Hubert Devonish
Dept. of Language, Linguistics & Philosophy
University of the West Indies
Mona, Jamaica.
&
Enita Castillo
Gwen Lizarraga High School
Belize City, Belize.

4th August 2002.
[Paper presented to 14th Biennial Conference, Society for Caribbean Linguistics, St. Augustine, Trinidad & Tobago, August, 2002].

Background

There tends traditionally to be two kinds of approach to linguistic description. The first is that done by non-native speaker linguist who collects data from native speakers and produces linguistic descriptions. These descriptions are not usually accessible, because of their technical format, to native speakers many of whom may have acted as informants. The other approach is for descriptions to be done by native speaker linguists, using data collected from members of their own speech community, supplemented by their own native speaker intuitions.

Through accident rather than design, the current project to describe Garifuna manages to bridge the two approaches. One of the participants in this activity, Enita Castillo, is a native speaker who, as well, has a solid grasp of the basics of the discipline of linguistics. The other participant, Hubert Devonish, is an academic linguist who is not a native speaker of the language but who has made some very faltering efforts to learn it as a foreign language. The immediate objective is to produce a grammatical description of the language that is faithful to the functioning of the language from the perspective of the native speaker. However, the description must go beyond that to reveal features which are characteristic of human languages generally.

The overall study has two sets of goals. The first set of goals are historical in nature. A comprehensive linguistic description of the language would form the basis for studying the linguistic history of Garifuna and related varieties spoken in St Vincent, Dominica and elsewhere in the Lesser Antilles. This would include the influences which these varieties would have received from Karina (Carib) and imported European languages such as French, Spanish and English. There is also the need to identify the influence which African languages, Bantu languages in particular, may have had on the development of Garifuna and related varieties. In addition, there is the issue of the impact on them of Caribbean Creole languages, notably Lesser Antillean and Haitian varieties of French lexicon Creole as well as English lexicon Creole. Most interesting, however, from the perspective of Caribbean Creole linguistics, is the totally unresearched issue of the role which Garifuna and related varieties may have had in the formation of Caribbean Creole languages.

The other set of goals is practical in nature. These involve producing a non-specialists’ description of the language which could be used in the formal education system as part of efforts to preserve and expand the domains of the language. Such uses include Garifuna being taught to native speakers as a school subject and the language being used as a formal medium of instruction in the formal education system. Most immediately, however, we aim to produce a teaching grammar of the language aimed at non-native speakers at university level.

The works of Taylor, as summarised in Taylor (1977, pp. 44-71) represent the main previous work on the syntax of Garifuna within a modern linguistic framework. As far as we can ascertain, his description was based on the variety of Garifuna spoken in Hopkins, Belize, around the middle of the 20th century. Our focus, by contrast, is the variety spoken in Seine Bight, with specific reference to the current speech and native speaker intuitions of Enita Castillo, one of the collaborators in this project.

The variety that is the basis for the description in this paper is distinct, in several ways, from that described by Taylor (1977, pp. 29-71). In the area of phonology, it is innovative. It seems to have applied historical phonological rules to forms similar to those described by Taylor to produce forms which deviate from historically more conservative dialects. For many words recorded by Taylor involving intervocalic /r/ as in erenga ‘tell’, the equivalent in the variety we describe is a form without intervocalic /r/ as in eenga [e:?ga]. Along similar lines, the form ariha ‘see’ in the dialect that Taylor describes appears in the Seine Bight variety as eiha [eiha]. This involves a deletion of the intervocalic /r/ and the raising of the vowel /a/ to /e/ under the influence of the high front vowel /i/ which now immediately follows. There are, as well, some possible differences in the morpho-syntax of the two varieties which will have to be explored as part of the overall description of the language.

Scope

This work builds on the description of the structure of simple sentences produced by Devonish & Castillo (2001). The focus is on the interaction between inflectional morphology and syntax. It is well established that languages may have quite distinct systems operating in their morphology as opposed to their syntax. A classic example is that of a language like French in which word-internal morpheme order for predicate stems is S(ubject) O(bject) Verb whilst order across the syntactic constituents is SVO. Garifuna, an agglutinating language, does have different sequences of subject, predicator and object constituents at the morphological and syntactic levels. The structural functions to which these two levels are put overlap considerably and frequently coexist within the same sentence. This raises the question of exactly how these two systems relate to each other, and the role that their interaction plays in the overall morpho-syntactic system of the language.

Analysis
Introduction

In Garifuna, relations between the root morphemes which make up a sentence are signalled both by word internal morphology and as by syntactic relations across word boundary. Not surprisingly, there is a high degree of interaction between these two levels of the grammar of the language.

Roots belong to one of three major lexical classes. These are (i) predicators, inclusive of auxiliaries, (ii) nominals and (iii) adjectives. Roots belonging to the first two major word classes can and do have pronominal affixes assigned to them. These affixes signal grammatical categories such as Subject, Object, number, gender, definiteness and possession. Seemingly, from an English perspective, adjectival concepts such as ‘old’, ‘dirty’, ‘good’, ‘heavy’, ‘long’, etc. are expressed in Garifuna by predicators, complete with pronominal and tense/aspect affixes. They are used attributively in a process which is closely linked to that of the formation of relative clauses in the language. Adjectives in Garifuna, by comparison with predicators and nominals, constitute a small and seemingly closed group. These include numerals and a small number of other roots. Adjectives are distinguished from predicators and nominals by not taking any pronominal affixes.

Transitivity

Intransitive predicators which actually occur in a sentence may be either (i) lexically specified to be intransitive and bear a subject but no object, or (ii) lexically specified as transitive, i.e. able to bear both subject and object, but occurring without an expressed object in the particular utterance.

The link between morphology and syntax which we shall be exploring here involve exploring the idea that, in Garifuna, sentences can be
(a) morphologically and syntactically transitive or
(b) morphologically intransitive but syntactically transitive or
(c) morphologically and syntactically intransitive.

Morphological and Syntactic Transivity

Let us first examine cases of (a). We see, in the first example of each pair in 1) below, that transitive predicators can take both pronominal prefixes and suffixes. These are used to signal respectively subject and object. Thus, in each of the first examples, eiha ‘see’, a transitive predicator, takes the agent prefixes, l- ‘third person singular masculine’ and t- ‘third person singular masculine’. We conclude that in these constructions, the prefix position represents the morphological subject. As suffixes signalling the recipient of the action, we see respectively –u ‘third person singular feminine’ and –i ‘third person singular masculine’. We conclude that these suffixes represent the morphological object.

By way of comparison, the second example in each pair shows the same predicator with the same affixes as the first. However, there are now expressed subjects and objects in the form, respectively, of nominal phrases, mutu ligiya ‘the man’ and mutu tugiya ‘the woman’. These phrases follow the predicator, inclusive of its affixes. We conclude that these phrases express subject and object at the syntactic level. The syntactic order is Subject and then Object. Each of the subject and object phrases includes the independent pronominal forms ligiya ‘he’ and tugiya ‘she’, which are here employed to mark the nominal phrase, the NP, as definite.

1) Morphologically and Syntactically Transitive Predicators
i.
L-eiha-u
He see her
‘He sees her’
vs.
L-eiha-u mutu ligiya mutu tugiya
He see her, person he, person she
‘The man sees the woman’
ii.
T-eiha-i
She see him
‘She sees him’
vs.
T-eiha-i mutu tugiya mutu ligiya
She see him, person she, person he
‘The woman sees the man’

We have here, so far, a model of a neat match between the signalling of subject and object at the morphological level and at the syntactic level. We will see the extent to which this image is maintained after considering further kinds of data.

Morphological Intransitivity and Syntactic Transitivity

We next look at cases of (b), sentences in which the predicator is morphologically intransitive but syntactically transitive. In 2) below, the predicator root, ichiga ‘give’, can take both a subject prefix, e.g. t- ‘she’ and object suffix i- ‘it/masc.’, to give t-ichiga-i, literally ‘he-give-it’. It should be mentioned here that non-human nominals are assigned masculine or feminine gender lexically. The item búngidu ‘bucket’ is masculine and, therefore, if the speaker was intending to refer to ‘bucket’, the suffix form would be –i ‘it/masc.’. By contrast, if the intended referent were gárada ‘book’, which is lexically specified as feminine, the pronominal suffix representing it would be –u ‘it/fem.’ Suffix. However, in the particularly sentence presented in 2), the predicator takes a morphological subject, the prefix t-, and no morphological object. This is in spite of the fact that it has a syntactic object, aban búngidu ‘a bucket’.

2) Morphologically Intransitive, Syntactically Transitive (with Subject Prefix)

T-ichiga ti aban búngidu t-un t-iraü
She-give (particle) a bucket her-to her-child
‘She gave a bucket to her daughter’ (Taylor, 1977, p. 122)
[A somewhat contextual translation by Taylor]

To understand the significance of the absence of an expressed morphological object in the above example, we have to understand the nature of the marking of definiteness and indefiniteness in Garifuna. The form aban ‘one, a’ functions as the indefinite marker in the above sentence. The use of this marker signals that the item, búngidu, ‘bucket’, is being introduced for consideration by the interlocutors in the discourse for the first time. If, at the same time, the appropriate morphological object, -i ‘it’, were employed, we would have a conflict at the level of definiteness. The pronoun affix, -i ‘it/masc.’, by its very nature refers to an entity already introduced into the discourse. There would hence be a clash between the morphological object affix and the indefinite noun phrase which functions as the object at the syntactic level. Where there is such a lack of correspondence, the predicator stem is left morphologically without an object.

The use of the aorist tense/aspect as in the examples in 3) below require the use of the auxiliary, -umut-, if both subject and object affixes are to be employed, as in 3)ii and 3)iii. These must be used when the object NP is definite, whether marked by the object pronominal suffix at the morphological level alone as in 3)ii or by both the object suffix and the pronominal definite marker, ligiya in the case of 3)iii. Nouns without any marking for definiteness, either at the morphological or syntactic levels, may be used to refer to the class to which an entity belongs, signalling the generic. We see this in the example 3)i below in which uraga ‘story’ simply refers to members of the class ‘story’, appearing with neither the pronominal form, ligiya, nor the morphological object suffix, -i ‘it/masc.’ The pronominal suffix which does occur, -u ‘she’, is actually here a subject suffix representing the female teller of the stories.

3) Morphologically Intransitive, Syntactically Transitive (with Aorist Tense and Subject Suffix)
i)
Enga-t-u uraga bun ligira buga
Tell-TNS-she story you-to other day
‘She told you stories (i.e. She was story telling) the other day’
ii)
enga t-umut-i úraga b-un ligira buga
Tell she-TNS-it/masc. story you-to other day
‘She told you the story the other day.’
iii)
enga t-umut-i úraga ligia b-un ligira buga
Tell she-TNS-it/masc. story the you-to other day
‘She told you the story the other day’

If we look at the syntactic object of 3)ii and 3)iii, uraga ‘story’ and uraga ligia ‘story it’ respectively, they both get translated as ‘the story’. The reason for this, we suggest, is that the pronominal object antecedent, i.e. –i ‘it’ in both cases, is definite and forces a definite interpretation on its object nominal referent, with or without the definite marker, ligiya. However, in the first example, 3)i, there is no object pronominal affix. The pronominal suffix is the pronoun subject, -u ‘she’. There is no definite pronoun acting as an antecedent for the nominal object, úraga. Neither is there a definite marker, ligiya following. This frees the object from any definite meaning and opens it to the generic meaning of ‘stories’ as in ‘story telling’ which is assigned it in 3)i.

This pattern seems to hold up even when the object slot in a main sentence is filled, not by an NP but by a subordinate sentence. In the first of the three sentences below, the subordinate sentence l-abosiha D. l-uagu E. ‘D. is boasting about E.’ is functioning as the syntactic subject of the predicator root, -afara- ‘kill’, and agrees with the morphological subject, l-. In the second, the same subordinate sentence functions as syntactic object of the predicator, this time agreeing with the –i object suffix of the predicator auxiliary. In the third example, even though the same tense information is being conveyed and the subordinate sentence is also the syntactic object, the use of the auxiliary, -umut- ‘aorist tense/aspect’, is avoided. This is so since such use would require both subject and object affixes. However, with the object affix having a meaning which signals definiteness, and given the generic meaning of the syntactic object, l-abosiha D. l-uagu E., ‘D. boast(ing) about E. on occasion, from time to time’, there could be no agreement on definiteness between a morphological object and a syntactic one. Hence, the auxiliary requiring a morphological object is avoided and a predicate stem selected which has no object suffix. The agreement problem is thus solved.

4) Morphological Intransitivity, Syntactic Transitivity (Subordinate Sentences as Objects
i)
L-afara-di-na l-abosiha D. l-uagu E.
It-kill-0-me he-boast D. him-about E.
‘D’s boasting about E. is going to kill me.’
ii)
Eiha n-umut-i l-abosiha D. l-uagu E.
See I-Aux-it he-boast D. him-about E.
‘I see/saw D. boasting about E.’
iii)
Eiha-ti-na l-abosiha D. l-uagu E.
See-TNS-I he-boast D. him-about E.
‘There have been times when) I saw D. boasting about E.’

The feature identified at the beginning of this section as (b), i.e. morphological intransitivity coupled with syntactic transitivity, functions in Garifuna as a device for marking objects as indefinite. It is clear that the system begins at the level of word-internal morphology and extends itself to the syntactic level, not just to NPs in object position but to subordinate sentences that function as objects of the predicator of a main sentence.

Morphological and Syntactic Intransitivity and Subject Fronting

There are then the cases of (c), sentences that are intransitive at both the morphological and syntactic levels, in which there is neither a morphological nor a syntactic object. The most frequent occurrences of this involve predicators that are, in our terms, lexically specified to bear only one argument, a subject argument. This is expressed morphologically by a suffix. In the examples below, we treat as intransitive Predicators both (a)chülüra ‘to arrive’ and. würiba- ‘bad’. These are respectively analysed as a verb and an adjective by other treatments of Garifuna grammar, notably (Taylor, 1977) and Cayetano (1993). However, both types behave morphologically and syntactically in identical fashion when bearing tense and aspect markers. What differentiates them is that the former type can occur without a tense or aspect marker, and with an a- verbal prefix, as in n-a-chülüra ‘I arrive’. Items of the würiba type have no such option.

5) Intransitive Predicators (with aorist –t-) and Pronominal Suffixes
i)
Chülü-t-i
Arrive-TNS-he
‘He arrived’

Chülü-t-i mutu ligia
Arrive-TNS-he person he
‘The man arrived’
ii)
würiba-t-u
bad-TNS-she
‘She is bad’.

würiba-t-u mutu tugia
bad-TNS-she person she
‘The woman is bad’.

The second example in each pair above, involving nominal phrases such as mutu ligia ‘the man’ and mutu tugia ‘the woman’, shows the predicator with a pronominal suffix which agrees with the noun phrase in number and gender. The pronominal suffixes as well as the NPs function as subjects of the predicator at the morphological and syntactic levels respectively. Taylor (1977, p. 51) points out that adjectival predicators, in our terms intransitive predicators of the würiba- type, retain their verbal attributes, i.e. tense-aspect markers and pronominal affixes, even when they are used attributively. He uses the examples in 6) below to demonstrate how the attributive is formed. The subject NP appears before rather than after the predicator when the predicator is being used to signal an attribute of an NP. In the attributive, however, the form of the predicator remains unchanged, inclusive of its tense-aspect marker, here -t- and pronominal subject suffix, in this case –u ‘she’.

6) The Fronted Subject NP Construction with Intransitive Predicators

Würibatu aba aufuri
‘An aunt is bad’
vs.
aban aufuri würibatu
‘A bad aunt’

In 7)i below, we have an intransitive predicator root, weiri- ‘big’ along with its tense marker affix, -t- and pronominal subject suffix, -i ‘it’. It occurs with a subject consisting of entire sentence, complete with its own intransitive verb, -abosiha ‘boast’, pronominal subject prefix, l-, expressed NP subject, D, and prepositional phrase consisting of pronoun prefix, l-, prepositional root, -uagu, and the NP, E, to which the prefix pronoun on the prepositional root refers. In 7)ii, we see not just an NP but the whole subordinate sentence, the subject of the main sentence, moved to before the predicator of the main sentence.

7) The Fronted Subject Subordinate Sentence Construction with Intransitive Predicators
i)
Weiri-t-i l-abosiha D. yebe l-uagu E.
Big-TNS-it he-boast D. ‘to no purpose’ he-about E.
‘D. boasted a lot about E. to no purpose.’
ii)
L-abosiha D. yebe l-uagu E. weiri-t-i
He-boast D. ‘to no purpose’ he-about big-TNS-it
‘D. boasted A LOT about E. to no purpose.’

The question arises as to whether there is much justification for describing the process of fronting of subject NPs as one of attributive formation. This arises since precisely the same process can be applied to entire subordinate sentences. Given the normal Predicator-Subject-Object, we would argue that the fronting process simply changes the focus of the sentence, away from the predicator to the fronted syntactic unit, irrespective of the composition of that subject. It should be noted that although, for reasons of exploring the Taylor (1977, p. 51) proposal for the formation of the attributive, we have focussed on subject fronting, objects in transitive sentences can also be fronted.

Extending Fronting to Relative Clause Formation

Relativisation is the classic operation in which syntactic and morphological systems interlock very closely. In Garifuna, it is a mere extension of the process of sentential focus already discussed. The items le ‘this (male) one’, to ‘this (female) one’ and ha ‘these’ can appear as demonstrative pronouns, alone performing the function of NP in the main sentence or as demonstrative pronouns in apposition to an immediately preceding NP with which it agrees in gender and number. In 8) below, we see le functioning in this role, along with a preceding syntactic NP subject, mutu, which it is in apposition to, as well as the morphological subject pronominal suffix, –i. It is this basic function which is presented in 8) below.

8)

Weiri-a-i mutu le
Old-ASP-he person this (male)
‘This person is old.’

The items le/to/ha play an important role in sentence subordination where the subordinate sentence modifies an NP. Such a sentence cannot modify a bound subject or object affix. Neither can it directly modify a noun. The example below is typical of this type of sentence subordination. We see le functioning simultaneously as a subject NP in the main sentence and as a fronted object NP in the subordinate sentence. In this double role, however, the demonstrative le ‘this one, the one near to me in space’ is transformed from its specifically spatial meaning and is interpreted instead as anaphoric, with the meaning ‘this one, the one which is close by in the text’. Here, the referent would be mutu and the preceding bound morpheme -i in the main sentence and the suffix –i on the predicator in the subordinate sentence which follows. The effect is to modify the subject of the main sentence, morphologically –i and syntactically, mutu. These are being modified by way of elaboration coming from the subordinate sentence. This sentence is underlined.

9) Relative Subordinate Sentence within Main Sentence Subject

Weiri-a-i mutu le b-arihu-ba-i ñei
Old-ASP-he person this (masc.) you-see-MOD-him there
‘The man (whom) you see there is old.’

Occurring with an immediately preceding NP is not a requirement in the relativisation process in Garifuna. In 10)i below, the relativiser, le, has as its only referent in the main sentence the morphological pronominal –i suffix on the auxiliary b-umut-i. It also simultaneously functions as the fronted subject of the subordinate sentence, t-eingu-ba-i. The examples in 10)ii and 10)iii follow a similar pattern. Here, the relativiser, le and to respectively, have as their only referents the morphological object pronoun suffix of the main sentence and of the subordinate sentence. The relativisers also serve as the fronted subject of the subordinate sentence.

10)
i.)
M-agambu b-umut-i le t-eingu-ba-i
Neg-hear you-TNS-it (masc.) this (masc.) she-say-TNS-it (masc.)
‘You didn’t hear what she said.’
ii)
Adüga b-a-i me le l-ubalit-i b-un
Do you-INJ-it (masc.) will this (masc.) he-tell-it (masc.) you-to
‘Do whatever he tells you to.’ [Female speaking.]
iii)
Adüga b-a-u me to l-ubalit-u b-un
Do you-INJ-it (fem.) will this (fem.) he-tell-it (fem.) you-to
‘Do whatever he tells you to.’ [Male speaking.]

There is a significant difference between 10)ii and 10)iii. This involves the treatment of the abstract ‘it’, object of the main sentence as in ‘Do it’, and of the subordinate sentence, that which the listener is being ordered to do, along with the relativiser which links them. In 10)ii, when the speaker is female, these are realised as masculine. By contrast, in 10)iii, when the speaker is male, the same abstract object, ‘it’, and the relativiser which links them, are realised as feminine. This is a feature which operates simultaneously at the morphological and syntactic levels. This difference in the treatment of pronouns of impersonal reference, as between male and female speakers, has been noted by Taylor (1977, p. 60). He states that these pronouns are treated as masculine by women and as feminine by men.

Conclusion

This represents a report on work in progress. We shall, therefore, simply try to comment on the significance or potential significance of what we have observed so far.

The neat picture presented by our examination of morphological transitivity co-occurring with syntactic transitivity, is not maintained in the light of further analysis. Morphological intransitivity coupled with syntactic transitivity produce a special device to deal with the alignment of the two levels when the syntactic object, whether an NP or a subordinate sentence, is indefinite. The device employed, to have a construction with no object suffix, points to the tentative conclusion that where there is a clash between the morphological and syntactic levels, the syntactic system takes precedence. One issue that arises, however, is why no such need for alignment seems to exist in the case of indefinite syntactic subjects in relation to matching pronominal subject affixes. What exactly is so special about definiteness of the object suffix that blocks its use with an indefinite object, whilst no such restriction exists in relation to subject pronominal affixes and their syntactic equivalents?

Another issue that arises is that of the relationship between intransitive predicators when they occur with post-posed subjects as compared with when they occur with pre-posed subjects. Having examined evidence not just involving NP subjects but also subordinate sentence subjects, we rejecting the idea that the pre-posing of syntactic subjects was purely a device for using such predicators in an attributive manner. We rather opt for a more general explanation which points to the use of pre-posed subjects whenever there is a wish to emphasise the subject and de-emphasise the predicator. Pre-posing the subject in order to use an intransitive predicator in an adjective-like or attributive manner is only one of a set of related uses to which this device is put in the language.

One of these additional uses is in relative clause formation. Whereas the attributive use of fronting seems to be limited to sentences in which an intransitive predicator is being employed, relative clause formation has a much wider distribution. Even though, in relativisation, the pronominal referent in the subordinate sentence is always expressed at the morphological level, there is need for a syntactic demonstrative such as le, which is fronted from within the subordinate sentence, for the process to take place. Even though Garifuna is a highly agglutinating language, the process of relativisation is clearly a syntactic rather than morphological one.

Finally, one probably significant observation is that fully formed sentences may operate in the role of subject and of object. The most widely distributed sentence type in this role, however, appears to be those without a tense or aspect marker, of the type. l-abosiha D. yebe l-uagu E. ‘D boasts a lot about E to no purpose’ presented in 7)i and 7)ii. More detailed investigation is needed in this area.

We get the sense that a significant portion of our analysis of sentence structure in Garifuna will involve trying to separate out the effects of morphology from those of syntax and then developing an understanding of how they interact with each other. The preceding discussion represents the first few tentative steps in that direction.

REFERENCES

Cayetano, E. 1993. The People’s Garifuna Dictionary. Belize: National Garifuna Council.

Devonish, H. & E. Castillo. 2001. 'The syntax of simple sentences in Garifuna’.  Belize: School for Continuing Studies Country Conference, Nov. 2001.

Taylor, D. 1977.  Languages of the West Indies. Baltimore: Johns Hopkins University Press.