Volume 12 Issue 2 (2014)
DOI:10.1349/PS1.1537-0852.A.443
Note: Linguistic Discovery uses Unicode characters
to represent phonetic symbols. Please see Optimizing Display
for requirements to accurately reproduce this page.
Emai’s Variable Coding of AdjunctsRonald P. Schaefer and Francis O. Egbokhare Southern Illinois University Edwardsville and University of
Ibadan This paper examines the morphosyntactic character of clauses
containing adjuncts in Emai (Edoid and West Benue Congo). In clauses differing
as to discourse function, adjunct coding is variable. Some adjunct types are
consistently structured as either head of a phrase or complement in a phrase
headed by a verb. Other adjuncts are coded more variably. In canonical
declarative clauses, they appear in postverbal position unmarked by a verb, but
in one or more noncanonical clause types, their clause requires a verb otherwise
latent. Resulting patterns are assessed from a perspective in Croft (2001),
where adjuncts are relations with their matrix clause as
argument. 1.
Introduction[1]The adjunct/argument distinction is widely recognized in
language description and explanation. It is most often associated with a
syntactic criterion; nonetheless, the semantic nature of adjuncts and arguments
has drawn some attention (Matthews 1981, 2007, Croft 2001). For the languages of
sub-Saharan Africa, adjunct and argument have received minimal scrutiny. Watters
(2000) notes that adjuncts (X) generally follow objects (O) in SVOX languages,
while in SOV languages they either precede V (SXOV) or follow V (SOVX). For this paper, we explore adjunct structures in Emai, a West Benue
Congo language within southern Nigeria’s Edoid group (Elugbe 1989,
Williamson and Blench 2000). Typologically, Emai is relatively strict SVO with
lexical and grammatical tone but little inflectional morphology and few
prepositions (Schaefer and Egbokhare 1999, 2007, to appear). Word order is
pervasive as a marker of grammatical relations. Regarding clause structure, Emai
is characterized by simple predicates as well as complex predicates consisting
of verbs in series, verbs in construction with postverbal particles, and verbs
in series with verbs and postverbal particles (Schaefer and Egbokhare 2010). In
addition, Emai shows an extensive array of preverbs (Schaefer and Egbokhare
2000), many of them adverb like, that affect interpretation of clausal event
(che ‘again,’ ya ‘almost,’ duu
‘for no reason,’ kakégbe ‘perseveringly,’
kpao ‘initially’) or a core participant (zemi
‘very many,’ gba
‘together’).[2] 2. Adjunct Character
Croft (2001) reviews a number of criteria proposed to
distinguish adjuncts from arguments. The classic syntactic criterion holds that
adjunct constituents are optional while arguments are obligatory. This seems
relatively straightforward. In (1a-b), an adjunct (e.g. in the park) is
peripheral to its associated verb, since it can be omitted without consequence
to grammaticalness. In contrast, arguments (George, the dog) are
obligatory relative to their predicate; argument omission results in
ungrammaticality (1a-c). (1a) George chased the dog in the park. (1b) George chased the dog. (1c) *George chased in the park. Questions arising from Croft’s analysis and
illustration are no doubt multiple. Two, however, concern us. One is whether the
distributional potential that characterizes locative (i.e. in the park),
applies equally to other adjunct types, for instance those expressing
temporality (for the afternoon in 2a) or manner (clumsily in 2b).
(2a) George chased the dog for the afternoon. (2b) George chased the dog clumsily. A second focuses on whether all adjunct types lead to
consistent morphosyntactic expression across canonical and noncanonical clause
types.[3] Directing these questions toward Emai, we find that adjuncts in clause
types differing as to discourse function attract nonuniform coding. Some adjunct
types across canonical and noncanonical clauses occur as either head of a phrase
or as complement in a phrase headed by a verb. Other adjuncts are coded more
variably. In canonical declarative clauses, they appear in postverbal position
unmarked by a verb, but in one or more noncanonical clause types, e.g.
imperative, interrogative or contrastive focus, their clause requires a verb
otherwise latent. As an initial sample of this variability, we present Table 1.
It reveals that outside of manner, which is consistently unmarked by a verb, and
reason, which is consistently verb marked (by re), adjuncts with a
locative or temporal character require, in addition to a main verb, a latent
verb such as re or
za.[4]
| DECL | IMP | INTER | CF | MANNER | verb AD | verb AD |
|
|
|
|
|
|
| TEMPORAL DEIXIS | verb AD | verb AD | re verb | AD verb |
|
|
|
|
| TEMPORAL BOUND | verb AD | re AD verb |
|
|
|
|
|
|
| TEMPORAL QUANTITY | verb AD | re AD verb | re verb |
|
|
|
|
|
|
|
|
|
|
| LOCATIVE | verb AD | verb AD | za verb | AD za verb |
|
|
|
|
| REASON | re AD verb | re AD verb |
| AD re verb |
|
|
|
|
|
Table 1: Emai adjunct (AD) occurrence with main verb and
latent verb (za or re) across clauses that
are canonical, i.e. declarative (DECL), and noncanonical, i.e. imperative (IMP),
interrogative (INTER) and contrastive focus (CF). 3. Emai Adjuncts and
ArgumentsArgument types in Emai exhibit rather consistent
distributional behavior compared to the more variable nature of adjuncts. A
direct object argument (óràn ‘wood’), whose
grammatical relation is morphologically unmarked but syntactically indicated by
word order, follows a transitive verb such as hian ‘cut’
(3a). Verbs like hian also occur in series with a transitive verb such as
re ‘take,’ which precedes its obligatory direct object
argument (ópìà ‘cutlass,’ 3b-c).
Both arguments relate to an overall event of cutting. (3a) | òjè | híán | óràn. / | *òjè | híán. |
| Oje | PRP.cut | wood | Oje | PRP.cut |
| ‘Oje cut wood. / Oje cut.’ |
(3b) | òjè | ré | ópìà | híán | óràn. |
| Oje | PRP.take | cutlass | cut | wood |
| ‘Oje used a cutlass to cut wood.’ |
(3c) | *òjè | ré | híán | óràn. |
| Oje | PRP.take | cut | wood |
| ‘Oje cut wood.’ |
As well, intransitive verbs in series such as za
‘be located’ and se ‘move up to, as far as,
reach, extend to’ precede their obligatory arguments marked by
locative preposition vbi (4a-b). (4a) | òjè | zá | vbí | áfúzé' | shán | sé | vbì | òkè. |
| Oje | PRP.be.located | LOC | Afuze | walk | move.up.to | LOC | Oke |
| ‘Oje walked from Afuze to Oke.’ |
(4b) | *òjè | zá | shàn | sé | vbì | òkè. |
| Oje | PRP.be.located | walk | move.up.to | LOC | Oke | | ‘Oje walked to Oke.’ |
Emai adjunct types display some distributional consistency
across clauses as well. In canonical declaratives they prototypically occupy
postpredicate position following an intransitive verb or direct object of a
transitive verb. As constituents adjuncts occur either unmarked or marked by
preposition vbi. Unmarked are temporal deictic adjuncts
(òdè ‘yesterday’ 5a) and temporal
quantity adjunct phrases (íkpédè
èéà ‘for three days’ 5b). Marked by
vbi are temporal ordinal adjunct phrases
(úkpédé lí
ózèéà ‘on the third day’ 5c)
and locative adjuncts (ímè ‘farm’ 5d).
(5a) | ójé | híán' | órán | òdè. |
| Oje | PAP.cut | wood | yesterday |
| ‘Oje cut wood yesterday.’ |
(5b) | ójé | híán' | órán | íkpédè | èéà. |
| Oje | PAP.cut | wood | days | three |
| ‘Oje cut wood for three days.’ |
(5c) | ójé | híán' | órán | vbí | úkpédé | lí | ózèéà. |
| Oje | PAP.cut | wood | LOC | day | R | third |
| ‘Oje cut wood on the third day.’ |
(5d) | ójé | híán' | órán | vbí | ímè. |
| Oje | PAP.cut | wood | LOC | farm |
| ‘Oje cut wood on the farm.’ |
A distinguishing feature of adjuncts, as opposed to
arguments, is their tonal impact on a preceding constituent. Relative to a
preceding verb argument (óràn), Emai adjuncts activate high
tone spread starting at argument right edge, as in órán
òdè (5a) versus óràn (3a).
Nonadverb constituents following a verb argument do not activate high tone
spread however. In (3b), where verb hian follows argument
ópìà, high tone spread is not activated.
Verbs in series thus do not trigger high tone spread. In addition, postverbal
particles that follow a direct object argument do not activate high tone spread
(6a-c), as shown by right edge low tone of úì
‘rope’ preceding Terminative lee ‘already,
finish,’ Change of Location o or Applicative
li. (6a) | òjè | híán | ólí | úì | léé. |
| Oje | PRP.cut | the | rope | TERM |
| ‘Oje finished cutting the rope / already cut the
rope.’ |
(6b) | òjè | híán | ólí | úì | ó | vbì | èvá. |
| Oje | PRP.cut | the | rope | CL | LOC | two |
| ‘Oje cut the rope into two.’ |
(6c) | òjè | híán | ólí | úì | lí | òhí. |
| Oje | PRP.cut | the | rope | APP | Ohi |
| ‘Oje cut the rope for Ohi.’ |
An important feature of nondeclarative clauses containing
Emai adjuncts pertains to their potential for variable coding. In information
question clauses, temporal deictic adjuncts represented by
òdè ‘yesterday,’ for example, correspond
to an interrogative proform (éghè
‘when’). Their predicate phrase requires the verb re
‘take’ as the initial verb in series (7), despite the fact that
re never occurs in the corresponding declarative clause, cf.
(5a).[5] (7) | éghè | ójé | ré' | híán | ólí | óràn? |
| time | Oje | PAP.take | cut | the | wood |
| ‘When did Oje cut the wood?’ |
| òdè. |
| yesterday |
| ‘Yesterday.’ |
A similar condition holds for a vbi marked adjunct of
place. In transitive declaratives, where a locative adjunct follows a verb and
its direct object, place marking is signaled by preposition vbi (5d). In
information questions, where the locative corresponds to a fronted interrogative
proform, the verb za ‘be located’ occurs in the matrix clause
and precedes other verbs in series. Thus in declarative (8a),
ímè ‘farm’ appears in postverbal position
preceded by preposition vbi; no za marks it. However, when adjunct
ímè corresponds to an interrogative proform
(ébé' ‘where’), za is obligatory in the
predicate phrase (8b). (8a) | ójé | híán' | órán | vbí | ímè. |
| Oje | PAP.cut | wood | LOC | farm |
| ‘Oje cut wood on the farm.’ |
(8b) | ébé' | ójé | zá' | híán | óràn? |
| where | Oje | PAP.be.located | cut | wood |
| ‘Where did Oje cut wood?’ |
Fronting of a locative adjunct in a contrastive focus clause
leads to similar marking by za. When ímè occurs in
focus position and precedes positive focus (PF) morpheme li, za is
required in the matrix clause as the initial verb in series. (9) | ímè | lí | ójé | zá' | híán | óràn. |
| farm | PF | Oje | PAP.be.located | cut | wood |
| ‘It was on a farm that Oje cut wood.’ |
In contrast, locative arguments of a verb do not give rise
to za. As an argument of verb o ‘enter’ in declarative
clauses, ímè in postverbal position follows preposition
vbi. (10) | òjè | ó | vbí | ímè. |
| Oje | PRP.enter | LOC | farm |
| ‘Oje entered the farm.’ |
In information question clauses, where
ímè corresponds to fronted interrogative proform
ébé' ‘where,’ za is disallowed (11a-b).
As well, occurrence of ímè in contrastive focus position
does not lead to a matrix clause predicate phrase with za as a verb in
series (11c). (11a) | ébé' | ójé | ó'-ì? |
| where | Oje | PAP.enter-F |
| ‘Where did Oje enter?’ |
(11b) | *ébé' | ójé | zá' | ó? |
| where | Oje | PAP.be.located | enter |
| ‘Where did Oje enter?’ |
(11c) | ímè | mè | lí | ójé | ó'-ì. |
| farm | my | PF | Oje | PAP.enter-F |
| ‘It was my farm that Oje entered.’ |
It is likewise the case for vbi marked locative
arguments (e.g. ùdékèn ‘wall’)
with transitive verbs (e.g. fi ‘hit,’ 12a). Neither
information question clauses (12b) nor negative contrastive focus clauses (12c)
with such verbs lead to verb za in series. (12a) | òjè | fí | úkpórán | vbì | ùdékèn. |
| Oje | PRP.hit | stick | LOC | wall |
| ‘Oje hit a stick on the wall.’ |
(12b) | ébé' | ójé | fí' | úkpóràn? |
| where | Oje | PAP.hit | stick |
| ‘Where did Oje hit the stick?’ |
(12c) | ùdékèn | kí | ójé | fí' | úkpóràn. |
| wall | NF | Oje | PAP.hit | stick |
| ‘It wasn’t on a wall that Oje hit the
stick.’ |
This brief overview of Emai locative and temporal forms
reveals adjuncts in nondeclarative clauses scaffolded by a latent verb. Scaffold
structures framed by a latent verb in series also affect adjuncts of temporal
quantity. In declarative clauses íkpédè
èéà ‘for three days,’ for instance, is
postverbal and unmarked by a preposition (13a). In information question clauses
(13b), where the temporal quantity constituent corresponds to a fronted
interrogative proform (íkpédé
ékà ‘how many days’), the verb re
‘take’ is obligatory and precedes other verbs in series. (13a) | ójé | híán' | órán | íkpédè | èéà. |
| Oje | PAP.cut | wood | days | three |
| ‘Oje cut wood for three days.’ |
(13b) | íkpédé | ékà | ójé | ré' | híán | óràn? |
| day | quantity | Oje | PAP.take | cut | wood |
| ‘For how many days did Oje cut wood?’ |
| íkpédè | èéà. |
| days | three |
| ‘For three days.’ |
Imperative clauses with adjunct
íkpédè èéà
grammatically mandate re as well; re must occur in
the matrix clause and in a position preceding other verbs in series (14a-b).
Temporal quantity adjuncts, though, are not found in contrastive focus
clauses. (14a) | rè | íkpédè | èéà | híán | óràn. |
| take | days | three | cut | wood |
| ‘Take three days to cut wood. / Cut wood for three
days.’ |
(14b) | *hìàn | órán | íkpédè | èéà. |
| cut | wood | days | three |
| ‘Cut wood for three days.’ |
Another adverbial adjunct type utilizing latent verb
re expresses temporal bounding for an event. In declarative
clauses, ékéín
íkpédè èéà
‘within three days’ is marked by preposition vbi (15a) and
occupies postpredicate position. In corresponding imperatives (15b),
re is obligatory with ékéín
íkpédè èéà as
complement; the resulting re
ékéín
íkpédè èéà
constituent precedes other verbs in series. Temporal bounding adjuncts
articulate neither information question nor contrastive focus clauses. (15a) | ólí | ómóhé | híán' | ólí | óràn | léé | vbí | ékéín | íkpédè | èéà. |
| the | man | PAP.cut | the | wood | TEMP | LOC | inside | days | three |
| ‘The man had finished cutting the wood within three
days.’ |
(15b) | rè | ékéín | íkpédè | èéà | híán | ólí | óràn | léé. |
| take | inside | days | three | cut | the | wood | TEMP |
| ‘Finish cutting the wood within three days.’ |
Not all adjuncts with temporal significance require latent
re. In fact, some evince no latent verb as scaffold. Adjuncts
expressing temporal ordinal relations (e.g.
úkpédé lí
ózèéà ‘on the third day’),
for instance, appear in postpredicate position marked by preposition vbi
(16a). Nonetheless without latent re, they retain postpredicate
position in imperative clauses (16b) and in contrastive focus clauses they
occupy focus position
(16c).[6] (16a) | ólí | ómóhé | híán' | ólí | órán | vbí | úkpédé | lí | ózèéà. |
| the | man | PAP.cut | the | wood | LOC | day | R | third |
| ‘The man cut the wood on the third day.’ |
(16b) | hìàn | ólí | órán | vbí | úkpédé | lí | ózèéà. |
| cut | the | wood | LOC | day | R | third |
| ‘Cut the wood on the third day.’ |
(16c) | úkpédé | lí | ózèéà | lí | ólí | ómóhé | híán' | ólí | óràn. |
| day | R | third | PF | the | man | PAP.cut | ólí | wood |
| ‘It was on the third day that the man cut the
wood.’ |
Adjuncts expressing a temporal frequency relation (e.g.
ìsèvá ‘twice’) also occur in
postpredicate position (17a), although not as a constituent marked by
preposition vbi. Without re or any other latent verb in the
matrix clause, temporal frequency adjuncts retain postpredicate position in
imperative clauses (17b), correspond in information questions to an
interrogative pronoun (ísékà ‘how often’
in 17c) and occupy contrastive focus position (17d). (17a) | ólí | ómóhé | é' | ólí | émáé | ìsèvá. |
| the | man | PAP.eat | the | food | twice |
| ‘The man ate the food twice.’ |
(17b) | è | ólí | émáé | ìsòkpá. |
| cut | the | wood | once |
| ‘Eat the food at once.’ |
(17c) | ísékà | ólí | ómóhé | é' | ólí | émàè? |
| how.often | the | man | PAP.eat | the | food |
| ‘How often did the man eat the food?’ |
(17d) | ìsèvá | lí | ólí | ómóhé | é' | ólí | émàè. |
| twice | PF | the | man | PAP.eat | the | food |
| ‘It was twice that the man ate the food.’ |
In stark contrast to these last two temporal types, adjuncts
expressing reason exhibit a more consistent verb scaffold pattern. Adjunct
reason constituents (e.g. òhíó
ísì òjè ‘because of Oje’) do not
occur in postpredicate position in declarative clauses (18a). Instead, as
complement they immediately follow verb re as the initial verb
phrase in series in declaratives (18b), in imperatives (18c), and when they
occupy contrastive focus position (18d), re is retained as the
initial verb in series.[7] (18a) | *ólí | ómóhé | gbé' | ófé | òhíó | ísì | òjè. |
| the | man | PAP.kill | rat | cause | ASS | Oje |
| ‘The man killed rats because of Oje.’ |
(18b) | ólí | ómóhé | ré' | òhíó | ísì | òjè | gbé | ófè. |
| the | man | PAP.take | cause | ASS | Oje | kill | rat |
| ‘The man killed rats because of Oje.’ |
(18c) | rè | òhíó | ísì | òjè | é | ólí | émàè. |
| take | cause | ASS | Oje | eat | the | food |
| ‘Use Oje as the reason for eating the food.’ |
(18d) | òhíó | ísì | òjè | lí | ólí | ómóhé | ré' | é | ólí | émàè | léé. |
| cause | ASS | Oje | PF | the | man | PAP.take | eat | the | food | TEMP |
| ‘It was because of Oje the man finished eating the
food.’ |
Differing from reason adjuncts and exhibiting a distinct
placement for their latent verb compared to temporal and locative types are
adjuncts conveying aspectual and temporal extent. Adjuncts of aspectual extent
(gbègbéí ‘completely’) occur in
postpredicate position in declarative (19a) and imperative (19b) clauses. In
information questions, where adjuncts of aspectual extent correspond to
interrogative proform (ébé' ‘how’) and an
accompanying manner (MAN) preverb
i,[8] their matrix clause
requires the postpredicate verb se ‘extend to, reach’
as the final verb in series (19c). Aspectual extent adjuncts fail to occupy
contrastive focus position. (19a) | ójé | ánmé' | óí | étò | á | gbègbéí. |
| Oje | PAP.scrape | her | hair | CS | completely |
| ‘Oje scraped off her hair completely.’ |
(19b) | ànmè | óí | étò | á | gbègbéí. |
| scrape | her | hair | CS | completely |
| ‘Scrape off her hair completely.’ |
(19c) | ébé' | ójé | í' | ánmé | óí | étò | sé? |
| how | Oje | PAP.MAN | scrape | her | hair | extend.to |
| ‘To what extent did Oje scrape her hair?’ |
| ó | ì | ànmè | óí | étó | á | gbègbéí. |
| he | NEG | scrape | her | hair | CS | completely |
| ‘He did not scrape off her hair at all.’ |
Other aspectual extent adjuncts include
jáún ‘completely, crisply’ (20a-b) and
sésésé ‘completely,
neatly’ (21a-b). In interrogative clauses they, too, give rise to latent
verb se. (20a) | ólí | ógó | tóó' | á | jáún. |
| the | bush | PAP.burn | CS | crisply |
| ‘The bush burned to a crisp.’ |
(20b) | ébé' | ólí | ógó | í' | tòò | sé? |
| how | the | bush | PAP.MAN | burn | extend.to |
| ‘To what extent did the bush burn?’ |
| ó | tóó' | á | jáún. |
| the | PAP.burn | CS | crisply |
| ‘It burned to a crisp.’ |
(21a) | ólí | ótóí | fúán-í' | sésésé. |
| the | ground | PAP.be.clean-F | completely |
| ‘The ground was absolutely clean.’ |
(21b) | ébé' | ólí | ótóí | í' | fùàn | sé? |
| how | the | ground | PAP.MAN | be.clean | extend.to |
| ‘To what extent was the ground clean?’ |
| ó | fúán-í' | sésésé. |
| it | PAP.be.clean-F | completely |
| ‘It was absolutely clean.’ |
Another adjunct type, temporal extent
(títítí ‘long time’), occurs in
postpredicate position in declarative (22a) clauses. In information questions,
where adjuncts of temporal extent correspond to interrogative proform
ébé' ‘how’ and its manner preverb i,
their matrix clause requires not only the extent verb se in series
but also the temporal verb tee ‘be long’ (22b).
Temporal extent adjuncts do not appear in imperative clauses or occupy
contrastive focus
position.[9] (22a) | ólí | ómóhé | múzán-í' | títítí. |
| the | man | PAP.wait-F | long.time |
| ‘The man waited for a long time. |
(22b) | ébé' | ólí | ómóhé | í' | mùzàn | téé | sè? |
| how | the | man | PAP.MAN | wait | be.long | extend.to |
| ‘How long did the man wait?’ |
|
| ó | múzán-í' | títítí. |
| the | PAP.wait-F | long.time |
| ‘He waited for a long time.’ |
A distinct and final pattern characterizes adjuncts of
manner. They reveal no evidence of a latent verb. Manner adjuncts
(kóíkóí ‘in a gulping fashion’)
occur in postpredicate position regardless of whether their clause is
declarative (23a) or imperative (23b). When manner adjuncts in information
questions correspond to interrogative proform (ébé'
‘how’) and manner preverb i (23c), their matrix clause fails
to show a latent verb like se, re or za.
Manner adjuncts of this type do not occupy contrastive focus position. (23a) | ólí | ómòhè | ò | ó | è | ólí | émáé | kóíkóí. |
| the | man | SC | C | eat | the | food | gulpingly |
| ‘The man is gulping the food / eating the food in a gulping
fashion.’ |
(23b) | è | ólí | émáé | kóíkóí. |
| eat | eat | food | gulpingly |
| ‘Gulp the food. / Eat the food in a gulping
fashion.’ |
(23c) | ébé' | ólí | ómóhé | ó | ò | í | è | émàè? |
| how | the | man | SC | H | MAN | eat | food |
| ‘How does the man eat food?’ |
| ó | ò | è | óí | kóíkóí. |
| he | H | eat | it | gulpingly |
| ‘He eats it in a gulping fashion. / He gulps it.’ |
4. DiscussionIn the preceding section we called attention to Emai’s
latent verb coding patterns for some clause structures that incorporate
adjuncts. For several adjunct types (locative, temporal bounded, deixis and
quantity as well as aspectual and temporal extent), one or more of their
noncanonical clauses, i.e. imperative, interrogative or contrastive focus, were
coded with a latent verb, whereas their canonical declarative clause was not.
Moreover, coding was not uniform across adjunct types, either by latent verb
form (re, za, se, tee) or
position (pre- versus post-matrix predicate). Still other adjunct types revealed
either no latent verb (manner, temporal frequency and temporal ordinal) or
consistently required a preceding verb (reason). As a summary of Emai adjunct
behavior and its accompanying clausal coding we present Table 2.
| DECL | IMP | INTER | CF | R-Q | MANNER | verb AD | verb AD | ébé' i verb |
| in situ |
|
|
|
|
|
| A EXTENT | verb AD | verb AD | ébé' i verb se
|
| in situ | T EXTENT | verb AD |
| ébé' i verb tee
se |
| in situ | |
|
|
|
|
| T FREQ | verb AD | verb AD | ísékà verb | AD li verb | ex situ | T ORDINAL | verb vbi AD | verb vbi AD |
| AD li verb | ex situ |
|
|
|
|
|
| T DEIXIS | verb AD | verb AD | éghè re
verb | AD li verb | ex situ | T BOUND | verb vbi AD | re AD verb |
|
|
| T QUANT | verb AD | re AD verb | íkpédè ékà
re verb |
| ex situ |
|
|
|
|
|
| LOCATIVE | verb vbi AD | verb vbi AD | ébé' za verb | AD li za verb | ex situ |
|
|
|
|
|
| REASON | re AD verb | re AD verb |
| AD li re verb | ex situ |
Table 2: Coding of Emai adjunct (AD) types (where A is
Aspectual and T is Temporal) by verbs se, tee,
re, and za relative to clause types declarative (DECL),
imperative (IMP), interrogative (INTER), contrastive focus (CF) and response to
information question (R-Q). At the outset we noted in passing Croft’s (2001)
comparison of adjunct and argument semantic character. His semantic analysis
emphasizes adjuncts as relations relative to their associated predication,
following theoretical arguments laid out by Langacker (1987: 214-216), who
posits a relation as existing when the definition of one concept inherently
requires reference to another concept. If adjuncts are inherently relations,
they are functions, i.e. predicates, that take an argument. With reference to
(24), in the park is then a predicate whose single argument is the event
of chasing. One and the same semantic component, i,e, chase, can thus be
a relation or a filler argument of a role in a relation. While chasing is a
relation with George and dog as filler arguments, chasing is also a filler
argument for the relation being-in-the-park. (24) George chased the dog in the park. Earlier, we identified two questions that derive from
Croft’s analysis and illustration. Do all adjunct types lead to consistent
morphosyntactic expression across clause types? And do all adjuncts manifest
similar distributional potential? More importantly for Croft’s analysis is
a third question: Is there morphosyntactic evidence to support the claim that an
adjunct is a relation and so can take an associated event as filler
argument? Recall Croft’s position that adjuncts are relations taking as
their filler argument a matrix clause event. The facts from Emai suggest that
not all adjunct constituents are relations vis-à-vis the matrix
predicate, i.e. predicates that take a matrix event as filler. Instead, a number
of adjunct types appear to be filler arguments for a latent verb that under
varying discourse conditions appears in series in the matrix predicate phrase.
It is these filler adjunct types, locative and temporal deixis for example, and
their latent verbs, za and re respectively, that as a
relation could take the matrix clause event as filler. Not all adjunct expressions serve as filler argument for a latent verb
however. Some appear to be relations that could directly take the matrix
predicate as filler. The clearest example of this adjunct type is manner;
adjuncts expressing temporal frequency and temporal ordinal sequence also appear
to be examplars of this type. Based on these distribution facts from Emai, one could formulate a
relation-filler cline for adjuncts in which the propensity to serve as a filler
argument increases while the propensity to serve as a relation decreases. The
most comprehensive filler adjunct would be REASON, which requires the verb
re in all clauses where its exponents occur. The next most
filler-like adjunct would be LOCATIVE, which requires latent verb za in
contrastive focus and interrogative clauses. The least filler-like and most
relation-like adjunct would be MANNER, which revealed no latent verb. TEMPORAL
would clearly be the most inconsistent class since temporal frequency and
temporal ordinal evince no latent verb, while temporal deixis, temporal quantity
and temporal bound lead to latent verb re in either interrogative
or imperative clauses or both. MANNER < TEMPORAL < LOCATIVE < REASON Extent adjuncts, however, exhibit unique properties, as
shown by their interrogative frame. They manifest a correspondence relation to
not only an information question word (ébé'
‘how’) and its preverb í but also a latent verb
(se) or verbs (se, tee) in
postpredicate position, neither of which surfaces in imperative or contrastive
focus clauses. The syntactic position of latent verbs associated with extent
adjuncts thus contrasts with the position of latent verbs for locative and
temporal adjuncts. Extent adjuncts would be troublesome for a relation-filler cline. They
consistently require a latent verb (or verbs) in interrogative clauses but
position it after, not before, the matrix predicate. Where would extent adjuncts
best fit on a relation-filler cline? Moreover, one wonders whether there might
be other linguistic evidence that would identify structural affinities between
or among Emai adjunct types. One fact to consider in this regard is shape of
adjunct response frame relative to its information question. In question-answer
discourse contexts, many adjunct types occur in response to an information
question as phrases isolated from clause structure, i.e. ex situ as
summarized in Table 2 and exemplified in (7), (8b), (13b) and (17c). Three
adjunct types, aspectual extent, temporal extent and manner, fail this test
however; each requires an in situ frame in which the respective adjunct
follows its matrix verb, as in (19c), (22b) and (23c), repeated here as (25),
(26) and (27). (25) | ébé' | ójé | í' | ánmé | óí | étò | sé? |
| how | Oje | PAP.MAN | scrape | her | hair | extend.to |
| ‘To what extent did Oje scrape her hair?’ |
| ó | ì | ànmè | óí | étó | á | gbègbéí. |
| he | NEG | scrape | her | hair | CS | completely |
| ‘He did not scrape her hair completely. |
(26) | ébé' | ólí | ómóhé | í' | mùzàn | téé | sè? |
| how | the | man | PAP.MAN | wait | be.long | extend.to |
| ‘How long did the man wait?’ |
|
| ó | múzán-í' | títítí. |
| he | PAP.wait-F | long.time |
| ‘He waited for a long time.’ |
(27) | ébé' | ólí | ómóhé | ó | ò | í | è | émàè? |
| how | the | man | SC | H | MAN | eat | food |
| ‘How does the man eat food?’ |
|
|
| ó | ò | è | óí | kóíkóí. |
| he | H | eat | it | gulpingly |
| ‘He gulps it.’ |
In situ responses as well as interrogative proform in
information questions (ébé' and i) thus suggest that
adjuncts with a postpredicate latent verb are similar to adjuncts that reveal no
latent verb; both are more relation like than filler like. Clearly, a simple
correlation between morphosyntactic properties and adjunct status as relation or
filler is not straightforward.[10]
Nonetheless, it does appear that while all adjunct expressions may be
relational, not all adjuncts are relations. Some adjuncts are fillers that
require a latent verb, especially in clause types outside the canonical
declarative. ReferencesCroft, William. 2001. Radical construction grammar. New York:
Oxford University Press. Elugbe, Ben. 1989. Comparative Edoid: Phonology and lexicon. Port
Harcourt: University of Port Harcourt Press. Langacker, Ronald. W. 1987. Foundations of cognitive grammar, Vol
I: Theoretical perspectives. Stanford: Stanford University Press. Matthews, Peter H. 1981. Syntax. Cambridge: Cambridge University
Press. -----. 2007. Syntactic relations. New York: Cambridge University
Press. Schaefer, Ronald P. and Francis O. Egbokhare. 1999. Oral tradition
narratives of the Emai people, Part I and Part II. Hamburg: LIT
Verlag. -----. 2000. Emai preverb order. Proceedings of the 2nd
World Congress of African Linguistics, Lepzig 1997, ed. by H. Ekkehard Wolff and
Orin D. Gensler, 733-746. Cologne: Rüdiger Köppe Verlag. -----. 2007. A dictionary of Emai: An Edoid language of Nigeria.
Cologne: Rüdiger Köppe Verlag. -----. 2010. On Emai ditransitive constructions. Studies in
ditransitive constructions: A comparative handbook, ed. by Andrej Malchukov,
Martin Haspelmath and Bernard Comrie, 115-145. Berlin: De Gruyter
Mouton. -----. To appear. A reference grammar for Emai: A Nigerian Edoid
language. Berlin: Mouton de Gruyter. Watters, John. R. 2000. Syntax. African languages: An introduction,
ed. by Bernd Heine and Derek Nurse, 194-230. New York: Cambridge University
Press. Williamson, Kay and Roger Blench. 2000. Niger Congo. African
languages: An introduction, ed. by Bernd Heine and Derek Nurse, 11-42. New York:
Cambridge University Press. Author’s Contact Information (corresponding
author): Ronald P. Schaefer
Center for International Programs
Southern Illinois University Edwardsville
Edwardsville, IL 62026-1616
rschaef@siue.edu
Francis O. Egbokhare
Department of Linguistics and African Languages
University of Ibadan
Ibadan, Nigeria
foegbokhare@yahoo.com
[1]Data incorporated in
this paper derive from research sponsored by the National Science Foundation,
(BNS #9011338 and SBR #9409552), the U.S. Department of State (College and
University Affiliations Program grant ASJY 1333), Southern Illinois University
Edwardsville, particularly its Distinguished Research Professor award, and the
University of Ibadan, Nigeria, particularly its Inaugural Lecture series. We
thank these institutions for their continued support, while not extending to
them any responsibility for our data interpretation. [2]Orthographic
conventions for Emai generally reflect Schaefer and Egbokhare (2007), where
o represents a lax mid back vowel, e a lax mid front
vowel, and vb a voiced bilabial approximant. For tone, acute accent marks
high, grave signals low, and acute followed by an apostrophe designates high
downstep. Abbreviations for grammatical morphemes used in this paper include:
APP = applicative, ASS = associative, C = continuous, CL = change of location,
CS = change of state, F = factative, H = habitual, ID = identity pronoun, IND =
indicative, LOC = locative, MAN = manner, NEG = negative, NF = negative focus,
PAP = past perfect, PF = positive focus, PRP = present perfect, R = relator, SC
= subject concord, TEMP = temporal perspective. [3]It is worthwhile to
note that English adjuncts differ in morphosyntactic expression as well, e.g.
noun preceded by a preposition (in the park) versus lexical adverb
(clumsily). [4]Evidence that za
and re are verbs and not preverbs emerges from tonal behavior in
Present Perfect aspect (Schaefer and Egbokhare to appear), where verb phrase
initial monosyllabic preverbs like Additive gbo ‘too, also’
(in addition to auxiliaries) show a high low falling tone (òjè
gbóò híán óràn [Oje ADD PRP.cut
wood] ‘Oje has cut wood too’) but phrase initial monosyllabic verbs
do not (òjè gbé ófè [Oje PRP.kill rat]
‘Oje has killed a rat’ not *òjè gbéè
ófè). [5]Temporal deictic
adjuncts occur in contrastive focus position, òdè
lí ólí ómóhé
é' ólí émàè [yesterday PF
the man PAP.eat the food] ‘It was yesterday that the man ate the
food,’ although not with a latent verb in series. Temporal deictic
òdè does not occur in imperative clauses; other
deictic adjuncts, e.g. ákhò ‘tomorrow’
and éènà ‘today,’ do, though not
with a latent verb. [6]Temporal ordinal
adjuncts occur in information question clauses with a complex structure,
although not one involving a latent verb in series. Ordinal adjuncts correspond
to a question frame marked by the identity pronoun í ID and the
verb yi ‘identify’ and its complement noun
édè ‘day.’ Ordinal adjuncts thus
do not correspond to éghè re
interrogatives, as temporal deictic adjuncts do. (i) | í | yì | édé | lí | ó | híán' | ólí | óràn? |
| ID | identify | day | R | he | PAP.cut | the | wood |
| ‘On which day did he cut the wood? |
| úkpédé | lí | ózèéà. |
| day | R | third |
| ‘On the third day.’ |
[7]Reason adjuncts
correspond to the complex interrogative frame émé'
ó ze khi ‘why,’ which does not involve a
latent verb in series. Instead, the erstwhile matrix clause occurs in an
embedded indicative marked clause (indicative complement khi) under a
cause verb, i.e. ze. It is this matrix cause verb that
syntactically corresponds to the reason adjunct. (i) | émé' | ó | zé-í' | khí | ólí | ómóhé | é' | ólí | émàè | léé? |
| what | it | PAP.cause-F | IND | the | man | PAP.eat | the | food | TEMP |
| ‘What caused the man to finish eating the food? |
|
|
| òhíó | ísì | òhànmì. |
| cause | ASS | hunger |
| ‘Because of hunger.’ |
[8]That manner i is
a synchronic preverb, not a verb, is supported by its distribution. It never
occurs as a simple predicate, transitive or intransitive, or as one constituent
of a complex predicate. It arises only in manner expressions where a manner
related constituent has been fronted, e.g. manner demonstrative adjunct
(ìyá ‘that way,’ ìná
‘this way’) or with information question word
ébé' ‘how.’ In contrast, se,
za and re occur as simple predicates or as constituents of
complex predicates (Schaefer and Egbokhare to appear). [9]Temporal extent
adjuncts fail to occur in contrastive focus and imperative clauses, as the
examples below illustrate. (i) | *tìtìtì | lí | ólí | ómóhé | múzán'-ì. |
| long.time | PF | the | man | PAP.wait-F |
| ‘It was for a long time that the man
waited.’ |
(ii) | *mùzàn | títítí. |
| wait | long.time |
| ‘Wait a long time.’ |
[10]The correlation
between position of latent verb relative to matrix verb, i.e. preceding vs
following matrix predicate, and shape of interrogative response allowed, i.e.
in situ vs ex situ, suggests that a more complex parameter than
relation/argument may be operating. For the moment, however, we leave this
notion unexplored. |