How Swahili packages meaning

The subject-concord prefix is obligatory on every finite verb; the free pronoun is not. Free pronouns are reserved for emphasis, contrast, or focus.

Class-1 human singular has a constructionally conditioned alternant: a- appears before a finite verb (Hamisi anasoma "Hamisi is reading"), while yu- appears before the locative copular stems -ko / -po / -mo and before non-verbal predicative complements of state (Hamisi yuko wapi? "Where is Hamisi?"; Hamisi yu mrefu "Hamisi is tall"). This distribution is based on Ashton 1944 p. 28, not on a simple vowel-vs-consonant rule.

The full paradigm, beyond persons 1–2, is driven entirely by noun class — the same slot that carries "I/you/we" also carries "the book", "the tree", "the water", with the prefix changing to match. There is no separate agreement system for human vs non-human subjects; it is all class concord.

The habitual tense -hu- (hu-zungumza "[one] speaks habitually") is the one form that takes no subject prefix at all — it is used for all classes and numbers and expresses generic or recurrent action detached from a specific subject (Ashton 1944 pp. 36, 38).

Swahili is an asymmetric one-object-prefix language: only one object-concord slot is available on the verb. If more than one argument is present, only the topical/given one gets infixed; the other stays as a full noun.

The object infix is not simply an animacy marker. Corpus studies (Seidl & Dimitriadis 1997; Mursell 2018) show it is driven by information-structural givenness: it is obligatory for left-dislocated objects, dropped object pronouns, applicatives, and written-style object relatives, but optional for definite/animate objects when they are focal or new.

Personal pronouns and proper names are inherently given/definite, so when they are objects, the object infix is used and the free pronoun may be doubled for emphasis (Ninakuona wewe "I see you — you yourself").

Indirect objects do not get a dedicated marker; instead, the verb is extended with the applicative suffix -i-/-e- (dimension 7: Voice), which lets the beneficiary occupy the normal object slot.

The -hu- habitual is the unique form that carries no subject prefix (Ashton 1944 pp. 36, 38). It expresses generic/recurrent action detached from a specific subject; historically a contraction of ni + ku (Ashton p. 38 fn).

Swahili additionally has slot-based conditional TAMs: -ki- (realis situative: "if/while"), -nge- (present counterfactual: "would"), -ngali- (past counterfactual: "would have"), and -ja- (negative perfect "not yet"). All occupy the same tense slot.

There is no morphological remote-past vs near-past split. The -li- / -me- contrast is more about event-completion-with-current-relevance (-me-) vs. perfective narrative past (-li-) than distance in time; corpus studies show the two alternate freely in a single event chain (notes §25, Zawawi 1971).

Time adverbs (jana "yesterday", kesho "tomorrow", sasa "now") co-occur with tense prefixes but never replace them — a fully tenseless finite clause is not possible in Standard Swahili.

Swahili does NOT have a dedicated habitual in -na-. The analytical habitual is -hu- (Ashton 1944 pp. 36, 38); -na- is present progressive / definite present. English translations collapse both — "I speak Swahili" can gloss either — so the Swahili contrast is easy to miss in translation.

The perfect -me- covers both resultative ("he is standing" from "he stood up and is now standing") and experiential ("I have read this"). On state verbs it reads as a current state; on process verbs it reads as completed-with-current-relevance.

The situative -ki- is one of Swahili's most productive subordinators: it introduces simultaneous background clauses ("while X-ing") and realis conditional clauses ("if X happens"), with the same form for both.

Because tense and aspect share the same prefix slot, the -ja- negative perfect ("not yet") and -sipo- negative situative ("if not") are also aspectual — they substitute for -me- and -ki- in their respective negative contexts.

The subjunctive final vowel -e is triggered by purpose clauses, polite commands, embedded hortatives, and the plural-imperative suffix -ni. Example: nataka aende "I want him/her/them to go" — aende has the subjunctive -e because it is the complement of "want".

The -nge- / -ngali- pair is a clean counterfactual system. -nge- is present-reference ("if I were speaking now"), -ngali- is past-reference ("if I had been speaking then"). Both can also appear as -ngeli- in some registers.

Modality that English expresses with "can, must, should, might" is handled lexically in Swahili: weza "be able to", lazima "it is necessary that" + subjunctive, taka "want", fa(h)a "be proper that" + subjunctive. Each takes an infinitive or a subjunctive complement rather than a modal auxiliary.

The imperative singular uses the bare stem with final -a (soma! "read!"). When an object prefix is added, the final vowel shifts to -e (niambie! "tell me!") — the same subjunctive shape. The plural/polite adds -ni after the subjunctive -e: someni! "read! (pl)".

Swahili negation has two structural positions. Finite main-clause negation uses the pre-subject ha- (fused to si- in 1SG). Subordinate and subjunctive negations use the internal slot -si- between SM and TAM — exactly the slot that NEG1 cannot reach.

The present-tense FV change -a → -i is what tells you it is a negative clause when the verb has no overt tense prefix. The other tenses keep -a but swap the TAM: past -li- → -ku-, perfect -me- → -ja-.

Arabic-origin verbs (samehe "forgive", rudi "return", jibu "reply", fikiri "think") end in -e, -u, or -i rather than -a, and they do NOT undergo the -a → -i flip in the negative — their final vowel stays fixed. This is the one systematic exception to the FV rule.

The relative-clause internal negator -si- combines with the relative -po- to give -sipo- "if not / unless" — effectively the negative counterpart of the situative -ki-.

Extensions can stack. APPL-CAUS and CAUS-APPL are both attested, and APPL-REC / REC-APPL are both attested, with the surface order reflecting semantic scope (Ngonyani 2016 corpus, notes §23). Fixed orders: REV-APPL, STAT-APPL, APPL-PASS. Ungrammatical: *APPL-REV, *APPL-STAT, *PASS-APPL.

The stative -ik-/-ek- is not a second passive — it expresses a property or end-state of the patient without an agent. Compare kitabu kimesomwa "the book has been read (by someone)" with kitabu kinasomeka "the book is readable / legible".

A single root can grow many derived verbs: soma "read" → someka "be readable", somea "read to/for", somesha "teach", somana "read each other", someshwa "be taught". This is why a Swahili dictionary entry looks so short: extensions generate most of the family of related verbs on the fly.

The applicative is not restricted to beneficiaries. It also introduces reasons ("because of"), instruments ("with"), and locations ("at/on"), depending on the verb and the noun argument that follows.

Evidentiality is almost universally optional in Bantu languages — there is no paradigm slot for it in the 10-slot verb template, and no obligatory particle.

The common hearsay construction inasemekana kwamba "it is said that" is morphologically a class-9 subject + present + the verb sema "say" with stative + reciprocal extensions — voice-system machinery doing evidential-ish work rather than a dedicated evidential.

Specific stories or historical reports often open with hapo zamani za kale "a long time ago" + a matrix verb like inasimuliwa "it is narrated" (passive of simulia "tell a story"), which cues the listener that the speaker is not the eyewitness.

A non-witnessed, surprising, or ironic tone can be signalled prosodically or lexically (ati "supposedly", kwa hakika "indeed"), but none of this is grammatically required.

Each class has its own demonstrative forms. Class 1: huyu / huyo / yule. Class 7 (things): hiki / hicho / kile. Class 2 (human plural): hawa / hao / wale. The three-way near / anaphoric / distal contrast is consistent across all classes.

Demonstratives follow the noun (watu hawa, kitabu hiki), not the other way around. Putting the demonstrative first is either emphatic/fronted or ungrammatical depending on the register.

Because object-concord infixes on the verb already track a specific / given object, they provide a second way to signal definiteness without an article: ninakisoma "I am reading it (the specific one)" is more definite-feeling than ninasoma kitabu "I am reading a/the book".

In noun-class 1, "referential" huyo is the form that means "the aforementioned X" — a crucial discourse tool for cross-clause reference that no English article cleanly maps onto.

Each class pair has a prototypical semantics: 1/2 people, 3/4 trees and powerful/active things, 5/6 fruits and paired body parts and masses, 7/8 tools and diminutives, 9/10 animals and loanwords (plus most inanimates without a clear prefix), 11/10 thin abstracts, 14 qualities (often no plural). But these are prototypes, not hard rules — loanwords get absorbed where they fit phonologically.

Arabic-origin adjectives (safi "clean", ghali "expensive", hodari "skilled") do not carry the concord prefix — they stay invariant. Native Bantu adjectives (-zuri, -kubwa, -dogo) always take the class prefix.

There is no dedicated dual form, no inclusive/exclusive "we", and no numeral classifier system. Number is entirely a property of the class prefix itself.

Numerals 1–5 and 8 take the class prefix and behave like adjectives (kitabu kimoja "one book", vitabu vitatu "three books"). Numerals 6, 7, 9, 10, and the Arabic-origin higher numerals (tisa, kumi, ishirini) are invariant and do not agree.

The pronominal possessor stems are: -angu (my), -ako (your sg), -ake (his/her/their sg), -etu (our), -enu (your pl), -ao (their pl). The class prefix on the linker is what changes per class — the possessor stem is invariant.

The same -a linker handles non-pronominal possession too: ch-a mwalimu "of the teacher", y-a watu "of people", w-a ndege "of the birds". So "the teacher's book", "the book of the teacher", and "a teacherly book" are all realised the same way — kitabu cha mwalimu.

Agreement is with the POSSESSED, not the possessor. Kitabu cha mwalimu — the linker ch- agrees with kitabu (class 7), not with mwalimu (class 1). This is the opposite of, say, English "the teacher's book" where the -'s agrees with the possessor.

Swahili has no single verb "to have" in the English sense. Possession of things is expressed with the verb kuwa na "be with": Nina kitabu = Ni- + na "with" + kitabu = "I am with a book" = "I have a book". Kinship and body parts use the same construction.

The locative -ni is semantically flexible: the same form nyumbani can mean "at home" (static), "to the house" (goal), or "from the house" (source), depending on the verb. The three locative classes 16/17/18 are what disambiguate: -po "at a specific point", -ko "in the general area / directional", -mo "inside".

The locative copula stems -ko / -po / -mo are the reason class-1 SM surfaces as yu- rather than a- in "Hamisi yuko wapi?" — SM concord with these stems uses the non-verbal alternant (Ashton 1944 p. 28; notes §4).

Body-part nouns often do double duty as spatial terms: juu "on top / above", chini "below" (chi "ground" + -ni), ndani "inside" (nda "inner" + -ni), nje "outside". These combine freely with ya (class 9 linker) to form compound spatial phrases: juu ya meza "on top of the table".

WALS 85A classifies Swahili as a prepositional language. The locative -ni is morphological, not a postposition — it is a class-changing suffix that redirects the noun into the locative-class concord system.

The focus-cleft word ni is historically and synchronically distinct from the subject prefix ni-. Written separately, it asserts "it is X (that …)" and requires the relative clause following it (ni Juma aliyenunua "it is Juma who bought" — note the -ye- relative infix).

Left-dislocating an object promotes it to topic and forces an object-concord prefix on the verb to cross-reference it — this is one of the obligatory contexts for OM (Mursell 2018; notes §24). Without the OM, the sentence is ungrammatical.

Question words stay in-situ — they do NOT need to move to a focus position. "Juma amenunua nini?" (Juma has bought what?) is perfectly grammatical. But the cleft construction can be used for emphasis: "Ni nini Juma alichonunua?" (literally "It is what that Juma bought?") = "What was it that Juma bought?".

Yes-no questions use the particle je at the start or rising intonation alone; neither strategy shifts word order.

The plural-for-respect strategy extends to the imperative: Someni! (read! pl/polite) is used both for a group and for a single elder. The bare singular Soma! is reserved for peers and children.

Shikamoo (from Arabic "I hold your feet") is the fixed junior-to-elder greeting, with the mandatory response marahaba from the elder. Failure to use shikamoo to an elder is a sharp breach of coastal etiquette.

Unlike Japanese or Korean, Swahili verbs do not inflect for register — there is no "polite" verb ending that shadows a "plain" one. The honorific work is done entirely by pronoun and formula choice.

Titles (Bwana "Sir", Bibi "Madam", Mzee "Elder", Mwalimu "Teacher", Ndugu "Brother/Sister") are frequently used instead of or alongside names in polite address. Teknonymy (Mama/Baba + child's name) is common in coastal and mainland traditional registers alike.