A scholarly examination of the theoretical debate in Yorùbá phonology between derivational tonal underspecification and constraint-based markedness faithfulness within Optimality Theory.
Yorùbá phonology displays a pronounced structural asymmetry among its three level register tones: High (H, òkè), Mid (M, àárín), and Low (L, ìsàlẹ̀) [S1, S2, S3]. While High and Low tones actively spread across morpheme boundaries, create dynamic tonal contours, and survive vowel elision, the Mid tone exhibits systematic inertness, failing to trigger phonological processes and disappearing under contact [S3, S4].
Linguists have explained this asymmetric behavior through two competing theoretical frameworks: derivational Tonal Underspecification, which treats the Mid tone as underlyingly absent and supplied by late default insertion [S3, S4], and Optimality-Theoretic Markedness Faithfulness, which fully specifies all tones underlyingly and derives their asymmetric preservation through harmonic scales and ranked faithfulness constraints [S5, S6].
This file examines the empirical evidence supporting both models, details their formal architectures, analyzes their treatment of vowel elision and tone insertion, reviews the historical evolution of the debate, and presents phonological paradigms with interlinear glosses and explicit analytical commentary.
Standard Yorùbá contrasts three level register tones [S1, S2, S3]:
´).`).In rule-based and autosegmental frameworks, tonal processes are expected to treat contrastive tones symmetrically unless specific structural conditions intervene [S4, S7]. In Yorùbá, however, empirical data from speech production, cliticization, compounding, and vowel hiatus reveal a scalar behavioral hierarchy [S3, S5]:
This three-way behavioral split creates what phonologists call the asymmetry problem [S4, S5]. The core theoretical question is whether the inactivity of the Mid tone results from its total absence in the underlying lexicon (underspecification) or from its status as the least marked element within an articulated constraint hierarchy (markedness faithfulness) [S3, S5, S6].
The classical model of Yorùbá tonal phonology was formulated within Autosegmental Phonology and Radical Underspecification Theory by Akinbiyi Akinlabi and Douglas Pulleyblank .
Under Radical Underspecification, underlying representations store only non-redundant, non-predictable phonological features [S3, S4, S9]. For a three-tone system like Yorùbá, underspecification theory posits that only two tonal values are present in underlying representation, or that binary tonal features are assigned asymmetrically [S3, S4]:
/H/ autosegment [S3, S4]./L/ autosegment [S3, S4].[ ] [S3, S4].Underlying Representation:
H L (none)
| | |
/ CV / / CV / / CV /
(High) (Low) (Mid)
Because Mid-toned vowels lack an underlying tonal node on the tonal tier, they cannot spread, cannot trigger the Obligatory Contour Principle (OCP), and do not block the interaction of adjacent specified tones [S3, S4, S7].
In this derivational framework, phonological rules operate exclusively on specified autosegments /H/ and /L/ [S3, S4]. At the end of the phonological derivation, any tone-bearing unit (TBU, èyà-ohùn) that remains unassociated receives a Mid tone through a language-specific Default Rule [S3, S4]:
$$\emptyset \rightarrow \text{M} \quad / \quad [\quad]_{\text{TBU}}$$
This rule guarantees that every surface syllable receives phonetic pitch interpretation without requiring the lexicon to store redundant Mid features [S3, S4, S8].
Akinlabi and Pulleyblank documented several empirical phenomena supporting the underlying absence of the Mid tone [S3, S4, S10]:
/H/ or /L/ tone [S3, S4].Beginning in the late 1990s and formalized in a pivotal 2004 paper, Douglas Pulleyblank re-evaluated the underspecification account, arguing that it suffered from theoretical and empirical deficiencies . Pulleyblank proposed that the asymmetries in Yorùbá tone are better explained within Optimality Theory (OT) through the interaction of universal markedness scales and tone-specific faithfulness constraints, building on frameworks developed by Paul de Lacy, Darin Howe, and Douglas Pulleyblank [S5, S6, S13].
Optimality Theory operates on the principle of Richness of the Base, which states that no constraints hold over underlying representations . Under Richness of the Base, a phonological theory cannot rely on the arbitrary absence of features in underlying representation to explain surface patterns, because the grammar must map any conceivable input (whether underspecified, fully specified, or redundantly specified) to the correct surface output [S5, S14].
Furthermore, Pulleyblank identified an empirical contradiction in the underspecification account, termed the epenthesis-versus-stability paradox [S5, S6]:
Under a standard markedness analysis without specialized faithfulness, the least marked tone (M) should be preferred everywhere, meaning that elision should favor preserving Mid over High or Low . Because natural language does the exact opposite, preserving High and Low at the expense of Mid, Pulleyblank demonstrated that tonal markedness must be directly encoded into faithfulness constraints [S5, S6].
In Optimality Theory, pitch levels are organized along a universal Harmonic Scale representing structural markedness [S5, S6]:
$$\text{Harmonic Scale: } \text{M} \succ \text{L} \succ \text{H}$$
This scale states that Mid tone is harmonic (least marked), Low tone is intermediate, and High tone is marked . Converted into standard markedness constraints, the ranking is strictly universal [S5, S6]:
$$*\text{H} \gg *\text{L} \gg *\text{M}$$
*H: Avoid High tone (most marked) .*L: Avoid Low tone .*M: Avoid Mid tone (least marked) .To explain why marked tones (H and L) survive deletion while unmarked tones (M) disappear, Pulleyblank adopted the Markedness as Faithfulness framework [S5, S6, S13]. Under this model, phonological grammars maintain higher faithfulness to phonologically marked features than to unmarked features [S5, S6, S13]:
$$\text{MAX}(\text{H}) \gg \text{MAX}(\text{L}) \gg \text{MAX}(\text{M})$$
MAX(H): Every High tone in the input must have a correspondent in the output (highest priority) [S5, S6].MAX(L): Every Low tone in the input must have a correspondent in the output [S5, S6].MAX(M): Every Mid tone in the input must have a correspondent in the output (lowest priority) [S5, S6].When vowel elision forces the loss of a tone-bearing mora, the constraint hierarchy $\text{MAX}(\text{H}) \gg \text{MAX}(\text{L}) \gg \text{MAX}(\text{M})$ guarantees that High and Low tones will be retained on the remaining mora, while Mid tone will be sacrificed without requiring Mid to be absent from the underlying representation .
Vowel hiatus resolution in verb-object compounds provides the definitive testing ground for comparing Underspecification and Markedness Faithfulness [S3, S4, S5, S11].
+-------------------+-------------------+----------------------+-------------------------+
| Construction | Morphemes | Surface Realization | English Gloss |
+-------------------+-------------------+----------------------+-------------------------+
| High + Mid | /wá/ + /ilé/ | [wálé] | look for a house |
| Mid + High | /jẹ/ + /ẹran/ | [jẹ́ran] | eat meat |
| Low + Mid | /kà/ + /ìwé/ | [kàwé] | read a book |
| Mid + Low | /wọ̀/ + /aṣọ/ | [wọṣọ̀] / [wọ̀ṣọ] | wear clothes |
| High + Low | /rí/ + /ọ̀bẹ/ | [rọ́bẹ̀] | see a knife |
+-------------------+-------------------+----------------------+-------------------------+
Consider the combination of the Mid-toned verb jẹ ("to eat") and the Mid-High noun ẹran ("meat") [S3, S4]:
/dʒɛ/ + /ɛran//dʒɛ/ has no tonal node [ ]. The noun /ɛran/ has no tone on the first syllable [ ] and a /H/ autosegment on the second syllable -ran [S3, S4]./ɛ/ [S4, S11]./dʒɛran/./ɛ/ remains tonally unassociated.Now consider a High verb and a Mid noun: wá ("to seek") + ilé ("house") [S3, S4]:
/wá/ carries /H/; /ilé/ carries [ ] on /i/ and /H/ on /lé/./a/ deletes. The floating /H/ tone from the verb links to the initial vowel /i/ of the noun [S3, S4].Under Optimality Theory, both inputs are fully specified :
Input: /wá/ (H) + /ilé/ (M-H)
+------------------------------------+------------+------------+------------+---------+
| Candidates | MAX(H) | MAX(L) | MAX(M) | *H |
+------------------------------------+------------+------------+------------+---------+
| a. [wā.lé] (M-H, delete H) | *! | | | * |
| b. ☞ [wá.lé] (H-H, delete M) | | | * | ** |
| c. [wà.lé] (L-H, insert L) | *! | | * | * |
+------------------------------------+------------+------------+------------+---------+
*H, but satisfies top-ranked $\text{MAX}(\text{H})$ .Input: /jẹ/ (M) + /ẹran/ (M-H)
In this configuration, where both interacting vowels carry Mid tone, $\text{MAX}(\text{H})$ and $\text{MAX}(\text{L})$ are inactive . The constraint $\text{MAX}(\text{M})$ decides between candidates, and the surviving vowel surfaces with Mid tone without requiring a derivational default insertion rule .
An enduring problem in Yorùbá tonology is the prohibition against High tones on word-initial vowels in indigenous nouns [S1, S3, S5, S15].
In standard Yorùbá, all indigenous nouns begin with a vowel prefix (or a syllabic nasal) [S1, S15]. These initial vowels surface exclusively with either a Mid tone (such as igbá, calabash; ọkọ, husband; ará, relative) or a Low tone (such as ìgbà, season; ọ̀wọ̀, respect; ìwé, book) [S1, S3]. Vowel-initial nouns carrying a High tone (such as áásì, acid, or áíkì, hiccup) occur strictly in loanwords borrowed from English or neighbouring languages [S1, S12].
+-------------------+-------------------------+---------------------------------+
| Phonological Type | Examples | Tone Structure of Initial Vowel |
+-------------------+-------------------------+---------------------------------+
| Indigenous Nouns | igbá, ọkọ, owó, ẹyẹ | Mid (unmarked) |
| Indigenous Nouns | ìgbà, ọ̀wọ̀, òwò, àlá | Low (marked with `) |
| Prohibited Form | *ígbá, *ọ́kọ, *ówó | High (systematically absent) |
| Loanwords | áásì, ágídí, bọ́ọ̀lù | High (permitted in borrowings) |
+-------------------+-------------------------+---------------------------------+
In the underspecification framework, Akinlabi and Pulleyblank explained this restriction by proposing a structural morpheme-structure condition [S3, S4]:
[ ] [S3, S4]./H/ to an initial nominal prefix, and because Mid is supplied to unassigned vowels by default insertion, indigenous nouns can surface only with initial Mid or Low tones [S3, S4].In the constraint-based framework, Pulleyblank demonstrated that positional markedness accounts for this distribution without ad-hoc morpheme-structure rules .
Word-initial vowels in Yorùbá lack an onset consonant [S5, S15]. Cross-linguistically, onsetless syllables occupy weak prosodic positions [S5, S6, S16]. By ranking a positional markedness constraint against High-toned onsetless syllables above general tonal faithfulness, the grammar excludes initial High tones in native vocabulary while allowing them in loans via high-ranked loanword faithfulness :
$$\text{COINCIDE}(\text{Onset}, \text{H}) \gg \text{MAX}(\text{H}) \gg *\text{H}$$
Under this constraint ranking, an underlying High tone on an onsetless initial vowel is forced to neutralize to Mid or Low, explaining the systematic gap in the native lexicon .
To evaluate this debate accurately, Yorùbá tonological phenomena must be distinguished across three independent descriptive registers: native speaker meta-knowledge, linguistic documentation, and colonial-era orthographic formulations.
Within Yorùbá oral and musical traditions, pitch levels are recognized as distinct expressive and communicative targets [S1, S8, S17]. In the verbal arts, such as ìjálá (hunters' chants), ẹsẹ̀ Ifá (divination poetry), and oríkì (praise poetry), speakers and chanters manipulate the three pitch registers with precision [S8, S17].
In the dùndún talking drum tradition, drummers squeeze the tension cords of the gángan drum to produce three distinct acoustic levels that directly emulate the High, Mid, and Low tones of spoken Yorùbá [S8, S17].
From the internal perspective of Yorùbá verbal art, the Mid tone is not perceived as an "absence" or "empty void", but as the baseline resting pitch (ohùn àárín) from which the voice rises to High (ohùn òkè) or descends to Low (ohùn ìsàlẹ̀) [S1, S17].
Academic phonologists across the twentieth and twenty-first centuries have established that while the Mid tone is phonetically real and serves as an acoustic reference level, it participates in fewer phonological alternations than High or Low [S1, S2, S3, S5, S8, S18].
Acoustic analyses conducted by Yetunde Laniran and George N. Clements confirmed that fundamental frequency ($F_0$) tracks of Mid tones exhibit flat, stable trajectories, whereas High tones trigger high-raising before Low tones, and Low tones trigger downstep across successive High tones [S8, S18].
The academic consensus confirms the empirical asymmetry of the three tones, even as theoretical explanations of that asymmetry remain actively contested [S3, S5, S19].
Early nineteenth-century missionary linguists struggled with the nature of Yorùbá tone [S1, S20]. In his 1843 and 1852 dictionaries and grammars, Bishop Samuel Ajayi Crowther recognized the existence of three vocal tones and devised the system of acute (´), grave (`), and unmarked vowels that forms modern standard Yorùbá orthography [S1, S20].
Crowther left the Mid tone unmarked primarily for typographic economy, to reduce the complexity of printing typesets and simplify handwriting [S1, S20]. Later colonial grammarians misinterpreted this orthographic convention as evidence that Yorùbá had only two "real" tones (High and Low) and that Mid was simply an unaccented English-like syllable [S20, S21].
Modern phonological theory cautions against conflating Crowther's practical orthographic choice with underlying phonological representations [S3, S5, S20].
The following oral texts illustrate the interaction of lexical tone, tonal stability, and vowel contact in performance.
Ìwà búburú ń ba orúkọ rere jẹ́;
Ẹni tó da omi síwájú á tẹlẹ̀ tútù.
Ìwà búburú ń ba orúkọ rere jẹ́;
Character evil PROG spoil name good eat-up;
Ẹni tó da omi sí-iwájú á tẹ-ilẹ̀ tútù.
Person REL-H pour water to-front FUT tread-ground cold.
Evil character destroys a good reputation; whoever pours water ahead will step on cool ground (translated by O. Awobuluyi and A. Bamgbose) [S1, S2].
A person's moral conduct determines their future reality. Destructive behavior erodes social standing, while constructive and generous actions ensure a peaceful and secure existence [S1, S2].
Used in elder discourse and family counseling as a serious moral admonition, warning younger members against reckless behavior and reminding them that current actions produce inescapable future consequences [S1, S2].
Spoken by a family elder during a dispute settlement where a youth has engaged in fraudulent business practices, warning the youth that damaged character cannot be repaired by financial gain [S1, S2].
Reinforces the foundational Yorùbá philosophical concept of ìwà (character), emphasizing that personal integrity (ìwà rere) outranks material wealth and social prestige [S1, S2].
Recorded in Ọ̀yọ́ dialects as Ẹni tó bá da omi síwájú ní ń tẹlẹ̀ tútù, inserting the conditional marker bá and the focus particle ní .
The phrase tẹ-ilẹ̀ ("tread ground") undergoes vowel elision where the Mid tone of the verb tẹ encounters the Mid-Low sequence of ilẹ̀. The initial Mid vowel of ilẹ̀ is elided, and the remaining vowel retains the Low tone of the root noun, yielding the compound tẹlẹ̀ (Mid-Low) [S1, S2, S3].
English terms like "evil" and "reputation" fail to capture the metaphysical weight of ìwà, which in Yorùbá thought represents the inner spiritual essence and moral alignment of the individual [S1, S2].
Ọ̀rúnmìlà, Bara Àgbọnnìrègún,
A-jẹ́-jù-oògùn, a-tọ́-fì-ara-tì-bí-òkè.
Ọ̀rúnmìlà, Bara Àgbọnnìrègún,
Ọ̀rúnmìlà, Great-Lord Àgbọnnìrègún,
A-jẹ́-jù-oògùn,
One-who-is-efficacious-surpassing-medicines,
a-tọ́-fì-ara-tì-bí-òkè.
one-sufficient-to-lean-body-against-like-mountain.
Ọ̀rúnmìlà, Master of Àgbọnnìrègún, who is more efficacious than physical medicines, a support as dependable as a mountain (translated by Wande Abimbola) .
Ọ̀rúnmìlà, the deity of wisdom and divination, provides spiritual stability and knowledge that surpasses the physical efficacy of herbs and healing charms .
Recited by babaláwo (diviners) as an invocatory praise poem (oríkì) at the beginning of an Ifá divination session to align the priest with divine wisdom .
Chanted during the Ìtẹ̀fá (Ifá initiation ritual) when the initiate approaches the sacred palm nuts (ikin Ifá) to receive their destiny odù .
Enforces the supremacy of spiritual insight and ethical alignment over mechanical reliance on protective charms .
In Èkìtì and Ìjẹ̀ṣà dialects, Bara is pronounced with archaic vowel lengthening as Bàbá-ara .
The epithet A-jẹ́-jù-oògùn contains a sequence of High tones (jẹ́, jù) clashing with the Low-High sequence of oògùn. The contraction preserves the High tone of jù and the Low-High pitch contour of oògùn, demonstrating tonal preservation across morphological compounding boundaries [S3, S5, S22].
The title Àgbọnnìrègún contains initiate-held etymological references that are not fully public; popular folk etymologies translate it as "the coconut tree that gives good shade", but liturgical tradition treats it as an esoteric divine name .
The table below summarizes the core differences between the two competing phonological frameworks in accounting for Yorùbá tonal phenomena [S3, S4, S5, S6, S13].
+---------------------------+----------------------------------+-------------------------------------+
| Theoretical Parameter | Tonal Underspecification (TU) | Markedness Faithfulness (OT-MF) |
+---------------------------+----------------------------------+-------------------------------------+
| Mid Tone Representation | Absent underlyingly: [ ] | Fully specified underlyingly: /M/ |
| Surface Mid Tone Source | Late default insertion rule | Faithful mapping via MAX(M) |
| Hiatus Tonal Survival | Floating H/L docks to empty vowel | High MAX(H)/MAX(L) outranks MAX(M) |
| Theoretical Framework | Derivational Autosegmental Rules | Non-derivational Constraint Ranking |
| Input Restrictions | Restricted by Morpheme Structure | Richness of the Base (unrestricted) |
| Epenthesis Mechanism | Inserted vowel is tonally empty | Markedness *H >> *L >> *M selects M |
| Non-Initial H in Nouns | Missing morphological H-prefix | Positional Markedness COINCIDE(O,H) |
| Dialectal Variation | Parameterized rule ordering | Constraint re-ranking |
+---------------------------+----------------------------------+-------------------------------------+
Both theoretical models account for the basic asymmetries of Yorùbá tonology, but each possesses distinct strengths and challenges [S3, S5, S19]:
The systematic study of Yorùbá tone began in the 1840s with Church Missionary Society linguists [S1, S20]. Bishop Samuel Ajayi Crowther's 1843 Grammar and Vocabulary of the Yoruba Language and his revised 1852 edition established the three-tone orthographic convention . Crowther recognized the lexical function of tone, using acute, grave, and unmarked vowels to differentiate words . However, missionary and colonial grammars throughout the late nineteenth century lacked a formal phonological theory, frequently confusing pitch registers with English stress patterns [S20, S21].
Across pre-colonial Yorùbá kingdoms, including Ọ̀yọ́, Ìjẹ̀bú, Èkìtì, and Òndó, tone systems evolved distinct dialectal variations [S1, S15, S23]. While standard Yorùbá (based largely on Ọ̀yọ́-Ìbàdàn dialects) developed three level register tones with inert Mid, southeastern dialects such as Ìkálẹ̀ and Ìjẹ̀bú maintained active Low-tone spreading and distinct tonal contour patterns in nominal compounds [S15, S23].
The Yorùbá civil wars of the nineteenth century (including the Owu wars, the Jalumi war, and the Kiriji/Ekitiparapo war between 1877 and 1893) caused massive population displacements and urban concentration in refugee centers such as Ìbàdàn, Abẹ́òkúta, and Ìjàyè . This intensive contact promoted dialect leveling and consolidated the three-tone register system of standard Yorùbá as the dominant koine across southwestern Nigeria .
In the mid-twentieth century, Ida C. Ward published Introduction to the Yoruba Language (1952), providing detailed phonetic transcriptions of Yorùbá pitch contours . Following Nigerian independence in 1960, Ayọ̀ Bamgboṣe published A Grammar of Yoruba (1966) and Ọladele Awobuluyi published Essentials of Yoruba Grammar (1978), establishing the modern descriptive foundation of Yorùbá morphophonology and identifying the inert nature of the Mid tone in vowel elision [S1, S2].
The rise of autosegmental phonology led to the formalization of Tonal Underspecification [S3, S4, S7]. Akinbiyi Akinlabi's 1985 doctoral dissertation Tonal Underspecification and Yoruba Tone at the University of Ibadan, alongside Douglas Pulleyblank's 1986 monograph Tone in Lexical Phonology, made Yorùbá the primary international archetype for radical underspecification in generative linguistics [S3, S4].
In 2004, Douglas Pulleyblank published "A note on tonal markedness in Yoruba" in the journal Phonology, demonstrating that tonal underspecification was theoretically problematic under Richness of the Base and proposing Markedness Faithfulness as an alternative . Subsequent research by Ọlanikẹ Ọla Orie, Ọládiípò Ajíbóyè, Rose-Marie Déchaine, and Bryan Gick has further investigated the syntax-phonology interface, acoustic properties, and constraint interactions of Yorùbá tones [S8, S11, S15, S18].
In Lucumí (Cuba) and Candomblé Ketu (Brazil), liturgical Yorùbá chants retained lexical and grammatical tone melodies across centuries of transatlantic transmission . While conversational Lucumí underwent Spanish contact-induced intonational changes, ritual sacred songs preserve distinct High, Mid, and Low tonal trajectories corresponding directly to Nigerian Yorùbá liturgical performance .
The debate over Yorùbá tonal representation has been shaped by prominent Nigerian and international phonologists:
While both theoretical frameworks successfully describe standard Yorùbá data, several questions remain unresolved in current scholarship:
How Yorùbá came to be written in Arabic and then Roman script, who decided the spelling rules, and why the subdots and tone marks are not optional.
A scholarly analysis of major reference grammars, foundational syntactic debates, and dialectological classifications in Yorùbá linguistics.
The Yorùbá sound system and its three tones, with worked minimal pairs showing how pitch alone changes a word's meaning.
An analysis of the Yorùbá three-level register tone system, demonstrating through minimal pairs and grammatical operations why tone is an essential phonemic component rather than optional decoration.
A comprehensive linguistic investigation into the Yorùbá High Tone Syllable (HTS), detailing its phonological behavior, syntactic distribution, dialectal variation, and competing theoretical models across six decades of scholarship.
A comprehensive linguistic investigation of Advanced Tongue Root (ATR) vowel harmony in Yorùbá, analyzing mid-vowel co-occurrence restrictions, high and low vowel asymmetries, morphological domain boundaries, and dialectal variations.