




































American International Journal of Education and Linguistics Research  

Vol. 1, No. 1; 2018 

Published by American Center of Science and Education, USA 

 

1 

 

 

Optimality Analysis of Vowel Harmony in ÀÍKA 
 

 

Ibikunle Abiodun Samuel 

Assistant Lecturer 

Department of Linguistics and Languages AAUA, Nigeria 

Adekunle Ajasin University, Akungba (AAUA) Ondo State Nigeria 

Email address: abiodun.ibikunle@aaua.edu.ng 

Phone numbers:+2348130585989/+2348053853645 

 

 

 

Received: November 9, 2018      Accepted: November 13, 2018             Online Published: November 26, 2018        

 

Abstract 

This paper gives an Optimality Theory (Henceforth OT) account of advanced tongue root (ATR) vowel harmony in 

ÀÍKA
1
, an Edoid

2
 language that consists of four speech forms spoken in Akoko-Edo area in Nigeria. The ATR 

harmony manifests within as well as across morpheme boundaries. The ATR harmony across morphemes affects the 

subject pronouns, prefixes as well as demonstrative pronouns because they are underspecified for ATR value while 

object pronouns are underlyingly specified. It is further noted that ATR has a morphological effect on the items it 

affects as it triggers phonological allomorphy in them. In addition to right-to-left spreading analysis in the literature 

(Abiodun 1999, Ibikunle 2014, and 2016), this research further reveals that there are pieces of evidence for left-to-

right spreading of harmonic value. More importantly, this analysis shows that OT is viable and problem-solving 

efficient compared with the Non-Linear or traditional generative account on Vowel Harmony system of the 

language.  

Keywords: Vowel harmony, ATR, Optimality theory, ÀÍKA, Pronouns. 

 

1. Introduction 

ÀÍKA is an Edoid language spoken in four different villages namely: ‘Ìígáú’ spoken in Àúga, Ìísheù spoken in Íshè, 

Ìkààn spoken in Kákùmo ̩̀  and ‘Ìyínno ̩̀ ’ spoken in Ayánrán. Ìkààn and Ìyínno ̩̀  are spoken in Akoko-Edo area in Edo 

State while Ìígáú and Ìísheù are spoken in Ondo State all in Nigeria.    

The motivation for this research paper is to show that ÀÍKA does not only operate a leftward but also a rightward 

movement in its ATR spreading technique. This is evident when someone considers the ATR harmony relationship 

between the plural demonstrative pronouns ‘these/those’ and the vowels of the preceding nominal item in the 

language. The vowel harmony of the language therefore supports the claim that ATR harmony is one of the features 

licensed to have a bidirectional spreading. 

This rightward spreading of the ATR harmony in question will pose difficulties for Autosegmental theory to analyze 

as shown in section 4.3. This being the case, the OT which is viable to account for the rightward spreading is 

employed in the analysis.   

   

2. Review of Relevant Literatures 

At the undergraduate, masters as well as Ph.D. levels, there have been some few works on Phonology of the 

language. The two known Ph.D. works on the language: Abiò̩ dun (1999) and Salffner (2009) are carried out using A 

Non-Linear (Autosegmental) Approach where both of them did not focus on the vowel harmony system of the 

                                                           
1
 This language is also known as Ùkààn/Ìkààn in the literature. (see Arohunmolase et al 2006, Abiodun 1999, 

Salffner 2009 among others) 

2 Arohunmolase et al (2006) classify the language under Benue Congo family alongside Akpes, Edoid, Defoid 

among others while Elugbe 2001 classifies it as the oldest branch of Edoid, more closely related to Akpes than to 

other Edoid languages. 

 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

2 

 

 

language. Abiodun’s work was on the comparative phonology and morphology of the language while Salffner’s 

work was purely on tone in the Phonology, lexicon and grammar of the language.  

For the vowel harmony in ÀÍKA, Abiodun (1999:87) says that ‘vowels within a phonological word agree with 

respect to the feature [ATR]…’ The language data he examined consist of subject pronouns, singular and plural 

prefixes and concord prefixes. Abiodun states that: ‘…subject pronouns, particularly singular forms, show 

allomorphic variation in all the dialects of Ukaan. The variation is normally determined by the [ATR] feature of the 

vowels in harmonic roots’. This is the case of leftward spreading of ATR while this paper reveals evidence of 

rightward spreading. 

 

According to Salffner (2009:55), the vowels of the language are distinguished by height and fronting as well as 

contrastive ATR value, length and nasality…the only gaps so far are /e͂:, ʊ͂:/. This research does not show any 

evidence of long vowels in the language; I considered those she transcribed as long vowels to be two similar but 

discreet vowels with discreet tones that occupy two different syllable peaks in sequence. 

Ibikunle (2014 and 2016) proposes a right-to-left analysis for the harmony system in the language. This research 

sees beyond this as the harmonic relationship between a noun and the plural demonstrative pronouns ‘these/those’ 

reveals a left-to-right spreading of ATR value. This later discovery supports the claim that ATR harmony is 

bidirectional in its spreading techniques. 

3. The ATR Harmony Process 

Many researchers have expressed their views on vowel harmony as a phonological process in human languages. 

Kenstowicz (1994: 347) captures the view of such researchers thus: 

 

"Vowel harmony is a phonological state in which the vowels in a given domain share or harmonize for a particular 

feature…typically all of the vowels of the language participate in the harmonic constraint. In addition, the harmony 

applies in an essentially unbounded fashion, affecting all the relevant vowels within the domain. Virtually any of the 

common features used to distinguish among vowels have been discovered to seat a harmonic system, including 

vowel height, backness, rounding, nasality and pharyngeal opening or [ATR]. 

 

Simply put, vowel harmony makes the vowels in a specific domain to sound alike. It is a kind of assimilation which 

requires that vowels in a certain domain should share some phonetic features such as vowel height, backness, 

rounding, nasality and pharyngeal opening or [ATR]. 

Research shows that, vowel harmony is a common phonological phenomenon in many languages in Africa while 

ATR is a common course. The tongue root is the principal actor in ATR harmony. The root of the tongue may 

project forward to expand the pharyngeal cavity (Agoyi 2008) 

3.1 Vowel Inventory of ÀÍKA 

The vowel inventory for AIKA includes contrast for three heights (high, mid, and low), two horizontal positions 

(front and back), and advanced-retracted. Advancement is the feature central to AIKA vowel harmony, and the 

table below is organized to convey that. 

Table 1   

[-ATR] Front  Central  Back  [+ATR] Front Central Back 

High ɪ    ʊ  High i  u 

Mid 

Low 

ɛ   

 

a 

 ɔ  Mid 

Low 

e  

 

 

o 

 

All of these 9 vowels are contrastive and they all have nasal counterparts. 

 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

3 

 

 

3.2 Vowel Under specification 

Under specification is good to use when some features of the surface segment is always determined by another 

segment. If it depends on something else, it has no identity of its own. In this case, it is assumed that the vowels of 

the subject pronouns, prefixes as well as demonstrative pronouns are underspecified for ATR in the language. 

Their ATR value is determined by the roots in their environment. An underspecified segment is writing with a 

technical symbol called an archiphoneme. This is demonstrated in section 4. 

4. The Data 

The data in the next two subsections show vowel harmony within as well as across morpheme boundaries 

4.1  Lexical Vowel Harmony 

The assumption here is that the last vowel in a nominal root is specified for ATR value but it only spread on the 

underspecified preceding vowels, affixes, subject pronouns or demonstrative pronouns around it. 

[ATR] Harmony within morphemes can be shown thus: 

Ìyínno ̩́  

1. SET A [+ATR]    SET B [–ATR] 

/Èhú/ =[èhú]  ‘rat’  /Ènɔ / =[ɛ nɔ ]  ‘meat’ 

/Ìhwe/ =[ìhwé] ‘seed’  /Àtɔ / =[àtɔ ]  ‘lie’ 

    /Òjĩ/ =[òji   ]  ‘child’  /Àʃʊ/ =[àʃʊ]  ‘cloth’ 

/Ùwé/ =[ùwè]  ‘leg’  /Ùfɪ / =[ʊ fɪ ]  ‘thing’ 

/Òjù/ =[òjù]  ‘wind’  /Òʃɛ / =[ɔ ʃε  ]  ‘mat’    

 

The data above reveal a harmony agreement between all the vowels in the lexical root as shown on the right. I 

assume that the second vowel predicts or determines the ATR value of the first (i.e. V2 is the trigger) that is why the 

first vowels are represented with archiphonemes at the left to show that they are underspecified. This assumption is 

based on two reasons: first, the language operates no suffixes but prefixes whose harmonic value is determined by 

the value of the root vowel. The second reason being that majority of the spreading techniques of the vowel 

harmony in the language are leftward with only one instance of rightward spreading. 

4.2 Harmony Across Boundaries 

ATR harmony affects the subject pronouns and the following lexical item such as verbs. This kind of harmony does 

not exist between the object pronouns and the preceding lexical item. This is to say that the directionality of the 

spreading is leftward (right-to-left) rather than rightward (left-to-right) in this regard.  

 

The data in 2a below illustrate how the subject pronouns are realized in sentences. 

 

  Ìígáú 

2. a [-ATR]     [+ATR] 

[ʤɛ  jáfí ɔ hɔ ]   ‘I licked up the soap’ [ʤe rúàg ùmó]   ‘I poured the water’ 

I  lick  soap     I   pour  water 

  

[hɛ   jɔ wɪ  lɛ bɔ]   ‘you called us’  [hè ʃéʤì ìh
w
é]   ‘you stole the money’ 

you call us     you steal money 

 

[ɔ   ɲàgɛ  awu]   ‘he tore the cloth’  [ò víè oʤòʤ]   ‘he descended from a rock’ 

he tear cloth     he descend rock 

 

[ɛ  jɔr séni]    ‘it flew away’  [è kũ lɛ bɔ]  ‘it tied us’ 

it fly far     it tie us 

 

This being the case, the ATR harmony has a morphological effect on the subject pronouns as they now have 

allomorphic variations. The subject pronouns are underspecified in the language as shown below: 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

4 

 

 

         Ikaan  Ìyínnó̩     Iigau   Ìsheu  Gloss 

2.b.(i) /ʤÈ/  /ʤÒ/   /ʤÈ/   /ʤÒ/ 

       [ʤè/ʤɛ ]            [ʤò/ʤɔ ]        [ʤè/ʤɛ ]          [ʤò/ʤɔ ]  I 

          (ii)  /hÈ/  /hÒ/   /hÈ/   /hÒ/  

       [ hè/hɛ ]           [hò/hɔ ]           [hè/hɛ ]            [hò/hɔ ]  you 

        (iii) /Ò/  /Ò/   /Ò/   /Ò/ 

       [ò/ɔ ]            [ò/ɔ ]            [ò/ɔ ]              [ò/ɔ ]  he 

        (iv)  /È/  /È/   /È/   /È/ 

         [è/ɛ ]  [è/ɛ ]   [è/ɛ ]   [è/ɛ ]  it 

 

4.3 The Plural Demonstrative Pronouns 

The data for these pronouns indicate an area where the language could spread its harmony in a rightward (left to 

right) direction. The data in 3a and b below indicate how the plural demonstrative pronouns are realized in noun 

phrases. 

Ìsheu     

3.a  [+ATR]     [-ATR] 

[èri    re]  ‘those elephants’  [àʃɛ   ra]  ‘those eggs’ 

[èhwí re] ‘those corpses’  [àgʊ  ra] ‘those thorns’ 

Iigau 

3.b [+ATR]     [-ATR] 

[eso rè] ‘those seeds’   [àkàʤɪ  rà] ‘those villages’ 

[èhú rè] ‘those days’   [àhʊ  rà]  ‘those arms’ 

The variants of the demonstrative pronouns is shown in 3c below 

Ìsheu    Iigau       

3.c  /rE/   /rE/    

[re/ra]   [re/ra]  ‘these/those’ 

Non-Linear/Autosegmental theory would fail in accounting for the final derivation of the plural demonstrative 

pronouns given in ‘3’ above. This is because, in Autosegmental analysis of this kinds of data, the spreading of [-

ATR] on the archiphoneme [rE] from the preceding syllable will rather generate *[rɛ] which is ill-formed in the 

language. The reason for this is simple. The archiphoneme [rE] could either be realized as [re] or [rɛ] after the 

spreading of an ATR value in Autosegmental analysis as in: 

 

3.d      [re] 

     /rE/ 

      *[rɛ]   



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

5 

 

 

The rule(s) that will further account for the derivation of [ra] from [rɛ] is not clear in Autosegmental analysis for the 

fact that the high vowels among others are not transparent in the language. This difficulty is unarguably resolved 

with the use of constraints in OT. We will explain how OT resolves this problem in section 6 using five constraints 

that includes the markedness constraint *[MID,-ATR] which states that “A mid vowel must not be produced with 

retracted tongue root”. 

5. Optimality Theory: An Overview 

It had been observed, since the early days of Generative Phonology, that many phonological rules were born out of 

the desire by Language to maintain some phonological constraints in a language. 

Optimality theory is the current standard phonological model. OT was formally proposed in the early 1990s with 

the work of Alan Prince, John McCarthy, and Paul Smolensky. In OT, all languages share the same set of 

aspirations. This set is finite and universal, but includes contradictions. It is the constraints and the ways they 

conflict that shapes language outputs. They form the basis of sound patterns. 

 

There are two major classes of constraints that pull the output in opposite ways: faithfulness constraints and 

markedness constraints. Faithfulness constraints make some comparison between the output and the input. For a 

faithfulness constraint to be satisfied, the output will have to resemble the input in some meaningful way. 

Markedness constraints demand that an output be easy to pronounce. Marked sounds and combinations of sounds 

are the less common ones and the harder ones to pronounce. For a markedness constraint to be satisfied, the output 

will need to have some property that makes it easy on the speaker. 

 

Faithfulness constraints are good for listeners. The more direct information they have about the input, the less work 

they have to do to form an interpretation of the sentence. Markedness constraints are good for speakers. The easier 

the phonetic sounds of the output are to say, the less work a speaker’s mouth and other articulators will have to do. 

In this way OT captures an intuitive fact about the social and communicative nature of language: phonology is a 

compromise between the priorities of the speaker and the listener. 

 

Each language resolves this universal conflict between markedness and faithfulness in a structured way. An 

infinite number of possible outputs, or candidates, will be graded on the constraints usually on a pass-fail basis. 

Since it is impossible to satisfy all the constraints, every language has its list of priorities. The candidate that 

performs the best wins and is the optimal output, hence the name Optimality theory. Even a winning candidate will 

have problems. However, the phonology of a language must dictate which of these problems it can ignore and 

which will be deal breakers. In practice, only the most conservative/compelling candidates and the constraints 

relevant for the problem at hand will be shown on an Optimality tableau. 

 

In Tableau 1, two imaginary outputs are evaluated on the imaginary constraint X. An asterisk marks each time A 

and B violate X. (In this case, A violates X zero times and B violates X one time) 

 

Tableau 1: Two outputs A and B evaluated by Constraint X 

/input/ Constraint X 

a. [output A]  

b. [output B] * 

 

In Tableau 2, the same imaginary outputs are evaluated on another imaginary constraint Y. An asterisk marks each 

time A and B violate Y. 

 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

6 

 

 

/input/ Constraint X Constraint Y 

☞ a. [output A]  * 

b. [output B]            *!   

 

/input/ Constraint Y Constraint X 

a. [output A]                               * !  

☞ b. [output B] 

 

 

 * 

 

 Tableau 2: Two outputs A and B evaluated by Constraint Y 

/input/ Constraint Y 

a. [output A]                         * 

b. [output B]  

 

In the next tableau 3, both constraints are laid side by side. If X is more important than Y (if X outranks Y, 

abbreviated X ≫ Y) then candidate A will beat B. A language like this will pronounce the input as output A. 

This fact is indicated by a hand that points to the winning output. When candidates lose, their downfall is 

marked with an exclamation point. 

 Tableaux 3:. How OT chooses the best output if X ≫ Y 

 

 

 

 

 

 

Another language might have the reverse set of priorities. With the reverse ranking, Y is more highly 

prioritized than X and it outranks X (Y ≫ X). A speaker of a language like this would rather pronounce the input as 

output B, as shown in Tableau 4: 

Tableau 4: How OT chooses the best output if Y ≫ X 

 

 

 

  

 

 

 

 

5.1   Relevant Constraints for the Vowel Harmony Analysis  

 

The relevant constraints for this analysis are selected from the literature. They are presented below: 

IDENT-IO (ATR)  Corresponding vowels in the input and output have 

  identical specifications for ATR. (McCarthy 1997,  

  Ballard 2010) 

 SPECIFY  All segments must be fully specified for features  

  (Ballard 2010) 

ROOT-AGREE[ATR] All the vowels in a root must share the same ATR value  

  (Williams 2009) 

ALIGN[ATR]-L/R  Align the left or right edge of the [ATR] element with  

  the Left or right edge of the prosodic word 

  (Oyebade 2008, Agoyi 2008, Williams 2009) 

*[MID,-ATR]  A mid vowel must not be produced with retracted   

         tongue root (Akinlabi 2009, Gambarage & Pulleyblank 2014) 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

7 

 

 

In order to define the grammar of the language, the constraints above are ranked as follows: 

 

SPECIFY>>ROOT-AGREE[ATR]>>[ATR]-L/R>>*[MID,-ATR]>> IDENT-IO(ATR). 

 

6. Analysis of Vowel Harmony in ÀÍKA using OT Tableau 
Tableau 5:. How OT chooses the best candidate in lexical vowel harmony 

/Èhú/ SPECIFY ROOT-AGREE[ATR [ATR]-L *[MID,-ATR] IDENT-IO(ATR) 

a. ɛhu              *!       *       *           ** 

b.Ehʊ       *!             *       *                    * 

c. ehʊ              *!       *                  ** 

d. ɛhʊ                    *!            ** 

e. EhU      *!*                   *                    ** 

☞f. ehu                      ** 

The optimal candidate or the correct output is [ehu] ‘rat’ in Ìyínno ̩̀  

 

Tableau 6: How OT chooses the best candidate in leftward spreading of ATR 

/E kũ lɛbɔ/ SPECIFY ROOT-AGREE[ATR [ATR]-L *[MID,-ATR] IDENT-IO(ATR) 

a. e kũ lebɔ              *!              *           ** 

b. ɛ kũ lebo                          *!            *           *** 

☞c.e kũ lɛbɔ                           **             * 

d. E kũ lɛbɔ      *!       *       **             

The optimal candidate is [e kũ lɛbɔ] ‘It tied us’ in Ìígáú 

 

Tableau 7: How OT chooses the best candidate in rightward spreading of ATR 

/Aʃɛ rE/ SPECIFY ROOT-AGREE[ATR [ATR]-R *[MID,-ATR] IDENT-IO(ATR) 

a. aʃe ra              *!       *                   *** 

☞b. aʃɛ ra                *              ** 

c. aʃɛ re         *!       *             ** 

d. Aʃɛ rE      *!*        *       *  

e. aʃe re              *!              *** 

f. aʃɛ rɛ              **!             ** 

g. aʃe rɛ              *!        *      *             *** 

h. aʃɔ rɛ                                **!             *** 

The optimal candidate is [aʃɛ ra] ‘these/those eggs’ in Ísheu 



www.acseusa.org/journal/index.php/aijelr            American International Journal of Education and Linguistics Research           Vol. 1, No. 1; 2018 

 

8 

 

 

The derivation of this output poses a threat to Auto segmental theory of phonology while it is easily resolved in OT 

with relevant constraints. 

 

7. Conclusion 

This research shows that ÀÍKA operates ATR harmony within as well as across morpheme boundaries. It further 

reveals that there is a piece of evidence for left-to-right spreading of harmonic value in addition to right-to-left 

spreading analysis in the literature on the language. Five relevant constraints are used where one of them is 

faithfulness while the remaining four are markedness constraints. More importantly, the paper shows that OT is 

viable and problem-solving efficient compared with the Non-Linear or traditional generative account on Vowel 

Harmony system of the language. 

 

References 

Abiodun, M. (1999). “A Comparative Phonology and Morphology of Ùkààn dialects of Old Akoko Division”. 

Ph.D. thesis, University  of Ilorin Nigeria. 

Agoyi, (2008). “The Phonology of Vowel Harmony in Àbèsàbèsì”. Ph.D thesis, Akungba: AAUA. 

Akinlabi, A. (2009). Neutral Vowels in Lokaa Harmony. Canadian Journal of Linguistics: 197-228. 

Ballard, L. (2010). “Akan Vowel Harmony in Optimality Theory”. M.A. thesis. Florida: University of Florida.  

Elugbe, B. (2001). “The Classification of Akpes and ÀÍKA” (Unpublished). 

Gambarage, J and D. Pulleyblank (2014). Vowel Harmony in Nata and Ikoma. UC Berkeley. 

Ibikunle, A. (2014). "An Autosegmental Phonology of Ìyínno ̩̀  Dialect of ÀÍKA". M.A. thesis. Ibadan: 

University of Ibadan. 

Ibikunle A. (2016). The Role of ‘No Crossing’ in the Vowel Harmony System of Ìyínno ̩́ . Journal of  Linguistics 

Association of Nigeria (JOLAN) Nigeria19 (1) 

McCarthy, J and A. Prince (1993). Generalized Alignment: Technical Report no.3 of the Rutgers Center for 

Cognitive Science, New Jersey 

McCarthy, J (1997). Process-specific Constraints in Optimality Theory. Amherst: UMass. 

Oyebade, F. (2008). A Course in Phonology. Ijebu-Ode: Sebiotimo Publications. 

Prince, A and Smolensky, P (1993). Optimality Theory: Constraint Interaction in Generative Grammar. Ms., 

Rutgers University, New Brunswick and University of Colorado, Boulder. 

Salffner, S. (2009). Tone in the Phonology, Lexicon and Grammar of Ìkààn. PhD Dissertation.London: University 

of London. 

Sasa, T. (2009). ‘Treatment of Vowel Harmony in Optimality Theory’. Ph.D Dissertation. Iowa University of Iowa.  

Williams, A. (2009). An Optimality Theory Analysis of Vowel Harmony in Ndura.GIAL: Occasional Papers 

 in Applied Linguistic 

 

 

Copyrights  

Copyright for this article is retained by the author(s), with first publication rights granted to the journal. This is an 

open-access article distributed under the terms and conditions of the Creative Commons Attribution license 

(http://creativecommons.org/licenses/by/4.0/). 

 

 


