in (2): (2) * T T T V That is, no single [+syllabic] segment may be associated with three tones. Verbs in Grebo may undergo any of a large set of morphological processes. Moreover, no phono/morphological processes that occur in other parts of speech in Grebo fail to occur in the verbal system. Therefore verbal morphology provides an ideal domain in which to investigate the Grebo interaction of phonology and morphology. Let us begin this investigation by looking at the Grebo system of vowel harmony. Innes (1962) divides the vowels into three groups: A. i, ĩ, u, ũ, ė, ȯ. B. e, ẽ, o, õ. C. ɛ, ɛ,̃ a, ã, ɔ, ɔ̃. Innes explains this grouping by noting that … it may be said that within stretches in which vowel harmony operates, vowels in any one set can occur together, those in A and C, B and C also occur together, but those in A and B are mutually exclusive. This harmony operates within all root morphemes, and extends to some affixes as well.

2

In fast speech, Grebo does exhibit some syncopated forms that have biconsonantal clusters ending in /w/ or /l/. No underlying forms have this cluster, however, just as in English there are no underlying word-initial /ft/ clusters yet some speakers pronounce the word “photography” as [ftagrəfi]. 3 Clements and Ford (1979) take “tone-bearing units” to be syllables or rhymes. For the purpose of simplicity, in this paper segments that are [+syllabic] will be considered tone-bearing elements.

3

Exactly what feature distinguishes the incompatible groups A and B is not entirely clear. Sapir (1931), in discussing the vowel system of Gweabo, a close neighbor of Grebo, makes a distinction between the vowels in groups B and C on one hand, and A on the other, calling the A vowels “muffled” and the B/C vowels “bright.” Sapir is uncertain exactly how to characterize this difference in phonetic terms, though he suggests that … ‘muffled’ vowels have a peculiar muffled or cavernous quality which is perhaps due to a resonance produced by the sagging of the cheek and lip muscles. For our purposes, it seems reasonable to take it that the bright/muffled distinction in question is the feature [Advanced Tongue Root], or ATR for short. The core of Grebo vowel harmony may basically be expressed by a constraint that at most one specification per morpheme is permitted on the [ATR] tier. Neutral vowels will be treated as being pre-specified [–ATR], and thus not subject to harmony. Only vowels of groups A and B will be “P-bearing elements”; that is, only members of groups A and B will have existence on the [ATR] tier, and even then only in root morphemes and those suffixes that are subject to vowel harmony. As examples, we may consider the two verbal suffixes +O1 “there, unspecific” and +dE1 “there exactly.”4 /O/ is an archiphoneme that is to be [+syllabic, –nasal, –low, –high, +back, +round, 1tone] but unspecified for [ATR], and the /E/ in +dE is [+syllabic, –nasal, –low, –high, –back, (redundantly)–round, 1tone] and similarly unspecified for [ATR]. By analyzing these suffixes like this, general principles of autosegmental phonology will themselves produce the correct surface allomorphs when these suffixes attach to any verb roots. For example, here these suffixes interact with du2 “pound” and te4 “descend”: duȯ21

“pound around there”

dudė21

“pound exactly there”

teo41

“descend around there”

tede41

“descend exactly there”

Let us now move on to consider nasality harmony. Innes proposes three generalizations about the surface distribution of nasality.5 First, vowels (by which is meant potentially nasal vowels, not /ė/ or /ȯ/) are always nasal following a nasal consonant. Second, in CVCV sequences where the second C is nasal, both the vowels are nasal. Third, in VV sequences, either both vowels are nasal or both are oral. An integer immediately following a Grebo expression denotes its tone pattern. For example, dudɔ21 “pound yesterday” has a high-mid tone on its first syllable and a high tone on its second syllable. Gliding tones are hyphenated, so gade3-23 “deny” has a rising mid tone on its first syllable and a low mid tone on its second syllable. Tonal information will sometimes be omitted when it is irrelevant to the discussion. 5 Innes uses these observations to simplify his orthography, omitting the tilde over a nasal segment when its nasality could be inferred by these generalizations. I will follow Innes’ orthographic convention in this respect, except in cases where I wish to call particular attention to a segment’s nasality. 4

4

In an autosegmental framework, these facts suggest the following treatment. Vowels are underlyingly unspecified for nasality, except for /ė/ and /ȯ/, which are always associated with [–nasal] on the nasality tier. Also, there are some exceptional root forms that contain vowels that are underlyingly associated with [+nasal]; for example, the verb pɔ̃4 “hunt.” Unlike vowels, all consonants are associated with features on the nasality tier. This effectively restricts the spreading of nasality harmony to a domain of a single syllable, except where a syllable has no consonant, or where that consonant is itself unspecified for nasality, as is the case with most suffix consonants. Autosegmental spreading occurs to associate those segments that are unspecified, by spreading first all [+nasal] specifications to whatever degree possible, and only then spreading the [–nasal] specifications. By this process all potential nasality-bearing elements come to be associated. While tone-bearing elements may receive more than one specification for tone, in Grebo no nasality-bearing elements may receive more than one specification for nasality. That is, Grebo subscribes to the following constraint: (3) * [α nasal]

[β nasal]

X Though intuitive, this constraint is not universal and must be stated as part of the grammar of Grebo. Some languages have prenasalized stops that may be analyzed as having “contour nasality.” Hart (1981) gives treatments of nasality in Guarani and Parintintin that make extensive use of this possibility. Most suffixes in Grebo are unspecified for nasality. A few are [–nasal], but suffixes are never associated underlyingly with [+nasal]; this results in affix nasality never determining the root nasality. Taken together, these principles will suffice to capture the same facts as Innes’ three generalizations, and also begin to give an explanation for them that goes beyond a mere recording of the observed data. With this analysis of nasality, the allomorphic variations of the following six verbal suffixes fall out straightforwardly: ɛ2

“today”

a2

“tomorrow”

ɔ3

agentive noun (analogous to the +er agent-forming suffix in English)

Da2

“before yesterday”

Dɔ1

“yesterday”

Dɔ2

“after tomorrow”

These suffixes have tones already attached to them and contain only neutral vowels, so their suffixation adds no additional P-bearing matter to either the tonal or the ATR tiers. All of these suffixes are unspecified for nasality, however, and so they will acquire their specification via 5

spreading from the root. Here are sample derivations for fininɔ “tie up after tomorrow” and bisėda “thanked before yesterday”: underlyingly: [–nasal] [+nasal] f i n i + D ɔ after [+nasal] spreading: [–nasal] [+nasal] f i n i + D ɔ after [–nasal] spreading: [–nasal] [+nasal] f i n i + D ɔ

[–nasal] [–nasal] [–nasal] b i s ė + D a [–nasal] [–nasal] [–nasal] b i s ė + D a [–nasal] [–nasal] [–nasal] b i s ė + D a

In each case, [+nasal] spreading occurs first, as much as can be done in accordance with the Well-Formedness Condition, and then any permissible [–nasal] spreading is done. No segment unspecified for nasality at the underlying level will fail to become specified through this process. Although this treatment of nasality suffices for describing the behavior of the suffixes given above, there is some indirect evidence that suggests a refinement. It is inelegant as well as redundant to expect every normal root consonant to be associated with its own nasality specification. Moreover, this system technically predicts that there is no general principle governing the nasality of sequences of consonants in Grebo words, an incorrect prediction. It is much more common in Grebo for words to have strings of consonants corresponding to the schema [–syll, –nasal]n[–syll, +nasal]m than the opposite. Put in simpler terms, there is a trend in Grebo words that once a nasal consonant appears, the remaining consonants will also be nasal. To accommodate this fact, at least partway, one potential approach would be to say that the first consonant in a word is associated with a specification on the nasality tier and the (possibly same) first nasal consonant is also so associated. Now the spreading conventions will give us words whose consonants conform to the “once nasal, always nasal” schema. This enables us to specify less structure directly in the lexicon, while at the same time predicting the correct generalization about the consonantal structure of Grebo words. Words that have other patterns than the basic schematic one outlined here should be more costly to record in the lexicon, and this is correctly captured in this system, for such words will require additional underlying nasality specifications. Now let us consider the interaction of tonality with Grebo verbs. Much inflectional information, especially distinctions among several moods, as well as many person and number distinctions, is conveyed by tone. For example:

6

bɔ2 du2 nɛ4

“let him pound it”

bɔ2 du1 nɛ4

“if he pounds it”

This role of tone, though quite important, is rigid and not predictable from more basic considerations; the tone patterns of each mood must be learned by rote and do not vary according to tonal context in any way. Since they pose no problems relevant here, these inflectional effects of tone will not be considered further. Tone does, on the other hand, interact quite fully with verbal affixes. Grebo monosyllabic verb roots, if classified according to tone, comprise four classes:6 Class 2-1.

CV

Class 2. CV

2 1 Class 3-2.

2

CV

Class 4. CV

3 2

4

There are a few verbs that do not fall into any of these classes, but according to Innes (1960) they are extremely rare, and even most of those are exceptional only in their citation forms, falling into one of the canonical tonal classes in all other respects. Disyllabic verb roots of the form CVCV fall into five classes, but only one of these is novel in any nontrivial sense. The first four are obvious extensions of the C-V skeleta of their monosyllabic counterparts, without any added tonal content:7 Class 21.

CVCV 2

Class 3-23.

Class 22. CVCV

1

CVCV

2 Class 44. CVCV

3 2 3

4

The remaining disyllabic verb class, 23, is genuinely primitively disyllabic even in its tonal form, unrelated to any of the monosyllabic verb classes. However, the 23 class has no special phonological or morphological status in the grammar, behaving in completely regular ways. To provide an example of tone interacting with Grebo verbal morphology, let us examine the process of converting verb roots into their “secondary stem” forms. Several Grebo suffixes occur Following Innes (1960, 1966), citation forms and their tones used throughout this paper are those of the simple imperative. 7 Actually, class 3-23, while clearly related to verb class 3-2, does not show as obvious a relationship to it as the other CVCV verb roots due to their corresponding CV roots. This peculiar status of the 3-2 glide exists throughout the grammar of Grebo, and we will merely note each apparent quirk in its behavior as we come upon it, returning later to give an overall explanation. 6

7

with either verb roots or their secondary stems, and some types of meaning (imperfective aspect, dative, causative, passive) may be carried only with the help of a verb’s secondary stem. While it contains a fair amount of irregularity, the secondary stem-forming process mainly involves vowel harmony, nasality harmony, and tonal spreading. Ignoring the irregular cases, the core phenomenon of the secondary stem is merely the addition of a front vowel counterpart to the root-final vowel, or adding nothing if the root-final vowel is already front. All other features of the root-final vowel remain the same, so we might well treat the secondary stem formation process as a reduplicative one, a possibility to which we will return later. For the present, we may analyze the secondary stem as adding a single vowel suffix that is specified [+syll, –back, –round] and unspecified for [nasal], [ATR], [high], [low] and tone. This predicts correctly that the secondary stem suffix will have the sort of distribution as below: to2

“string”

toe22

du2

“pound”

dui22

gȯ3-2

“moisten”

gȯė32

po4

“explode”

poe44

We assume that tone is assigned from left to right, as a one-to-one mapping from tonemes to syllabic segments, until (a) we run out of tonemes, in which case the final tone spreads to as many syllabic segments as remain in the expression, or (b) we run out of syllabic segments, in which case the remaining tones associate to the final syllabic segment, subject to the triple tone association constraint put forward above. This single characterization of the secondary stem-forming process accounts for all its regular surface manifestations, except that it does not dictate when the suffix is not to be added. Here we shall just stipulate that the suffixation will apply only when it adds a vowel other than the root-final one. Later we shall see two alternative ways of stating this restriction. Verbs of tonal class 3-2 behave as we would expect with regard to secondary stem-formation tone; this is the only morphological process with which the 3-2 tone interacts in an unexceptional way. This section has given a brief description of the sound system of Grebo, presented its systems of vowel harmony, nasality, and tonality and shown how these operate to determine the surface forms of many verbal affixes, according to standard autosegmental principles. 2.

Thornier Affixes and Reduplicative Processes

The verbal affixes analyzed in the prev...