View source for In-depth annotation of multi-verb constructions in Èdó

By 

[[User:Ota Ogie|Ota Ogie]]

Researcher affiliated with the Department of Language and Communication Studies. 

NTNU -
e-mail:ota.ogie@hf.ntnu.no

(alternatively you can use the TC-internal e-mail to contact me)


This page is under construction

'''INTRODUCTION'''

Type craft is a tool for annotation and sharing natural language paradigms online and is relevant in the documetation of minority or endangered languages [[media: Selected slides ISFITovation2009- Type Craft Software for documentation of minority languages.pdf]].
The main purpose of annotation is to describe the meaning and grammatical structure of languages.
Type craft features the following tiers arranged in the following order: 
1. Latinsed orthography
2. A tier for morphological analysis
3. A tier for meaning
4. A tier for glossing of morphological features
5. A tier for part of speech information. 

Syntactic and semantic information about construction types are provided by templates and encode also information provided in the tiers described above. The basic structural parts of a template are referred to as '''slots'''. In the slot specification, the following conventions are observed:
* Slots are interconnected by '-' (hyphen).
* Distinct items inside a slot are interconnected by '_' (underline).
* An item label containing neither ‘-‘ nor ‘_’ is an uninterrupted string of letters. If it acts as a complex label, the internal composition is indicated by alternation between small and capital letters (however, no labels are distinguished in terms of CAP vs. not). 

Constructions with a Verb as head have a Template structure with maximum seven slots:
* Slot 1: POS of the head, and diathesis information (e.g.: 'V_pas' for passive diathesis; 'V' alone if the construction is active).
* Slot 2: Valency: transitivity specification - intr, tr, ditr, ... (if the construction is passive, the valence given is that of its corresponding active form).
* Slot 3: Dependent Specification: comments on syntactic and referential properties of specific arguments .
* Slot 4: Participant Roles.
* Slot 5: Aspect and aktionsart and are written in CAPS.
* Slot 6: Situation Type: a label for the situation type expressed by the construction, written in CAPS
* Slot 7: Provides a linking type between slot 6 situation type and the specifications in slots 2-4.
Slots 1 and 2 are obligatorily filled, the others not. A slot not specified is not displayed(Cf. Hellan 2008 and Hellan and Dakubu 2009). I have represented this information under the tier for construction description. 

Main background theoretical assumptions are drawn from the following sources; implemented Head-Driven Phrase Structure Grammars for Norwegian (Hellan 2003) and Ga (Hellan 2007) a Kwa language spoken in Ghana; an HPSG account of argument realization patterns  (Beermann, Hellan and Sætherø 2003); and a theory of event structure template (Pustejovsky 1995, 2006).

The tier for glossing morphological features in my classification provides information in the following order:number,grammatical functions, case and thematic role information (except in instances where Typecraft does not provide gloss tags for some thematic roles and where I have improvised by providing such information on the meaning tier. The thematic role information is then represented before all gloss tags in the output construction). Number and case information are morphologically realised on pronouns and number is also realised morphologically on a small subset of nouns in Èdó. This is discussed further below.  Arguments of verbs are classified  in terms of their grammatical functions and theta roles and are lexically specified. They are represented for all examples.   
The approach, I adopt is a combination of James Pustejovsky`s framework  whereby thematic roles are defined by their position in an event template and definitions based on the referential properties that can be associated with each role bearer as in Gruber (1965),Hellan (2007)and Hellan and Dakubu (2009). 
The term ''agent'' is used as in Jackendoff 1992.He recognizes three kinds of agents: the doer of an action, the initiator of an action and the instigator of an action. In my analysis, the instigator is the precipitator of an action. The term  ''affected''  is used basically as defined by Jackendoff 1992 for the term ''patient''. My use of the term also incorporates Larson 1984`s definition of an affected entity. I use the term  ''Theme'' basically in the sense of Gruber 1965 and Jackendoff 1992, that is it is the entity that undergoes physical motion, or whose location is being asserted.    
To aid typological classification, I have used the convention of representing grammatical fuctions on noun heads only. This is not theoretically or language internally motivated but is necessary since type craft does not provide for this information to be represented on the whole noun phrase. Case information is represented in addition for pronouns. For example pronouns realised as value for the attribute SUBJ constraining the type ''val'' are classified as having nominative case and are token identified with the value of the attribute SUBJECT constraining the type ''qval'' expressing the subject grammatical function. In the following example, the first layer represents global tag information, the second layer, the Latin orthography, the third layer word division, the fourth layer the morphological description, the fifth layer represent both the meaning and gloss information, the sixth layer the part of speech information and the seventh layer the literal translation. 


v-tr-suAg_obThincrem-COMPLETED_MONODEVMT 
<Phrase>8159</Phrase>

''v'' in Slot1 in the example above states that the head of the construction is a verb. ''tr'' in Slot2 states that the verb is transitive, ''suAg_obThincrem'' in Slott 4 states that the NP that is the subject of the construction bears an agent theta role and the object an incremental theme theta role. Lastly slot 5 gives the information that the aktionsart of the construction is completed monodevelopment.


In the following I discuss first general features of Èdó language and then multi-verb constructions in Èdó. 


'''Èdó language: some basic facts'''

Èdó is a tone language spoken in Èdó state in Mid-Western Nigeria and belongs to the Edoid language group (Elugbe1979). It is a head initial SVO language with an open syllable system with no consonant clusters. The language has seven oral and five nasal vowels. Nasal vowels are distinguished from their oral counterparts by the orthographic convention of an [n] consonant after the nasal vowel. Their phonetic representations are as follows (Ogie 1987):

                       
Front close vowel:        [ i ]/ [ ĩ ]                              

Back close vowel:         [ u ]/ [ ũ ]                              

Front half close vowel:   [ e ]                                    

Back half close  vowel:   [ o ]                                    

Front half open  vowel:   [ ε ]/ [ ε̴ ]                             

Back half open vowel:     [ Ɔ ]/ [ Ɔ͠ ]                               

Central open  vowel:      [ a ]/ [ ã ]                                 



The front and back half open vowels [ε] and [Ɔ] have the following orthographic representations respectively ''ẹ'' and  ''ọ''.

I now discuss tone. There are two basic tones in Èdó: high ( ó) and low (È). Nominal heads bear constant tones while verbal heads bear relative tones. By relative tones, I mean grammatically and lexically constrained tonal realization. Tones in Èdó have contrastive functions and may serve to distinguish meaning in minimal contrastive (MC) forms as depicted in the following table (Agheyisi 1990:18):
{| border="5" cellspacing="5" cellpadding="2"
|-
! MC-1 lexeme 1
! MC-1 lexeme 2
! MC-2 lexeme 1
! MC-2 lexeme 2
! MC-3 lexeme 1
! MC-3 lexeme 2
! MC-4 lexeme 1
! MC-4 lexeme 2
|-
| <phrase>8178</phrase>
| <phrase>8177</phrase>
| <phrase>8175</phrase>
| <phrase>8176</phrase>
| <phrase>8172</phrase>
| <phrase>8171</phrase>
|<phrase>8174</phrase>
| <phrase>8173</phrase>
|}
In addition to the two basic tones above, Èdó language has two other tones that are derived. The falling tone and the downstepped high tone. The falling tone is derived when a high tone precedes a low tone as in Ódò [ódô] ''Potash'' and Áwà [áwâ] ''hour'' in the examples above. 
In Èdó due to the open syllable pattern of the language when two vowels occur adjacent across word boundaries contraction takes place and tone spreading and tone simplification processes apply. The downstepped high tone is derived when a word ending with a high tone bearing syllable combines with a noun with an initial low tone bearing syllable across word boundaries. In summary, the downstepped high tone is created through the processes of vowel assimilation and contraction (Agheyisi 1990)/vowel elision (Omozuwa 1992), tone spreading and tone simplification. The following example illustrates this:

/òwá/ 'house'+ / èbé / 'book'   [òwá!bé]

<phrase>8180</phrase>

In annotating Èdó data, the following glosses have been incorporated into the set of  glossing tags to account for tonal features in the language.

'''Glossing tag'''	'''Tag description'''

H			        high tone 

!H			        downstep high

L			        low tone




'''''Tense'''''


The glossing tags H, !H and L discussed above play a significant role in tense distinction. In addition, a gloss tag RT (Relative Tone) was introduced to account for the past tense suffix ''–rV''.

Tense is marked on the first verbal element after the subject NP, that is, the AUX/ADV if present in a clause or otherwise on the verb.
It may be realized as tones: past tense high(       ) and present tense low(       ) or high
(       ), a suffix: past ''–rV'', or a lexical item: future tense ''ghá''.  Tone marking on verbs encodes tense and transitivity information. For CV intransitive verbs, present tense is marked as a high tone and for transitive verbs it is marked as a low tone.

v-intr-suNom_Ag- ACTIVITY 
<Phrase>8291</Phrase>

v-tr-suNom_ag_obThincrem-NONCOMPLETED_MONODEVMT/HAB 
<Phrase>8284</Phrase>

For CV transitive verbs with direct objects realized in canonical positions, past tense is marked as a high tone.

v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT
<Phrase>8285</Phrase>

v-tr-suAg_obAcc_Aff-ACCOMPL- AFFECT
<Phrase>8296</Phrase>

The past suffix attaches to intransitive verbs and transitive verbs when their objects are not realized or are focalized in non-canonical positions. The suffix has seven allomorphs -re (dẹ́-rè ''"bought"''), -ri (fí-rì ''"threw"''), -ru (sún-rùn "smooth"), -ro (ròró-rò ''"thought"''), 
-rin (hín-rìn ''"climbed"''), -run (súnrùn'' "smooth"'') and ren (gbẹ́n-rẹ̀n ''"wrote"''). It harmonizes with the final vowel of the verb stem it is affixed to.

v-intr-suNom_ag_ACTIVITY 
<Phrase>8293</Phrase>

v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT
<Phrase>8286</Phrase>

v-tr-suAg_obAcc_Aff-ACCOMPL- AFFECT
<Phrase>8297</Phrase>


For transitive CVCV verbs marking past time, the final syllable obligatorily has a high tone while the initial syllable may be high or low. This also applies to the intransitive CVCV verb in the past and in addition the ''-rV'' suffix is attached to the verb stem. Transitive CVCV verbs marking present tense have a sequence of low low tone marking while intransitive verbs have low high tone marking. I have adopted the convention of tone marking the final syllable only.
Past tense in some preverbal CVCV adverbs is marked as a high-downstepped-high tone and a low-high pattern in the present.The following examples illustrates this. 
 
v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT
<Phrase>8283</Phrase>

v-tr-suNom_ag_obThincrem-NONCOMPLETED_MONODEVMT
<Phrase>8282</Phrase>

'''''Verbs, number and iteration representation'''''

Most verbs in Èdó can be inflected for the purpose of pluralizing nouns that occur with them or to mark repeated action.The verb stem to which a plural suffix attaches always bears a low tone. In addition, the last vowel on the verb root determines the form of the vowel on the plural suffix. There are six allomorphs of the plural suffix -lẹ (dẹ̀-lẹ ''"buy iterated"'' ), -le (sè-le ''"sew iterated"''), -lo (fì-lo ''"throw iterated"''), -lọ(tà-lọ ''"talk iterated"''), -nẹ (gbẹ̀n-nẹ ''"write-iterated"'') and -nọ (tìn-nọ ''"fly iterated"'').When the oral alveolar lateral consonant [ l ] in -lẹ and -lọ occurs in nasal context, it changes to the nasal alveolar consonant [ n ] as in   ''  gbẹ̀n-nẹ'' and ''tìn-nọ''. Verbs ending with the oral vowels [ Ɔ ] and [ a ] take the 
''- lọ'' suffix. Those ending with [ i ], [ u ],[ o ] take the ''-lo'' suffix, those ending with 
[ e ] take the ''-le''suffix and those ending with [ ε ] take the ''-lẹ'' suffix. Verbs ending with the nasal vowels [ ĩ ],[ ũ ],[ Ɔ͠ ],[ ã ] take the '' -nọ'' suffix and those ending with [ ε̴ ] take the '' -nẹ'' suffix (Ogie 1987:44-45).
    
For transitive verbs with a participant bearing the grammatical function of the direct object, the direct object is interpreted as plural when the verb has plural suffixation. In addition, the event may be interpreted as iterative.

v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT 
<Phrase>8285</Phrase>

v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT
<Phrase>8288</Phrase>

v-tr-suNom_ag_obThincrem-NONCOMPLETED_MONODEVMT 
<Phrase>8294</Phrase>


For verbs with only one argument which bears the subject grammatical function, the subject is interpreted as plural in the presence of plural suffixation on the verb and with the exception of accusative verbs,the event depicted by the verb may be interpreted as iterated event.

v-intr-suNom_ag_ACTIVITY
<Phrase>8290</Phrase> 

v-intr-suNom_ag_ACTIVITY 
<Phrase>8292</Phrase>



'''''Order of plural and past suffixes affixation'''''

Intransitive verbs and transitive verbs with focalized objects that encode past tense and plural information have the following order of affixation: Verbs stem + Plural suffix + past tense suffix.

v-tr-suNom_ag_obThincrem-COMPLETED_MONODEVMT
<Phrase>8289</Phrase>


'''''Noun and number representation'''''

In addition to interpreting number representation of nouns from verb morphology as discussed above, count nouns in Èdó may be modified by specifier premodifiers and quantifier and numeral post modifiers to encode number representation. Mass nouns may be modified only by quantifiers while proper nouns do not take number modification (Ogie 1987:49-60).The following examples of count nouns illustrate this:

v-intr-suTh-STATIVE-LOCATION
<Phrase>7995</Phrase>

v-intr-Th-STATIVE-LOCATION
<Phrase>8507</Phrase>

v-intr-Th-STATIVE-LOCATION
<Phrase>7996</Phrase>

v-intr-Th-STATIVE-LOCATION
<Phrase>8154</Phrase>

Nouns in Èdó generally do not inflect for number or case. However, a sub-set of nouns designating humans may inflect lexically to represent number. The strategy employed is alternation of the nouns initial vowels. The nouns in the singular form fall into two categories: those with ''ọ'' as initial vowel and those with ''o'' as initial vowel (Agheyisi 1990:38-39).

v-intr-Th-STATIVE-LOCATION
<Phrase>7999</Phrase>

v-intr-Th-STATIVE-LOCATION 
<Phrase>8000</Phrase>


v-intr-Th-STATIVE-LOCATION
<Phrase>8001</Phrase>


v-intr-Th-STATIVE-LOCATION
<Phrase>8002</Phrase>

The inflection of the human count nouns for number is not a random one. The count nouns that take the back half close vowel [ o ] in the singular form changes to the front close vowel [ i ] in the plural form while those that take the back half open vowel  [ Ɔ ] changes to the front half close vowel [ e ].The shifting of the vowels is from back to front. The singular nouns take the back vowels in the initial position and the plural nouns take the front vowels. The inflection of these nouns for number suggest that sometime in the distant past,Èdó must have had a noun class system like that of the Bantu and maybe, due to the innovation of the language this class system was lost 
(Ogie 1987:53-55).  

I have given an overview of the syllabic structure, tone, tense and number representation on verbs and nouns in Èdó. I now discuss immediately word formation processes.


'''''WORD  FORMATION PROCESSES'''''

There are two word categories in Èdó: content and function words. Content words consist of nouns, verbs, adjectives and adverbs while function words consist of determiners, prepositions, pronouns and complementizers. Nouns and verbs make up more than 80% of the content words in Èdó (Agheyisi 1990:29). Three strategies are used in word formation processes: affixation, compounding and reduplication. Content words may or may not be derived while function words generally consist of basic words.In the following I discuss first content words and then function words. 


'''''Derivational processes in content words'''''

''Nouns''

Nouns are formed from verb stems (basic or derived) through the processes of affixation and compounding. First I discuss Affixation and then compounding. 


Three affixation strategies are employed in noun derivation:prefixation, suffixation and circumfixation.

I now discuss prefixation.The following examples are from Agheyisi (1990:32-33) and Omoruyi (1986:17-18)and illustrates prefixation strategy for noun derivation in Èdó.

<phrase>8451</phrase>
<phrase>8450</phrase>
<phrase>8449</phrase>
<phrase>8447</phrase>
<phrase>8448</phrase>
<phrase>8446</phrase>
<phrase>8445</phrase>

The seven oral vowels above can function as nominal prefixes. While there is generally no phonetic restriction as to the affixation of these vowels to particular verb stems, they have restrictions in terms of their semantic function when attached to verb stems. Prefixation of the vowels ''i'' , ''e'','' o'' and ''ẹ'' to verb stems have as output abstract nouns, prefixation of the vowel'' u''  yields as output instrumental nouns while prefixation of the vowel ''a'' and ''ọ'' yields as output agentive nouns.

Turning now to suffixation, a suffix ''mwèn'' is affixed to a clausal stem consisting of a subject noun + verb construction to derive nouns.  The following examples from Agheyisi (1990:33-34) illustrate this:

<phrase>8513</phrase>
<phrase>8512</phrase>
<phrase>8511</phrase>

The output noun usually expresses idiomatic or metaphoric meanings. Crucially, the suffixation process triggers tonal change in the clausal stem as follows: 

(i)	Low tones on the subject noun.

(ii)	High tones on the verbs.


I now discuss the process of circumfixation.Gerunds are formed by the circumfixation of the circumfix ''Ù-mwẹ̀n '' to a verb stem. This is a phenomenon found also in the following Edoid languages: Dẹgẹma, Uvbiẹ, Boko, Yẹkhee and Ẹmhalhe (Elugbe 1984. cf. Omoruyi 1986:25). The following examples are from Omoruyi (1986:25-26).

<phrase>8517</phrase>
<phrase>8514</phrase>
<phrase>8515</phrase>
<phrase>8516</phrase>
<phrase>8520</phrase>


The derived gerund has a fixed tonal pattern as seen in the above examples. The circumfix ''Ú-mwẹ̀n'' retains it’s tone while the verb stem has a high tone.

Most verbs in Ẹ̀dó are capable of forming gerunds and this is crucial for predicate cleft where by the clefted verb yields a gerundive or nominal in the specifier position of the CP ( Agheyisi 1990, Stewart 1998, Ogie 2009). The following example illustrates this:

v-intrImplobj-su_Ag- ACHVMNT 
<Phrase>8519</Phrase>

v-intr-su_ag-ACTIVITY  
<Phrase>8521</Phrase>

v-intr-su_ag-ACTIVITY 
<Phrase>8522</Phrase>

Also derived nominals may function as cognate objects that serve to further qualify the verbs they occur with (Ogie 1987:32). The following examples illustrate this.

v-intr-su_ag-ACTIVITY
<phrase>8523</phrase>

v-intr-su_ag-ACTIVITY
<phrase>8524</phrase>

[[References for research on Èdó language by Ota Ogie]]