SIL Electronic Working Papers 1998-005, October 1998
Copyright © 1998 Howard Shelden and Summer Institute of Linguistics, Inc.
All rights reserved.

First presented at the Seminar for Research of Eastern Indonesia, Manado, Indonesia, July 23-29, 1985.

Transitivity and Galela pronominal reference[1]

Howard Shelden


1. Introduction
2. Decreasing transitivity
3. Increasing transitivity
4. Conclusion


This paper explores the relationship between the system of pronominal reference and the transitivity of clauses in discourse in the Galela language of North Halmahera. I show that the deletion of a pronominal verb prefix correlates with decreased clause transitivity, and thus with greater backgroundedness. Furthermore, I suggest that at very low levels of transitivity, possessive pronouns partially merge with pronominal verb prefixes. Finally, I show that the addition of a pronominal verb prefix is the best indicator of increased clause transitivity, and thus of greater foregroundedness.

1. Introduction

Hopper and Thompson (1980) have shown that transitivity is a crucial notion for understanding the interrelationships between several properties of clauses. Furthermore, they have demonstrated that the discourse functions of grounding correlate with the transitivity of clauses. That is, clause features which correlate with high transitivity encode foregrounded discourse information. Conversely, clause features which correlate with low transitivity encode backgrounded discourse information.

Any given speaker can word a clause in a number of ways which reflect varying degrees of transitivity. Depending on the degree of foregrounding (or backgrounding) he wishes to communicate in relation to the rest of his message, he will choose an appropriately higher (or lower) degree of transitivity for that clause. The speaker's choice often involves selection of a particular word from verb classes such as transitive, intransitive, and stative, as well as such factors as word order and the presence or absence of nominals. It is also common for lexical items to be flexible with regard to their category of transitivity. For example, the verb 'move' is intransitive in 'John moved' but transitive in 'John moved the car'. In addition to this kind of lexical flexibility with regard to transitivity, most languages have certain morphosyntactic devices which modify the transitivity of verbs.

The Galela language[2] has a complex verb morphology and thus several morphosyntactic devices which affect transitivity. Most of these have been but briefly described in earlier works (van Baarda 1895, 1908; van der Veen 1915; H. Shelden 1991), and that which interests us presently is the system of pronominal reference, specifically the pronominal verb prefixes[3] and the possessive pronouns.

In this paper we will explore the relationship between this system of pronominal reference on the one hand and the transitivity of clauses on the other. We will attempt to show that pronominal verb prefixes (and, in a similar but limited way, possessive pronouns) are often the only indicator on the verb itself of the degree of transitivity in the clause. Further, we will attempt to show that where other verb affixes indicate the degree of transitivity in the clause, there is a correlation with the pronominal verb prefixes.

The manner of presentation is as follows. First, we assume a typical (highly) transitive verb and discuss various degrees of decreasing transitivity as indicated in the pronominal reference system. Secondly, we assume a verb of inherent low transitivity and discuss various degrees of increasing transitivity as indicated in the pronominal reference system.

2. Decreasing transitivity

A Galela speaker who wishes to decrease the transitivity of a particular clause, and thus reduce its foregrounding within the discourse, has several alternatives. He may create a reflexive or reciprocal construction (pending semantic constraints). He may delete either the object (§2.2) or the subject (§2.3). Or he may recode the clause as a possessive noun phrase (§2.4).

2.1. Reflexives and reciprocals

For a typical transitive verb which is marked for both the subject and the object as in 1a we will use the term doubly-marked transitive. Reflexives and reciprocals are special cases in which subject and object are co-referential. These require subject markers only (no object markers) and are thus lower in transitivity than doubly-marked transitive verbs.

(1) a. to-ni-sano
      'I ask you.'

    b. to-ma-sano
      'I ask myself.'    

In the first example, to- 'I' marks the subject of both 1a and 1b. In 1a the object is marked by ni- 'you OBJECT'. In 1b the reflexive marker ma- signals that the subject and the object are co-referential and thus the object need not (indeed may not) be marked. Thus reflexives such as 1b carry only a single (subject) pronominal verb prefix and are less transitive than doubly-marked transitives such as 1a.

Reciprocals are a special case of reflexive construction in which the subject is plural and the action is done reciprocally by each member of the subject to the other member. As a special case of reflexive thus, it uses the reflexive marker ma-, which is then followed by the reciprocal marker ri-[4].

(2) a. mi-ma-sano
      'We ask ourselves (collectively).'

    b. mi-ma-ri-sano
      'We ask each other.'        

In both 2a and 2b, mi- 'we exclusive' marks the subject of the clause. In the reciprocal example, 2b, the members of the subject act individually to ask each other, whereas in the reflexive example, 2a, they act collectively to ask themselves as a group. Thus the reflexive marker ma- plus the reciprocal marker ri- creates a degree of transitivity somewhere between doubly-marked transitives and reflexives.

2.2. Object marker deletion

While reflexives and reciprocals can be said to delete the object marker from the verb, they nonetheless have other markers which indicate the object by signalling the co-referentiality of the subject and the object. There are other cases in which the object marker is simply deleted, with no concomitant change in derivational morphosyntax. This reduces the topicality or importance of that participant (D. Shelden 1986). It also reduces the degree of transitivity of that clause and thus the degree of foregrounding within the discourse. Two of D. Shelden's examples serve us well here.

(3) a. Aweng wo-mi-sasano 
       Aweng he-her-ask.questions
      'Aweng asked her questions.'

    b. de   mo-Ø-sano
       then she-Ø-ask
      'Then she asked (him).'

Both 3a and 3b were taken from a narrative in which a woman, mi- 'her' in 3a and mo- 'she' in 3b, visited Aweng's store. Clause 3a is climactic, thus highly foregrounded by marking both the subject and the object on the verb. No other verb in the short narrative is a doubly-marked transitive verb. Clause 3b for example, which occurs earlier in the narrative, is lower in transitivity and thus backgrounded by deleting the object marker which would otherwise have been wi- 'him'. The reduplication of the verb root sano 'ask' to sasano in 3a merely indicates that many questions were asked, and has no effect on the object marker.

2.3. Subject marker deletion

As the non-topicality of a participant can cause an object marker to be deleted, so can it cause a subject marker to be deleted. The backgrounding effect is similar, although the resulting stative clause is less transitive than the (active) intransitive clause which results from object marker deletion. D. Shelden 1986 provides 4 as well.

(4) ka-dahu          de  o nyawa  Ø-ni-sano 
    at-down.location and a person Ø-you.OBJ-ask
   'At a northern (i.e. 'down') location somebody asked you.'      

Clause 4 is backgrounded by deleting the subject marker so that the otherwise transitive verb is reduced to a stative verb form. In this case the subject referent, o nyawa 'a person', is non-topical and thus deletable for purposes of decreasing transitivity. A doubly-marked transitive form here would have wo- 'he' or mo- 'she' in place of Ø 'deleted prefix' (similar to 3a).

2.4. Possessive noun phrases

Noun phrases might at first appear to be out of place in a discussion of transitivity. Nonetheless, language after language can nominalize verbs, which then approach the lowest limit of transitivity—if we allow the term to be stretched that far.

Table 1: Galela possessive pronouns compared with stative pronominal verb prefixes

1 exclusive (a)i i- *minga > mi(a) mi-
1 inclusive     nanga na-
2 (a)ni ni- *ninga > ni(a) ni-
3 feminine (a)mi mi-    
3     manga ya-
3 masculine (a)wi wi-    
non-human ma Da- ma Da-

Van der Veen has discussed the similarity between the possessive pronouns and the pronominal prefixes which occur on stative verbs (1915:195-6). Table 1 compares them in parallel columns. The parentheses enclose the optional portion which is occasionally omitted. The asterisks indicate older forms which have been replaced by those to the right of the wedge (van Baarda 1908:68; van der Veen 1915:196). Van der Veen also posits -nga as a fossilized pronominal pluralizer, citing comparative data from Ternate (p.183). Removing this pluralizer, and taking the shortened singular forms, we thus arrive at forms identical with most of the stative pronominal verb prefixes. We will look at the exceptions in a moment, but first I give an example of those which are similar in this way.

(5) a. awi siri 
       his sick
      'his sickness'

    b. wi  siri 
       his sick 
      'his sickness' 
    c. wi-siri 
      'He is sick.'

Examples 5a and 5b are equivalent possessive constructions; they vary freely, although 5a is more common. Example 5c is a stative construction. Both 5b and 5c are phonologically identical, the orthographic distinction (one vs. two words) merely preserving the grammatical distinction. This, however, raises the question of whether they are indeed one construction. In other words, should the shortened possessive form 5b be reanalyzed as a stative form, as in 5c? Very few lexical items normally occur either as a possessed noun or as a stative verb, due to semantic constraints.

We now return to the question of whether it is permissible to discuss the transitivity of possessive noun phrases (NPs). If there is a sense in which possessive NPs are more verb-like than other NPs, then we can claim that they are also more transitive than other NPs. Indeed, there is a sense in which the possessor is in a state of possession. Conversely for stative verbs (lowest transitivity of true 'verbs'), there is a sense in which the patient possesses the state. We thus find some semantic similarity between possessive NPs and stative verb phrases to parallel the apparent structural similarity between possessive pronouns and stative pronominal verb prefixes.

We now handle the formal exceptions from Table 1, i.e. non-human and third person plural. In another paper, I have proposed a category of fourth person which, for Galela transitive pronominal verb prefixes, encompasses both non-human and what would otherwise be third person plural. However, this category of fourth person includes only non-human for stative verbs. It is useful to view speech participants in a hierarchy of agency as in Table 2.

Table 2: Pronominal verb prefixes ranked in a hierarchy of agency (H. Shelden 1991)

speaker addressee other (non-core) participants
singular plural non-human
1p 2p 3p 4p for transitive pronominal verb prefixes
<--high agency-----------------------------------------------------------low agency-->

Participants to the left are more likely to be found as agent of an action than are those to the right. In this hierarchy, speaker is labeled as first person, addressee is labeled as second person, and all others to the right are labeled as third (or fourth) person. Agency is a correlate of high transitivity (Hopper and Thompson 1980:252) and thus it is not surprising that the third person plural and non-human possessive pronouns, low in agency, do not behave quite as verb-like or transitively as do the other possessive pronouns.

3. Increasing transitivity

We have seen that a Galela speaker who wishes to decrease the transitivity of a particular clause for discourse purposes may do so by selecting a less transitive pronominal reference, sometimes with a corresponding derivational (or syntactic) change and sometimes without. Likewise, a Galela speaker who wishes to increase the transitivity of a particular clause, and thus increase its foregrounding within the discourse, may select a more transitive pronominal reference.

3.1. Verbalized nouns

It is quite common for Galela nouns to be used as verbs. In this case the subject is always marked with a pronominal verb prefix, with or without corresponding derivational changes.

(6) po-tagi      po-Soatobaru 
    we.incl.-go  we.incl.-Soatobaru
   'We went to Soatobaru.'  

Clause 6 is a sequence of two verbs, each composed of the subject marker po- 'we inclusive' and a stem. But notice that the first stem, tagi 'walk', is unambiguously an intransitive verb, while the second stem, Soatobaru 'name of village', is normally a noun. It is the presence of the subject marker here which verbalizes the noun Soatobaru[5]. Example 7 is perhaps even more convincing.

(7) So  ni-goroba         eko  hiwa? 
    so  or   not
   'So will you go by ox cart or not?'  

The word goroba 'cart' (from Malay gerobak) is normally a noun. However, adding a pronominal verb prefix as in 7 causes it to be used as a verb. In this particular example, we find corroborating evidence that the resulting construction is indeed a verb. Hiwa 'not' negates only predications; Da-sowo 'it-wrong' would be required in order to negate a nominal ('So is it your ox cart or not?'). Thus it is the presence of the pronominal verb prefix which verbalizes the noun in 7. Furthermore, if verbs are inherently more transitive than nouns, it follows that adding a pronominal verb prefix to a noun increases its transitivity.

3.2. Constructions with the causative prefix

The causative prefix is the most common of the derivational prefixes that affect transitivity. Pronominal verb prefixes interact identically with all of them, so causative examples serve our purposes here. The causative prefix has several functions (van Baarda 1895:380, 1908:89-99), one of which is to transitivize stative and intransitive verbs.

The verb siri 'be hurt' is a stative verb and normally marks its patient with an object marker. Thus in 8a, the prefix mi- 'her', which would also be used to mark the object of a transitive verb, marks the grammatical subject (semantic patient) of the clause mi-siri 'She is hurt'.

(8)a. mi-siri
      'She is hurt.' 
   b. wo-mi-si-siri
      'He hurt her.'

In 8b, however, the same verb is marked with both subject (wo- 'he') and object (mi- 'her') markers and also with the causative prefix si-. This is a transitive usage of the verb siri 'be hurt' signalled both by the use of the causative and by the use of the transitive pronominal verb prefix (i.e. subject marker) in addition to the stative pronominal verb prefix (i.e. object marker). From 8b we could conclude that transitivity is increased either by the causative or by the pronominal reference or by both. To clarify this we shall examine a transitivized intransitive verb.

The verb tagi 'walk' is an intransitive verb and thus normally takes only a subject marker. Thus in 9a, the subject o Robi 'Robby' is marked on the verb tagi 'walk' with the subject marker wo- 'he'.

(9) a. o Robi  wo-tagi 
       a Robby he-walk
      'Robby is walking.' 

    b. o Robi  to-wi-si-tagi 
       a Robby I-him-CAUS-walk 
      'I am helping (baby) Robby walk.'

In 9b, however, the same verb (tagi 'walk') is marked with both subject (to- 'I') and object (wi- 'him') markers and also with the causative prefix si-. The causative prefix increases the potential transitivity of the verb tagi 'walk' and the presence of two pronominal verb prefixes exploits this potential. The term potential transitivity here implies that either of the pronominal verb prefixes can yet be deleted as we saw in sections 2.2. and 2.3., which would result in an intransitive (or less transitive) utterance. It is the presence of both pronominal verb prefixes which ultimately increases the transitivity of the utterance.

4. Conclusion

In this paper we have explored several facets of the relationship between Galela pronominal reference and the transitivity of clauses. We have attempted to show that the pronominal verb prefixes are a reliable indicator of the degree of transitivity and are thus important devices in the foregrounding or backgrounding of speech utterances in Galela discourse.

The critic may respond that the pronominal verb prefixes are no more than inflectional reflexes of the verb's nominal arguments, and that it is these nominal arguments which determine the transitivity of the clause. However, the deletability of subject or object markers is independent of the presence or absence of nominal referents. Indeed, the semantic role of nominal arguments depends on these prefixes (D. Shelden 1986) and it should, therefore, not be surprising if clause transitivity as well depends on these prefixes.

Furthermore, the striking similarity of the possessive pronouns to the pronominal (stative) verb prefixes, on the one hand, and the ease with which nouns are verbalized by simply adding a pronominal verb prefix, on the other hand, suggest a transitivity continuum which places nouns at the low (zero) terminus.


1 An earlier draft of this paper was presented at the Seminar for Research of Eastern Indonesia, which was held in Manado, Indonesia, 23-29 July l985, and sponsored jointly by Leknas-LIPI and Sam Ratulangi University. The paper has not been substantially revised since then. It consequently reflects a viewpoint (Hopper and Thompson 1980) that has undergone significant development since the publication of that pioneering work. I wish to thank Frances Ingemann and Michael Martens for suggesting several improvements. I, of course, bear the ultimate responsibility for applying or misapplying their suggestions. This research would have been impossible without the permission and approval of various levels of the Indonesian government and the cooperative agreement between Pattimura University in Ambon and the Summer Institute of Linguistics (UnPatti/SIL) which sponsored the field work for this study.

Most of the examples were taken from transcribed text material and all were checked with native speakers. Much credit should go to my Galela neighbors and the countless other Galela speakers who taught me their language while I lived there.

2 The Galela language has been classified as a member of the Mainland Family of the North Halmahera Stock in the West Papuan Phylum (Wurm 1982:205). It is spoken by approximately 80,000 Galelans who live on north Halmahera Island around Galela Bay and Lake Galela and on the south side of Morotai Island in the north Moluccan Islands of Indonesia.

Field work to study the Galela language under the auspices of the Summer Institute of Linguistics and Pattimura University began in October 1983. The data which formed the basis for this analysis was gathered during several field trips totalling twelve months in the village of Duma, at the interior end of Lake Galela.

The phonemes of Galela are:

vowels /i/, /e/, /a/, /u/, /o/
voiceless stops /p/, /t/, /tS/, /k/
voiced stops /b/, /d6/, /d5/, /dZ/, /g/
fricatives /¸/, /s/, /h/
nasals /m/, /n/, /ø/, /N/
lateral /l/
flap /R/
semivowels /y/, /w/.

The /d6/ is realized by a voiced alveolar laminal stop, whereas the /d5/is realized by an voiced alveolar apical stop.

The practical orthography used in this paper is based on the Indonesian orthography and is the same as the phonemic orthography except that:

D is used for /d6/
d is used for /d5/
f is used for /¸/
c is used for /tS/
j is used for /dZ/
ny is used for /ø/
ng is used for /N/
r is used for /R/.

3 What I describe as pronominal verb prefixes were described as separate words by van Baarda (1895:37). The reasons for this departure are not relevant to the present discussion. The following table sets out the pronominal verb prefixes for intransitive and stative verbs in what is a typical split S-marking system (Shelden 1991). Subject prefixes agree with the subject of an intransitive or transitive verb. Object prefixes agree with the object of a transitive verb or with the patient of a stative verb.

1 exclusive to- i- mi- mi-
1 inclusive     po- na-
2 no- ni- ni- ni-
3 feminine mo- mi-    
3     yo- ya-
3 masculine wo- wi-    
non-human i- Da- i- Da-

4 Technically, ri- is the dual form of reciprocal and kV- (i.e. ka-, ko-, ki-, ku-, ke-) is the plural form. For details of these and more complex derivational devices involving reciprocals, see van Baarda (1908:117-22).

5 This is not a locative construction. Locatives in Galela have an entirely different structure. See van der Veen (1915:95-97).

5. References

Baarda, M. J. van. 1895. Een Galelareesch-Hollandsche Woordenlijst. 's-Gravenhage: Martinus Nijhoff.

--------. 1908. Leiddraad bij het bestudeeren van't Galela'sch Dialekt, op het Eiland Halmaheira. 's-Gravenhage: Martinus Nijhoff.

Hopper, Paul J. and Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56:251-300.

Shelden, Deidre. 1986. Topical and non-topical participants in Galela narrative discourse. Papers in New Guinea linguistics, No. 25.233-48. Pacific Linguistics, A-74.

Shelden, Howard. 1991. Galela pronominal verb prefixes. Papers in Papuan linguistics No. 1.161-175, Pacific Linguistics, A-73.

Veen, H. van der. 1915. De Noord-Halmahera'se taalgroep tegenover de Austronesiese talen. Leiden: L. van Nifterik.

Wurm, Stephen A. 1982. Papuan languages of Oceania. Tübingen: Gunter Narr.

Date created: 22-Oct-1998

[SILEWP 1998 Contents |SILEWP Home | SIL Home]