The schedule is posted below. Note that times are in Eastern US Time.
(During the workshop, this will be UTC−04:00.)
The current time for the Eastern US is given below.
| Friday March 19 | ||
| 11:00 – 11:15 | — PɸF Welcome — | |
| 11:15 – 12:00 | Laura McPherson | Edges and (anti-)alignment in Poko: Why we need autosegmental representations |
| 12:05 – 12:50 | Jochen Trommer | Gestalt Contours as Subtonal Morphology in Gaahmg |
| 12:50 – 1:30 | — Break for Lunch/Coffee/Dinner — | |
| 1:30 – 2:15 | Adam Jardine | Tone: computation, representation, and learning |
| 2:20 – 3:05 | Gabriela Caballero | Tonal exponence and lexical-grammatical tone interactions in San Juan Piñas Mixtec |
| 3:05 – open-ended | — Mingling over Zoom — | |
| Saturday March 20 | ||
| 10:30 – 11:15 | — Mingling over Zoom — | |
| 11:15 – 12:00 | Christopher Green & Nicola Lampitelli | When exception masks intention: structural conditions on tonal neutralization in Somali subject marking |
| 12:05 – 12:50 | Hiroto Uchihara | Otomanguean ‘inflectional tones’ = ‘grammatical tones’? |
| 12:50 – 1:30 | — Break for Lunch / Tea Time / Dinner — | |
| 1:30 – 2:15 | Larry Hyman | Allomorphy and Tonal Opacity at the Phrase Level in Kuki-Thaadow |
| 2:20 – 3:05 | Hannah Sande | Not all morphology is item-based: Evidence from three tonal processes |
| 3:05 – open-ended | — Mingling over Zoom — | |
| Sunday March 21 | ||
| 10:30 – 11:15 | — Mingling over Zoom — | |
| 11:15 – 12:00 | Laura Downing | Outermost does not always win: melodic tone assignment in Chichewa verb paradigms |
| 12:05 – 12:50 | Marjorie Pak | Logoori verb tones and the role of morphophonology |
| 12:50 – 1:30 | — Break for Lunch / Tea Time / Dinner — | |
| 1:30 – 2:15 | Nicholas Rolle & Florian Lionnet | Phantom structure: A representational account of floating tone association |
| 2:15 – 2:45 | — Wrap-up — | |
| 2:45 – open-ended | — Mingling over Zoom — | |
Abstracts
Gabriela Caballero
“Tonal exponence and lexical-grammatical tone interactions in San Juan Piñas Mixtec”
While phonological description and analysis of Oto-Manguean tone systems spans several decades, more recent work addresses the role that tone plays in the verbal inflectional systems in these prosodically complex languages (Cruz 2011, Palancar & Avelino 2015, Campbell 2016, Kim 2016, Palancar et al. 2016, Uchihara & Tiburcio Cano 2020, Palancar 2021, inter alia). This talk presents results of ongoing research conducted with Michelle Yuan, Claudia Juárez Chávez and students (UC San Diego) on the tonal and morphosyntactic properties of San Juan Piñas Mixtec (SJPM), a previously undocumented variety of Mixtec spoken in Oaxaca and diaspora communities in California. As documented for other Mixtec varieties, SJPM features a tonal inventory with three contrastive tones, and there is evidence for tonal underspecification, a large number of tone sandhi processes, and floating tones that yield complex phrasal morphotonemic patterns. Tone in SJPM also carries a significant functional load, and is involved in the exponence of transitivity alternations, TAM and polarity contrasts, which in turn interact with mood-based segmental stem alternations attested in some verb classes. Crucially, some morphosyntactic categories allow a choice between a segment+tone exponent and a tonal exponent, with choice partially determined by tonal context, but also attested as free variation. This talk provides evidence that the tonal properties of inflected verbs in this language can be explained as resulting from the “layering” of tonal and segmental exponents moving outwards from the root and the application of general tone processes in a morphemic tone analysis. We analyze TAM and polarity tonal exponents as floating tones aligned to the left edge of the verb complex and show evidence that the tonal grammar of SJPM prioritizes faithful realization of lexical tones; the availability of segment+tone exponents and purely tonal exponents of morphosyntactic categories and the possibility of having complex grammatically derived contours allows preservation of most lexical and grammatical contrasts in verbal paradigms. Finally, while the system can be largely described in autosegmental terms in a straightforward fashion, we discuss cases where tonal patterns in SJPM challenge a morphemic tone analysis and suggest how these patterns may arise via paradigm leveling.
Laura Downing
“Outermost does not always win: melodic tone assignment in Chichewa verb paradigms”
Many African tone languages have tonal morphemes and/or morphemes that introduce construction-specific grammatical tone patterns. As work like Hyman (2016), Inkelas (1998), Odden & Bickmore (2014) and Rolle (2018) shows, grammatical tone patterns have a number of properties that challenge a concatenative view of morphology. They associate a High tone or a tone pattern to a position or to positions that are not local to the sponsoring morpheme; they can also delete the tone of their morphological base; and they can introduce constructionspecific tonal processes. In addition, more than one inflectional morpheme can influence the grammatical tone pattern of a word or a construction. The question then arises, as Hyman (2016) puts it, if grammatical tone patterns are in conflict, which one wins? Work like Hyman (2016), Inkelas (1998, 2018), McPherson & Heath (2016) and Rolle (2018) has taken the strong position that tonal dominance is defined by the morphosyntactic hierarchy: grammatical tones affect the base they take morphosyntactic scope over, and the hierarchically outermost dominant grammatical tonal morpheme wins.
This talk shows that grammatical tone patterns in Chichewa (Bantu N.31; Downing & Mtenje 2017; Hyman & Mtenje 1999; Kanerva 1990; Myers & Carleton 1996) are problematic for the outermost dominance principle. Indeed, often a single morpheme does not alone determine the tone pattern in a particular verb paradigm. And an innermost dominant morpheme can have an effect on the overall tone pattern, overriding the predicted effect of an outer dominant morpheme. These points can be briefly illustrated with verb paradigms related to the present habitual (PH) aspect, segmentally marked by the prefix -ma-. The affirmative PH suggests that the PH prefix is tonally dominant, as the contrast between High-toned roots and toneless roots is neutralized (acute accents indicate High tone): ndí-ma-fotokóoza ‘I explain, toneless’ vs. ndí-ma-tambaláala ‘I stretch legs, High’. The root tone contrast reappears in the negative PH, however, even though the negative prefix sí- is tonally neutral (it simply realizes a High tone): sí-ndí-má-fotokooza ‘I don’t explain, toneless’ vs. sí-ndí-má-tambaláala ‘I don’t stretch legs, High’. The intensive derivational suffix also affects the verb tone pattern. In the negative PH intensive, the tonal contrast in the root is neutralized: sí-ndí-má-fotokoz-eéts-á ‘I don’t explain a lot, toneless’ vs. sí-ndí-má-tambalal-iíts-á ‘I don’t stretch legs a lot, High’. In these last two forms, we see an inside out effect of tonal dominance. The intensive conditions a dominant, neutralizing tone pattern on the stem that is different from the dominant, neutralizing tone pattern of the outer PH prefix.
In sum, Chichewa grammatical tone patterns provide a challenge to what Spencer (2004) calls a “Radical Agglutination” approach to morphology: the phonological form of inflected words is not always predicted by a one-to-one match between meaning/grammatical function, output form and hierarchical constituent structure. Instead, a combination of morphological features holistically predicts the inflectional tone pattern associated with a particular word form. I will argue that paradigm-based, realizational models like that of Ackerman & Stump (2004), Bond (2016), Spencer (2004) and Stump (2016) best account for the Chichewa data. More specifically, I will propose that while the segmental exponents of verb paradigms are built up from the stem outwards, in 3 morphologically-defined “blocks”, the tonal realization for each block is determined by combinations of morphemes relevant for the entire verb word.
Christopher Green & Nicola Lampitelli
“When exception masks intention: structural conditions on tonal neutralization in Somali subject marking”
This talk investigates the tonal and segmental exponents associated with subject marking (SM) in Somali and the role that prosodic structure plays in its realization. High tone is culminative on a given Somali word, but there are two grammatical conditions revealing that it is not obligatory; one of these is SM, and the other is in present and past tense realis mood verbs. Our focus here is primarily on the prosodic realization of SM, and namely the intentional removal/loss of High tone from a given word that it entails in most, but not all instances. SM may also be exponed segmentally. Briefly, SM’s segmental realization is contingent on the presence vs. absence of an independently motivated catalectic V slot at the right edge of some stems. When such a slot is present, SM is typically realized by -i.
Important to this talk is that SM’s segmental and prosodic exponents do not have an immediate bearing on one another, despite the fact that both ultimately target the final element of a DP. In analyzing the various outcomes of SM, we build upon analyses offered by Green & Morrison (2016, 2018) in illustrating that properties characteristic of particular Somali morphemes, and importantly their prosodic subcategorization frame, greatly influence a word’s surface form and accordingly its tonal behavior. We illustrate that SM can target for removal only a H tone associated with the head of a DP final prosodic word (PWd). If H tone falls outside of this PWd window, it fails to be removed, resulting in apparent exceptions to SM’s prosodic imperative.
Several previous accounts of Somali SM have been proposed, but each suffers to some extent in its assumption that SM’s prosodic exponence is exclusively autosegmental, e.g., by a morphological L tone primitive or some feature that overrides High tone. Such assumptions are not particularly well-motivated relative to other alternations found in the language. The alternative that we propose here is grounded in Cophonology Theory (Inkelas 2008, 2014), which permits more flexibility in teasing apart phonological processes that may or may not overtly involve affixation. We appeal to recent work by Sande, Jenks & Inkelas (2020) in treating SM as being composed of three types of phonological information: a string of segments or feature set, a prosodic subcategorization frame, and a constraint (re)ranking. Crucial to our analysis is a SM-specific subcategorization frame and re-ranking of constraints that allows us to formalize Somali’s preference in this particular morphological configuration to neutralize certain H tones, but not necessarily enforce their removal wholesale. Such an approach also easily captures SM’s ability to affect non-local, or non-adjacent H tones, provided they reside within its PWd scope.
Larry Hyman
“Allomorphy and Tonal Opacity at the Phrase Level in Kuki-Thaadow”
In this paper I address two rather clear cases of phrase-level tonal opacity in Kuki-Thaadow (KT), a Kuki-Chin (Tibeto-Burman) language spoken in NE India and Myanmar. As is well-known, postlexical tonal processes involving multiple words are both more widespread and more varied than corresponding segmental effects. Much of the work and proposed solutions to opacity concerns segmental interactions that either both occur in the word-level phonology, or where one process is lexical and the other postlexical (although not necessarily affecting words in contact). An oft cited counterfeeding example comes from (chi-)Mwera (Kenstowicz & Kisseberth 1977), where a nasal prefix causes a following voiced consonant to delete (e.g. /m+ba/ -> m-a) and a following voiceless consonant to become voiced (e.g. /m+pa/ -> m-ba). In this case the observed counterfeeding can be accounted for if voiced consonant deletion is “lexical” and voiceless consonant voicing is “postlexical”. There are still other possibilities when the interactions all occur within the word, e.g. strata, cyclicity, exceptionality, reference to specific morphemes etc. Such interpretations are less available when the interactions are between phrasal rules of tone spreading, tone shifting, tone reduction, tone insertion etc. which are quite common. In this context, KT offers an ideal tone system for studying opacity at the phrase level: (1) Almost all words are monosyllabic and monomorphemic, hence all tonal interaction is phrasal. (2) Tones spread in a quite general and familiar way from one word to the next. (3) All of the tone rules apply within the same intonational phrase. 4) The rules are exceptionless. 5) The interaction of these tone rules create opaque outputs. For example, although both H tone spreading (/H/ + /L/ -> H + HL) and L tone spreading (/L/ + /H/ -> L + LH) affect virtually any word combination within the intonational phrase, sequences of H + L + H + L are commonly attested on successive words in outputs. This is the first opacity that I will address, which occurs quite generally in the language, particularly since any HL or LH contour tone must be simplified in pre-final position. The second concerns the /HL/ vs. /L/ tonal allomorphs of the two “function words” hlon ‘dual’ and ding ‘future, benefactive’. These tonal allomorphs have the unusual property of being sensitive to the prior application of H tone spreading onto the preceding word, suggesting “late allomorph selection”, but participate in the “later” process of contour simplification producing opacity. Most work addressing whether allomorph selection must precede phonology vs. be determined in parallel with the phonology has been limited to the word level. KT thus shows that the same issue arises at the phrase level where we have fewer options.
Adam Jardine
“Tone: computation, representation, and learning”
Hyman (2011) states that “anyone who is interested in the outer limits of what is possible in phonology would … be well-served to understand how tone systems work.” Tone is thus an important empirical target for theoretical computational phonology, which studies the computational laws that delineate possible phonological patterns from impossible ones (Heinz 2018). In this talk, I survey results showing that computationally, tone is distinct from segmental phonology. I then show that, in understanding this difference, we find new arguments for melodic representations of tone, including their necessity for learning tone patterns. In sum, computational phonology has much to learn from, and much to contribute to, the study of tone.
Laura McPherson
“Edges and (anti-)alignment in Poko: Why we need autosegmental representations”
The tone system of Poko (Skou, PNG) is built off of three tonal primitives—L, M, H—in addition to unspecified syllables. However, Poko displays unusual restrictions on L and H: L can never be final and H can never be initial. First, I demonstrate that these restrictions cannot be captured with positive constraints on alignment but rather require markedness constraints banning L and H from their respective edges. Second, the distribution of floating tones points to the need for two domains of evaluation: associated syllables and the autosegmental tone melody (cf. Jardine 2017). Neither domain on its own can account for the inventory of surface tone patterns. I present a constraint-based analysis, showing how “edge constraints” like Non-Initial and Non-Final indexed to these different domains can account for Poko lexical tone. Ultimately, these distributional facts provide further validation of an autosegmental treatment of tone.
Marjorie Pak
“Logoori verb tones and the role of morphophonology”
This paper takes a hard look at whether we need to invoke morphophonology—phonology conditioned by specific morphemes—to analyze verb tones in Logoori (Bantu>Luyia, Kenya). Typical of Luyia, Logoori has a rich tense-mood-aspect (TMA) system and two verb classes marked by distinct tone melodies:
| (1) | Class A ‘fly’ | Class B ‘take’ |
| a. infinitive | -bʊrʊk-a LLL | -vʊ́gʊr-a HLL |
| b. consecutive | -bʊ́rʊ́k-a HHL | -vʊgʊ́r-á LHH |
| c. neg. subjunctive | -bʊ́rʊ́k-a HHL | -vʊgʊr-a LLL |
| d. middle future | -bʊ́rʊ́k-á HHH | -vʊ́gʊ́r-á HHH |
Traditionally (Odden 2018, Goldsmith 1991), these melodies are derived by assuming (i) underlying H on V1 of Class B (underlined), (ii) morphophonological H-lowering that targets this H in specific TMAs (1b)-(1c), and (iii) a floating-(H) suffix inserted in some TMAs, which docks on μ2 or final V and spreads left. Importantly, H-lowering in (ii) is morpheme-specific: it can’t be explained as e.g. *HH repair (cf. (1c)).
Morphophonological rules have an uneasy theoretical status given that they directly violate a principle of (morpho)syntax-free phonology (see e.g. Haugen 2015, Bermúdez-Otero 2012); even theories that permit them tend to use them judiciously (Embick & Shwayder 2018, Welmers 1973:132). I consider a reanalysis of (1) in which Class B is distinguished by underlying accent on V1 and all verb tones are inserted as TMA exponents: e.g. [CONSECUTIVE] is exponed as floating (H) in Class A, floating (L)(H) in Class B. Essentially, the reanalysis recasts H-lowering as allomorphy, in the spirit of Rolle 2018, Lieber 1992:170.
The reanalysis runs into trouble with a few TMAs where the placement of tones is irregular, e.g. (1d) where floating H docks on μ3 instead of μ2. Goldsmith 1991 deals with this by invoking σ1 extrametricality in the middle future, which can be effected by another morphophonological rule (see Noyer 2013). In the reanalysis, however, there is no obvious solution: what we need is a special instruction to override regular tone-alignment principles, but special instructions essentially amount to morphophonological rules.
This case study underscores one of the notable properties of tone—the degree to which it showcases the autonomy of autosegmental tiers (Hyman 2011)—demonstrating that not only tones themselves, but also the grammars that determine where tones are placed, can vary in morpheme-specific ways.
Nicholas Rolle & Florian Lionnet
“Phantom structure: A representational account of floating tone association”
Floating tones are defined as tones in a representation which are not associated to a tone-bearing unit (TBU). This representation may be either part of the underlying form of a morpheme, or arise at some derivational stage. If a constraint against floating tones is ranked high enough, at a point in the derivation floating tones will be either associated to some specific TBU and realized, or will be deleted/unrealized.
This paper focuses on the association of floating tones, illustrated with Bantu languages. One type is PHONOLOGICAL ASSOCIATION, where the TBU to which the floating tone associates is determined by the general phonological grammar. Typically, this can be analyzed as default/unmarked position, common examples including association to a stressed syllable, or to a TBU demarcating a domain edge (e.g. ultimate or penultimate position in a phonological word). Another type is TARGETED ASSOCIATION, where the floating tone targets some specific numerical position within a domain, e.g. a floating H associates to the fourth TBU (τ ) of the stem.1 Under this latter type (known in the Bantu literature as ‘melodic tone’), association is idiosyncratic to the sponsoring morpheme, and does not (necessarily) target a default/unmarked position.
Marlo et al. (2015) and Paster (2019) take this latter type as evidence for counting in grammar, a position we argue against here. Instead, to account for the difference between two types of floating tone association patterns we propose a novel representation we call PHANTOM STRUCTURE. The main intuition behind phantom structure is that certain morphemes require structure to be present in order for their sponsored floating tone to be realized, but do not and cannot provide this structure themselves. We implement this representation within Optimality Theory (Prince & Smolensky, 2004[1993]), via two partially overlapping correspondence relations (McCarthy & Prince, 1995): one between the actual, substantive input structure and output (IO-CORRESPONDENCE), and another between phantom and output structure (PHO- CORRESPONDENCE). We refer to the two inputs as the substantive plane and the phantom plane, respectively. In the input, the floating tone is actually pre-associated to a specific TBU position in the phantom plane. A simple constraint grammar requires the merger of these two input planes in the output, and results in the association of the floating tone to the equivalent numerical position in the output as in the phantom plane input.
Hannah Sande
“Not all morphology is item-based: Evidence from three tonal processes”
Introduction: There is a long-standing debate in the phonological literature about whether phonological alternations specific to a morphosyntactic context should be analyzed as trig- gered by an underlying phonological item, or by a morpheme-specific rule or constraint (Hockett 1954, Anderson 1992, Bermudez-Otero 2012, Trommer & Zimmermann 2014). Here I provide evidence from morphologically specific tonal alternations in Guébie (Kru), Kipsigis (Nilotic), and Amuzgo (Oto-Manguean) that not all tonal morphology is item-based.
Guébie scalar tone shift: Imperfective aspect in Gu´ebie is exponed as a scalar shift in tone from input to output (Sande 2017, 2018). There are four contrastive level tones 1-4, where 4 is high. The first tone of a verb surfaces one step lower in imperfective contexts than elsewhere. If the verb has a low default tone, then the final tone of the preceding subject raises in imperfective contexts. The tone shift cannot be restated as assimilation or dissimilation between subject and verb tones; sometimes the subject and verb tones in imperfective contexts are more disparate than elsewhere (/Subj2 Verb2/ [Subj2 Verb1]), and sometimes they are more similar (/Subj2 Verb3/ [/Subj2 Verb2]). Additionally, the result of the imperfective shift can be lowering of an input tone (/Subj2 Verb2/ [Subj2 Verb1]), or raising (/Subj2 Verb1/ [Subj3 Verb1]). For these reasons, Sande (2018) argues against an item-based analysis of the Gu´ebie imperfective tone shift.
Kipsigis tonal polarity: In Kipsigis, the nominative form of a modifier surfaces with ex- actly opposite (polar) tones of the default form (Kouneli & Nie 2020). There are two tone heights in the language, H and L. All default H’s in the word surface as L in the nominative and vice versa. Additionally, there is a regular flattening of LH tones to H (we never see a rising contour tone on the surface). Default HL contours on modifiers surface as H in the nominative, predicted by tonal polarity (HL LH) + flattening (LH H). No other H and L affixes or floating tones in the language have a polarizing effect, so the tonal polarity facts cannot be derived from a floating tonal morpheme. For example, the plural suffix surfaces as L after both H and L tones in non-nominative forms: HL-L→HLL and H-L→HL in plural contexts; H-H→HH and LL-H→LLH in secondary contexts.
Amuzgo doubly-conditioned tone: Surface tone on verb stems in Amuzgo is jointly de- termined by the lexical class of a verb and the person and number features of the subject (Kim 2016). There are eight contrastive tone melodies in Amuzgo: H, M, M+, L, L+, HM, HL, MH. All eight possible melodies are attested in 3rd person singular forms of verbs. How- ever, the tone in first and second persons is not predictable given the 3rd person form, or vice versa (Kim 2016). Additionally, there is no consistent type of tonal alternation (lowering, raising, etc.) associated with any person/number combination. Kim (2019) argues that the tonal alternations cannot be due to a combination of underlying items, and must be the result of a cophonology sensitive to both lexical class and subject features.
Implications for the item-versus-process debate: Item-based accounts do not provide any additional insight into scalar tone shifts like in Gu´ebie, or across-the-board tonal po- larity as in Kipsigis. Instead, allowing for cophonologies or morpheme-specific tonal anti- faithfulness constraints in the phonological component makes better predictions (see Sande 2018, Kim 2019, and Kouneli & Nie 2020).
Jochen Trommer
“Gestalt Contours as Subtonal Morphology in Gaahmg”
In this talk, I provide a new type of evidence for sub-tonal features (Yip 1980, 2002, Bao 1999) from tonal morphology in the Eastern Sudanic language Gaahmg (Stirtz 2011): contour formation where specific morphological categories change the underlying tone of a base word to a falling contour, but with different absolute tone values (High-Mid, Mid-Low, and High-Low) depending on the input tone. I show that the three different morphophonological gestalt contours in Gaahmg can be captured succinctly via feature affixation using the tone geometry proposed by Snider (1990, 1998, 1999) (Register Tier Theory), and that this analysis receives independent support by the general morphophonology of the language, especially by a pervasive sub-tonal assimilation (lowering) process, and context-specific tonal modification triggered by morphological context. The presentation thus strengthens recent arguments by McPherson (2016) and Meyase (2020) that the skepticism of Hyman (2010) and Clements et al. (2011) against the featural decomposition of tones is premature: Tone languages show systematic and consistent evidence for subtonal representations.
Hiroto Uchihara
“Otomanguean ‘inflectional tones’ = ‘grammatical tones’?”
Otomanguean languages represent oft-cited cases of ‘paradigmatic tones’ (Palancar & Léonard 2016; Rolle 2018; Baermann et al. 2019): languages which “(i) show extensive inconsistency with grammatical categories; (ii) show extensive inconsistency across roots/stems in parallel paradigms; and (iii) there being little positive evidence for determining the underlying tone of the root/stem (Rolle 2018: 109)”. Such complex ‘inflectional tones’ (Palancar & Leonard 2016) have been reported for Tlapanec (Suárez 1983), Mazatec (Pike 1948; Jamieson 1997; Léonard and Fulcrand 2016; Nakamoto 2020), Amuzgo (Kim 2016; Palancar 2021), Chatino (Woodbury 2019), and Chinantec (Baerman & Palancar 2015). Table 1 from Huehuetepec Tlapanec (cf. Uchihara & Tiburcio Cano 2020) illustrates this, where the agent person categories, except for the 2SG which has the prefix ta-, are encoded solely by tonal alternations. Here, it appears that grammatical categories (in this case, agent person) and stems do not have consistent exponents, and thus it appears challenging to determine the underlying tone of the stem.
Table 1. Illustration of Huehuetepec Tlapanec verb conjugation
| 1SG | 2SG | 3SG | 1PL.IN | |
| ‘hobble’ | nì-ràhmúúʔ | nī-tā-rāhmúúʔ | nī-rāhmúúʔ | nī-rāhmúúʔ |
| ‘breathe’ | nì-gìxíʔ | nī-tā-gíxíʔ | nī-gíxíʔ | nī-gíxíʔ |
| ‘cry’ | nì-mbìyàʔ | nī-tā-mbīyàʔ | nī-mbīyàʔ | nī-mbīyàʔ |
| ‘tie up’ | nì-ráhmāā | nī-tā-ráhmāā | nī-ráhmāā | nī-ráhmāā |
| 1PL.EX | 2PL | 3PL | |
| ‘hobble’ | nì-ràhmúúʔ | nì-ràhmúúʔ | nī-ràhmúúʔ |
| ‘breathe’ | nì-gìxíʔ | nì-gíxíʔ | nī-gìxíʔ |
| ‘cry’ | nì-mbìyàʔ | nì-mbìyáʔ | nī-mbīyàʔ |
| ‘tie up’ | nì-ráhmāā | nì-ràhmáā | nì-ráhmāā |
Even though Tlapanec superficially resembles those which manifest ‘paradigmatic tones’, I will argue in fact there is positive evidence for postulating lexical tones of stems and affixes. That is, Tlapanec tonal alternations are due to phonologically conditioned tonological operations, which are “a tonological operation conditioned by a particular phonological context” (Rolle 2018: 18), and not grammatical tone, defined in Rolle (2018: 19), as “a tonological operation which is not general across the phonological grammar, and is restricted to the context of a specific morpheme or construction, or a natural class of morphemes or constructions”. Thus, the tonal alternation due to agent person categories is triggered by the underlying tones of the prefixes, along with the phonologically conditioned tonological operations, such as tone spreading and floating tone docking, that can be accounted for with a consistent ranking of cross-linguistically common tonal constraints. I will show this by demonstrating that these tonological operations are not restricted to the context of a specific morpheme, but are more general in the language. I will further argue that such an analysis is more adequate than a paradigmatic tone analysis, in that it can explain and predict exactly in which cells free variation is expected, the number of patterns of tonal alternation, and the patterns of syncretism.
The Tlapanec situation no way represents Otomanguean languages; more in-depth analyses may reveal that other Otomanguean languages are also susceptible to such an alternative, phonologically conditioned tonological operation analysis. On the other hand, other Otomanguean languages, such as Mazatec (Nakamoto 2020) or Chinantec, may manifest true cases of paradigmatic tones, where there is no justification for postulating lexical tones of stems. Between these two extremes, there could also be yet other Otomanguean languages such as Amuzgo (Kim 2016; Palancar 2021), Teotitlán Zapotec (Uchihara & Gutiérrez 2020), which may have sufficient justification for postulating lexical tones of the stems but some operations need to be considered as instances of grammatical tone. Finally, I explore the possibility that the degree of paradigm complexity, that is the entropy measures of the inflectional classes (Ackerman & Malouf 2013), may correlate with how ‘phonological’ the alternations are.
