Corps de l’article


In West Greenlandic, the verb moods are divided more or less neatly into superordinate moods, including the indicative, interrogative, imperative, and optative moods, and the subordinate moods, including what are termed the causative, conditional, contemporative, and participial moods. The latter two subordinate moods are particularly interesting for the range of functions they fulfill in the language. The participial, for example, is the head of modifying clauses such as relative clauses and adverbial clauses of time or manner; argument clauses, in particular object clauses; and focus constructions involving particles such as sunaaffa, variously glossable as ’it turns out that,’ ’suddenly,’ or ’wow!’; and it serves as a basis for nominalization.

The role and importance of the participial within the verb mood system has changed considerably within the historical period of West Greenlandic. One of the more noticeable changes has been the gradual development of shared function between the participial and the contemporative. For example, from perhaps the first half of the 19th century, the mood of preference as the head of object clauses shifted from almost exclusively participial to either participial or contemporative (a trend noted by Kleinschmidt 1851: 76). In a slightly later but probably related development, the participial and contemporative came to be associated with switch-reference, with the former indicating switch-subject and the latter same-subject across two clauses (Berge 2000). In the following example from the mid-19th century (from HE[1]), a participial object clause is preferred; in the modern language, the contemporative would be:

(1)  tauna maliit ama takkogene ungnirpuk
taanna Maliit aamma taku-gini / -llugit unner-voq and see-4sg.subj / 3pl.obj.part /
3pl.obj.ct say -3sg.ind
’that Maliit also said she saw them’

This example not only illustrates participial use in object clauses, but more specifically, a participial object clause in construction with a verb of elocution. In fact, a close inspection of object clauses in the historical period suggests that not all verbs of elocution behave alike in their requirements of participials in object clauses. At a time when the changes outlined above were taking place, a certain conservatism could be noted in the use of the participial with one verb of elocution, unnerpoq ’to say about, tell, narrate,’ both in its function as head of an object clause and in its coreferential subject marking capability.

Verbs of elocution typically introduce direct or indirect speech constructions. In many languages, special grammatical forms exist to signal indirect speech; thus the use of the subjunctive in German, for example. Perhaps because of such forms in familiar languages, grammarians of Greenlandic have listed ways of forming indirect speech constructions, but no particularly special construction for indirect speech has been noted. In examining historical changes in the use of the participial in object clauses, however, I have found several distributional patterns which suggest that, in fact, there may have been a distinct indirect speech construction in old West Greenlandic. In addition to uncovering an archaic construction, these patterns also reveal important preferences in both older and more modern West Greenlandic narrative style. Both distributional patterns and discourse preferences have been missed in grammatical descriptions of the language, especially as our understanding of reported speech comes from descriptions which, until the second half of the 20th century, made only passing mention of reported speech, and in which examples of reported speech are presented out of context to illustrate other grammatical phenomena.

Working with a variety of texts, including oral and written personal recollections, traditional tales, and letters, written by native Greenlanders (with the exception of Pok’s Book, thought to have been written by Poul Egede) spanning the historical period of Greenlandic, from the 18th century to the present, and representing almost 3000 clauses, I found important differences in syntactic requirements among the most common verbs of elocution. The majority appear to allow a number of different object clause types, and indeed, a variety of construction types. The verb oqarpoq ’say,’ for example, allows object clauses headed by participial, contemporative, or causative verbs, and can refer to direct or indirect speech, as well as to neither in particular. One verb in particular, however, unnerpoq ’say about, tell,’ is never found, even in modern times and despite diachronic changes in the use of the participial, with anything but indirect speech and participial object clauses. Texts from all periods show a distinct preference for direct over indirect speech as a method of narrating reported events. The verb unnerpoq fell into disuse by the turn of the 20th century, and no other verb of elocution has shown such a fixed set of syntactic requirements. It appears, therefore, that a distinct construction reflecting a dispreferred narrative stylistic option has been lost.

In this paper, the direct and indirect speech constructions and the participial’s role as head of object clauses in these constructions will be examined in order to show these patterns and their effects on our understanding of West Greenlandic discourse, and most especially, oral or orally-based narrative discourse. First, I will summarize the methods of reporting speech in West Greenlandic and briefly review the published literature on verbs of elocution in the language. These descriptions have focused on object clause requirements of verbs of elocution or on changes in these requirements. I will then present data to show that there are substantive differences in the development, use, and grammatical requirements of various verbs of elocution. What has been missing from traditional descriptions is the contextual information necessary to detect these differences. Finally, therefore, I will suggest some benefits to the study of discourse in West Greenlandic.

Direct and indirect speech strategies in West Greenlandic

The grammatically-unchanged presentation of speech as it was produced, that is, with the same verb mood, tense, pronouns, and so forth (e.g., ’he said: "I am going to the store now"’) is referred to as direct speech. In contrast, reported speech that is presented with grammatical modifications, rather than as it was originally said by the speaker, is known as indirect speech; most commonly, this entails differences in verb mood and deixis (e.g., ’he said he would go to the store then / now’). The reported speech, whether direct or indirect, is a functional direct object of the verb of elocution. It may be syntactically marked as a direct object as well, where the verb is syntactically transitive, or it may be unmarked and simply juxtaposed, where the verb is intransitive. Fortescue (1995) explores the various syntactic and morphological methods of indicating both direct and indirect speech in West Greenlandic, and they are summarized here, although the examples are taken from other sources. These methods are found in both older and more modern varieties of the language. For direct speech, the most common method is the use of a verb like oqarpoq ’he said’ followed by the speech itself, i.e. a syntactic strategy:

(2)  direct speech — syntactic method

Soligok Polarlone Kevlak Okarpuk anigadlait (HE)
suli-gooq pulaar-luni Qillaq oqar-voq ani-gallar-git visit-4sg.ct Qillaq say-3sg.ind go.out-imp.softener-2sg.imp
’while she was still visiting, Qillaq said: "go out!"’

Morphologically, there is an affix -Vr- which can attach to a limited number of common words or expressions:

(3)  direct speech — morphological method

’he said "boat"’

For indirect speech, again, an independent reportative verb such as oqarpoq ’he said’ or unnerpoq ’he related’ can be used, with an object clause, most commonly in the participial or contemporative mood (i.e. depending on switch-reference), but sometimes in the causative:

(4)  indirect speech — syntactic method

Nuka oqarpoq qasoqaluni (Langgård and Langgård 1988: 67-68)
Nuka oqar-voq qasu-qi-luni
Nuka say-3sg.ind be.tired-intns-4sg.ct
’Nuka says / said she is / was very tired’

Morphologically, an affix -nerar- ’say that’ can be attached to the verbal component of the indirect speech:

(5)  indirect speech — morphological method

pitsaanirarpaat (Fortescue 1984: 3)
be.good-say.that-3pl.subj / 3sg.obj.ind
’they say that it is good’

Finally, a quotative enclitic -gooq indicates that what was said was originally said by someone else (e.g., ’someone else said it might rain,’ ’it is said that it might rain,’ etc.). Rather than emphasizing the speech, the enclitic deemphasizes the responsibility of the speaker for the speech; this enclitic is not under consideration here. There are also various ways of indicating direct and indirect questions, which I leave out of consideration here. There are both structurally and semantically different consequences in the use of morphological versus syntactic means of expressing reported speech, at least some of which were presented in Fortescue (1995). Because of the particular cooccurrence of object clauses, and thus participials, with the syntactic constructions, the focus of this article is on these rather than on the morphological means.

There is yet another possibility for discussing reported speech. Indirect speech still conveys a fairly accurate idea of the speech itself. There is also the possibility of mentioning the subject matter of the speech without reproducing the speech itself, either directly or indirectly. For lack of a better term, I call this the subject matter of speech (e.g., ’he talked / said something about going to the store’):

(6)  subject matter of speech

amagok tauna Ottorkasovak Okaluktualirpuk (HE)
aamma-gooq taanna utoqqaq-suaq oqaluttuar-ler-voq
and-and tell.story-begin-3sg.ind
’and that old man began to tell a story / also told a story’

arni Okaluktuvarlogo
arni oqaluttuar-lugu
mother.4sg.pos.abs tell.story-begin-3sg.obj.ct
’telling a story about his own mother’

In West Greenlandic, it is generally expressed as a nominal object of a verb of elocution. Although commonly not considered in discussions of reported speech, I find the concept of subject matter of speech useful in understanding different construction types relating to reported speech.

What the sources say about verbs of elocution

It is true that examples of syntactic direct and indirect speech such as these are found in all periods of documented West Greenlandic; however, some fundamental syntactic changes which occured in the historical period, in particular the overlapping of functions between the participial and the contemporative and the loss of participial subject coreferential morphology, directly affected object clause constructions, including those headed by verbs of elocution. Some of these effects are evident in the different grammatical descriptions spanning the historical period.

Few of the sources on Greenlandic directly address direct speech, presumably because it is fairly straightforward; so the focus of any discussion of the verbs of elocution has been with respect to indirect speech. The verbs of elocution are generally considered a subset of experiential verbs, such as verbs of saying, thinking, feeling, seeing, and so forth; consequently, they are generally described together with other experiential verbs. In fact, most 18th and 19th century sources do not distinguish experiential verbs from others which take object clauses.

In the very early sources, Top (see Bergsland and Rischel 1986), Egede (1760) and Fabricius (1801)[2], examples involving verbs of elocution are found, but they are presented to illustrate grammatical features such as the use of coreferential and noncoreferential pronominal inflection and the use of particular verb moods rather than direct or indirect speech. Object clauses of experiential verbs, including verbs of elocution, are found with verbs in the participial and causative moods (ex. 7-8), as well as with nominalized participials in the instrumental case (ex. 9)[3].

(7)  verb + part

tunniomarine unnerpok (Fabricius 1801: 376)
tuni-juma-ginni unner-voq
give-want-4sg.subj / 3pl.obj.part say-3sg.ind
’he(i) says he(i) wants to give to them’

(8)  verb + ca

unniorame okallukpok (Egede 1760: 192; Fabricius 1801: 378)
unior-gami oqaluC-voq
miss(a.shot) say.about-3sg.ind
’he(i) says he(i) missed the shot’

(9)  verb + inst-mik

ermiksumik unnerput
ermiC-soq-mik unner-vut
wash-part-sg.inst say-3pl.ind
’they(i) say [about themselves] that they(i) have washed themselves’

This last example is one member of an odd pair of constructions mentioned by Top and later Egede and Fabricius. In its transitive form, unnerpoq is unremarkable:

(10)  transitive singular verb + part

ermiksok unnerpa
ermiC-soq unner-vaa
wash-3sg.part say.about-3sg.subj / 3sg.obj.ind
’he(i) says that he(j) has washed himself(j)’

(11)  transitive plural verb + part

ermiksut unnerpei
ermiC-sut unner-vai
wash-3pl.part say.about-3sg.subj / 3pl.obj.ind
’he(i) says that they(j) washed themselves(j)’

But in its intransitive, or reflexive, form, and only in 3rd person, the participial object is identified as a nominal; in the singular, it is in the relative case, and in the plural, it is in the singular instrumental case, as in example 9:

(12)  intransitive singular verb + sg.rel

ermiksup unnerpok
ermiC-soq-p unner-voq
wash-part-sg.rel say-3sg.ind
’he(i ) says [about himself] that he(i) has washed himself’

No other independent source exists to corroborate this, at least in the singular form. Fabricius lists the same examples as Egede, but in his dictionary (Fabricius 1804: 530), he gives an example of an intransitive form with a contemporative, which seems to directly contradict this rule:

(13)  intransitive plural verb + ct

Ajunginiaromaudlutik unnerput
ajor-nngit-niar-juma-lutik unner-vut
be.bad-neg-fut-want-4pl.ct say-3pl.ind
’they(i) say [about themselves] they(i) will try to be good’

This is an inconsistency which may affect the validity of these reflexive constructions or which may represent the beginning of the change in favour of contemporative objects. Example 13 is unusual in several respects for this period; it is the only experiential verb I am aware of with a contemporative object, and it is the only early example of unnerpoq ’say about, tell’ with a contemporative; even Egede’s dictionary of 1750 only has participial examples for this entry. This construction was not originally identified as belonging exclusively to any particular subset of verbs, but rather as a strategy for reflexive verbs in general; yet the only examples given are from the experiential verbs oqarpoq ’say,’ unnerpoq ’say about, tell,’ and misigilerpoq ’experience.’ Fortescue (1984: 47) even suggests that it is a construction specific to the first two verbs, or in other words, to particular verbs of elocution. From a structural point of view, the instrumentally-marked object (cf. ex. 9) is nothing more than the singular antipassive object of an intransitive experiential verb. It is commonly found throughout the 19th century. The ergative-marked object (cf. ex. 12) remains questionable. Bergsland (1976) has discussed a possible explanation for this; but the data must in any event be taken with a grain of salt. It seems that Egede mistranslated some of his Greenlandic examples[4]:

(14)  mistranslation

ermikane unnerpok ’he(i) says that he(j) washed himself(j)’ (Egede 1760: 198)
ermiC-gaani unner-voq
wash-3sg.subj / 4sg.obj.part say-3sg.ind
’he(i) says that he(j) washed him(i)’

Kleinschmidt (1851: 76) writes that by his time, participial object clauses have started to be replaced by contemporative and causative ones, particularly in subject coreferential cases and especially in conjunction with 1st and 2nd person objects. It is especially coreferentiality which seems to be the driving consideration here, since by this time, subject coreferential forms of the participial have become rare. There is evidence from my texts that this change started to occur in the period between the 1820s and the 1850s. There is a particularly clear indication of this in the following difference between a Kragh (1820s) manuscript of Qaqitsoq’s story and the later (1850s) recopying by seminary students for Rink’s publication of the same story in KO:

(15)  (from Kragh, 1820s)

say-totally-4pl.subj / 3sg.obj.part

(16)  (from Rink’s seminary students, 1850s)


However, this cannot have been more than a tendency at the time, since many of the stories collected by Rink nevertheless have these coreferential forms, and they are contemporaneous with Kleinschmidt. Thus:

(17)  coreferential participial still common in the 1860s (from Bergsland 1955: 46; from KO III 34)

igluni tammartuq uyarini unnirlugu
illoq-ni tammar-soq ujar-gini unner-lugu
cousin-4sg.pos.abs look.for-4sg.subj / 3sg.obj.part
’saying of him [his cousin] that he was looking for his cousin who was lost’

With respect to the special intransitive form of the indirect speech construction with the ergative case, Kleinschmidt notes it but it is clear that he finds it odd (see Bergsland 1976: 14): the only examples he gives are variations of those found in the previous grammars, and he provides alternate forms which he claims are more usual in his day. However, both singular and plural forms of verbs of elocution with the instrumental nominalized participial are attested at this time. Kleinschmidt again suggests that the contemporative object clause is more common:

(18)  replacement of nominalized participial with contemporative

aggísassumik oKarput ’sie sagten, dass sie (selbst) kommen würden’ (Kleinschmidt 1851: 76)
aggísavdlutik oKarput ’sie sagten kommen zu wollen’ (Kleinschmidt 1851: 76)
aggi-ssa-soq-mik / -lutik oqar-vut
come-fut-part-sg.inst / -3pl.ct say-3pl.ind
’they said that they themselves would come’

Rasmussen (1888: 196) makes the same observation: instrumental objects are possible, but the contemporative object clause is preferred (in the following example, this is the very environment in which Egede and Fabricius would have listed a nominalized participial in the relative case):

(19)  replacement of nominalized participial with contemporative (old singular reflexive)

unigkumanngitsumik únerpoq ’han omtalte sig ikke villende blive’ (Rasmussen 1888: 196)
uniC-juma-nngit-soq-mik unner-voq
stay-want-neg-part-sg.inst say-3sg.ind
’he said [about himself] that he didn’t want to stay’

unigkumanani únerpoq ’han sagde, at han ikke vilde blive’ (ibid.)
uniC-juma-nani unner-voq
stay-want-4sg.neg.ct say-3sg.ind
’he said that he [himself] didn’t want to stay’

By the 20th century, it appears that the participial is not used in reflexive and subject coreferential cases[5] and the choice of contemporative or participial is dependent on subject coreference or lack thereof (Bergsland 1955: 49-50; Fortescue 1984: 40; Langgård and Langgård 1988: 67-68; see ex. 1). Fortescue (1984: 40) does, however, mention the use of nominalized participials in the instrumental with unnerpoq ’say about, tell,’ and earlier with oqarpoq ’say.’ He suggests that unnerpoq with a following instrumental object is still used, although the contemporative would be more colloquial. In fact, the example he gives is exactly that of Rasmussen (1888). Newer grammars, such as Langgård and Langgård (1988), do not even mention unnerpoq, a reflection of its essentially non-existent role in modern Greenlandic speech.

From the sources, therefore, it seems that verbs of elocution (as with other experiential verbs) underwent changes in their requirements of object clauses, following changes in the use of the participial verb mood to express subject coreferentiality. However, there are grounds for more closely examining this category of verbs. In my texts, there is an unusually high number of coreferential participial forms in object clauses of verbs of elocution, and more particularly with unnerpoq. It turns out that in context, unnerpoq is far more conservative that the descriptions given above would lead one to imagine. Further, there are non-negligible differences between constructions involving oqarpoq and other verbs of elocution, and those involving unnerpoq, as I will show in the following sections. I focus on verbs of elocution rather than more generally on experiential verbs or other verbs which allow object clauses, as the construction which requires attention is specific to reported speech, and as the narrative texts show a preference for verbs of elocution over more general experiential verbs in reporting interactions between participants.

Verbs of elocution in the texts — oqarpoq and its derivatives

There are many different ways in West Greenlandic of expressing what would be translated as ’say’ in English. By far, the most frequent and most important in narratives is with the use of oqarpoq and its derivatives, including oqaatigaa ’say something about him / it,’ oqaluppoq ’speak,’ oqaluttuarpoq ’talk about something,’ etc. I therefore take this group of verbs as representative of average verbs of elocution in reported speech constructions; other, much less frequently used verbs do show similar characteristics. In all texts prior to the mid-19th century, oqarpoq and its derivatives clearly take participials or, less frequently, causatives as heads of object clauses; thus, to see the effects of the encroachment of the contemporative in this position, the examples presented below are largely taken from the period immediately after the noted change. Immediately obvious from the texts is the range of complement types that can occur with oqarpoq or its derivatives. Thus, these verbs are regularly found with direct objects which are clauses, nominalized participials, noun phrases, or which are simply unexpressed. The examples presented below illustrate these complement types; they are found from the earliest texts to the most recent.

The effect of the derivational morphology on the simple stem oqar- ’say’ is to manipulate which of several possible objects can be in a direct relationship with the verb. For example, oqarfigaa ’talk to someone’ specifies who was talking and the person being talked to; oqaluttuarpaa ’talk about something’ on the other hand specifies who was talking and what was being talked about. Thus, depending on the particular derivation, the verb can take an object clause for the purposes of expressing indirect speech, or it may not be able to from a strictly structural point of view, and the clause containing the indirect speech will be juxtaposed. This can account for the frequent combination of oqarpoq ’say’ and its derivatives with another verb of elocution, particularly unnerpoq ’say about, tell’ in all periods in which unnerpoq is still in use. An early illustration of this is found in example 20, from Egede’s 1744 translation of Matthew and quoted by Bergsland (1976: 14). Example 21 is from the end of the period in which unnerpoq is used:

(20)  frequent combination of oqarpoq with other verbs of elocution

kingorna Jesup Ajokarsukene okarbigilerpai
kingorna Jesu-p ajoqersugaq-t-ni oqarfigi-ler-vai
after Jesus-rel disciple-pl-4sg.posm.abs / 3pl.obj.ind
’afterwards Jesus talked to his disciples’

Jerusalamut pissirsub unnertlune
Jerusalem-mut pi-ssa-soq-p unner-luni
Jerusalem-term do-fut-part-rel say.about-4sg.ct
’[saying himself] that he would go to Jerusalem’

(21)  oqarpoq with other verbs of elocution

Egedevligok Okarfiga (HE)
Egede-p-li-gooq oqarfigi-vaa / 3sg.obj.ind
’but Egede said to them’

taimak penavejangitomik ungnerlone
taamak pi-navianngit-soq-mik unner-luni
thus do-absolutely.not-neg-part-sg.inst say-4sg.ct
’saying he certainly didn’t want to do it’ [i.e. ’talking about absolutely not wanting to do it’]

Koisimagamek Tokotinauvejangitomek unirlone
kui-sima-gamik toqutsi-navianngit-soq-mik unner-luni murder / kill-absolutely.not-part-sg.inst say- 4sg.ct
’saying that because they [Egede’s family] had been baptized he certainly didn’t want to kill’ [i.e. ’talking about not wanting to kill because they had been baptized’]

Oqarpoq ’say’ is also used with an object clause for indirect speech, but this use is infrequent. In fact, this verb is disproportionally more frequently used for direct speech than indirect speech, and there are no good textual examples of this use before the mid-19th century. Even thereafter, oqarpoq is almost always found with direct speech. Examples 22 and 23 illustrate the use of oqarpoq with following participial and contemporative object clauses:

(22)  oqarpoq + participial object clause for indirect speech

aiso Okarput nangmasimata elat anelerlone (HE)
asu oqar-vut naammassi-mmata ila-at ani-ler-luni
well say-3pl.ind part-3pl.pos / sg.posm.abs go.out- begin-4sg.ct
’well, they said when they were finished, one of them was going out’

avatarsovak Tomaramiok
aavataq-suaq tummar-gamiuk step.on-4sg.subj /
’when he trod on the big float’

qiver-riallar-gamiuk /
’as soon as he bent over it’

angoagsorsagata avata
angu-sussaa-gaq-ata aavataq-a / sg.posm.rel float- 3sg.pos / sg.posm.abs
’the float of the one who was going to catch seals’

elais usorotagaraut
ila-asa usoruut-tigi-gaat
part-3pl.pos / pl.posm.rel / 3obj.part
’the others praised him [the shaman]!’

(23)  oqarpoq + contemporative object clause for indirect speech

taimaidlogingok nejovertob Okarfegai (HE)
taamaaC-lugit-gooq niuertoq-p oqarfigi-vai trader-sg.rel / 3pl.obj.ind
’while it was like this, the trader said to them’

Kub Senanut Tevaijartortarkudloget
kuuk-p sini-anut tiva-jartor-qqu-lugit
river-sg.rel side / edge-3sg.pos / sg.posm.term dance.with.drum-
’they could go have a drum dance by the side of the river’

Derivatives of oqarpoq ’say,’ particularly those of oqaluppoq ’speak’ behave as if they were more remotely connected to means of expressing speech. For example, oqaluppoq ’speak’ (intransitive) is only infrequently found with a nominalized participial with instrumental case marking -mik, in juxtaposition with a following verb of elocution, and with participial object clauses. These complement types are less frequently found in the 20th century, but they are nevertheless still possible. More often than not, however, oqaluppoq ’speak’ and words deriving from it tend not to indicate speech but rather the subject matter of the speech (e.g., ’he told about old people,’ ’he talked about witches,’ etc.). Transitive derivations tend to have nominal objects, as with oqaluttuarpoq / paa ’talk about something,’ or oqaluttuuppaa ’tell a story to someone’ and both transitive and intransitive forms can take juxtaposed participial or contemporative clauses, as the following examples show, but they do not require direct or indirect speech, or for that matter, an overtly expressed subject matter of speech. In example 6 above, a juxtaposed contemporative clause contains the expressed subject matter of speech; in example 24 below, it is unexpressed:

(24)  Vb + juxtaposed CT, unexpressed subject of speech

elani orniglogit Okaluktudluget (HE)
ila-ni orniC-lugit oqaluttuuC-lugit
part-4sg.pos.abs meet-3pl.obj.ct
’meeting her relatives to tell them [about it, i.e. seeing a ghost, in the preceding narrative]’

(25)  Vb + juxtaposed direct speech construction

allijortorkagat (HE)
aliortor-qi-vagut / see.a.ghost-intns-1pl.subj / 3pl.obj.ind[6]
’"we have seen a ghost!"’

esivsodlogo Okaluktûpa (HE)
isussuuC-lugu oqaluttuuC-vaa / 3sg.obj.ind
’whispering to him, she told him’ [i.e. ’she told him whispering "we have seen a ghost!"’]

Oqaluppalaar- ’tell a story’ is most often used as a nominal; as a verb, it is found with or without nominal objects:

(26)  Vb + nominalized direct object

angaijorkaûnût Okalupalarotigilirkarpa (HE)
angajoqqaat-nnut oqaluppalaaq-ut-gi-ler-qqaar-vaa
parents-1sg.pos.term story-own-have-begin-first-3sg.subj / 3sg.obj.ind
’she told to my parents her stories’

’about when she first began to know things’

(27)  Vb + no object

amale attataga Abaram Okalupalartok (HE)
aamma-li ataata-ga Abraham oqaluppalaar-soq
and-but father-1sg.pos / sg.posm.abs Abraham tell.story-3sg.part
’my grandfather Abraham tell stories’

tusarnaar-sar-vara / 1sg.obj.ind
’I used to listen to’

In all of these, by far the most common clause types are participials and contemporatives. Egede (1760: 192) gives examples of subject matter of speech constructions, or as he describes it, when the meaning is that one tells about how something happened, with oqaatigi- ’say something about him / it’ and oqaluC- ’speak’ and a following causative verb mood (see example 8 above). In my corpus, there are a few causative clauses in texts from the 18th and early 19th centuries which are questionably object clauses of oqarpoq ’say.’ In example 22 above, the question is ultimately whether the "object" of oqarput ’they said’ is one of the directly following causative clauses, or the final participial. Fortescue (1984) and others suggest that the causative is possible in the modern language, and causatives are found in my texts with other less commonly found verbs of elocution. Given this and given the range of complement types that oqarpoq and its derivatives can head, there is no a priori reason to suggest that oqarpoq cannot take causative object clauses, although these are largely for subject matter of speech. Thus, it appears that oqarpoq with its following object clause has never been a fixed construction, the verb allows a variety of options in its object clauses, including the contemporative, the participial, and the causative verb moods, in addition to the nominalized instrumental object, and it has a wide range of uses, from indicating direct or indirect speech to noting the subject of the speech itself. In all respects, oqarpoq behaves as a normal experiential verb requiring an object, and no differently from the second most commonly used verb of elocution, the general all-purpose verb pivoq ’do.’ However, it appears that oqarpoq gradually developed from primarily head of a direct speech construction to head of more general reported speech constructions; this slow expansion occured during the 19th century. Further, there has always been a strong tendency for the more derived forms to more remotely represent speech. By preference, to this day oqarpoq is used for direct speech; given that it is by far the most common verb of elocution, it also follows that syntactic direct speech is preferred to indirect speech as a narrative technique[7].

Verbs of elocution in the texts — unnerpoq

In the corpus under consideration, unnerpoq ’say about, tell’ is only found in texts from 1765 (the earliest known extant text written by a native Greenlander) through the 1860s. None of the texts after that have it, although it is frequently found in grammars and dictionaries in isolated examples of verbs of elocution or participial constructions, and so forth. Most of these examples have been quoted and requoted from the earliest sources, with minor changes thought to reflect grammatical developments. In all of the actual texts from the period prior to the turn of the 20th century, unnerpoq only takes either an object clause with the participial verb mood or a nominalized participial in the instrumental case, and it only indicates indirect speech. It is so regular, in fact, that although Fortescue (1995) writes that oqarpoq precedes an embedded participial and unnerpoq can do so, from my data, I would have to say that it is rather the reverse.

The vast majority of examples of unnerpoq are found with participial object clauses. In the two versions of Pok’s Book (from 1760 and 1857), we see about four instances of unnerpoq for indirect speech, all with participial object clauses. From the Greenlander Jacob Poulson’s letter of 1765, there is one example of indirect speech using unnerpoq, also with participial:

(28)  unnerpoq + participial

ajorartogut kaktogullo unneraratigut (Jacob Poulson, 1765)
ajorsar-sugut kaaC-sugut-lu unner-sari-gaatigut
suffer.want-1pl.part be.hungry-1pl.part say.about-hab-3sg.subj / 1pl.obj.part
’he said about us that we are hungry and suffer want’

In one text (HE), unnerpoq is used with a participial object clause at least 10 times (three with coreferent subject), and with the instrumental nominalized participial about five times (four with coreferent subject); but it is never once used with other verb moods or nominal constructions. The transitive unnerpaa only shows up with participial object clauses, as in example 29; the intransitive unnerpoq can take either a nominalized participial in the instrumental, as in example 30, or a transitive participial object clause, as in example 1.

(29)  transitive verb + participial object clause:

taimaitok umidlartunga uniraranga (HE)
taamaattoq uumi-llar-sunga unner-sari-vaannga
yet / still say.about-hab- 3pl.subj / 1sg.obj.part
’yet they said about me that I was unhappy because I missed [her]’

(30)  intransitive verb + instrumental object:

amali attarsovara Egidi (HE)
aamma-li aata-rsuaq-ga Egede
and-but grandfather-big-1sg.pos / sg.posm.abs Egede
’and my grandfather Egede’

nungmigìk enusomek ungnirpuk
Nuuk-mi-gooq inuu-soq-mik unner-voq live / be.born-part-sg.inst say.about-3sg.ind
’said he was [supposedly] born in Nuuk’

Note that in each case, the indirect speech is both functionally and structurally an object. Thus, the object of the transitive verb of elocution is an object clause. In the second case, the object clause of the intransitive verb is given antipassive marking, i.e. the instrumental case-marking -mik, just as if it were a normal antipassive object. In the last case, it is not possible to nominalize a transitive participial, and so the object is maintained as a verbal clause.

There is another possible difference between the use of the nominalized participial in the instrumental case and the participial object clause. In the former, the instrumental may sometimes be translated as the subject matter of speech, as in example 21; in the latter, the participial clause is the indirect speech. The English translations of many of the examples given previously of these nominalized participials are accurate but not literal translations of the original German or Danish. Kleinschmidt notes the use of the instrumental as a subject of speech:

(31)  instrumental object as subject:

átánik oKalugput, ’sie reder von seehunden’ (Kleinschmidt 1851: 85)
aataaq-nik oqaluC-vut
saddle-back.seal-pl.inst talk.about-3pl.ind
’they are talking about saddle-back seals’

pilíssaminik univkárpoK, ’er erzählt von seinen thaten’ (ibid.)
pi-ler-ssaq-minik unikkaar-voq
do-begin-pass.part-4sg.pos.inst tell.about-3sg.ind
’he tells about his deeds’

Perhaps this is related to the use of the nominalized participials in the instrumental case as adverbials or adjectives:

(32)  nominalized participials in the instrumental case as adverbials

kigaitsumik aggerpoK (Kleinschmidt 1851: 86)
kigaaC-soq-mik agger-voq
be.slow-part-sg.inst approach-3sg.ind
’he approaches slowly’

Another example is found in Jacob Poulson’s letter of 1765:

(33)  nominalized participials in the instrumental case as adverbials

opernartomik okausikarngarniarmet
uppernar-soq-mik oqaaseq-qar-ngaar-niar-mat
’although he really has words that can be believable’

There are three cited examples of unnerpoq with a contemporative object clause. In example 13, I noted Fabricius’ dictionary entry for unnerpoq and some reasons for questioning it, although it may represent the beginnings of the grammatical change. Another is cited by Kleinschmidt, and is in fact his modernized version of an example cited by Fabricius using by then obsolescent coreferent pronominal forms in the participial; I find it particularly interesting that no new examples of a contemporative object clause are given in this context:

(34)  unnerpoq + contemporative

tunniomarine unnerpok (Fabricius 1801)
tuniumavdlugit únerpoK (Kleinschmidt 1851)
tuni-juma-ginni / -lugit unner-voq
give-want-4sg.subj / 3pl.obj.part / -3pl.obj.ct say.about-3sg.ind
’he said he wanted to give them something’

The third example is by Rasmussen (1888) and is again unoriginal. All examples are decontextualized, that is, they are single sentence examples illustrating sentences with object clauses and the at the time modern preference for contemporative over participial verb moods. It is also noteworthy that none of the contemporary texts appear to have examples of unnerpoq with a contemporative object clause; and these include the texts collected by Kragh and Rink and published as KO, and my primary text, Hans Egede’s narration[8].

These examples might suggest that the use started changing at the same time it also started becoming obsolescent at around Kleinschmidt’s time. This is also suggested by the following pair, in which the original verb of saying has been replaced by oqarpoq:

(35)  obsolescence

iserbigingikitik unnerput (Fabricius 1801)
iserfigíngíkitik oqarput (Kleinschmidt 1851: 75)
iser-fik-gi-nngit-gitik unner / oqar-vut / 3sg.obj.part say.about- / say- 3pl.ind
’they(i) said they(i) hadn’t come in to them(j)’ [i.e. entered into their abode]

By the turn of the 20th century, there are no obvious contextual examples of unnerpoq. Interestingly, native speakers today accept as understandable and correct the use of contemporative and causative verb moods with unnerpoq. However, without exception, all speakers I have consulted with feel the verb is obsolete. Thus, modern judgments made about an obsolete verb, like the decontextualized examples of Kleinschmidt and Rasmussen, reflect modern preferences in the use of the verb moods but say nothing about the indirect speech construction as it was when actually in use[9].

One of the most obvious differences between unnerpoq and other verbs of elocution is that unnerpoq is only found with indirect speech or, arguably, the subject matter of speech. Everything, in fact, points to the existence of a distinct set of syntactic preferences relating to indirect speech. Thus, where indirect speech is indicated, morphological means are preferred over syntactic means (and these are outside consideration here). Where indirect speech is indicated syntactically, unnerpoq is clearly preferred to other verbs of elocution; in fact, it may have been the primary indicator of indirect speech before the mid-19th century. Finally, in object clauses of the verb unnerpoq, participials are not only preferred to other verb moods, but appear to be required, even at a time when the participial is being replaced by the contemporative in other environments.

Conclusions and directions for further study

It appears that unnerpoq is not just a preferred, but indeed a primary indication of indirect speech in archaic West Greenlandic. It also has to compete with the preferred method of indicating reported speech, which is direct speech. Other verbs of elocution are used for direct speech, preferentially verbs based on the stem oqar- ’say.’ Although all methods of indicating reported speech are present throughout the historical period, changes in object clause construction and coreference marking on verbs affected these methods in different ways. Thus, the more commonly used direct speech constructions prove to be more flexible, allowing contemporative mood to replace coreferential participial mood as head of object clauses, and gradually expanding to include indirect speech and a variety of complementation types. The verb unnerpoq, however, proves less flexible, maintains its requirement for participially-based object clauses indicating indirect speech, and gradually loses ground as coreferential participial forms become obsolete and more general requirements of object clauses change. The gradual disappearance of unnerpoq parallels the gradual expansion of oqarpoq to fill this niche. No comparable syntactic indirect speech construction seems to exist today. Further, oqarpoq and its derivatives did not become more likely to indicate indirect speech than previously. Because direct speech is greatly preferred to indirect speech in both the written and the oral texts from all periods, the loss of a preferred but nevertheless not obligatory indirect speech construction (in the sense that it was not the only option for expressing indirect speech) has not led to the compensatory creation of a new construction.

If unnerpoq was in a fixed construction with a participial object clause to denote indirect speech, this is not to suggest that it was the obligatory method of indicating indirect speech, but rather the preferred syntactic method. It was a feature of older Greenlandic; the origins of this indirect speech construction, however, are obscure. No other Inuit dialect seems to have a comparable indirect speech construction, at least from the rather sparse descriptions available. This may be a reflection of the lack of thorough linguistic descriptions available for many of the Inuit dialects, or it may be instead a reflection of innovation in Greenlandic. However, as Dorais (1996) has suggested, the distinct Inuit dialects appear to be of relatively recent date (ca. 16th century), leaving little time for such an independent development. Thus, the tendency which would have allowed for this construction would have had to be present already in proto-Inuit.

Little enough is suggested from a separate categorization of verbs of elocution. Where verbs of elocution in other dialects have been addressed in the literature, there seems to be no requirement for a participial object clause, although a participial is possible (cf. especially Bourquin 1891; Schneider 1976). The stem unner- is apparently only found in Greenlandic and the Canadian Inuit dialects (and perhaps in Sireniki Yup’ik, according to Fortescue et. al. 1994), and no remarks have been made about it in descriptions of Canadian dialects. However, the role of the participial in all dialects, indeed in most of the Eskaleut languages, is consistent, despite current obfuscation in Canadian and Alaskan Inuit dialects, where the participial and indicative moods have largely fused. The participial plays an important role in narratives, where it is used for describing perfective actions or states, for observational constructions, where events or situations are observed as a consequence of certain action, and so forth (cf. Bergsland 1997; Hinz 1944; Jacobson 1995; Mennecier 1995; etc.). Many early grammars of Greenlandic associated the participial with the past tense or perfective aspect, and this association is noted time and again for other Inuit dialects and languages. For example, in Alutiiq and Siberian Yup’ik (de Reuse 1994), using the participial is a common way of expressing past tense in narratives, including the past of actions not seen by the narrator. It is understood today that the participial in West Greenlandic is not an indication of tense per se; however, there is clearly some connection with the conditions that presuppose perfective aspect or past tense, i.e. something happened and it is being talked about as a fait accompli. One possible reason, therefore, for the development of an indirect speech strategy in West Greenlandic which might have required the participial is that indirect speech by definition is a paraphrase of what was said elsewhere or at another time. It is parallel to indications to the listener that something was not directly seen or heard; this was not directly said. The importance of the participial in the indirect speech construction, therefore, has some basis in pan-Inuit discourse preferences, and changes in usage of the participial can therefore reasonably be expected to affect a method of expressing indirect speech which relies on the participial.

The rather narrow focus of this article has been on whether or not one is justified in claiming the existence of an indirect speech construction based on the textual evidence in the historical period of West Greenlandic; I claim there is. In studying this construction, however, it has also become clear that there is a marked preference for direct speech strategies, at least in certain kinds of texts. In fact, some support for this already comes from Frederiksen (1954: 18), who notes that speeches in older Greenlandic literature are most frequently indicated by the use of a verb like oqarpoq and following direct speech, and in modern Greenlandic by visual representations (colons, quotation marks) but nevertheless by direct speech. Frederiksen, however, is concerned with literature in particular, rather than linguistics, and it therefore still comes as a revelation of preferences in language use and points to the still vastly unexplored area of discourse studies in West Greenlandic (and, for that matter, in the Inuit language in general).

It is also interesting that observations of construction preferences for many languages show a marked preference for direct speech strategies, whether or not these result in exact reduplication of the actual speech (Linguistlist query summary 4.303, 25 April, 1993). This preference, however, is strongly affected by such factors as discourse type, orality versus written literature, and so forth. The study I have presented here covers a variety of textual types, but these are primarily narratives; they include oral and written narratives, personal recollections, and correspondence. Modern literary traditions, prepared speech (e.g., newspaper reports, etc.), and other manifestations of an important change from an oral to a literacy-dependent culture might quite possibly show different strategies, both because of the rise of a modern literary scene as well as from Danish influence on this scene. In any case, this study highlights the need both for revisiting old data and for future studies of discourse in West Greenlandic.

Grammatical information: sg = singular, PL = plural, subj = subject, obj = object; nominal inflection: abl = ablative, abs = absolutive, loc = locative, rel = relative, inst = instrumental, term = terminalis, via = vialis; verbal inflection: ca = causative, ct = contemporative, ind = indicative, imp = imperative, part = participial; derivation: fut = future, hab = habitual, intns = intensifier, neg = negative, nomz = nominalizer, pass = passive, perf = perfective, pos = possessor, posm = possessum.