In linguistics and etymology, suppletion is traditionally understood as the use of one word as the inflected form of another word when the two words are not cognate. For those learning a language, suppletive forms will be seen as "irregular" or even "highly irregular". For example, go:went is a suppletive paradigm, because go and went are not etymologically related, whereas mouse:mice is irregular but not suppletive, since the two words come from the same proto-Indo-European ancestor.
The term "suppletion" implies that a gap in the paradigm was filled by a form "supplied" by a different paradigm. Instances of suppletion are overwhelmingly restricted to the most commonly used lexical items in a language.
An irregular paradigm is one in which the derived forms of a word cannot be deduced by simple rules from the base form. For example, someone who knows only a little English can deduce that the plural of girl is girls but cannot deduce that the plural of man is men. Language learners are often most aware of irregular verbs, but any part of speech with inflections can be irregular.
For most synchronic purposes—first-language acquisition studies, psycholinguistics, language-teaching theory—it suffices to note that these forms are irregular. However, historical linguistics seeks to explain how they came to be so and distinguishes different kinds of irregularity according to their origins.
Most irregular paradigms (like man:men) can be explained by phonological developments that affected one form of a word but not another (in this case, Germanic umlaut). In such cases, the historical antecedents of the current forms once constituted a regular paradigm.
Historical linguistics uses the term "suppletion"to distinguish irregularities like person:people or cow:cattle that cannot be so explained because the parts of the paradigm have not evolved out of a single form.
Hermann Osthoff coined the term "suppletion" in German in an 1899 study of the phenomenon in Indo-European languages.[1] [2] [3]
Suppletion exists in many languages around the world.[4] These languages are from various language families: Indo-Aryan, Dravidian, Arabic, Romance, etc.
For example, in Georgian, the paradigm for the verb "to come" is composed of four different roots (and).[5]
Similarly, in Modern Standard Arabic, the verb ("come") usually uses the form for its imperative, and the plural of ("woman") is .
Some of the more archaic Indo-European languages are particularly known for suppletion. Ancient Greek, for example, has some twenty verbs with suppletive paradigms, many with three separate roots.
In English, the past tense of the verb go is went, which comes from the past tense of the verb wend, archaic in this sense. (The modern past tense of wend is wended.) See Go (verb).
The Romance languages have a variety of suppletive forms in conjugating the verb "to go", as these first-person singular forms illustrate (second-person singular forms in imperative):[6]
| Language | Imperative | Present | Subjunctive | Future | Preterite | Infinitive | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| French | French: va, French: vas-y | 1 | French: vais | 1 | French: aille | 4 | French: irai | 2 | French: allai | 4 | 4 | |
| Romansh (Sursilvan) | Romansh: va | 1 | Romansh: mon | 6 | Romansh: mondi | 6 | - | - | 2 | |||
| Sardinian (Logudorese) | Romansh: bai | 1 | Romansh: ando | 3 | Romansh: andaia, Romansh: andaio | 3 | - | - | 3 | |||
| Italian | Italian: vai, Italian: va, Italian: va' | 1 | Italian: vado, Italian: vo | 1 | Italian: vada | 1 | Italian: andrò | 3 | Italian: andai | 3 | 3 | |
| Occitan (Languedocien) | Occitan (post 1500);: vai | 1 | Occitan (post 1500);: vau | 1 | Occitan (post 1500);: ane | 3 | Occitan (post 1500);: anarai | 3 | Occitan (post 1500);: anèri | 3 | 3 | |
| Catalan | Catalan; Valencian: vès | 1 | Catalan; Valencian: vaig | 1 | Catalan; Valencian: vagi | 1 | Catalan; Valencian: aniré | 3 | Catalan; Valencian: aní | 3 | 3 | |
| Spanish | Spanish; Castilian: vetú | 1 | Spanish; Castilian: voy | 1 | Spanish; Castilian: vaya | 1 | Spanish; Castilian: iré | 2 | Spanish; Castilian: fui | 5 | 2 | |
| Spanish; Castilian: andávos | 3 | |||||||||||
| Portuguese | Portuguese: vaitu | 1 | Portuguese: vou | 1 | Portuguese: vá | 1 | Portuguese: irei | 2 | Portuguese: fui | 5 | 2 | |
| Portuguese: idevós | 2 | |||||||||||
Many of the Romance languages use forms from different verbs in the present tense; for example, French has French: je vais ‘I go’ from Latin: vadere, but French: nous allons ‘we go’ from Latin: ambulare. Galician-Portuguese has a similar example: Galician: imos from Latin: ire ‘to go’ and Galician: vamos from Latin: vadere ‘we go’; the former is somewhat disused in modern Portuguese but very alive in modern Galician. Even Galician: ides, from Latin: itis second-person plural of Latin: ire, is the only form for ‘you (plural) go’ both in Galician and Portuguese (Spanish Spanish; Castilian: vais, from Latin: vadere).
Sometimes, the conjugations differ between dialects. For instance, the Limba Sarda Comuna standard of Sardinian supported a fully regular conjugation of Sardinian: andare, but other dialects like Logudorese do not (see also Sardinian conjugation). In Romansh, Rumantsch Grischun substitutes present and subjunctive forms of ir with vom and giaja (both are from Latin vādere and īre, respectively) in the place of mon and mondi in Sursilvan.
Similarly, the Welsh verb Welsh: mynd ‘to go’ has a variety of suppletive forms such as Welsh: af ‘I shall go’ and Welsh: euthum ‘we went’. Irish Welsh: téigh ‘to go’ also has suppletive forms: Welsh: dul ‘going’ and Welsh: rachaidh ‘will go’.
In Estonian, the inflected forms of the verb Estonian: minema ‘to go’ were originally those of a verb cognate with the Finnish Finnish: lähteä ‘to leave’, except for the passive and infinitive.
In Germanic, Romance (except Romanian), Celtic, Slavic (except Bulgarian and Macedonian), and Indo-Iranian languages, the comparative and superlative of the adjective "good" is suppletive; in many of these languages the adjective "bad" is also suppletive.
| + good, better, best | ||||||
| Language | Adjective | Etymology | Comparative | Superlative | Etymology | |
|---|---|---|---|---|---|---|
| Germanic languages | ||||||
| English | good | Proto-Germanic
cognate to Sanskrit: gadhya||what one clings to|script=Latn | better | best | Proto-Germanic:
cognate to Sanskrit: bhadra "fortunate" | |
| Danish | Danish: god | Danish: bedre | Danish: bedst | |||
| German | German: gut | German: besser | German: besten | |||
| Faroese | Faroese: góður | Faroese: betri | Faroese: bestur | |||
| Icelandic | Icelandic: góður | Icelandic: betri | Icelandic: bestur | |||
| Dutch | Dutch; Flemish: goed | Dutch; Flemish: beter | Dutch; Flemish: best | |||
| Norwegian Bokmål | Norwegian: god | Norwegian: bedre | Norwegian: best | |||
| Norwegian Nynorsk | god | betre | best | |||
| Swedish | Swedish: god | Swedish: bättre | Swedish: bäst | |||
| Romance languages | ||||||
| French | French: bon | Latin: bonusfrom Old Latin: Latin: duenos
| French: meilleur |
| ||
| Portuguese | Portuguese: bom | Portuguese: melhor | ||||
| Spanish | Spanish; Castilian: bueno | Spanish; Castilian: mejor | ||||
| Catalan | Catalan; Valencian: bo | Catalan; Valencian: millor | ||||
| Italian | Italian: buono | Italian: migliore | ||||
| Celtic languages | ||||||
| Scottish Gaelic | Gaelic; Scottish Gaelic: math | Proto-Celtic Celtic languages: *matisfrom Proto-Indo-European: * "ripen", "mature" | Gaelic; Scottish Gaelic: feàrr | Proto-Celtic Celtic languages: *werrosfrom Proto-Indo-European: * "peak" | ||
| Irish | Irish: maith | Irish: fearr | ||||
| Breton | Breton: mat | Breton: gwell, Breton: gwelloc'h (1) | Breton: gwellañ (1) |
| ||
| Welsh | Welsh: da | Proto-Celtic: Celtic languages: *dagos "good", "well" | Welsh: gwell (1) | Welsh: gorau (2) | ||
| Slavic languages | ||||||
| Polish | Polish: dobry | Proto-Slavic: Slavic languages: *dobrъ | Polish: lepszy | Polish: najlepszy | Proto-Indo-European *, * "behoof", "boot", "good" | |
| Czech | Czech: dobrý | Czech: lepší | Czech: nejlepší | |||
| Slovak | Slovak: dobrý | Slovak: lepší | Slovak: najlepší | |||
| Ukrainian | Ukrainian: добрий | Ukrainian: ліпший | Ukrainian: найліпший | |||
| Serbo-Croatian | dobar | bolji | najbolji | Proto-Slavic: Slavic languages: *bolьjь "bigger" | ||
| Slovene | Slovenian: dober | Slovenian: boljši | Slovenian: najboljši | |||
| Russian | Russian: хороший|khoroshiy|label=none | probably from Proto-Slavic: Slavic languages: *xorb[12] | Russian: лучше|luchshe|label=none | Russian: (наи)лучший|(nai)luchshiy|label=none | Old Russian лучии, neut. лучеOld Church Slavonic: Church Slavic; Old Slavonic; Church Slavonic; Old Bulgarian; Old Church Slavonic: лоучии "more suitable, appropriate" | |
| Indo-Iranian languages | ||||||
| Persian | Persian: خوب|khūb|label=none pronounced as /[xʊb]/ | probably cognate of Proto-Slavic Slavic languages: *xorb (above). Not a satisfactory etymology for Persian, Old (ca.600-400 B.C.);: beh; but see comparative and superlative forms in comparison to Germanic | Persian: خوبتر|xūb-tar|label=none or Persian: بِهْتَر|beh-tar|label=none | Persian: خوبترین|xūb-tarīn|label=none or Persian: بِهْتَرين|beh-tarīn|label=none | From Proto-Indo-Iranian *Hwásuš "good". Not a cognate of the Germanic forms above. | |
| Non-Indo-European languages | ||||||
| Georgian | კარგი, k'argi [kʼäɾgi]. | possibly an Iranian borrowing via Old Armenian կարգ (karg, “order”). | უკეთესი, uk'etesi [uk'e̞tʰe̞si]. | საუკეთესო, sauk'eteso [säuk'e̞tʰe̞so̞]. | From Proto-Georgian-Zan *ḳet- “to add, mix”. | |
| + bad, worse, worst | Language | Adjective | Etymology | Comparative | Superlative | Etymology | |
|---|---|---|---|---|---|---|---|
| Germanic languages | |||||||
| English | bad | Uncertain, possibly from OE English, Old (ca.450-1100);: bæddel ("effeminate man, hermaphrodite, pederast"), related to OE English, Old (ca.450-1100);: bædan ("to defile") < Proto-Germanic Germanic languages: *baidijaną ("constrain, cause to stay") In OE English, Old (ca.450-1100);: yfel was more common, compare Proto-Germanic Germanic languages: *ubilaz, Gothic Gothic: ubils (bad), German German: übel (evil / bad) Eng evil | worse | worst | From Proto-Germanic *wirsizô, *wirsistaz. | ||
| Old Norse | Norse, Old: vándr | From Proto-Germanic *wanh-. | Norse, Old: verri | Norse, Old: verstr | |||
| Icelandic | Icelandic: vondur | Icelandic: verri | Icelandic: verstur | ||||
| Faroese | Faroese: óndur | Faroese: verri | Faroese: verstur | ||||
| Norwegian Bokmål | Norwegian: ond, Norwegian: vond | Norwegian: verre | Norwegian: verst(e) | ||||
| Norwegian Nynorsk | vond | verre | verst(e) | ||||
| Swedish | Swedish: ond | Swedish: värre | Swedish: värst | ||||
| Danish | Danish: ond | Danish: værre | Danish: værst | ||||
| Romance languages | |||||||
| French | French: mal | Latin: malus | French: pire | Latin: peior, cognate to Sanskrit Sanskrit: padyate "he falls" | |||
| Portuguese | Portuguese: mau | Portuguese: pior | |||||
| Spanish | Spanish; Castilian: malo | Spanish; Castilian: peor | |||||
| Catalan | Catalan; Valencian: mal | Catalan; Valencian: pitjor | |||||
| Italian | Italian: male | Italian: peggiore | |||||
| Celtic languages | |||||||
| Scottish Gaelic | Gaelic; Scottish Gaelic: droch | Proto-Celtic Celtic languages: *drukos ("bad") < (possibly) PIE * ("to deceive") | Gaelic; Scottish Gaelic: miosa | Proto-Celtic Celtic languages: *missos < PIE *- ("to change") | |||
| Irish | Irish: droch | Irish: measa | |||||
| Welsh | Welsh: drwg | Welsh: gwaeth | Welsh: gwaethaf | Proto-Celtic Celtic languages: *waxtisamos ("worst") | |||
| Slavic languages | |||||||
| Polish | Polish: zły | Proto-Slavic Slavic languages: *zel | Polish: gorszy | Polish: najgorszy | compare Polish Polish: gorszyć (to disgust, scandalise) | ||
| Czech | Czech: zlý (špatný) | Czech: horší | Czech: nejhorší | ||||
| Slovak | Slovak: zlý | Slovak: horší | Slovak: najhorší | ||||
| Ukrainian | archaic Ukrainian: злий | Ukrainian: гірший | Ukrainian: найгірший | ||||
| Serbo-Croatian | zao | gori | najgori | ||||
| Russian | Russian: плохой (Russian: plokhoy) | probably Proto-Slavic Slavic languages: *polx | Russian: хуже (Russian: khuzhe) | Russian: (наи)худший (Russian: (nai)khudshiy) | Old Church Slavonic Church Slavic; Old Slavonic; Church Slavonic; Old Bulgarian; Old Church Slavonic: хоудъ, Proto-Slavic Slavic languages: *хudъ ("bad", "small") | ||
Celtic languages:
| + small, smaller, smallest | Language | Adjective | Comparative / superlative |
|---|---|---|---|
| Irish | beag (Old Irish bec < Proto-Celtic *bikkos) | níos lú / is lú (< Old Irish laigiu < Proto-Celtic *lagyūs < PIE *h₁lengʷʰ- ("lightweight")) | |
| Welsh | bach (< Brythonic *bɨx < Proto-Celtic *bikkos) | llai / lleiaf (< PIE *h₁lengʷʰ- (“lightweight”)) |
| + great, greater, greatest | Language | Adjective | Comparative / superlative |
|---|---|---|---|
| Irish | mór (< Proto-Celtic *māros < PIE *moh₁ros) | níos mó / is mó < Proto-Celtic *māyos < PIE *meh₁-) | |
| Welsh | mawr (< Proto-Celtic *māros < PIE *moh₁ros) | mwy / mwyaf < Proto-Celtic *māyos < PIE *meh₁-) |
In many Slavic languages, great and small are suppletive:
| + small, smaller, smallest | Language | Adjective | Comparative / superlative |
|---|---|---|---|
| Polish | mały | mniejszy / najmniejszy | |
| Czech | malý | menší / nejmenší | |
| Slovak | malý | menší / najmenší | |
| Slovene | majhen | manjši / najmanjši | |
| Ukrainian | малий, маленький | менший / найменший | |
| Russian | маленький (malen'kiy) | меньший / наименьший (men'she / naimen'shiy) |
| + great, greater, greatest | Language | Adjective | Comparative / superlative |
|---|---|---|---|
| Polish | duży | większy / największy | |
| Czech | velký | větší / největší | |
| Slovak | veľký | väčší / najväčší | |
| Slovene | velik | večji / največji | |
| Ukrainian | великий | більший / найбільший |
In Albanian there are 14 irregular verbs divided into suppletive and non-suppletive:
| Verb | Meaning | Present | Preterite | Imperfect | |
|---|---|---|---|---|---|
| Albanian: qenë | to be | Albanian: jam | Albanian: qeshë | Albanian: isha | |
| Albanian: pasur | to have | Albanian: kam | Albanian: pata | Albanian: kisha | |
| Albanian: ngrënë | to eat | Albanian: ha | Albanian: hëngra | Albanian: haja | |
| Albanian: ardhur | to come | Albanian: vij | Albanian: erdha | Albanian: vija | |
| Albanian: dhënë | to give | Albanian: jap | Albanian: dhashë | Albanian: jepja | |
| Albanian: parë | to see | Albanian: shoh | Albanian: pashë | Albanian: shihja | |
| Albanian: rënë | to fall, strike | Albanian: bie | Albanian: rashë | Albanian: bija | |
| Albanian: prurë | to bring | Albanian: bie | Albanian: prura | Albanian: bija | |
| Albanian: ndenjur | to stay | Albanian: rri | Albanian: ndenja | Albanian: rrija |
Ancient Greek had a large number of suppletive verbs. A few examples, listed by principal parts:
In Bulgarian, the word Bulgarian: човек|chovek|label=none ("man", "human being") is suppletive. The strict plural form, Bulgarian: човеци|chovetsi|label=none, is used only in Biblical context. In modern usage it has been replaced by the Greek loan Greek, Modern (1453-);: хора|khora|label=none. The counter form (the special form for masculine nouns, used after numerals) is suppletive as well: Bulgarian: души|dushi|label=none (with the accent on the first syllable). For example, Bulgarian: двама, трима души|dvama, trima dushi|label=none ("two, three people"); this form has no singular either. (A related but different noun is the plural Bulgarian: души|dushi|label=none, singular Bulgarian: душа|dusha|label=none ("soul"), both with accent on the last syllable.)
In English, the complicated irregular verb to be has forms from several different roots:
This verb is suppletive in most Indo-European languages, as well as in some non-Indo-European languages such as Finnish.
An incomplete suppletion exists in English with the plural of person (from the Latin Latin: persona). The regular plural persons occurs mainly in legalistic use. More commonly, the singular of the unrelated noun people (from Latin Latin: populus) is used as the plural; for example, "two people were living on a one-person salary" (note the plural verb). In its original sense of "populace, ethnic group", people is itself a singular noun with regular plural peoples.
Several irregular Irish verbs are suppletive:
There are several suppletive comparative and superlative forms in Irish; in addition to the ones listed above, there is:
See main article: Latin conjugation. Latin has several suppletive verbs. A few examples, listed by principal parts:
In some Slavic languages, a few verbs have imperfective and perfective forms arising from different roots. For example, in Polish:
| Verb | Perfective | ||
|---|---|---|---|
| to take | Polish: brać | Polish: wziąć | |
| to say | Polish: mówić | Polish: powiedzieć | |
| to see | Polish: widzieć | Polish: zobaczyć | |
| to watch | Polish: oglądać | Polish: obejrzeć | |
| to put | Polish: kłaść | Polish: położyć | |
| to find | Polish: znajdować | Polish: znaleźć | |
| to go in/to go out (on foot) | Polish: wchodzić, wychodzić | Polish: wejść, wyjść | |
| to ride in/to ride out (by car) | Polish: wjeżdżać, wyjeżdżać | Polish: wjechać, wyjechać |
In Polish, the plural form of Polish: rok ("year") is Polish: lata which comes from the plural of Polish: lato ("summer"). A similar suppletion occurs in Russian: год|god ("year") > Russian: лет|let|label=none (genitive of "years").
The Romanian verb Romanian; Moldavian; Moldovan: a fi ("to be") is suppletive and irregular, with the infinitive coming from Latin fieri, but conjugated forms from forms of already suppletive Latin sum. For example, Romanian; Moldavian; Moldovan: eu sunt ("I am"), Romanian; Moldavian; Moldovan: tu ești ("you are"), Romanian; Moldavian; Moldovan: eu am fost ("I have been"), Romanian; Moldavian; Moldovan: eu eram ("I used to be"), Romanian; Moldavian; Moldovan: eu fusei/fui ("I was"); while the subjunctive, also used to form the future in Romanian; Moldavian; Moldovan: o să fiu ("I will be/am going to be"), is linked to the infinitive.
In Russian, the word Russian: человек|chelovek|label=none ("man, human being") is suppletive. The strict plural form, Russian: человеки|cheloveki|label=none, is used only in Orthodox Church contexts, with numerals (e. g. Russian: пять человек|pyat chelovek|label=none "five people") and in humorous context. It may have originally been the unattested Russian: *человекы|*cheloveky|label=none. In any case, in modern usage, it has been replaced by Russian: люди|lyudi|label=none, the singular form of which is known in Russian only as a component of compound words (such as Russian: простолюдин|prostolyudin|label=none). This suppletion also exists in Polish (Polish: człowiek > Polish: ludzie), Czech (Czech: člověk > Czech: lidé), Serbo-Croatian (Serbian: čovjek > Serbian: ljudi),[17] Slovene (Slovenian: človek > Slovenian: ljudje), and Macedonian (Macedonian: човек > Macedonian: луѓе).
Strictly speaking, suppletion occurs when different inflections of a lexeme (i.e., with the same lexical category) have etymologically unrelated stems. The term is also used in looser senses, albeit less formally.
The term "suppletion" is also used in the looser sense when there is a semantic link between words but not an etymological one; unlike the strict inflectional sense, these may be in different lexical categories, such as noun/verb.[18] [19]
English noun/adjective pairs such as father/paternal or cow/bovine are also referred to as collateral adjectives. In this sense of the term, father/fatherly is non-suppletive. Fatherly is derived from father, while father/paternal is suppletive. Likewise cow/cowish is non-suppletive, while cow/bovine is suppletive.
In these cases, father/pater- and cow/bov- are cognate via Proto-Indo-European, but 'paternal' and 'bovine' are borrowings into English (via Old French and Latin). The pairs are distantly etymologically related, but the words are not from a single Modern English stem.
The term "weak suppletion" is sometimes used in contemporary synchronic morphology in reference to sets of stems whose alternations cannot be accounted for by synchronically productive phonological rules. For example, the two forms child/children are etymologically from the same source, but the alternation does not reflect any regular morphological process in modern English: this makes the pair appear to be suppletive, even though the forms go back to the same root.
In that understanding, English has abundant examples of weak suppletion in its verbal inflection: e.g. bring/brought, take/took, see/saw, etc. Even though the forms are etymologically related in each pair, no productive morphological rule can derive one form from the other in synchrony. Alternations just have to be learned by speakers — in much the same way as truly suppletive pairs such as go/went.
Such cases, which were traditionally simply labelled "irregular", are sometimes described with the term "weak suppletion", so as to restrict the term "suppletion" to etymologically unrelated stems.