Authors: David Inman, Marine Vuillermet
Acknowledgments: Selma Hardegger, Anna Graff
Singular-plural verbal stem alternation (henceforth Sg-Pl alternation) is present in a language if there are any lexically specific sets of verbal stems whose distribution can be determined solely by the number of an argument (S, A or P). In example (1), the shape of the verbs expressing ‘die’ in Yaqui [yaqu1251] (Uto-Aztecan; Mexico, United States) changes depending on the number of participants involved in the event, muuku- with a singular participant, and koko- with a plural participant (Lindenfeld 1973: 52):
| (1) | Yaqui (Uto-Aztecan) | |||
|---|---|---|---|---|
| a. | huʔu | čuuʔu | muuku-k | |
| this | dog | die(SG)-PERF | ||
| ‘The dog died.’ | ||||
| b. | hu-me | čuʔu-m | koko-k | |
|---|---|---|---|---|
| this-PL | dog-PL | die(PL)-PERF | ||
| ‘The dogs died.’ | ||||
We chose the term singular-plural verbal stem alternation because formally our definition encompasses two types of lexically limited alternation: root suppletion, and limited regular patterns. Both patterns are exemplified in Wari’ [wari1268] (Chapacuran; Brazil; Birchall et al in review), where some verbal alternations are suppletive (14 pairs) and some use a (lexically) limited morpheme (27 pairs), as illustrated in Table 1.
| Root suppletion | (limited) Regular pattern -rVCV- | ||||
|---|---|---|---|---|---|
| SG | PL | SG | PL | ||
| ‘stand’ | xat | koko’ | ‘hang’ | wa | warawa |
| ‘run’ | mao | mama’ | ‘walk’ | xut | xuruxut |
| ‘die’ | mi’ | kono’ | ‘kill’ | topa’ | toparapa’ |
| ‘take’ | an | kut | ‘pull out’ | pit | piripit |
Root suppletion (henceforth suppletion) is when the verb roots are fully suppletive, as with muuku ~ koko 'die’ in Yaqui (Uto-Aztecan) in (1), or where no pattern is recognizable across multiple verb pairs. A lexically limited regular pattern (henceforth regular pattern) is when verb stems alternate by some segmentable but unproductive morphology, including reduplication. Our definition requires that the alternation be lexically limited. An example of semantically limited morphology, which is definitionally excluded from this survey, is the morpheme koo in Magdalena Peñasco Mixtec [magd1235] (Otomanguean; Mexico), which marks a plural subject only for verbs of Motion (Erickson de Hollenbach 2013: 48).1
Note that the difference between regular pattern and suppletion is sometimes unclear, and can be considered as two ends of a continuum. In Western Keres [west2632] (Keresan, United States), many verb forms seem regular, but are not in their entirety: -u̓:bəN ~ -âaʔabəN ~ -je̓ebəN ‘enter’ and -jaʔac̓íN ~ -âaʔác̓iN ~ -jéedyuB ‘arrive’ share the same dual prefix, but not quite the same plural prefix. Furthermore, the root for ‘enter’ is consistent across all numbers while that for ‘arrive’ is suppletive in the plural. Miller (1965: 59) notes that “the [phonological] changes are [sometimes] so great that the [verb pairs] can be considered suppletive” and “there is [...] no clear-cut difference between irregular [our limited regular pattern] and suppletive changes.” We followed his analysis and considered all of these verbs suppletive (61, the highest number in our sample), since no pattern was synchronically straightforwardly retrievable.
Note that languages displaying verbal number as a regular morphological process available to all verbs are captured in another feature set, namely Monoexponential Verbal Plurals. A very particular kind of edge case is a language like Nisga'a [nisg1240] (Tsimshianic; Canada), which has pluralizing morphology for all verbs, but its form is unpredictable and lexically conditioned. Tarpent (1987) describes the system in great diachronic detail, showing that the plural system developed in stages, with later morphology modifying or adding to existing plural (or distributive) morphology, resulting in a contemporary system with at least eight strategies for marking a plural absolutive argument: lV- prefixing, tx̌a·- prefixing, Cix- prefixing, qa- prefixing, full stem reduplication, irregular reduplication (i.e. with phonologically irregular onsets or codas), partial reduplication, and full (form) reduplication. In addition to these patterns, there are some verbs that undergo true root suppletion to mark plurality. Every verb marks the plural of its absolutive argument using one of these strategies, but there is no way to predict which strategy a verb will use.
Because these inflectional classes divide up the whole lexicon, we consider that there is a single plural paradigm which applies to every word. They have thus been coded in the monoexponential verbal plurals feature set despite the plural form being lexically conditioned by the (plural) inflectional class of the target verb. Nisga’a was still considered as exhibiting Sg-Pl alternation because of its few suppletive verb pairs. The semantics of these pairs also fell in line with the verb semantics generally found in Sg-Pl alternation.
We only considered lexically basic verb pairs, i.e. we did not consider derived verb forms. An example of this is the Hopi [hopi1249] (Uto-Aztecan; United States) pair wari(k-) ~ yùutu(k-) ‘run’, which has many derivations, such as wawarna ~ yuyutna ‘train for racing’ and wárikiwta ~ yùutukiwta ‘be crazy or wild’ (Hopi Dictionary Project 1998). One reason for considering lexically basic verb pairs only is consistency across languages and grammars: Languages more prone to morphological derivation would logically display more pairs than languages with less derivation, and grammars (as opposed to complete dictionaries) are more likely to list lexically basic alternations than derived ones.
Note that in some cases, a probable diachronic link is not easily retrievable, as in the Hokkaido Ainu [ainu1240] (Ainu; Japan) verb pairs a ~ rok ‘sit’ and as ~ roski ‘stand’ (Nakagawa 2022). In such cases, we did count both pairs.
Semantically, our comparative concept of Sg-Pl alternation excludes event number, and only targets participant number. Not only are participant and event number theoretically separable, but there are languages which encode the two separately. This is the case in Seri [seri1257] (isolate; Mexico; Marlett 1981: 96, 224), where different morphological processes encode event number (repetition) and participant number independently, giving rise to a four-way distinction for most verb forms. The full morphology governing these forms is quite complex, but event and participant number can be marked by affixation processes alone (as in ‘talk about’, Table 2) or by affixation and suppletion (as in ‘go to’, Table 3).2
| ‘talk about’ (single P argument) | ‘talk about’ (multiple P arguments) | |
|---|---|---|
| one event | -šaχʷ | -šaχʷ-t |
| multiple events | -šaχʷ-tim | -šaχʷ-toɬka |
| ‘go to’ (single S argument) | ‘go to’ (multiple S arguments) | |
|---|---|---|
| one event | -ya:i | -oši:t |
| multiple events | -ya:i-tim | -oši:t-am |
These categories are conceptually different, and we only consider participant number in our comparative concept. This definition also entails that we exclude cases where additional semantic factors distinguish the verbs. As defined by (Mithun 2022), this means that alternating verbs must be “basic-level terms” and not in a synchronic hypernym/hyponym relationship (even if, as Mithun suggests, some verb pairs originate from a historic hypernym/hyponym relationship).
An example of a verbal pair that fails to meet this criterion is English kill ~ massacre. The verb massacre necessarily involves a plural object, but it is not a basic-level term and is a hyponym of kill. More than just a plural object, massacre also provides information about brutality, intentionality, a relatively short span of time, and other semantic properties.
Another type of alternation that we exclude definitionally is suppletion that encodes person along with number. In Table 4, the conjugation of the verb ‘eat’ in Malinaltepec Me’phaa [mali1285] (Otomanguean; Mexico; Suárez 1983: 160) does not count as Sg-Pl stem alternation. Though the paradigm shares the same stem amongst all plural subjects, the forms for singular subjects change based on person.
| sg | pl | |
|---|---|---|
| 1 | -kho2 | -pho2 |
| 2 | -’co2 | -pho2 |
| 3 | -kho2 | -pho2 |
The phenomenon of singular-plural stem alternation was probably first mentioned by Boas (1911: 381) in his Tsimshian sketch under the term “irregular plural”, but little attention was paid to the phenomenon until the 1980s. Several crosslinguistic works on verbal number have appeared since: - Mithun (2022) on about 13 Californian languages; - Booker (1982) on about 32 North American languages; - Krasnoukhova (2022) on 70 South American languages; - Durie (1986) on a worldwide sample of 40 languages; - Veselinova (2013) on a worldwide sample of 193 languages
All studies including North American language data emphasize its overwhelming presence in North America, and Mithun (2022) demonstrates that it is an areal pattern across languages from different families in Northern California. Krasnoukhova's (2022) survey shows that verbal number is also very present in South America, but the phenomenon she surveyed is verbal number more generally: it includes both regular and irregular patterns, expressing event and/or participant number, and be they expressed by reduplication, morphological markers or stem alternation.
Each language in the sample was coded for the 22 features detailed below, which target:
Individual verb pairs were, in addition, entered into a separate spreadsheet. Each entry in this database is a single verbal pair, and includes the meaning of the pair, the type of alternation (suppletion or regular pattern, and bi- or tripartite pattern, see features SgPl-02, SgPl-04, and SgPl-18), a field for comments, and the identification code of the language the pair occurs in. Additional columns for semantic categorization were added post-hoc, based on a semantic gestalt that arose from the table.
{ yes | no }
If <no> to SgPl-01, then the state is <NA> for all other features, except for SgPl-19.
Mushin (2012) reports no Sg-Pl stem alternation.
According to Nakagawa (2022), some verb forms in Hokkaido Ainu are suppletive (2), and others have an -n/-p alternation for sg/pl (3).
| (2) | Suppletion | ||
|---|---|---|---|
| a. | an ~ oka | ‘be, exist; also continuous auxiliary’ | |
| b. | arpa ~ paye | ‘go’ | |
| c. | ek ~ arki | ‘come’ | |
| d. | as ~ roski | ‘stand’ | |
| e. | a ~ rok | ‘sit’ | |
| f. | omanan ~ pay(e)oka | ‘wander’ | |
| g. | rayke ~ ronnu | ‘kill’ | |
| h. | uk ~ uyna | ‘take’ | |
| i. | anu ~ ari | ‘put’ | |
| (3) | Regular -n ~ -p alternation | ||
|---|---|---|---|
| a. | riki-n ~ riki-p | ‘go up’ | |
| b. | ra-n ~ ra-p | ‘go down’ | |
| c. | sa-n ~ sa-p | ‘go forward’ | |
| d. | maka-n ~ maka-p | ‘go back’ | |
| e. | ya-n ~ ya-p | ‘land, go toward shore’ | |
| f. | ahu-n ~ ahu-p | ‘go in’ | |
According to Nichols (2011: 313), "a few verbs undergo agreement with the number of the S/O". (See Table 5 in SgPl-03 below for a list.)
{ NA | yes | no common pattern }
The presence of two alternating pairs that follow the same pattern is sufficient for analyzing them as a regular pattern.
According to Valenzuela (2003: 150), Shipibo-Conibo has two Sg-Pl alternating verb pairs, and they are fully suppletive.
| (4) | a. | jo- ~ be- | ‘come’ |
|---|---|---|---|
| b. | ka- ~ bo- | ‘go’ |
Yuracaré has an atypical regular formation: 4 (out of 11) plural roots are built from the reduplication of the first syllable: letu ~ lele ‘sit down’, chittu ~ chitchi ‘cross’, wita ~ wiwi ‘arrive’, sheta ~ sheshe ‘sit down’.
Nichols (2011: 313) describes 23 Sg-Pl pairs whose "main formations are suppletion, ablaut of stem vowel, and change of stem-final -ll to -xk".
{ NA | _number_ }
For four verb pairs, van der Voort (2004: 384) notices "obscure reduplication patterns" illustrated in example (5).
| (5) | Regular patterns | ||
|---|---|---|---|
| a. | bui- ~ bu'bui- | ‘to go outside’ | |
| b. | oja- ~ oʔo'ja- | ‘to go’ | |
| c. | tsje- ~ tsi-tsje- | ‘to grab’ | |
| d. | dai- ~ da-'dai- | ‘to take’ | |
Nichols (2011: 313f) describes a total of 23 Sg-Pl alternating pairs among the present, nominalized and converb verbal forms. We only considered the forms in the present and ignored the other, non-verbal derivations, resulting in 21 pairs. Out of these 21 pairs, 18 display regular patterns, distributed over 8 distinct stem modifications, as illustrated in Table 5, which includes ablaut, changes in stem-final consonants, and sometimes both. Three pairs, given at the end of the table, are suppletive.
| SG | PL | |
|---|---|---|
| V ablaut | ||
| ‘start, go’ | d.oal | d.oul |
| ‘run’ | d.od | d.oud |
| d → V + lx | ||
| ‘go’ | d.uoda | d.olx |
| ‘run out’ | hwed | hwelx |
| CC → V + C | ||
| ‘cast’ | qoss | qous |
| ‘throw’ | toss | tous |
| ‘stand up’ | ott | out |
| qq → x | ||
| ‘take’ | d.oaqq | d.oax |
| dzh → V + sh | ||
| ‘lie’ | d.udzh | d.uush |
| ∅ → sh | ||
| ‘sit down’ | xou | xoush |
| ll → xk | ||
| ‘be contained’ | d.oall | d.oaxk |
| ‘drive’ | loall | loaxk |
| ‘insert’ | d.oll | d.oxk |
| ‘hang up’ | oll | oxk |
| ‘throw, cast’ | qoll | qoxk |
| ‘lie’ | ull | uxk |
| ll → V + xk | ||
| ‘lay’ | d.ull | d.oxk |
| ‘put, lay’ | tull | toxk |
| Suppletion | ||
| ‘die’ | le | d.ou |
| ‘burst’ | eqq | hwelx |
| ‘lie’ | ull | d.aada |
As exemplified in (4) above, Shipibo-Conibo only has two suppletive verb pairs, and no regular pattern (Valenzuela 2003: 150).
{ NA | yes | no }
According to Heath (1984: 488f), Wubuy only has regular patterns.
Kwaza displays suppletive verb pairs (Voort 2004: 387).
| (6) | Suppletion | ||
|---|---|---|---|
| a. | a'sa- ~ (u?)u'ja- | ‘to end’ | |
| b. | hy- ~ tow- | ‘to move’ | |
As exemplified in (4) above, Shipibo-Conibo has suppletive verb pairs (Valenzuela 2003: 150).
{ NA | _number_ }
According to Heath (1984: 488f), Wubuy only has regular patterns.
As exemplified in (4) above, Shipibo-Conibo has two suppletive verb pairs (Valenzuela 2003: 150).
Kwaza has six suppletive verb pairs (van der Voort 2004: 384), two of which are illustrated in (5).
{ NA | yes | no }
As shown in (4), Shipibo-Conibo has two suppletive verb pairs expressing Motion (Valenzuela 2003: 150).
Kwaza has six suppletive verb pairs (van der Voort 2004: 387), and two of them belong to the Motion category (‘to cross’ and ‘to move’), and two to Posture (‘to lie down’ and ‘to sit’). The last two have uncategorized semantics ('to end' and to 'sting on head').
{ NA | intransitive | transitive | both }
The only alternating verb pairs reported by Valenzuela (2003: 150) in Shipibo-Conibo are ‘go’ and ‘come’, both intransitive verbs, as given in (4).
Esselen has only one Sg-Pl alternating verb pair, the transitive verb ‘bring’ (Shaul 2019: 350).
| (7) | aha/la- ~ čup/hu- | ‘to bring’ |
As demonstrated in the list in (2) and (3) above, Hokkaido Ainu has both intransitive and transitive alternating pairs (Nakagawa 2022).
In Kwaza, all suppletive pairs are intransitive, but some regular patterns are transitive, as illustrated in (5) with tsje- ~ tsi-tsje- ‘to grab’ and dai ~ da-'dai- ‘to take’ (van der Voort 2004: 384).
{ NA | yes | no }
We intentionally group together Spontaneous and Caused posture (as well as both types of motion) verbs, and investigate separately the transitivity status of the alternating pairs (SgPl-17), since we don't want to mix semantic and syntactic categorization. As a matter of fact although spontaneous vs caused motion semantics often align with intransitive vs transitive motion verbs, there are discrepancies. In some languages, the verb ‘go across’ is syntactically transitive (with the entity being crossed encoded as a direct object), yet semantically, it is nevertheless a spontaneous motion event. Spontaneous and Caused posture (and motion) are not distinguished in this question, but they are separated in the associated verb list.
Ainu has two verb pairs expressing Posture (Nakagawa 2022).
| (8) | a. | a ~ rok- | ‘sat’ |
|---|---|---|---|
| b. | as ~ roski- | ‘stood’ |
Kwaza has two verb pairs expressing Posture (van der Voort 2004: 385)
| (9) | a. | bu?ũ'rjỹ- ~ bu?ũmũ'rjỹ- | ‘to sit’ |
|---|---|---|---|
| b. | ũi- ~ u'mũi- | ‘to lie down, sleep’ |
Shipibo-Conibo has only two alternating verb pairs (Valenzuela 2003: 150), which belong to Motion, as illustrated in (4).
{ NA | _number_ }
Even if two of the alternating verb pairs in Wubuy are Posture verbs, they all follow a regular pattern (Heath 1984: 488f).
Shipibo-Conibo has only two suppletive verb pairs (Valenzuela 2003: 150), which belong to Motion, as illustrated in (4).
Ainu has two suppletive verb pairs expressing Posture (Nakagawa 2022), illustrated in (2).
Kwaza has two suppletive verb pairs expressing Posture (van der Voort 2004: 385), illustrated in (5).
{ NA | yes | no }
The two pairs described in Shipibo-Conibo are ‘come’ and ‘go’, illustrated in (4).
Valentine (2001: 328f) only lists verb pairs not expressing Motion.
Seifart (2005) does not report Sg-Pl alternation.
{ NA | _number_ }
Alternating verb pairs in Ottawa involve reduplication only, which is a regular pattern (Valentine 2001: 328f).
The two pairs described in Shipibo-Conibo are ‘come’ and ‘go’ illustrated in (4), both classified into the Motion verbs.
{ NA | none | ; separated list of [location, expected, property, other] }
We have listed in the remarks the verb pairs according to the following semantic categories: - location (e.g. ‘be located’, ‘be inside’, ‘stay’, ‘grab’, ‘hold’) - expected (‘kill’, ‘die’, ‘cry’, ‘sleep’, ‘eat’) - property (property concept semantics, e.g. ‘be small’, ‘be kind’) - other (none of the above)
Note that this feature is restricted to the list of alternating verbs. This means that if adjectives formed a distinct part of speech category, they were not listed under Property in this feature, but in SgPl-21.
The verb ‘exist’ has been listed in the Other category, unless it is encoded by a Posture verb, like t̓á· ~ wán ‘to sit, to exist’ in Nisga'a (Tarpent 1983: 199-200), or by a Location verb, like an ~ oka ‘be, exist’ in Hokkaido Ainu (Nakagawa 2022). Such existential verbs have been listed in the Posture and Location categories respectively, so as to not be counted twice.
According to Nakagawa (2022), the other salient semantics are the following:
| (10) | a. | Location | |
|---|---|---|---|
| ũan ~ oka | ‘be, exist; also continuous auxiliary’ |
| b. | Expected | ||
|---|---|---|---|
| rayke ~ ronnu | ‘kill’ |
According to van der Voort (2004: 384f), the other salient semantics are the following:
| (11) | a. | (Caused) Location | |
|---|---|---|---|
| tsje- ~ tsi-tsje | ‘grab’ | ||
| b. | Other | ||
|---|---|---|---|
| esa'hỹ-kuty- ~ tsādy-kuty- | ‘sting on the head’ | ||
| a'sa- ~ (u?)u'ja- | ‘end/leave/terminate/obliterate’ | ||
The two pairs described in Shipibo-Conibo involve two Motion verbs only, namely ‘come’ and ‘go’, illustrated in (4).
One of the regular pattern verb pairs in Wubuy is ‘to see’ (Heath 1984: 488f).
{ NA | _number_ }
All three alternating verb pairs in Wubuy follow a regular pattern (Heath 1984: 488f).
The two pairs described in Shipibo-Conibo involve ‘come’ and ‘go’, illustrated in (4).
van der Voort (2004: 384f) does not report any suppletive verb pairs expressing ‘cry’, ‘die’, ‘kill’, ‘sleep’ and ‘eat’.
According to Nakagawa (2022), rayke/ronnu ‘kill’ is the only suppletive verb pair expressing one of the five semantics targeted here.
{ NA | _number_ }
Mushin (2012: 180f) reports no Sg-Pl stem alternation.
The two suppletive verb pairs described by Valenzuela (2003: 150) involve ‘come’ and ‘go’.
According to Nakagawa (2022), an ~ oka ‘be, exist’ is the only suppletive verb pair different from the most frequent semantics listed in (2).
van der Voort (2004: 384f) reports two such suppletive verb pairs, namely tsje- ~ tsi-tsje ‘grab’ and esa'hỹ-kuty- ~ tsādy-kuty- ‘sting on the head’, a'sa- ~ (u?)u'ja- ‘end/leave/terminate/obliterate’.
{ NA | yes | no }
Mushin (2012: 180f) reports no Sg-Pl stem alternation.
Park (2012: 266, 273) reports two Posture verb pairs with the same plural form: naagí ~ gáá ‘be sitting’, maagí ~ gáá ‘be lying’, and three further Posture and Location verbs with the same plural form: nahgú ~ áhgu ‘be, continue in a sitting position’, mahgú ~ áhgu ‘dwell, stay, live’, háhgu ~ áhgu ‘be about’.
van der Voort (2004: 384f) does not report any shared plural form among its alternating verb pairs.
{ NA | unaccusative | unergative | both }
Unaccusative intransitive verbs like ‘to fall’ have patient-like subjects. Unergative intransitive verbs like ‘to run’ have agent-like subjects.
Crevels (2006: 165) reports only one alternating verb pair, oli/soloh ‘fall’, which is unaccusative.
The only pairs (‘come' and ‘go') are unergative (Valenzuela 2003: 150).
The list reported in Rice (1989: 790f) includes both unergative intransitive verb pairs like ‘be seated’, ‘go (controlled)’ and ‘swim’, as well as unaccusative ones like ‘go (non-controlled)’, ‘fall’, ‘sleep’ and ‘die’.
{ NA | A | P | either | both }
<either> signifies that the plural argument depends on the verb, i.e. it is always A for some verbs and always P for other verbs. <both> signifies that the same verb root can be used for plural A or P depending on the context. This state was not attested in our sample.
Hopi is P-dominant, and the alignment is overwhelmingly absolutive. However, the A role is the trigger for ngöyva ~ ngööngöya ‘chase’ and lalvaya ~ lavayta ‘talk about’ (Hopi Dictionary Project 1998: 320, 321).
In Imonda, either A or P arguments may be pluralized, depending on the verb pairs. A arguments are pluralized in ‘eat’, ‘not like’, ‘make netbag’, while P arguments are pluralized in ‘follow’, ‘cut’, ‘hear’, ‘see’. Also, ‘stick into’ has a specific plural form if both A and P are plural: shi (A+P SG) ~ eh (P.PL) ~ ih (A+P.PL).
van Gijn (2006: 191-192) has two transitive alternating verb pairs, and they encode the number of the P argument.
| (12) | a. | mii-∅ |
|---|---|---|
| take.SG-3 | ||
| ‘He takes it.’ |
| b. | ma-pu-∅ | |
|---|---|---|
| 3PL-take.PL-3 | ||
| ‘He takes them.’ |
| c. | mii-∅=w | |
|---|---|---|
| take.SG-3=PL | ||
| ‘They take it.’ |
Marlett (1981: 89-107, 220) reports that -Qaχš ~ -Qakx ‘hit with stick’ has irregular root suppletion for form 3 (plural subject, singular action) and for form 4 (plural subject, pluractional), which means that it has one verb that suppletes for A, while all other suppletive verb pairs are intransitive.
{ NA | ; separated list of [ SG-PL, PAUCAL-MANY, INDIVIDUAL-COLLECTIVE, SG-DU-PL, SG-PAUCAL-PL ] }
The semantics of number is often left unmentioned in grammatical descriptions, and the differentiation between singular-plural and individual-collective number are often uninvestigated. The semantics of number were determined only through remarks in the grammar specifically dealing with these words, and not a reanalysis of the lexicon, with a bias for a singular-plural distinction, if no detailed description of number was present.
In Yuracaré (van Gijn 2006: 191-192), the semantic contrast is not regular throughout the pairs: - SG-PL for (at least) Motion verbs ‘go’, ‘go in’, ‘arrive’, and ‘go across’, whose distribution is explicitly described: “singular subjects trigger singular stems and plural participants trigger plural stems.” - other verbs like ‘fall’ alternate according to a PAUCAL-MANY distinction: in (13), the subject ‘two books’ is still compatible with a singular verb stem.
| (13) | a. | lëshie | ti-wadernu | dele-∅ | mesa=jsha |
|---|---|---|---|---|---|
| two | 1SG-book | fall.SG-3 | table=ABL | ||
| ‘My two books fell from the table.’ | |||||
| (13) | b. | bëmë | ti-wadernu | ñeta-∅ | mesa=jsha |
|---|---|---|---|---|---|
| many | 1SG-book | fall.PL-3 | table=ABL | ||
| ‘My many books fell from the table.’ | |||||
Seiler (1985: 82, 86) reports the following number semantics: - SG-DU-PL: ‘put up’, ‘hang up’, ‘fill in’; - PAUCAL-MANY: e.g. ‘be/stand’; - SG-PL: e.g. ‘hold’.
Despite most examples being described as SG/DU/PL in Rice (1989: 790f), they are described as SG/PAUCAL/PL in Corbett (2000: 249; our bold): “There may be three different verb forms, or two, or indeed the normal case which is for there to be no distinction of this type based on number. The traditional labels are again slightly misleading, in that the dual is not strictly for two but may be used for a small number (Keren Rice, personal communication).”
Nuu-chah-nulth has clear Sg-Pl alternation for several of its verbs, and without the additional information described below, we would preferentially code the number distinction as SG-PL. However, corpus data leads us to believe that the distinction is rather INDIVIDUAL-COLLECTIVE, as the putative “singular” verbs can occasionally be used with plural participants, specifically in contexts of non-collective action. An example of this can be found with the verb pair mat- ~ hu- ‘fly’ in Sapir and Swadesh (1955: 13, 32), in which a group of geese first fly together as a collective (hu-) and then, later in the story, fly in a disorganized and individual manner (mat-) due to being blinded (14).
| (14) | a. | ʔuunuuƛ | ʔiiḥ | ʔuusa-ʕaƛ | hu-kʷis-saƛ=!aƛ=quu |
|---|---|---|---|---|---|
| because | big/very | loud-make.a.sound | fly.PL-up-on.beach.PF=NOW=PSSB.3 |
| ʔuyi | ʔuunuuƛ | ʔiiḥ | ʔaya-L.uu | hu-kʷiƛ | ḥuqsim | ||
|---|---|---|---|---|---|---|---|
| at.a.time | because | big/very | many-as.one | fly.PL-PF | goose |
| ‘There were so many geese flying up from the beach that it made a loud sound.’ | |||||
| b. | ʔink-R<c>L.sawiƛ=!aƛ | maamaati | ḥi-siła-iičiƛ=!aaḥ=!a | |
|---|---|---|---|---|
| fire-at.eyes.PF=NOW | bird | unable.to-do-PF-IRREALIS=NOW |
| saya=ʔii | mat-aa | tak-ʕaaʔatu-LS=!aƛ | ʔuunuuƛ | ||
|---|---|---|---|---|---|
| far=go.to | fly.SG-CONT | straight-downward-ONGOING=NOW | because |
| ʔink-R<c>L.sawiƛ=!aƛ=qa | ||
|---|---|---|
| fire-at.eyes.PF=NOW=EMBED |
| ‘The birds, blinded by the fires, were unable to fly far.’ | |||||
Note that the determination we have made about the number semantics of Nuu-chah-nulth Sg-Pl alternation was only possible due to our access to a large corpus.
{ yes | no }
Elements encoding number only (i.e. monoexponential plurals) are excluded. Elements encoding an extra semantic distinction like person are also excluded. The verbal morphology targeted by SgPl-19 is, for instance, the Ainu pair of causative forms -re ‘CAUS.SG’ ~ -yar ‘CAUS.PL’ (see below).
The causative has two forms: -(r)e~te for singular causees, and -(y)ar for plural ones (Shibatani 1999: 48-49).
| (15) | a. | ne | |
|---|---|---|---|
| b. | ne-re | ‘cause X to become Y’ | |
| c. | e | ‘eat’ | |
| d. | e-re | ‘cause X to eat’ | |
| e. | kor | ‘have’ | |
| f. | kor-e | ‘give’ |
| (16) | a. | hopum-pa | ‘people get up’ |
|---|---|---|---|
| b. | hopum-pa-yar | ‘cause people to get up’ | |
| c. | sitoma | ‘to be afraid’ | |
| d. | sitom-yar | ‘cause people to become afraid’ | |
| e. | nukar | ‘see’ | |
| f. | nukar-ar | ‘cause people to see’ |
The imperfective narrative has a Sg and a Pl form, nĩgnĩ ~ nỹgnĩ (Gonçalves 2011: 235).
{ NA | yes | no }
For Mithun (1988: 214), “the primary function of stem alternation is not to enumerate entities, but to quantify the effect of actions, states, and events” (see also Durie 1986 and Veselinova 2013). Consequently, we expected the marking of Sg-Pl alternation to sometimes also mark iterativity and event repetition.
The verbal pairs that alternate via regular pattern sometimes can indicate event repetition, although proper suppletion only signifies participant number (Birchall et al in review).
The verbal pairs in Shipibo-Conibo only signify the number of participants involved and transitivity (Valenzuela 2003: 150).
{ NA | yes | in property concept verbs only | no }
Frequent suppletive adjectives are ‘small’, ‘long’, ‘short’, ‘tall’, ‘extreme’, ‘ugly’, ‘sick’, ‘dead’ (Booker 1982). In cases where no suppletive adjectives were mentioned in the grammar and a dictionary was available, we checked in the dictionary for these adjectives, to confirm that there was no suppletion. As with verbs, we considered a limited set of adjectives with an irregular Sg-Pl morphology as long as the set was not semantically limited.
If a language uses verbs to express such concepts and those verbs alternate according to number, then we coded <in property concept verbs only>, a special case which allowed us to separately track Sg-Pl alternation in a distinct adjective class (<yes>), as well as its presence in property words, regardless of their part-of-speech (<yes> and <in property concept verbs only>).
In Grass Koiari, some adjectives have a specific plural form achieved by reduplication. However, all DIMENSION, AGE and VALUE adjectives are targets of this reduplication phenomenon, which is semantically (not lexically) limited and thus is excluded from our definition of Sg-Pl alternation (Dutton 2003: 339; Dutton 1996: 43).
In Krongo, several property verbs are sensitive to number: ‘be short’, ‘be pointy’, ‘be short’, ‘be pointy’, ‘be dull’, ‘be big/tall’, ‘be big/fat’, ‘be large’, ‘be narrow’, ‘be half full’, ‘be empty.’
Nuu-chah-nulth has a limited set of adjectives that have singular and plural forms, drawn from a large inventory of irregular plural marking for nouns. These are ʔiiḥ ~ ʔaʔiiḥ ‘big’, ƛaʔuu ~ ƛaƛuu ‘other’, ƛuł ~ ƛutƛuuł ‘good’, maquuł ~ maatquuḥ ‘blind’, kʷisḥii ~ kʷiy̓aasḥi ‘different’, kʷatyiik ~ kʷay̓aatyik ‘heavy’, cumaa ~ cucum ‘full’, and perhaps a few others (Inman 2021). Most adjectives however do not have plural forms.4
{ Sg-Pl | Sg-Pl and irregular MonPl | irregular MonPl | none }
| Sg-Pl | if SgPl-01 is <yes> and MonPl-01 is <no> |
| Sg-Pl and irregular MonPl | if SgPl-01 is <yes> and MonPl-01 is <yes> |
| irregular MonPl | if SgPl-01 is <no> and MonPl-01 is <no> |
| none | if SgPl-01 is <no> and MonPl-01 is <no> |
{ yes | no }
| yes | if SgPl-01 is <yes> or SgPl-19 is <yes> |
| Sg-Pl and irregular MonPl | if SgPl-01 is <no> and SgPl-19 is <no> |
This feature unites languages with Sg-Pl alternation in lexical verbs and languages with Sg-Pl alternation in verbal morphology. The data suggest that these two phenomena are diachronically linked. Indeed, in our sample only 6 languages display Sg-Pl alternation in verbal morphology only, while the 27 other languages which have it display Sg-Pl alternation in both lexical verbs and verbal morphology. In addition, in several cases, the alternating verbal morphology clearly comes from alternating verb roots, as in Yaqui, where the associated motion morpheme -sime ~ -saka was grammaticalized from the verb pair for ‘go’ síme ~ sáka (Dedrick and Casad (1999:293-294).
{ NA | yes | no }
| yes | if SgPl-16 is <unergative> or <both> |
| no | if SgPl-16 is <unaccusative> |
| NA | if SgPl-16 is <NA> |
{ NA | yes | no }
| yes | if SgPl-16 is <unaccusative> or <both> |
| no | if SgPl-16 is <unergative> |
| NA | if SgPl-16 is <NA> |
{ NA | yes | no }
| yes | if SgPl-18 contains <SG-DU-PL> or <SG-PAUCAL-PL> |
| no | if SgPl-18 does not contain <SG-DU-PL> or <SG-PAUCAL-PL> |
| NA | if SgPl-18 is <NA> |
This section is a summary of our findings for the survey on the ATLAs sample, (for more detailed information, see Inman & Vuillermet, submitted).
Distribution: Our data show that the phenomenon is fairly frequent around the world, in about a third of the languages. Confirming previous studies, SgPl stem alternation (SgPl-01) is overwhelmingly present in North America, in over half (52%) of the 115 North American languages in our sample. Less expectedly, South America is the macroregion with the lowest percentage of cases (17%), but a cluster of languages around the Brazilian state of Rondônia (part of the proposed Guaporé-Mamoré linguistic area (Crevels and van der Voort (2008)) stands out as a hotspot of Sg-Pl alternation. Coastal eastern Papunesia is another strong language cluster displaying the phenomenon.
Typological tendencies: Most languages displaying the phenomenon have a handful of alternating verb pairs (SgPl-02 & 04), and their number distinction overwhelmingly contrasts singular vs plural (SgPl-18). (Individual vs collective semantics are however plausibly underrepresented in existing grammatical descriptions.) The data confirmed previous works (Kinkade 1981; Booker 1982; Durie 1986; Mithun 1988; Veselinova 2006, Veselinova 2013; Krasnoukhova 2022) that the verb pair semantics primarily target Motion Events (Talmy 1972) and the following five verbs ‘die’, ‘kill’, ‘cry’, ‘sleep’ and ‘eat’ (SgPl-06; only six languages do not). Our detailed investigation further showed that semantically specific verbs spread across close neighboring languages, and reveal highly local areas of SgPl alternation across North America, as well the two local areas mentioned above in South America and Papunesia (SgPl-06, 08, 10 & 12).
Most languages have alternating intransitive verb pairs only, or both intransitive and transitive ones (SgPl-07). If the verb pair is transitive, its number value targets the P argument (SgPl-17). Often described as impossible in the literature, we still found five languages with transitive-only verb pairs, and also five (almost non-overlapping) languages whose number value targets the A argument.
We (wrongly) expected that languages which lack Sg-Pl stem alternation would also lack such alternating suffixes (SgPl-19). One reason for this is that we thought that such suffixes would be grammaticalized from Sg-Pl stems. However, this feature was coded for all languages, irrespective of their state in SgPl-01, and the results show that these two phenomena are not as intertwined as we expected.
Similarly, much of the literature expects a correlation between participant number and event number. However, we did not find very many languages in which the same form was used to mark both participant number and aspect (SgPl-20).
Conceptualization: David Inman, Marine Vuillermet
Data collection: Anna Graff, Selma Hardegger, Marine Vuillermet, David Inman
Supervision of data collection: David Inman, Marine Vuillermet
Birchall, Josh, David Inman & Marcelina Oro Waram Xijein. 2025. Verbal number in Wari'. International Journal of American Linguistics tbd. tbd.
Boas, Franz. 1911. Tsimshian. In Franz Boas (ed.), Handbook of American Indian languages: Volume 1, 283–422. Washington: Government Printing Office.
Booker, Karen M. 1982. Number suppletion in North American Indian languages. Kansas Working Papers in Linguistics 7. 15–29.
Corbett, Greville C. 2000. Number. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press.
Crevels, Mily. 2006. Verbal Number in Itonama. In Grażyna J. Rowicka and Eithne B. Carlin (eds.), What's in a Verb?, 159–170. LOT, Utrecht University.
Crevels, Mily & Hein van der Voort. 2008. The Guaporé-Mamoré region as a linguistic area. In Muysken, Pieter (ed.), From linguistic areas to areal linguistics, 151–179. Amsterdam/Philadelphia: John Benjamins.
Dedrick, John M. & Eugene H. Casad. 1999. Sonora Yaqui language structures. Tucson: The University of Arizona Press. xxiii+411pp.
Durie, Mark. 1986. The Grammaticization of Number as a Verbal Category. In Vassiliki Nikiforidou, Mary VanClay and Niepokuj, Mary and Feder, Deborah (eds.), Proceedings of the twelfth Annual Meeting of the Berkeley Linguistics Society, 15-17 February 1986, 355–370. Berkeley, California: Berkeley Linguistics Society Publications.
Dutton, Tom E. 1996. Koiari. (Languages of the World/Materials, 10.) München: Lincom. 77pp.
Dutton, Tom. 2003. A Dictionary of Koiari, Papua New Guinea: with grammar notes. (Pacific Linguistics, 534.) Canberra: Research School of Pacific and Asian Studies, Australian National University. xxvi+424pp.
Hollenbach, Elena Erickson de. 2013. Gramática del mixteco de Magdalena Peñasco (Sa'an Ñuu Savi). México, D.F.: Instituto Lingüístico de Verano. 475pp.
Gonçalves, Solange Aparecida. 2011. Tempo, aspecto e modo em contextos discursivos no Kaingang Sul (Jê). Universidade Estadual de Campinas dissertation. (xx+294pp.)
Heath, Jeffrey. 1984. Functional Grammar of Nunggubuyu. (AIAS new series, 53.) Canberra: Australian Institute of Aboriginal Studies. xiv+664pp.
Project, Hopi Dictionary. 1998. Hopi dictionary=hopiikwa lavaytutuveni: a Hopi-English dictionary of the Third Mesa dialect with an English-Hopi finder list and a sketch of Hopi grammar. In Kenneth C. Hill and Emory Sekaquaptewa and Mary E. Black and Ekkehart Malotki (eds.) The University of Arizona Press.
Inman, David. 2021. The Exponence and Development of Plurals in Nuuchahnulth. Papers for the 56th International Conference on Salish and Neighbo(u)ring Languages. pp. 111–133.
Kinkade, M Dale. 1981. Singular vs. plural roots in Salish. Anthropological Linguistics 23(6). 262–269.
Krasnoukhova, Olga. 2022. Number in the languages of South America. In Paolo Acquaviva and Michael Daniel (eds.), Number in the World's Languages: A Comparative Handbook, 609–670. Berlin, Boston: De Gruyter Mouton.
Lindenfeld, Jacqueline. 1973. Yaqui Syntax. (University of California Publications in Linguistics, 76.) Berkeley and Los Angeles: University of California Press. ix+162pp.
Marlett, Stephen A. 1981. The Structure of Seri. San Diego: University of California at San Diego dissertation. (xii+229pp.)
Miller, Wick R. 1965. Acoma grammar and texts. (University of California Publications in Linguistics, 40.) Berkeley and Los Angeles: University of California Press. 259pp.
Mithun, Marianne. 1988. Lexical categories and the evolution of number marking. In Michael Hammond (ed.), Theoretical morphology: Approaches in modern linguistics. Nueva York: Academic Press.
Mithun, Marianne. 2022. Sitting and talking together: packaging meaning into verbs with the neighbors. Linguistic Typology 26(2). 375–402.
Mushin, Ilana. 2012. A grammar of (Western) Garrwa. (Pacific Linguistics, 637.) Berlin, Boston: De Gruyter Mouton. xv+476pp.
Nakagawa, Hiroshi. 2022. 17 Verbal number. In Bugaeva, Anna (ed.), Handbook of the Ainu Language, 573–612. De Gruyter Mouton.
Nichols, Johanna. 2011. Ingush Grammar. (University of California Publications Series.) University of California Press. xviii+806pp.
Park, Indrek. 2012. A grammar of Hidatsa. Indiana University dissertation. (xxiii+557pp.)
Rice, Keren. 1989. A Grammar of Slave. (Mouton Grammar Library, 5.) Berlin, New York: Mouton de Gruyter. xliv+1370pp.
Sapir, Edward & Morris Swadesh. 1955. Native accounts of Nootka ethnography. (1-6.) Indiana University, Research Center in Anthropology, Folklore, and Linguistics.
Seifart, Frank. 2005. The Structure and Use of Shape-Based Noun Classes in Miraña (North West Amazon). 32. Nijmegen: Radboud Universiteit Nijmegen dissertation. (xiii+375pp.)
Seiler, Walter. 1985. Imonda, a Papuan Language. (Pacific Linguistics: Series B, 93.) Canberra, Australia: Canberra: Research School of Pacific and Asian Studies, Australian National University. v+236+21pp.
Shaul, David L. 2019. Esselen Studies: Language, Culture, and Prehistory. (LINCOM Studies in Native American Linguistics, 83.) München: LINCOM. ix+431pp.
Shibatani, Masayoshi. 1999. The languages of Japan. (Cambridge Language Surveys.) Cambridge: Cambridge University Press. xv+411pp.
Suárez, Jorge A. 1983. La Lengua Tlapaneca de Malinaltepec. México: Instituto de Investigaciones Filológicas, Universidad Nacional Autónoma de México. 641pp.
Talmy, Leonard. 1972. Semantic structures in English and Atsugewi. University of California. Berkeley dissertation.
Tarpent, Marie-Lucie. 1983. Morphophonemics of Nisgha plural formation: A step towards Proto-Tsimshian reconstruction. Kansas Working Papers in Linguistics 8(2). University of Kansas. Linguistics Graduate Student Association.
Tarpent, Marie-Lucie. 1987. A grammar of the Nisgha language. University of Victoria dissertation.
Valentine, J. Randolph. 2001. A reference grammar of Nishnaabemwin. Toronto: University of Toronto Press. xxxvii+1100pp.
Valenzuela, Pilar M. 2003. Transitivity in Shipibo-Konibo Grammar. Eugene: University of Oregon dissertation. (xxxiv+1029pp.)
Gijn, Rik van. 2006. A Grammar of Yurakaré. Nijmegen: Radboud Universiteit Nijmegen dissertation. (xi+361pp.)
Veselinova, Ljuba N. 2006. Suppletion in verb paradigms: Bits and pieces of the puzzle. (Typological studies in language, v. 67.) Amsterdam; Philadelphia: J. Benjamins.
Veselinova, Ljuba N. 2013. Verbal Number and Suppletion (v2020.3). In Matthew S. Dryer and Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Zenodo.
Voort, Hein van der. 2004. A Grammar of Kwaza. (Mouton Grammar Library, 29.) Berlin, New York: Mouton de Gruyter. xxxviii+1026pp.
At least in Magdalena Peñasco Mixtec, this semantic restriction appears to be motivated by the historical origin of koo from koyo ‘pour’, or metaphorically ‘movement in a group, pouring out’ (Erickson de Hollenbach 2013: 151). ↩
Note that in Table 2, suffixes encode both participant and event number. In Table 3, suffixes again encode both participant and event number, but root alternation encodes participant number only. ↩
SgPl-21 is about alternation in adjectives, but only if it is also present in verbs. As a result, if alternation occurs only in a part of speech distinct from verbs, namely adjectives, then it is not reported in this feature set. ↩
Note that there is a strong preference in Nuu-chah-nulth for anti-agreement, i.e. to indicate plural only once per referent. It is unusual for these plural adjectives to co-occur with plural marking on the noun, and thus a full list of all alternating adjectives remains elusive. ↩