Sotho verbs: Difference between revisions

Content deleted Content added

Inline

Revision as of 11:12, 30 October 2016

Sesotho verbs are words in the language that signify the action or state of a substantive, and are brought into agreement with it using the subjectival concord. This definition excludes imperatives and infinitives, which are respectively interjectives and class 14 nouns.

In the Bantu languages, verbs often form the centre of a complex web of regular derivational patterns, and words/roots belonging to many parts of speech may be directly or indirectly derived from them. Not only may new verbs be derived using a large number of derivational suffixes, nouns (and, iteratively, the other parts of speech that derive from them), some imperative interjectives and, to a lesser extent, ideophones may be formed by simple morphological devices.

Varieties

Verb stems may be divided into four varieties:

Regular stems beginning with a consonant and ending in a vowel
Monosyllabic verbs
Vowel verb stems begin with a vowel
Derived verbs constructed from other verbs, noun roots, adjectival roots, and ideophones by suffixes.

Regular verbs are those beginning with a consonant and ending in the vowel /ɑ/. The final /ɑ/ may change into every vowel except the near-close near-back vowel (/ʊ/) through inflexion or derivation. The verb root is the atomic part of the verb, which does not change (save for some purely phonetic changes) and Bantu languages share numerous similar verb roots (with predictable sound changes between languages).

Stem see, from root , also existing as isiZulu -bon-, Swahili -on-, Tshivenda -vhon-, Chishona -von-, Chilamba -won- etc. Proto-Bantu *-bon-

Monosyllabic stems may be classified into several categories:

The i-stems have a typical /i/ in derivatives, and /u/ in the passive
[t͡ɬʼɑ] come ⇒ Perfect [t͡ɬʼile], Causative [t͡ɬʼisɑ], Passive [t͡ɬʼuwɑ]

[jɑ] go ⇒ Perfect [ile], Causative [isɑ], Passive [uwɑ]

[xɑ] draw water ⇒ Perfect [xile], Causative [xisɑ], Passive [xuwɑ]
The e-stems^[1] have a typical near-close near-front /ɪ/ in their derivatives
[t͡ʃʰɑ] burn ⇒ Perfect [t͡ʃʰɪlɛ], Causative [t͡ʃʰɪsɑ], Passive [t͡ʃʰɪwɑ]

[ʒɑ] eat ⇒ Perfect [ʒɪlɛ], Causative [ʒɪsɑ], Passive [ʒɪwɑ]
The "velar" e-stems^[1] have labialized onsets, and have similar forms to other e-stems but have a near-close near-back vowel /ʊ/ in the passive
[nʷɑ] drink ⇒ Perfect [nʷɪlɛ], Causative [nʷɪsɑ], Passive [nʊwɑ]
There are three defective stems, ending in a vowel other than /ɑ/. The first two of these verbs are very common among the Bantu languages^[2]
[ʀɪ] say ⇒ Perfect [it͡sʼe], No causative, Passive [tʰʷɪ]

[lɪ] be; very restricted in use (only used in the participial sub-mood of certain copulatives)

[t͡ʃʰɔ] say so ⇒ Perfect [t͡ʃʰɪlɔ] / [t͡ʃʰʊlɔ], No causative, No passive

Vowel verb stems are conjugated as regular verbs but are put into a separate class due to being uncommon in Bantu languages (and, in some languages but not in Sesotho, causing changes to concords and other formatives prefixed to them).^[3] Class 1 and 5 nouns derived from these verbs do not cause any velarization to the prefix. The Proto-Bantu reconstructions of many of these verbs suggests that they originally began with *g (or sometimes *j), which "protected" the vowel.

avoid (as a taboo)

travel

hear, sense

construct

strike, punish

Tones

Verbs fall into only two categories when it comes to their tones: L verbs and H verbs. The difference lies in whether the "underlying tone" of the verb's first syllable is high or null (under-specified). Thus, all verbs of a certain length in the same tonal category are pronounced with similar tonal patterns under the same grammatical circumstances.

What the verbal tone system lacks in variety, however, it more than makes up for in complexity. The tones of the syllables of the verbs regularly change under varying grammatical environments, with the high tones being manipulated by "tonal rules", and the tones associated with certain syllables being changed by numerous "tonal melodies."

Verbal derivatives

Various derivatives may (recursively) be formed from verbs by means of several suffixes (called "extensions"). Each derived verb is as much an authentic verb as the original.

In the following sections, "polysyllabic" generally means "of more than two syllables."

Verbs are derived primarily through suffixes, some of which are no longer active ("dead").

Verb derivations using the root [ǃɛtʼ]^[4]
Type	Suffix	Valency change	Example	Meaning
Simple	[ɑ]	0	[ǃɛtʼɑ]	finish
Passive	[wɑ]	–1	[ǃɛtʼʷɑ]	be finished
Passive	[uwɑ]	–1	[ǃetʼuwɑ]	be finished
Neutro-active	[ɑhɑlɑ]	intr.	[ǃɛtʼɑhɑlɑ]	finishable
Neutro-passive	[ɛhɑ]	intr.	[ǃɛtʼɛhɑ]	finishable
Applied	[ɛlɑ]	+1	[ǃɛtʼɛlɑ]	finished for
Causative	[isɑ]	+1	[ǃetʼisɑ]	cause to finish
Intensive	[isisɑ]	0	[ǃetʼisisɑ]	finish intensely
Perfective	[ɛl̩lɑ]	0	[ǃɛtʼɛl̩lɑ]	finish completely
Reciprocal	[ɑnɑ]	–1	[ǃɛtʼɑnɑ]	finish each other
Associative	[ɑhɑnɑ]	–1	[ǃɛtʼɑhɑnɑ]	(be finished together)
Reversive	[ʊl̩lɑ]	0	[ǃetʼʊl̩lɑ]	unfinish
Augmentative	[ʊl̩lɑ]	0	[ǃetʼʊl̩lɑ]	(finish extensively)
Extensive	[ɑkʼɑ]	0	[ǃɛtʼɑkʼɑ]	(finish repeatedly and extensively)
Diminutive	(see text)	0	[ǃɛtʼɑǃɛtʼɑ]	finish a little
Positional (dead)	[ɑmɑ]	0	([t͡sʼʊʀɑmɑ])	(squat)
Stative extensive (dead)	[ɑlɑ]	0	([ʀɔbɑlɑ])	(sleep)
Contactive (dead)	[ɑʀɑ]	0	([fupʼɑʀɑ])	(clench the hand)

The passive indicates that the subject is acted upon by the agent, just like the "passive voice" in English. The agent is indicated by the copulative prefix [kʼɪ] although passives may also be used idiomatically without an agent.

The suffix may be either [wɑ] (Proto-Bantu *-u-) (short passive) or [uwɑ] (long passive).

The following rules are applied to form the passive:

The long passive is formed simply by changing the final [ɑ] to [uwɑ]
[bʊpʼɑ] mould ⇒ [buˌpʼuwɑ] be moulded
Many verbs accept the short passive suffix by simply becoming labialized
[et͡sʼɑ] do ⇒ [et͡sʼʷɑ] be done
Palatalization occurs where necessary (when the final consonant is /pʼ/, /pʰ/, /b/, or /f/)
[hɑpʼɑ] win ⇒ [hɑpʃʼʷɑ] / [hɑt͡ʃʼʷɑ] be won
Velarization occurs where necessary (when the final consonant is /m/ or /ɲ/)
[t͡sʼit͡sʼiɲɑ] move slightly ⇒ [t͡sʼit͡sʼiŋ̩ŋʷɑ] be moved slightly
Monosyllabic e-stems suffix [ewɑ] (except the velar e-stems ending in [wɑ], which suffix [ʊwɑ]) and i-stems suffix [uwɑ]
[fɑ] give ⇒ [fuwɑ] be given
Verbs ending in [jɑ] replace it with ['uwɑ]
[ɬʷɑjɑ] select, indicate ⇒ [ɬʷɑ'uwɑ] be selected
Verbs ending in [u'ɑ] replace it with [u'uwɑ]
[t͡sʼu'ɑ] judge, condemn ⇒ [t͡sʼu'uwɑ] be judged

It is very rare to have other verbs derived from the passive through suffixes.

This suffix has the effect of decreasing the valency of the verb and giving it an agentive import.

In the most formal standard language, the perfect of the passive is generally formed by inserting [ilʷ] before the final vowel of the perfect form (that is, the passive suffix has to come after the perfect suffix). In non-standard common speech, however, the perfect of the passive may alternatively be formed by using the long passive with the final vowel changed to the final vowel (usually [e]) of the verb's perfect. Additionally, in non-standard speech the perfect passive of verbs ending in a [mɑ] that changes to [ŋʷɑ] in the passive replace it with [ŋ̩ŋʷe].

The passive is used more commonly in Sesotho than the English "passive voice." Consider the following example:

Sesotho (passive) , English (active) "Who will feed the dog?"

The alternatives are more complex in their respective languages:

Sesotho (normal) , English (passive) "The dog will be fed by whom?"

Passive verbs are rare in the Niger–Congo family outside the Bantu sub-branch.

The neutro-active indicates an intransitive state without reference to the agent determining the condition. It can be approximated in English by using "get" or "become." It is, however, distinct from the passive. It indicates a current state of being done or being doable.

The suffix is [ɑhɑlɑ]. Only transitive vers may take this suffix.

accomplish ⇒ (currently) get finished, take place

do ⇒ be done

This suffix has the effect of making the valency of the verb 0, even if the original verb had two objects. The resultant verb is completely intransitive and cannot assume any objects even if they are prefixed.

The perfect of verbs ending with this suffix is achieved by changing the final [ɑlɑ] to [et͡sʼɪ].

This extension is quite rare in the Bantu language family as a whole.

The neutro-passive indicates an intransitive state without reference to the agent determining the condition. It can be approximated in English by the suffix "-able." It is, however, distinct from the passive. It indicates that the verb has the potential of being doable, but not necessarily currently.

The suffix is [ɛhɑ] (Proto-Bantu *-ik-,^[5] with an irregular vowel shift). Only transitive vers may take this suffix.

disperse ⇒ be (potentially) spillable, become scattered

do ⇒ be doable, become done

This suffix has the effect of making the valency of the verb 0, even if the original verb had two objects. The resultant verb is completely intransitive and cannot assume any objects even if they are prefixed.

The past tense of verbs ending with this suffix is formed in the general way by replacing the final vowel with [ile].

The applied indicates an action applied on behalf of or with regard to some object. It can be approximated in English by prepositions and prepositional phrases such as "for" and "towards."^[6]

The suffix is [ɛlɑ] (Proto-Bantu *-id-, with an irregular vowel shift^[7]). Sometimes this extension is doubled to [ɛl̩lɑ], causing the verb to look like a perfective form but with an applied meaning.

The following rules apply when forming the applied:

Usually one simply suffixes [ɛlɑ]
[bɑt͡ɬʼɑ] to search for ⇒ [bɑt͡ɬʼɛlɑ] search on behalf of
Verbs ending in [jɑ] replace it with [ɛlɑ]
[t͡sʼɑmɑjɑ] walk ⇒ [t͡sʼɑmɑ'ɛlɑ] walk on behalf of, towards
Verbs ending in [lɑ] preceded by an open vowel (/ɛ/, /ɑ/, or /ɔ/) elide the middle /ɛ/ and contract to [l̩lɑ]
[ŋɔlɑ] write ⇒ [ŋɔl̩lɑ] write to/for
Verbs ending in [lɑ] preceded by a closed vowel (/i/, /ɪ/, /ʊ/, or /u/) don't contract
[hʊlɑ] grow ⇒ [hʊlɛlɑ] grow for/towards
Polysyllabic verbs ending in [sɑ], [t͡sʼɑ] (most), [t͡sʼʷɑ], [n̩t͡sʰɑ], and [ɲɑ] cause the [lɑ] to alveolarize to [t͡sʼɑ]
[et͡sʼɑ] do ⇒ [et͡sʼet͡sʼɑ] do for
Polysyllabic causative verbs ending in [t͡sʼɑ] replace it with [let͡sʼɑ], reversing an original alveolarization^[8]

work ⇒ work for

The applied increases the valency of verbs; intransitive verbs may become transitive in the applied, and transitive verbs may become doubly transitive

live ⇒ live for

say something ⇒ tell someone something (two objects)

The past tense of verbs ending with this suffix changes the [elɑ] to [et͡sʼɪ].

The causative indicates an action caused to happen by some agent. It can be approximated in English by using "cause to."

The suffix is [isɑ] (Proto-Bantu long causative *-îc- + short causative *-î- ⇒ *-îcî-).

The following rules apply when forming the causative. Most complications are caused by the original Proto-Bantu "short causative" *-î- being absorbed into the preceding consonant (Sesotho does not allow palatal glides):

Usually one simply suffixes [isɑ]
[et͡sʼɑ] do ⇒ [et͡sʼisɑ] cause to do
Verbs ending in [jɑ] replace it with [isɑ]
[t͡sʼɑmɑjɑ] walk ⇒ [t͡sʼɑmɑ'isɑ] cause to walk
Some verbs ending in a [t͡sʼɑ], which is an alveolarization of an original [lɑ], revert the alveolarization, ending in [disɑ]
[sebet͡sʼɑ] work ⇒ [sebedisɑ] use
Monosyllabic e-stems suffix [esɑ] and i-stems suffix [isɑ]
[nʷɑ] drink ⇒ [nʷɪsɑ] cause to drink
Verbs ending in [ɲɑ] and disyllabic verbs ending in [nɑ] contract and cause nasalization resulting in [n̩t͡sʰɑ]
[bɔnɑ] see ⇒ [bon̩t͡sʰɑ] show
The original Proto-Bantu short causative suffix causes some verbs ending in [lɑ] and [nɑ] to change to [t͡sʼɑ] and [ɲɑ] respectively (in common non-standard speech all verbs ending in [nɑ] are changed to [n̩t͡sʰɑ])
[kʼɔpʼɑnɑ] meet ⇒ [kʼɔpʼɑɲɑ] join
Most verbs ending in [ʊhɑ] and [uhɑ] change the [hɑ] to [sɑ]. This is also due to the Proto-Bantu short causative (Proto-Bantu *-k- + short causative *-î- + final *-a ⇒ *-kîa, which appears as Sesotho [sɑ])
[ɑlʊhɑ] go to graze ⇒ [ɑlʊsɑ] herd

Often the causative verb has a meaning implying "help to do"

build ⇒ help to build ⇒ help each other to build ⇒ neighbour (since traditionally neighbouring houses would share a wall and yard, which the owners would build together)

The causative may increase the valency of verbs

know something ⇒ cause someone to know something

Usually the perfect is formed by further suffixing [it͡sʼe], but if the derivation alveolarized an original final [lɑ] to [t͡sʼɑ] then the alveolarization is reversed, resulting in final [dit͡sʼe]. If the suffix changed final [nɑ] to [ɲɑ] then the perfect is formed by replacing this final syllable with [n̩t͡sʼe].

The intensive indicates intensity or quickness of action.

The suffix is simply a doubling of the causative suffix ([isisɑ]) and the first syllable therefore follows similar phonetic rules as the causative. Sometimes, the suffix [isɑ] is used instead, resulting in causative and intensive verbs looking the same.

look for ⇒ investigate, search thoroughly

do ⇒ do intensely

The perfective indicates an action that has been carried out to completion or perfection.

The suffix is simply a doubling of the applied suffix ([ɛl̩lɑ]). It must therefore not be confused with the applied form of verbs ending in [ɛlɑ].

step on ⇒ oppress, coerce

do ⇒ do thoroughly

A further intensification of meaning is achieved with the suffixes [elet͡sʼɑ] ([ɛl̩lɑ] + Proto-Bantu *-îa) and [el̩let͡sʼɑ] ([ɛl̩lɛlɑ] + Proto-Bantu *-îa), a compounding of intensive and perfective suffixes. These verbs tend to denote meanings indicating specific purpose, and it is not unlikely that they are in fact intensifications of the applied suffix [ɛlɑ] instead (though the verb's valency is not increased).

shout ⇒ call out, scream

obscure, screen ⇒ protect

Though one might expect this suffix to form the perfect by replacing the [ɛl̩lɑ] with [elet͡sʼɪ], it often appears as [el̩let͡sʼɪ] instead, even in standard speech.

The reciprocal denotes a reciprocated action.

It is formed by suffixing [ɑnɑ] (Proto-Bantu *-an-^[9]).

It is usually used with plural subjects and plural concords, and has the effect of decreasing the valency. However, an object (the second subject) as well as a singular subject may still be used if the object is prefixed with the conjunctive enclitic [lɪ] (and, with); that is, they have a conjunctive import

speak ⇒ cause to speak ⇒ they communicate, I communicate with him

do ⇒ do (to) each another

Often this suffix is used when there is no chance that two subjects are involved in reciprocating the action. In this case it simply converts the verb from transitive to conjunctive import, with a minor modification of meaning (the action is slightly extended in time, or indicates a habit of the actor)

look at, search for ⇒ / I am looking for them (class 8 or 10 object)

The perfect is usually formed by changing the final vowel to [ɪ], though if the original verb was monosyllabic then the perfect replaces the [nɑ] with [n̩ne]

The associative indicates that two or more subjects are associated together in the action of the verb.

It is formed by suffixing [ɑhɑnɑ].

This derivative formation is not regularly used with most verbs.

attach, hook ⇒ be attached to each other, telecommunicate with one another

do ⇒ ( (be done together)

The perfect simply replaces the final vowel with [ɪ]

The reversive (or inversive) indicates an entire reversal of an action.

It is formed by suffixing [ʊl̩lɑ] (Proto-Bantu *-udud-) although several other dead formations exist, showing two sets of derivations into intransitive, transitive, and causative. These extensions, or at least their short forms as found in other languages (Proto-Bantu intransitive *-uk- and transitive *-ud-), are sometimes called the "separative" instead.

**Dead reversive forms**
Type	Intransitive	Transitive	Causative
Short	[ʊhɑ]	[ʊlɑ]	[ʊsɑ]
Full	[ʊlʊhɑ]	[ʊl̩lɑ]	[ʊlʊsɑ]

do ⇒ undo

Though the theory (and standard grammar) would dictate that this suffix forms its perfect by changing to [ʊlʊt͡sʼɪ], it often appears as [ʊl̩lʊt͡sʼɪ] instead, even in standard speech.

The augmentative is a largely dead formation signifying an augmentation or extension of a verb.

It is indicated by suffixes similar to the dead full formation of the reversive ([ʊlʊhɑ], [ʊl̩lɑ], and [ʊlʊsɑ]).

set apart ⇒ be distinct

The extensive indicates performing the action repeatedly or extensively.

It is formed with the suffix [ɑkʼɑ]^[10] but is limited in scope. It is primarily used with verbs signify discrete actions, causing them to be continuous or habitual. It is also sometimes heard doubled as [ɑkʼɑkʼɑ], with the same meaning.

jump ⇒ prance about

do ⇒ do repeatedly

The perfect of this extension simply suffixes [ile].

The diminutive indicates an action done "a little."

It is indicated by reduplication,^[11] the form being determined by the length of the verb:

Disyllabic verbs repeat the entire stem
[et͡sʼɑ] do ⇒ [et͡sʼɑ'et͡sʼɑ] do slightly
Monosyllabic verbs are repeated with the near-close near-front vowel (/ɪ/) between the stems.^[12] This form is almost never used
[ʒɑ] eat ⇒ [ʒɑ'ɪʒɑ] eat a little
Polysyllabic verbs duplicate the first two syllables of the stem
[fumɑnɑ] find ⇒ [fumɑfumɑnɑ] find somewhat

Note that this derivation pattern, like all other uses of reduplication in Bantu languages, is also sometimes used to indicate an intensification and/or repetition of an action—in these cases the actual meaning must be determined from context.

After the reduplication, the new verb may only have an underlying high tone on the first syllable (that is, only the phones of the first syllable are repeated, but not its tone).

The positional is a dead stative^[13] formation found in many verbs, mostly indicating bodily positions.

It is marked by the suffix [ɑmɑ] (Proto-Bantu *-am-). Originally, this suffix was not used to derive new meanings as such, but rather to emphasise the stative positional nature of the verb.

The perfect of these verbs changes the [ɑmɑ] to [ɑmɪ] and indicates a continuous, current action instead of a completed one. Past tense may be indicated by multi-verbal conjugation.

lie face downwards ⇒ He is lying face down, He did assume a lying position, He was lying

The stative extensive is a dead stative^[13] formation found in a few miscellaneous verbs, united by the fact that they all indicate states.

It is marked by the suffix [ɑlɑ] (Proto-Bantu *-ad-). Originally, this suffix was not used to derive new meanings as such, but rather to emphasise the stative nature of the verb.

The perfect of these verbs changes the [ɑlɑ] to [et͡sʼɪ] and indicates a continuous, current action instead of a completed one. Past tense may be indicated by multi-verbal conjugation.

wonder ⇒ They are in awe, They did become amazed, They were amazed

The contactive is a dead formation found in a few verbs, all indicating touch or contact of some sort.

It is marked by the suffix [ɑʀɑ] (Proto-Bantu *-at-). Originally, this suffix was not used to derive new meanings as such, but rather to emphasise or intensify the contactive nature of the verb.

The perfect of these verbs changes the [ɑʀɑ] to [eʀɪ] and indicates a continuous, current action instead of a completed one. Past tense may be indicated by multi-verbal conjugation.

wear ⇒ We are clothed, We did become dressed, We were dressed

Compounding of extensions

A verb may assume more than one extension, giving it a correspondingly more complex meaning.

watch X ⇒ causative cause Y to watch X ⇒ causative-applied cause Y to watch X on behalf of Z ⇒ causative-applied-reciprocal cause Y to watch X on behalf of each other

Though it may appear that the possibilities are endless, the truth is that the depth is limited by various factors. Apart from the obvious constraints of semantics (whether a complex meaning actually makes any sense and serves any possible purpose) and markedness (how strange and complex the verb sounds to the native speaker), there are also restrictions on the order of the extensions.

If an extension increases the valency of a verb, any objects of the original verb are demoted and the new object is made principal.

I watch the fields ⇒ I cause the children to watch the fields ⇒ I cause the children to watch the fields on behalf of the old woman (highly marked)

If an objectival concord is used instead of an object, the concord agrees with what would have been the principal object. Additionally, if the original object was also only indicated by an objectival concord, then it becomes demoted to an absolute pronoun (Sesotho verbs may only have one objectival concord).

I watch them ( fields) ⇒ I cause them ( children) to watch them () ⇒ I cause them () to watch them () on behalf of her ( grandmother, old woman) (highly marked)

(Note how the infix^[14] [ɑ] disappears when the verb is followed by a direct object, even if it is not the object indicated by the concord.)

Like all other Bantu languages, Sesotho has inherited certain restrictions on the order of the extensions. The most basic rule (which is broken by very few languages) is that the passive and the short causative^[15] always follow all the other extensions (including the perfect [il], which is always used with the final vowel [e]). Although it is probable that Proto-Bantu had fairly strict restrictions on the order of the other extensions, these rules have been relaxed somewhat in modern Bantu languages.

For example, since the causative [is] is normally ordered closer to the verb stem than the reciprocal [ɑn] (or indeed, most other extensions), to form the causative of the reciprocal the (dead) short causative (Proto-Bantu *-î-) is usually used instead, therefore palatalizing the reciprocal to [ɲ]. Various other unexpected palatalizations and alveolarizations brought on by combinations of the causative with other extensions may be similarly explained by the action of the short causative either replacing the normal causative, or being used together with the long causative around another extension (causative + other extension ⇒ [is] + other extension + *-î- ⇒ [is] + modified extension).

Certain extensions (intensive, perfective, associative, reversive, and augmentative) are obviously fossilised compound extensions. Often a derived verb may continue being used while the original verb disappears from the language.

Note that, since prefixes are of the shape CV or V (where C represents a consonant and V a vowel), verb roots end without the final vowel, prefixes are of the shape (VC)* (* indicates possible repetition) and the final vowel simply has shape V, this and other structures reinforce the open syllable structure of the Bantu languages, and very few languages have broken it.

Non-verbal derivatives

Verbs may also, to a lesser degree, be derived from nouns, qualificatives, and ideophones.

Denominative verbs are stative verbs^[13] derived from nouns and qualificatives.

They are formed by suffixing [fɑ] (dead) or [fɑlɑ] to the stem, giving a verb meaning "become...."

intelligence ⇒ become intelligent

soft (relative) ⇒ become soft

The monosyllabic adjectival roots (except [ŋ̩] some, and [nɛ] four) become nasalized before assuming the suffix. Furthermore, the vowel of [t͡ɬʼɛ] (beautiful) changes to [t͡ɬʼɑ]:

ugly ⇒ become ugly

new ⇒ become renewed

beautiful ⇒ become beautiful

black ⇒ become black

Causatives are formed regularly by changing the [fɑlɑ] suffix to [fɑt͡sʼɑ]. Perfects are formed regularly by changing the [fɑlɑ] suffix to [fet͡sʼɪ].

This extension (the long [fɑlɑ]) is quite rare in the Bantu languages, though all languages have a few verbs in this form even if it may no longer be active.

Deideophonic verbs are formed rather irregularly from disyllabic ideophones.

They are miscellaneous in nature and are formed by the addition of several suffixes such as [hɑ], [lɑ], [t͡sʼɑ], [sɑ], [mɑ], [t͡sʼɛhɑ], [bɑlɑ], [kʼɑ] etc.

However, common across the Bantu language group are the forms [hɑ] for the intransitive, [lɑ] for the transitive, and [t͡sʼɑ] for the causative. Additionally, the causative of the intransitive may be formed regularly with the suffix [sɑ], but this is usually not done if the causative [t͡sʼɑ] form is regularly used.

of turning over ⇒ flip over, experience a car accident

of being completely full ⇒ burst open

of picking up and throwing forward ⇒ (of large herbivores) stand up and move in a certain direction, and initiate a process or plan

When forming these verbs, the tone of the first syllable of the verb (its characteristic tone) corresponds to the tone of the first syllable of the part of the original word used to form the verb (usually the root, but a complete noun for monosyllabic roots). Thus verbs derived regularly from monosyllabic stems are all L verbs (due to the null toned prefix).

easy, easyness [ _ _ ¯ ] ⇒ become easy (L verb)

angry, anger [ _ ¯ _ ] ⇒ become angry (H verb)

strong, strength [ _ _ ] ⇒ become strong (L verb)

Inflexion during conjugation

In addition to the verbal derivatives, the following changes may occur to the stem's suffix [ɑ], during conjugation:

The [ɑ] changes to -e ([ɪ]) to form the perfect subjunctive tense and certain tenses of the negative conjugation. This vowel always causes the syllable carrying it to assume a high tone.
The [ɑ] changes to -e ([ɛ]) to form the present-future tense of the subjunctive mood.
The [ɑ] becomes [ɑŋ̩] to form the plural of the imperative and certain relative tenses.
The [ɑ] becomes [ile] to form the perfect stem. Various phonological situations may change this basic construct.

The general rules for the formation of the perfect are varied due to various mostly phonological interactions with the suffix:^[16]

Generally, [ile] is suffixed
[ʀɛkʼɑ] buy ⇒ [ʀekʼile] bought
Verbs ending in [jɑ] replace it with [ile]
[t͡sʼɑmɑjɑ] go ⇒ [t͡sʼɑmɑ'ile] went
For monosyllabic stems, i-stems suffix [ile] and e-stems suffix [ɪlɛ]
[nʷɑ] drink ⇒ [nʷɪlɛ] drank
Disyllabic verbs ending in [mɑ] change it to [m̩me]
[ʀʊmɑ] send ⇒ [ʀʊm̩me] sent
Polysyllabic verbs ending in [sɑ], [t͡sʼɑ] (most), [t͡sʼʷɑ], and [n̩t͡sʰɑ] cause the [ile] to alveolarise to [it͡sʼe]
[ɬɑt͡sʼʷɑ] wash ⇒ [ɬɑt͡sʼit͡sʼe] washed
Verbs ending in [nɑ] of more than one syllable and disyllabic reciprocal verbs change the [nɑ] to [n̩ne]
[binɑ] sing ⇒ [bin̩ne] sang
and so forth...

For all verbs, however, the past tense may also be indicated with the simple [ile] (past subjunctive) multi-verbal conjugation, although its meaning does diverge somewhat from that of the perfect (especially with stative verbs)

They did sing

Conjugation

Verbal conjugation is by far the most complex and varied topic in the Bantu languages. The tenses are conjugated by means of prefixes and infixes^[14] indicating person, mood, implication, and aspect.

There are two conjugations, the positive and negative, and most tenses have corresponding forms in each. The language recognises four moods: the indicative, the subjunctive, the potential, and the participial sub-mood (infinitives are nouns and imperatives are interjectives). The moods may be divided into tenses according to time (remote past, immediate past, present, immediate future, and remote future) and implication (simple, progressive, and exclusive), which may be further subdivided according to aspect into indefinite, continuous, and perfect.

There are also many often complex compound tenses, indicated by changes in tone and the use of deficient verbs (multi-verbal conjugations).

Import refers to how the object of the verb is indicated.

Verbs can be either:

Intransitive, with no direct object
[kʼɪ'ɑtʰʊlɑ] I become quiet
Transitive, with a single direct object
[kʼɪ'ɑ'ʊlebʊhɑ] I thank you
Ditransitive, with two objects
[kʼɪfɑ ŋʷɑnesʊ mʊfɑhɔ] I give my sibling food for the journey
Locative, with a locative adverbial construction often indicated by [ŋ̩] / [eŋ̩]
[kʼɪkʼɛnɑ lɑpʼeŋ̩] I enter my home
Agentive verbs (usually passives), which need a copulative^[17] used as an agent adverb indicated by [kʼɪ]
[bɑtʰusʷɑ kʼiˌbukʼɑn̩t͡sʼwɪ] They are helped by the dictionary
Instrumental verbs, which use an instrumental adverb indicated by [kʼɑ]
[ʀɪ'ɛtʼɑ kʼɑkʼolo'i] We travel by car
Conjunctive verbs (mostly reciprocals), which use the conjunctive proclitic [lɪ]
[ʀiˌdumɛl̩lɑnɑ lɪbɔnɑ] We agree with them

Many verbs can have more than one import ([t͡sʼɑmɑjɑ] (walk) can be locative, instrumental, or conjunctive; [bu'ɑ] (speak) can be intransitive, transitive, instrumental, or conjunctive) and verb derivatives can also change the import of the stem.

Many shades of meaning are achieved by the employment of deficient verbs in multi-verbal conjugations. Many tenses and moods may only be formed in this manner.

The verbal complex

In the Bantu languages, the typical full structure of verbs, excluding contractions, is as follows (the * indicates possible iteration):

PI — SC — NEG — TM — AM — OC — ROOT — EXT* — FV

 │    │     │    │    │    │      │     │     │

 │    │     │    │    │    │      │     │   final vowel

 │    │     │    │    │    │      │  extensions

 │    │     │    │    │    │  vb. root

 │    │     │    │    │  objectival concord

 │    │     │    │  aspect marker

 │    │     │  tense marker

 │    │   negative

 │  subjectival concord

pre-initial morpheme

In Sesotho, as with most other Bantu languages, this has been modified somewhat, resulting in the following structure ("I shall no longer look on his behalf"):

[hɑkʼɪsɑt͡ɬʼɑmʊʃɛbɛlɑ]

PI — SC — A1 — A2 — OC — ROOT — EXT* — FV

 │    │    │    │    │      │     │     │

 │    │    │    │    │      │     │   final vowel

 │    │    │    │    │      │  extensions

 │    │    │    │    │  vb. root

 │    │    │    │  objectival concord

 │    │    │  second verbal auxiliary

 │    │  first verbal auxiliary

 │  subjectival concord

pre-initial morpheme

Though indicative tenses form their negatives with the prefix [hɑ], many other moods and tenses form their negatives with an infix (either [sɑ] or [sɪ], depending on the specific tense). The verbal auxiliary infixes are used to indicate tense, certain forms of the subjunctive, progressive implication, the potential mood, as well verb focus in the present indicative tense. The verbal infixes always follow the simple infixes, though there are some instances where two simple infixes are used at the same time.

The extensions include suffixes used in verbal derivatives as well as the perfect [il] (which is always followed by the final vowel [e]).

With the exception of the verb root, each of these formatives is monosyllabic, but in Sesotho some verbal infixes (those that are contractions) and extensions (those that are obvious compoundings of earlier forms) also have more than one syllable.

Additionally, the structure (obj conc. + stem) is often called the "macrostem" in various syntactical and tonal theories.

Many aspects and tenses are indicated by multi-verbal conjugations and, with the exception of the subjectival concord, the root, and the final vowel, most of these formatives are not always necessary. Note that infinitives and imperatives (both of with do not have subjectival concords) may be considered separate parts of speech (nouns/gerunds and interjectives). Deficient verbs are never used with objectival concords, and the use of the other formatives with them is also limited.

This structure obviously ignores any possible enclitics that is suffixed.

Tenses

The Sesotho tense system is somewhat less complex (though not necessarily less complicated) than that of other Bantu languages. Whereas many Bantu languages clearly divide the time into remote past, immediate past, present, immediate future, and remote future, not all Sesotho moods divide very clearly between immediate and remote tenses, and the differences in meaning are not as great.

**Examples indicating the tenses (positive simple indicative mood)**
Tense	Example
Present	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I know the truth
Past perfect	[kʼɪt͡sʼiˌbile n̩nɪtʼɪ] I knew the truth
Immediate past	[kʼɪt͡sʼʷɑt͡sʼɪbɑ n̩nɪtʼɪ] I just recently knew the truth
Immediate future	[kʼiˌ'ilot͡sʼɪbɑ n̩nɪtʼɪ] I shall know the truth soon
Future	[kʼɪt͡ɬʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I shall know the truth

Moods

There are basically four moods.^[18]

The indicative mood indicates what is, was, or will be. It uses the basic subjectival concord.
The potential mood indicates that an action is possible. It uses similar concords to those of the subjunctive.
The participial sub-mood is so-called since it has forms corresponding to the tenses of both the above moods (most of the indicative, but only the present potential). It is widely used after certain conjunctives, in forming the complements of numerous multi-verbal tenses, and in the formation of relative clauses.
The subjunctive mood is used in subordinate or consecutive constructions, in many cases being parallel in usage to the Latin subjunctive.

**Examples indicating the moods (present tense)**
Mood	Positive	Negative
Indicative	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I know the truth	[hɑkʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I do not know the truth
Potential	[ŋ̩kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may know the truth	[ŋ̩kʼekʼe kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may not know the truth
Participial	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] ...while I know the truth	[kʼɪsɑt͡sʼɪbɪ n̩nɪtʼɪ] ...while I do not know the truth
Subjunctive	[kʼɪt͡sʼɪbɛ n̩nɪtʼɪ] ...so I may know the truth	[kʼɪsɪt͡sʼɪbɪ n̩nɪtʼɪ] ...so I may not know the truth

Implication

Within the indicative and participial moods, tenses may be further sub-divided according to the implication of the action.

The simple implication indicates an action in no way qualified.
The progressive implication indicates an ongoing action.
The exclusive implication indicates an action that has not been happening until now.

**Examples indicating implication (indicative mood)**
Implication	Example
Simple	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I know the truth
Progressive	[kʼɪsɑt͡sʼɪbɑ n̩nɪtʼɪ] I still know the truth
Exclusive	[kʼɪse kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I now know the truth

Aspects

The tenses may be further divided according to the aspect of the action. In Sesotho there are at least three aspects, the definite, the continuous, and the perfect.

**Examples indicating aspect (with a multi-verbal indicative past tense)**
Aspect	Example
Definite	[kʼiˌ'ile kʼɑt͡sʼɪbɑ] I did know
Continuous	[kʼɪne kʼɪt͡sʼɪbɑ] I knew
Perfect	[kʼɪne kʼɪt͡sʼiˌbile] I had known

Deficient verbs

Deficient verbs, so called because they require a subordinate or complementary verb to complete their action, are used to form many tenses and to impart certain shades of meaning. They form part of multi-verbal conjugations consisting of a string of verbs, each with its own subjectival concord.

Deficient verbs, being "deficient", are never used alone. Many of them are irregular in form and have irregular inflexions. Monosyllabic deficient verbs are never used with the penultimate [ɪ] that is sometimes used with normal verbs (not to be confused with the indefinite concord).

Many of these verbs seem radical in nature, while others (especially those with complex implications) are obviously derived from certain extant normal verbs (but are used with slightly different meanings). What distinguishes the deficient usage of these normal verbs is the fact that they are followed directly by another verb and affect its meaning (and only the main verb may carry an objectival concord).

I no longer know

I knew

I shall (at some specific time) know

I should/would have known

I may indeed know

I will at least know

I may still know

I nearly knew

I shall no longer know

I have to know

Notes

^ ^a ^b Simple phonotactic explanations may make these apparent irregularities more understandable.
Almost all the non-velar e-stems are palatal or postalveolar in nature. This may be due to an original palatal glide being "absorbed" into the original consonant of the verbs (the alveolar /s/ also has similar origins). In Sesotho, the palatal /j/ may not be followed by the vowels /i/ or /u/ and these become weakened to /ɪ/ and /ʊ/. The original passive suffix (still used in Setswana and many Northern Sotho languages) was [iwɑ], and so the suffixes are weakened to [ɪlɛ], [ɪsɑ] and [ɪwɑ].
Apparently the velar e-stems use the modern [uwɑ] passive instead, and due to phonotactic restrictions occasionally applied on the labial approximant /w/, similar to those on the palatal, together with the fact that labialized consonants may not be followed by back vowels, the suffixes are weakened to [ɪlɛ], [ɪsɑ] and [ʊwɑ].
^ The first two verbs, together with the copulative verb [nɑ] (indicating possession, with a conjunctive import), are used in many Bantu languages in generally restricted circumstances.
There exists certain "defective" verbs across most Bantu languages: Proto-Bantu *-di (Sesotho participial copulative [lɪ]), *-ti (Sesotho [ʀɪ]), and *-na (Sesotho [nɑ]). Additionally, a common variant of *-ti — *-tio — appears as Sesotho [t͡ʃʰɔ] — an irregular palatalization (when an alveolarization would have been expected) possibly due to the verb being borrowed from some Nguni language (it does not exist in most other Sotho–Tswana languages).
These are distinguished from other verbs in that they are normally not used with many of the affixes in the verbal complex. For example, though they are all transitive and are therefore used with objects, they never take objectival concords (in Sesotho and many other languages, [ʀɪ] may take an objectival concord when used with ideophones). Additionally, except for Sesotho [ʀɪ] and [t͡ʃʰɔ], they may not be used as infinitives.
Even though they have these peculiarities and, except for [nɑ], they do not end with the typical vowel, Bantuists consider them verbs because they may be used with subjectival concords.
The highly irregular passive of [ʀɪ] may be due to Nguni -thiwa (most other Sotho–Tswana languages use -riwa instead).
^ The specific label comes from early descriptions of isiZulu grammar, where it was discovered that, apart from simply looking different from other verbs, vowel verbs are also conjugated slightly differently from normal verbs under certain situations, and many of them have alternative (and, at least in modern popular urban isiZulu, more common) forms with the initial vowel deleted. Though isiZulu has five vowel phonemes (plus two allophones), vowel verbs in that language may only begin with the vowels /ɑ/, /ɔ/, and /ɛ/ (written 'a', 'o', and 'e' respectively in its orthography). There is no similar restriction in Sesotho.
^ The traditional verb root used to demonstrate these derivations (and form their traditional names; by using the class 5 noun prefix [lɪ] and changing the final vowel to [i]) is [et͡sʼ] (do, act, make). The problem with this root is that the /t͡sʼ/ consonant tends to greatly complicate the forms of the derivatives (due to alveolarization), and it has been felt wise in this table to use a verb root with more neutral sounds. Note that the ɛ in [ǃɛtʼɑ] regularly undergoes vowel raising when followed by certain vowels and consonants, but this is a simpler and more predictable phenomenon than the various complications brought on by the /t͡sʼ/ consonant.
Compare this with the situation in Arabic where the verb فَعَلَ (Faʿala "he did") traditionally used to indicated the various verb forms often confuses non-native learners due to the ʿAin ع sound being confused with that of the Hamza ء that appears in some forms (such as the causative stem IV أَفْعَلََ 'Afʿala).
^ Proto-Bantu also had an allomorphic non-productive suffix (called the "impositive"), which, instead of meaning "to be put in a state" (the meaning of the neutro-passive), actually meant "to put in a state." As with the productive suffix, the vowel has undergone an irregular vowel shift in Sesotho, but unlike the productive suffix the /h/ reflex from Proto-Bantu *k is sometimes weakened (through voicing) and elided, thus resulting in verbs that end with [ɛ'ɑ]
Proto-Bantu *-janik- spread to dry out in the sun ⇒ Sesotho [ɑnɛhɑ]

Proto-Bantu *-jambik- cook ⇒ Sesotho [ɑpʼɛhɑ] (slightly archaic)
^ It is possible that in pre-Proto-Bantu the various meanings of this suffix where indicated by several different forms that eventually merged and became the single *-id-.
^ This "irregular" shift, which causes the applied suffix to look like it came from Proto-Bantu *-ed- instead of *-id-, is found in many other Bantu languages.
^ This may, as usual, be better explained by looking to the original Proto-Bantu morphology. The Proto-Bantu short causative *-î- alveolarizes the Sesotho /l/ to /t͡sʼ/. Additionally, there were, and (in almost all Bantu languages) still are, fairly strong restrictions on the ordering of the verbal extensions — the most basic restriction being that the short passive and the short causative *-û- always follow the other extensions (isiXhosa is an example of a language that allows other extensions to follow the passive).

So with these facts in mind, this particular morphological rule for forming the applied may be explained by simply saying that the short passive that alveolarized the original /l/ is removed from its current position (thus reversing the alveolarization) and placed after the applied extension to conform to the rules about extension ordering (with the added side-effect that the /l/ in the applied extension is now alveolarized).
^ Many researchers believe that this suffix might be from an earlier preposition *na (cf. the conjunctive clitic *na-, corresponding to the Sesotho proclitic [lɪ]), which became grammaticalised and attached to the verb. That is, Verb-a na X ("Verb along with X") became Verb-ana X in pre-Proto-Bantu (with the same meaning) and this was eventually used as a productive suffix.
^ Comparison with other languages shows that this is actually irregular as one would expect it to appear as [ɑhɑ] in Sesotho.
^ The verb thus formed is a full, independent verb, rather than a mere repeating of the verb's syllables. In particular, the derived verb belongs to the same tonal class as the original, with the underlying tone on the first syllable of the stem not copied when the syllable is copied. This is overwhelmingly the case in the Bantu languages, although there is one notable exception, viz. Chichewa.
^ The verb thus formed is a full, independent verb, rather than
^ ^a ^b ^c These "stative" verbs in Sesotho all have a strong inceptive feel to them. That is, instead of simply meaning "to be x", they actually mean "to become x", with the actual stative "to be x" meaning achieved by using the "present stative" tense (which is formed in exactly the same manner as the perfect for non-stative verbs). The "stative" label comes from the fact that in many other Bantu languages these verbs do actually indicate present states, not a continuous inceptive event.
^ ^a ^b The use of this term in Bantu linguistics means "formatives placed in the middle of a word" and not the more common "formatives placed in the middle of a morpheme." Bantu languages, being agglutinative, construct words by placing affixes around a stem, and if an affix is always placed after other affixes but before the stem (such as in certain verb tenses and moods) then it is usually called an "infix."
^ See the note above on the alveolarization of the applied.
^ In the formation of the perfect of many of the derivational suffixes listed earlier, many of the complications are caused by a process known as "imbrication" where the perfect suffix [il] loses its consonant and the vowel is placed before the previous consonant, thus causing changes to the previous vowel (and to the following consonant).

In isiZulu the forms are very predictable, with suffixes of the form aCa generally changing to eCe (aCa + ile ⇒ aiCe ⇒ eCe).
^ The fact that this is indeed the simple copulative (and not just a prefix that happens to be allomorphic with it) is evidenced by looking at these verbs in a language such as isiZulu where the simple copulative is much more complicated and yet coincides perfectly with the marking of the objects of agentive verbs.

^ This is Doke's analysis.

The issue of how many moods individual Bantu languages have exactly is not entirely settled due to their complex morphologies. A more inclusive scheme for Sesotho (and the one taught to first language speakers in school) is as follows:

**Sesotho moods**
Mood ([sɪkʼɑ'ʊ])	Positive ([tʼumɛlɔ])	Negative ([tʼɑtʼʊlɔ])
Indicative ([n̩nɪtʼɪ])	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I know the truth	[hɑkʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I do not know the truth
Potential ([xɔnɛhɔ])	[ŋ̩kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may know the truth	[ŋ̩kʼekʼe kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may not know the truth
Situative	[kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] ...while I know the truth	[kʼɪsɑt͡sʼɪbɪ n̩nɪtʼɪ] ...while I do/did not know the truth
Subjunctive ([tʼɑkʼɑt͡sʼɔ])	[kʼɪt͡sʼɪbɛ n̩nɪtʼɪ] ...so I may know the truth	[kʼɪsɪt͡sʼɪbɪ n̩nɪtʼɪ] ...so I may not know the truth
Relative/Qualificative ([xetʰi])	[jɑt͡sʼɪbɑŋ̩ n̩nɪtʼɪ] ...who knows the truth	[jɑsɑt͡sʼɪbɪŋ̩ n̩nɪtʼɪ] ...who does not know the truth
Consecutive ([tʼɑtʼɛlɑnɔ])	[kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] ...and then I knew the truth	[kʼɑsɪt͡sʼɪbɪ n̩nɪtʼɪ] ...and then I did not know the truth
Habitual ([t͡ɬʼʷɑɛlɔ])	[kʼɪje kʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I often know the truth	[hɑŋ̩kʼe kʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I do not often/often do not know the truth
Infinitive ([hʊ])	[hʊt͡sʼɪbɑ n̩nɪtʼɪ] To know the truth	[hʊsɪt͡sʼɪbɪ n̩nɪtʼɪ] To not know the truth
Imperative ([tʼɑ'ɛlɔ])	[t͡sʼɪbɑ n̩nɪtʼɪ] Know the truth	[sɪt͡sʼɪbɪ n̩nɪtʼɪ] Do not know the truth

According to Doke's analysis, the situative is just the participial sub-mood, the qualificative is a form of the participial, the consecutive is the past subjunctive (used when telling stories, it sounds as if the story-teller is using the present tense to describe a past action, but the subjectival concords differ from those of the indicative present), the habitual is a multi-verbal tense using a specific set of deficient verbs (Group III in Doke's classification) followed by a perfect subjunctive (and its negative uses another deficient verb in the same group); while the infinitive and most imperatives are not verbal moods (they are separate parts of speech and cannot be used as the predicate of a sentence, though imperatives can form interjectival sentences and there is a form of the subjunctive that could alternatively be interpreted as an imperative using subjectival concords).

The analysis is further complicated by the seemingly unpredictable form of the negative for each tense of each verb. The one point most Bantuists seem to agree upon is that, apart from the indicative mood, Bantu languages also have a subjunctive formed (usually) by changing the final vowel of the verb to *-e (which corresponds to Sesotho /ɛ/).

References

Coupez, A., Bastin, Y., and Mumba, E. 1998. Reconstructions lexicales bantoues 2 / Bantu lexical reconstructions 2. Tervuren: Musée royal de l’Afrique centrale.
de Schryver, G. M., and Prinsloo, D. J. 2000. Towards a sound lemmatisation strategy for the Bantu verb through the use of frequency-based tail slots — with special reference to Cilubà, Sepedi and Kiswahili. Makala ya kongamano la kimataifa Kiswahili 2000. Proceedings: 216–242, 372.
Doke C. M. 1963. Text Book of Zulu Grammar. Cape Town.
Doke, C. M., and Mofokeng, S. M. 1974. Textbook of Southern Sotho Grammar. Cape Town: Longman Southern Africa, 3rd. impression. ISBN 0-582-61700-6.
Downing, L.J, 2001. Tone (Non-) Transfer in Bantu Verbal Reduplication. Typology of African Prosodic Systems Workshop. Bielefeld University, Germany.
Güldemann, T. The history of quotative predicates: Can lexical properties arise out of grammatical construction?
Hyman, L. M. 2003. Segmental phonology. In D. Nurse & G. Philippson (eds.), The Bantu languages, pp. 42–58. London: Routledge/Curzon.
Hyman, L. M. 2007. Niger–Congo verb extensions: Overview and discussion. In D. L. Payne and J Pen̈a (eds.), Selected proceedings of the 37th Annual Conference on African Linguistics, 149-163. Somerville, MA: Cascadilla Proceedings Project.
Lodhi, A. Y. 2002. Verbal extensions in Bantu (the case of Swahili and Nyamwezi). In Africa & Asia, No 2, 2002, pp 4–26. Department of Oriental and African Languages, Göteborg University.

External links

[e_stems-1] Simple phonotactic explanations may make these apparent irregularities more understandable.
Almost all the non-velar e-stems are palatal or postalveolar in nature. This may be due to an original palatal glide being "absorbed" into the original consonant of the verbs (the alveolar /s/ also has similar origins). In Sesotho, the palatal /j/ may not be followed by the vowels /i/ or /u/ and these become weakened to /ɪ/ and /ʊ/. The original passive suffix (still used in Setswana and many Northern Sotho languages) was [iwɑ], and so the suffixes are weakened to [ɪlɛ], [ɪsɑ] and [ɪwɑ].
Apparently the velar e-stems use the modern [uwɑ] passive instead, and due to phonotactic restrictions occasionally applied on the labial approximant /w/, similar to those on the palatal, together with the fact that labialized consonants may not be followed by back vowels, the suffixes are weakened to [ɪlɛ], [ɪsɑ] and [ʊwɑ].

[mono_stems-2] The first two verbs, together with the copulative verb [nɑ] (indicating possession, with a conjunctive import), are used in many Bantu languages in generally restricted circumstances.
There exists certain "defective" verbs across most Bantu languages: Proto-Bantu *-di (Sesotho participial copulative [lɪ]), *-ti (Sesotho [ʀɪ]), and *-na (Sesotho [nɑ]). Additionally, a common variant of *-ti — *-tio — appears as Sesotho [t͡ʃʰɔ] — an irregular palatalization (when an alveolarization would have been expected) possibly due to the verb being borrowed from some Nguni language (it does not exist in most other Sotho–Tswana languages).
These are distinguished from other verbs in that they are normally not used with many of the affixes in the verbal complex. For example, though they are all transitive and are therefore used with objects, they never take objectival concords (in Sesotho and many other languages, [ʀɪ] may take an objectival concord when used with ideophones). Additionally, except for Sesotho [ʀɪ] and [t͡ʃʰɔ], they may not be used as infinitives.
Even though they have these peculiarities and, except for [nɑ], they do not end with the typical vowel, Bantuists consider them verbs because they may be used with subjectival concords.
The highly irregular passive of [ʀɪ] may be due to Nguni -thiwa (most other Sotho–Tswana languages use -riwa instead).

[v_vb_orig-3] The specific label comes from early descriptions of isiZulu grammar, where it was discovered that, apart from simply looking different from other verbs, vowel verbs are also conjugated slightly differently from normal verbs under certain situations, and many of them have alternative (and, at least in modern popular urban isiZulu, more common) forms with the initial vowel deleted. Though isiZulu has five vowel phonemes (plus two allophones), vowel verbs in that language may only begin with the vowels /ɑ/, /ɔ/, and /ɛ/ (written 'a', 'o', and 'e' respectively in its orthography). There is no similar restriction in Sesotho.

[vb_deriv_trad-4] The traditional verb root used to demonstrate these derivations (and form their traditional names; by using the class 5 noun prefix [lɪ] and changing the final vowel to [i]) is [et͡sʼ] (do, act, make). The problem with this root is that the /t͡sʼ/ consonant tends to greatly complicate the forms of the derivatives (due to alveolarization), and it has been felt wise in this table to use a verb root with more neutral sounds. Note that the ɛ in [ǃɛtʼɑ] regularly undergoes vowel raising when followed by certain vowels and consonants, but this is a simpler and more predictable phenomenon than the various complications brought on by the /t͡sʼ/ consonant.
Compare this with the situation in Arabic where the verb فَعَلَ (Faʿala "he did") traditionally used to indicated the various verb forms often confuses non-native learners due to the ʿAin ع sound being confused with that of the Hamza ء that appears in some forms (such as the causative stem IV أَفْعَلََ 'Afʿala).

[impos-5] Proto-Bantu also had an allomorphic non-productive suffix (called the "impositive"), which, instead of meaning "to be put in a state" (the meaning of the neutro-passive), actually meant "to put in a state." As with the productive suffix, the vowel has undergone an irregular vowel shift in Sesotho, but unlike the productive suffix the /h/ reflex from Proto-Bantu *k is sometimes weakened (through voicing) and elided, thus resulting in verbs that end with [ɛ'ɑ]
Proto-Bantu *-janik- spread to dry out in the sun ⇒ Sesotho [ɑnɛhɑ]

Proto-Bantu *-jambik- cook ⇒ Sesotho [ɑpʼɛhɑ] (slightly archaic)

[app_meanings-6] It is possible that in pre-Proto-Bantu the various meanings of this suffix where indicated by several different forms that eventually merged and became the single *-id-.

[app_vwl_shift-7] This "irregular" shift, which causes the applied suffix to look like it came from Proto-Bantu *-ed- instead of *-id-, is found in many other Bantu languages.

[caus_rev_alv-8] This may, as usual, be better explained by looking to the original Proto-Bantu morphology. The Proto-Bantu short causative *-î- alveolarizes the Sesotho /l/ to /t͡sʼ/. Additionally, there were, and (in almost all Bantu languages) still are, fairly strong restrictions on the ordering of the verbal extensions — the most basic restriction being that the short passive and the short causative *-û- always follow the other extensions (isiXhosa is an example of a language that allows other extensions to follow the passive).

So with these facts in mind, this particular morphological rule for forming the applied may be explained by simply saying that the short passive that alveolarized the original /l/ is removed from its current position (thus reversing the alveolarization) and placed after the applied extension to conform to the rules about extension ordering (with the added side-effect that the /l/ in the applied extension is now alveolarized).

[an_source-9] Many researchers believe that this suffix might be from an earlier preposition *na (cf. the conjunctive clitic *na-, corresponding to the Sesotho proclitic [lɪ]), which became grammaticalised and attached to the verb. That is, Verb-a na X ("Verb along with X") became Verb-ana X in pre-Proto-Bantu (with the same meaning) and this was eventually used as a productive suffix.

[ext_loan-10] Comparison with other languages shows that this is actually irregular as one would expect it to appear as [ɑhɑ] in Sesotho.

[11] The verb thus formed is a full, independent verb, rather than a mere repeating of the verb's syllables. In particular, the derived verb belongs to the same tonal class as the original, with the underlying tone on the first syllable of the stem not copied when the syllable is copied. This is overwhelmingly the case in the Bantu languages, although there is one notable exception, viz. Chichewa.

[12] The verb thus formed is a full, independent verb, rather than

[stat_vb_inc-13] These "stative" verbs in Sesotho all have a strong inceptive feel to them. That is, instead of simply meaning "to be x", they actually mean "to become x", with the actual stative "to be x" meaning achieved by using the "present stative" tense (which is formed in exactly the same manner as the perfect for non-stative verbs). The "stative" label comes from the fact that in many other Bantu languages these verbs do actually indicate present states, not a continuous inceptive event.

[infix-14] The use of this term in Bantu linguistics means "formatives placed in the middle of a word" and not the more common "formatives placed in the middle of a morpheme." Bantu languages, being agglutinative, construct words by placing affixes around a stem, and if an affix is always placed after other affixes but before the stem (such as in certain verb tenses and moods) then it is usually called an "infix."

[sht_caus_above-15] See the note above on the alveolarization of the applied.

[suff_perf-16] In the formation of the perfect of many of the derivational suffixes listed earlier, many of the complications are caused by a process known as "imbrication" where the perfect suffix [il] loses its consonant and the vowel is placed before the previous consonant, thus causing changes to the previous vowel (and to the following consonant).

In isiZulu the forms are very predictable, with suffixes of the form aCa generally changing to eCe (aCa + ile ⇒ aiCe ⇒ eCe).

[cop_pref-17] The fact that this is indeed the simple copulative (and not just a prefix that happens to be allomorphic with it) is evidenced by looking at these verbs in a language such as isiZulu where the simple copulative is much more complicated and yet coincides perfectly with the marking of the objects of agentive verbs.

[alt_moods-18] This is Doke's analysis.

The issue of how many moods individual Bantu languages have exactly is not entirely settled due to their complex morphologies. A more inclusive scheme for Sesotho (and the one taught to first language speakers in school) is as follows:

Sesotho moods
Mood ([sɪkʼɑ'ʊ]) Positive ([tʼumɛlɔ]) Negative ([tʼɑtʼʊlɔ])

Indicative ([n̩nɪtʼɪ]) [kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] I know the truth [hɑkʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I do not know the truth

Potential ([xɔnɛhɔ]) [ŋ̩kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may know the truth [ŋ̩kʼekʼe kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] I may not know the truth

Situative [kʼɪt͡sʼɪbɑ n̩nɪtʼɪ] ...while I know the truth [kʼɪsɑt͡sʼɪbɪ n̩nɪtʼɪ] ...while I do/did not know the truth

Subjunctive ([tʼɑkʼɑt͡sʼɔ]) [kʼɪt͡sʼɪbɛ n̩nɪtʼɪ] ...so I may know the truth [kʼɪsɪt͡sʼɪbɪ n̩nɪtʼɪ] ...so I may not know the truth

Relative/Qualificative ([xetʰi]) [jɑt͡sʼɪbɑŋ̩ n̩nɪtʼɪ] ...who knows the truth [jɑsɑt͡sʼɪbɪŋ̩ n̩nɪtʼɪ] ...who does not know the truth

Consecutive ([tʼɑtʼɛlɑnɔ]) [kʼɑt͡sʼɪbɑ n̩nɪtʼɪ] ...and then I knew the truth [kʼɑsɪt͡sʼɪbɪ n̩nɪtʼɪ] ...and then I did not know the truth

Habitual ([t͡ɬʼʷɑɛlɔ]) [kʼɪje kʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I often know the truth [hɑŋ̩kʼe kʼɪt͡sʼɪbɪ n̩nɪtʼɪ] I do not often/often do not know the truth

Infinitive ([hʊ]) [hʊt͡sʼɪbɑ n̩nɪtʼɪ] To know the truth [hʊsɪt͡sʼɪbɪ n̩nɪtʼɪ] To not know the truth

Imperative ([tʼɑ'ɛlɔ]) [t͡sʼɪbɑ n̩nɪtʼɪ] Know the truth [sɪt͡sʼɪbɪ n̩nɪtʼɪ] Do not know the truth

According to Doke's analysis, the situative is just the participial sub-mood, the qualificative is a form of the participial, the consecutive is the past subjunctive (used when telling stories, it sounds as if the story-teller is using the present tense to describe a past action, but the subjectival concords differ from those of the indicative present), the habitual is a multi-verbal tense using a specific set of deficient verbs (Group III in Doke's classification) followed by a perfect subjunctive (and its negative uses another deficient verb in the same group); while the infinitive and most imperatives are not verbal moods (they are separate parts of speech and cannot be used as the predicate of a sentence, though imperatives can form interjectival sentences and there is a form of the subjunctive that could alternatively be interpreted as an imperative using subjectival concords).

The analysis is further complicated by the seemingly unpredictable form of the negative for each tense of each verb. The one point most Bantuists seem to agree upon is that, apart from the indicative mood, Bantu languages also have a subjunctive formed (usually) by changing the final vowel of the verb to *-e (which corresponds to Sesotho /ɛ/).

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]