# Swahili

*A reference grammar*

SVO like English and nothing else like it: a dozen or more noun classes agreeing across the whole clause, prefixes rather than suffixes, and adjectives, numerals and demonstratives all following the noun.  
Generated 25 October 2026.

## What this is

This describes **a human language**, carried in **vocal** form and heard rather than read. Its smallest contrastive unit is called a *sound* throughout.

This is a language in the full sense: every one of Hockett's design features is present.

| Design feature | Present? |
|---|---|
| Duality of patterning | yes |
| Displacement | yes |
| Productivity | yes |
| Arbitrariness | yes |
| Cultural transmission | yes |
| Discreteness | yes |

## At a glance

SVO word order, prepositions, adjectives after nouns, agglutinative, no case marking, no articles, a passive, pro-drop.

Written left to right in an alphabet, with 27 distinct sounds and 27 words so far.

## 1. Phonology

### Consonants (22)

| IPA | Written | Sound |
|---|---|---|
| p | p | voiceless bilabial plosive |
| t | t | voiceless alveolar plosive |
| k | k | voiceless velar plosive |
| b | b | voiced bilabial plosive |
| d | d | voiced alveolar plosive |
| ɡ | g | voiced velar plosive |
| m | m | voiced bilabial nasal |
| n | n | voiced alveolar nasal |
| ɲ | ny | voiced palatal nasal |
| ŋ | ng | voiced velar nasal |
| f | f | voiceless labiodental fricative |
| v | v | voiced labiodental fricative |
| θ | th | voiceless dental fricative |
| ð | dh | voiced dental fricative |
| s | s | voiceless alveolar fricative |
| z | z | voiced alveolar fricative |
| ʃ | sh | voiceless postalveolar fricative |
| h | h | voiceless glottal fricative |
| l | l | voiced alveolar lateral approximant |
| r | r | voiced alveolar trill |
| j | y | voiced palatal approximant |
| w | w | voiced labial-velar approximant |

### Vowels (5)

| IPA | Written | Sound |
|---|---|---|
| i | i | close front unrounded vowel |
| u | u | close back rounded vowel |
| e | e | close-mid front unrounded vowel |
| o | o | close-mid back rounded vowel |
| a | a | open front unrounded vowel |

### Word shape

Syllables are `(C)(C)V` — `C` a consonant, `V` a vowel, parentheses an optional slot.

A word runs from 2 to 6 syllables.

Stress falls on the second-to-last syllable.

### Stress, tone and length

| | |
|---|---|
| Does vowel length change a word's meaning | No — length is not contrastive |
| Does consonant length change a word's meaning | No |
| Does pitch on a syllable change a word's meaning | No tone |
| Where does stress fall | Always the same syllable — first, last, penultimate |

### The sounds themselves

| | |
|---|---|
| Are there nasal vowels distinct from oral ones | No |
| How do stops contrast | Voiced against voiceless — /b/ against /p/. English, French |
| Are there ejective or implosive consonants | No |
| Are there retroflex consonants | No |
| Are there uvular consonants | No |
| Are there clicks | No |

### What happens when sounds meet

| | |
|---|---|
| Do the vowels in a word have to match | No |
| What happens to a voiced consonant at the end of a word | Nothing — it stays voiced |
| Does the language repeat part of a word to change its meaning | No |

### Syllable and word shape

| | |
|---|---|
| How many consonants may start a syllable | Up to two — Spanish, Greek |
| How may a syllable end | Open only — every syllable ends in a vowel |
| Can a consonant be a syllable on its own | Yes |
| Is there a minimum size a word must reach | Yes — a word must be at least two syllables or one heavy one |
| Does syllable weight matter | No |
| How long is a typical word | Four or more — agglutinating languages routinely |
| May a word begin with a vowel | Yes |

### Rhythm and timing

| | |
|---|---|
| What is the rhythm of the language | Syllable-timed — every syllable takes about the same time. Spanish, Italian, Japanese |
| Do unstressed vowels weaken | No — every vowel keeps its quality |
| Which syllable carries the accent, when it is fixed | The second from last — Polish, Swahili |
| Do long words carry a secondary stress | No |

### Intonation

| | |
|---|---|
| What does the pitch do at the end of a statement | Falls |
| What does the pitch do in a yes/no question | Rises |

### The shape of the inventory

| | |
|---|---|
| How large is the sound inventory meant to be | Average — around 25 to 35 |
| How many vowels | Five — the world's commonest system |

### Sounds changing in context

| | |
|---|---|
| Do neighbouring consonants become more alike | In place — "in-possible" becomes "impossible" |
| Is a vowel inserted to break up an awkward cluster | Yes |
| Are sounds dropped in fast or casual speech | No |
| What happens to a borrowed word with foreign sounds | It is remade in native sounds |
| Does the writing show predictable sound changes | No — spelling keeps the underlying form |

## 2. Writing system

The script is **alphabetic**: one glyph per sound, vowels written as fully as consonants.

- Written left to right.
- Words are separated by spaces.
- A sentence ends with `.`.

*No glyphs have been drawn yet, so the language can only be shown transliterated.*

### Marks on the page

| | |
|---|---|
| Does the script have two cases | Yes — capitals and small letters |

### Layout and division

| | |
|---|---|
| How are words separated | By spaces |
| How much punctuation is there | A full set — stops, commas, quotes, questions |
| Is there a mark for a new section or paragraph | No |
| How closely does spelling follow sound | Very closely — Finnish, Spanish, Korean |
| Are there conventional abbreviations | Yes |

### Numbers and dates

| | |
|---|---|
| How are numbers written | With digits, positionally |
| Which way do multi-digit numbers run | The same way as the text |

### How the script came to be

| | |
|---|---|
| Where did the script come from | Imposed from outside |
| What was it first written on | Paper or parchment — free-flowing |
| Do letters join or merge when adjacent | No |
| Do letters change shape by position in the word | No |
| Does the direction alternate line by line | No — every line runs the same way |
| Has the spelling been reformed | Once |
| Is there a calligraphic tradition | No |
| Is there a shorthand or cursive for fast writing | No |
| What is the script called | Latin |

## 3. Transliteration

Graphemic: one sound, one spelling, and the original can be recovered. Sounds with no Latin letter are written with two letters.

How the language is written back in Latin letters. Each sound's own spelling is in the phonology tables above.

- Morphemes run together, with no hyphens.
- The first word of a sentence and every proper noun take a capital.

### Names and borrowings

| | |
|---|---|
| How are foreign names written | Kept in their own spelling |
| Are romanised proper nouns marked | Yes — with a capital |

### Where systems differ

| | |
|---|---|
| May diacritics be dropped when typing | Yes — the bare letters are acceptable |
| Is there more than one accepted romanisation | One standard |


**The romanisation is not reversible.** "ng" for ŋ reads equally as n + g "th" for θ reads equally as t + h "dh" for ð reads equally as d + h "sh" for ʃ reads equally as s + h

## 4. Grammatical categories

What the language distinguishes. How each is expressed is in the next section.

| Category | Distinguished? |
|---|---|
| Number | singular against plural |
| Definiteness | no |
| Case | no |
| Tense | past, present, future |
| Aspect | progressive against simple |
| Agreement | the verb agrees with its subject |
| Copula | "be" is a verb |
| Voice | active, passive ("-wa", agent by "na") |
| Pro-drop | a pronoun subject may be dropped |

### Number beyond singular and plural

| | |
|---|---|
| What number distinctions are there | Singular against plural |
| Does "we" distinguish whether the hearer is included | No — one word for both |
| How many cases does the language distinguish | None |
| Is there a way to say "X and those with X" | No — a plural means more of the same thing |

### Gender and noun class

| | |
|---|---|
| Does the language sort nouns into classes | Many classes, by shape and kind — Bantu languages have a dozen or more |
| Do adjectives and verbs change to match the noun's class | Yes |
| How does a noun get its gender | By meaning alone — what the noun refers to decides |

### Mood and modality

| | |
|---|---|
| Which moods does the verb distinguish | Statement, command, and a subjunctive for the unreal |
| What aspects does the verb distinguish | A full set — habitual, iterative, inceptive and more |
| Does the verb change according to who is being addressed | No |
| Does negating a sentence change anything else about the verb | Yes — tense, aspect or agreement is marked differently under negation |
| Is subject agreement shown in every tense | Yes — every tense shows it |
| Does a command use a special form of the verb | No — the ordinary verb, with the subject left out |
| Is there a form for "may it be so" — a wish | No |

### Evidence and certainty

| | |
|---|---|
| Must a speaker say how they know | No — it is optional or absent |

### Voice and valency

| | |
|---|---|
| Is there a way to say "make someone do it" on the verb | Yes — a causative form |
| Can the verb promote an oblique to object | Yes — an applicative |
| Where is the relationship between words marked — on the head or on the dependent | On the head — agreement on the verb, as many American languages do |

### Time, more finely

| | |
|---|---|
| Does the past distinguish how long ago | No — one past |
| Is the future a tense, or a mood | A tense — grouped with past and present |
| Is there a perfect — a past with present relevance | Yes — "I have eaten" against "I ate" |
| Does the verb say whether the next clause has the same subject | No |
| Does the verb agree with the object as well as the subject | Yes |
| Is plural marking optional | No — obligatory whenever plural |
| Can part of a sentence be negated on its own | No — negation is always the whole clause |
| Does negation take two parts | No — one marker |
| Who can be commanded | The hearer only |

### Definiteness and specificity

| | |
|---|---|
| What does the article system track | There is no article |
| Do possessed nouns change form | No |
| What kind of classifier does counting need | None |

## 5. Morphology

The language is **agglutinative**: affixes stack up, each carrying one meaning.

| Category | Form | Attached as | Glossed |
|---|---|---|---|
| Plural | `wa--` | prefix | PL |
| Past | `li--` | prefix | PST |
| Future | `ta--` | prefix | FUT |
| Present | `na--` | prefix | PRS |
| Progressive | *not yet created* | prefix | PROG |
| Agreement, 1sg | `ni--` | prefix | 1SG |
| Agreement, 2sg | `u--` | prefix | 2SG |
| Agreement, 3sg | `a--` | prefix | 3SG |
| Agreement, plural | `tu--` | prefix | PL |
| Negation | `ha--` | prefix | NEG |
| Question | `je` | separate word | Q |

On a verb, aspect sits closest to the root, then tense, then agreement.

### How large that makes the paradigm

Your verb paradigm has at least 3 tense x 2 aspect x 3 person x 2 number x 2 voice x 2 polarity = 144 cells. 93% of the 146 languages in UniMorph have a verb paradigm no larger. At the rate real languages reuse forms — 0.881 distinct forms per cell — about 127 distinct forms would carry those 144 cells, with 17 filled by a form already doing another job. English "walked" is past and participle at once. At least, because this multiplies only the categories the app models: sentence markers you invented yourself, and any agreement richer than the app's yes/no, add cells this count cannot see.

### Pronouns

| | |
|---|---|
| Do third-person pronouns distinguish gender | By noun class |
| Is there more than one word for "you" | One — English |
| Where does a possessive marker attach | Suffixed to the thing possessed — Turkish ev-im "house-my" |
| Do adpositions change form for their object | No — the adposition is invariant |
| How is "someone" or "something" formed | From a generic noun — "person", "thing" |
| Is "this" the same word standing alone as it is before a noun | Different words entirely |
| Is the third-person pronoun related to a demonstrative | Unrelated — separate words with separate histories |
| How does a pronoun become plural | A plural marker used only on pronouns |

### Demonstratives and deixis

| | |
|---|---|
| How many degrees of distance do demonstratives have | Three — here, there, yonder. Spanish, Japanese, Latin |
| Do demonstratives mark elevation or direction | No |

### Numerals and counting

| | |
|---|---|
| What does the counting system group by | Tens |
| Where do numerals sit relative to what they count | After — dogs three |
| Are there separate ordinal numbers | Yes — first, second, third |
| Is there a way to say "two each" | Yes, by saying the numeral twice |

### Making new words

| | |
|---|---|
| How are new words most often made | By affix — happy, unhappiness |
| How is "one who does X" formed | m- |
| How is a diminutive formed | ki- |
| How is an abstract noun formed from an adjective | u- |
| Are there infixes — pieces inserted inside a word | Yes |
| Are the commonest words irregular in a way no rule explains | No — everything is regular |
| Does a verb change shape for how many are doing it | No |

### Clitics

| | |
|---|---|
| Where do clitics attach | To a particular kind of word |

### Mood and modality on the verb

| | |
|---|---|
| The form for ability or circumstance | weza |
| The form for possibility | labda |
| The form for obligation | lazima |

## 6. Syntax

### The clause

Basic order is **SVO**. Adpositions come before their noun, and adpositional phrases sit on the same side of the verb as the object does.

A manner adverb comes after the verb.

### Inside the noun phrase

| Modifier | Position |
|---|---|
| Adjective | After the noun |
| Numeral | After the noun |
| Article, demonstrative | After the noun |
| Possessor | After the possessed |

Where several modifiers fall on the same side, they mirror outward: the adjective sits nearest the noun, then the numeral, then the article.

### Negation

Negation is a prefix on the verb.

### Questions

A yes/no question takes a particle at the start of the sentence.

### Clauses inside clauses

A relative clause goes **after** the noun it modifies. Nothing marks the position the clause is missing: the gap is simply left open.

Relativisation reaches down to **direct objects** on the accessibility hierarchy, so the following can be relativised:

- yes — subjects — "the man **who saw me**"
- yes — direct objects — "the book **that she read**"
- **no** — indirect objects — "the child **I gave it to**"
- **no** — obliques — "the house **I live in**"
- **no** — possessors — "the man **whose dog barked**"

*Keenan and Comrie (1977): every language relativises an unbroken run from the top of this list, and none skips a position.*

A clause serving as an object is introduced by a complementiser **kwamba**, at the start of its clause.

A circumstance clause is opened by a subordinator; the general-purpose one is **wakati**. It goes after the main clause.

The nominaliser is **ku-**, attached to the subordinate verb.

A subordinate verb is **balanced**: it carries the same marking a main verb does.

### Joining things together

| | |
|---|---|
| Can unlike things be joined — a noun and a clause | Yes |
| Can a repeated verb be left out of the second half | No — the verb is repeated |
| Do multiple negatives reinforce rather than cancel | Yes — "I didn't see nobody" means I saw nobody |
| How is a negative command formed | A dedicated prohibitive form — Latin "noli", Greek "mē" |
| Is the word joining two nouns the same as the one joining two clauses | Yes — one word for both |
| Is the word for "and" also the word for "all" | No — separate words |

### Comparison

| | |
|---|---|
| How is a comparison made | An "exceed" verb — "the cat is big exceeds the dog". West Africa, Mandarin |
| What is the "exceed" verb | kuliko |
| How is a superlative made | The comparative plus "all" — "bigger than all" |
| How is "as big as" expressed | A dedicated equative marker |
| What is the equative marker | sawa na |
| Where does a degree word like "very" go | After the adjective — big very |
| Can an adjective stand without its noun — "the big one" | Yes, but it takes a marker |

### Having and belonging

| | |
|---|---|
| How is "I have a book" expressed | "I am with a book" — many African languages |
| Are body parts and kin possessed differently from objects | No — one construction for everything |
| Is a temporary possession marked differently from a lasting one | No — one construction for both |
| Is a possession that has ended marked differently from a current one | No — the tense of the clause does it |
| How is "there is a cat" expressed | The ordinary copula plus a location |
| How does the language say "the house is big" | Some adjectives take a copula and others behave as verbs |
| Is "she is a doctor" built the same way as "she is in the house" | The same — one copula covers both |
| Are there nouns that must always be possessed | No — any noun can stand alone |

### Emphasis and topic

| | |
|---|---|
| How is one part of a sentence emphasised | By moving it to a dedicated position |
| Is there a dedicated position for the topic | Yes — the topic comes first, whatever its role |

### Reflexives and reciprocals

| | |
|---|---|
| Is the emphatic "herself" the same word as the reflexive one | No — separate words |

## 7. Semantics

| | |
|---|---|
| Colour terms | the full set: black, white, red, green, yellow, blue and beyond |
| Counting | base ten |
| Kinship | Iroquois type |
| Motion | verb-framed: the path is in the verb ("exit running") |
| Possession | possession is marked one way for everything |
| Classifiers | none |

### Ideophones and expressive words

| | |
|---|---|
| Are there ideophones — words that depict rather than describe | A large open class covering sound, movement, texture and feeling |
| How do ideophones fit into a sentence | Like adverbs |
| Do ideophones break the ordinary sound rules | Yes — they use sounds and shapes no other word may |
| Are nouns and verbs clearly different classes | Yes |
| Are adjectives their own word class | Yes, but a small closed set — Igbo has about eight |

### Space and motion

| | |
|---|---|
| How is location described | Relative to the speaker — left, right, in front |

### Time

| | |
|---|---|
| Is the future ahead and the past behind | No — the past is ahead, because it can be seen |
| Is a repeated time said differently from a single one | No — the same construction, and context does the rest |
| How is time divided | By the sun, tides or seasons |

### What the language has words for

| | |
|---|---|
| How is the body divided into named parts | Finely — separate words for hand and arm, foot and leg |
| Is there one verb for eating and drinking, or two | Two |
| How are siblings named | By relative age — elder and younger, regardless of sex |
| Are maternal and paternal grandparents named differently | No |
| Are left and right named after anything | With moral or ritual weight attached |
| How many basic colour terms are there | Five or six |
| Is blue distinguished from green | No — one word covers both |
| How rich is the vocabulary for smell | Poor — smells are described by their source |
| Are red and yellow separate colours | Separate words |

### Borrowing

| | |
|---|---|
| How does the language take in foreign words | Freely, adapted to its own sounds |
| Which languages does it borrow from, and for what | Arabic, then English |

### People, places and things

| | |
|---|---|
| Is "with a friend" the same word as "with a knife" | No — company and instrument take different words |

## 8. Pragmatics

| | |
|---|---|
| Politeness | not grammatically marked |
| Evidentiality | not marked |
| Topic | shown by position |
| Focus | shown by position |
| Pronoun omission | the subject may be dropped |

A separate register exists for addressing particular people; small particles manage the flow of conversation.

### Addressing people

| | |
|---|---|
| How are people addressed | By kin term, extended to non-kin |
| Are some names or words forbidden in some company | No |

### Doing things with words

| | |
|---|---|
| What does a greeting ask about | How you are |
| How is a request usually made | Directly, with an imperative |
| Do listeners signal attention while someone else speaks | Yes, frequently — Japanese aizuchi |
| What does silence mean in conversation | Nothing — it is comfortable |
| Does "yes" agree with the question or with the fact | With the question |
| Do speakers use clicks outside words | Yes — to show feeling, as English tsk-tsk does |

### Registers and styles

| | |
|---|---|
| How many distinct registers are there | Formal and informal |
| Is there a separate language or register for ritual | Yes |
| Do speakers use different forms depending on who they are | No |
| Is there a distinct way of speaking to children | Yes — a separate vocabulary |
| Is reported speech marked as such | Yes — a quotative particle or verb |
| How is turn-taking managed | Overlap is normal and cooperative |

## 9. Typology

In one line: **SVO, agglutinative, prepositional, with no case marking, pro-drop**.

### Profile

| Feature | This language | Share of real languages | Rarity | Source |
|---|---|---|---|---|
| Order of Subject, Object and Verb | SVO | 35.5% | common | WALS, 1376 coded |
| Order of Adposition and Noun Phrase | Prepositions | 43.2% | very common | WALS, 1184 coded |
| Order of Adjective and Noun | Noun-Adjective | 64.3% | very common | WALS, 1367 coded |
| Order of Numeral and Noun | Noun-Numeral | 52.7% | very common | WALS, 1154 coded |
| Order of Genitive and Noun | Noun-Genitive | 37.5% | common | WALS, 1249 coded |
| Order of Relative Clause and Noun | Noun-Relative clause | 70.3% | very common | WALS, 824 coded |
| Prefixing vs. Suffixing in Inflectional Morphology | Strongly prefixing | 6% | rare | WALS, 969 coded |
| Fusion of Selected Inflectional Formatives | Exclusively concatenative | 75.8% | very common | WALS, 165 coded |
| Alignment of Case Marking of Full Noun Phrases | Neutral | 51.6% | very common | WALS, 190 coded |
| Expression of Pronominal Subjects | Subject affixes on verb | 61.5% | very common | WALS, 711 coded |
| Polar Questions | Question particle | 61.3% | very common | WALS, 955 coded |
| Definite Articles | No article | 31.9% | common | WALS, 620 coded |
| Coding of Nominal Plurality | Plural Prefix | 11.8% | uncommon | WALS, 1066 coded |
| Numeral Classifiers | Absent | 65% | very common | WALS, 400 coded |
| Negative Morphemes | Negative affix | 34.1% | common | WALS, 1157 coded |
| Position of Tense-Aspect Affixes | Tense-aspect prefixes | 13.5% | uncommon | WALS, 1131 coded |
| Position of Case Affixes | No case affixes | 36.8% | common | WALS, 1031 coded |
| Zero Copula for Predicate Nominals | Impossible | 54.7% | very common | WALS, 386 coded |
| Semantic Distinctions of Evidentiality | No grammatical evidentials | 43.3% | very common | WALS, 418 coded |
| Numeral Bases | Decimal | 63.8% | very common | WALS, 196 coded |
| Coding of Evidentiality | No grammatical evidentials | 43.3% | very common | WALS, 418 coded |
| Politeness Distinctions in Pronouns | No politeness distinction | 65.7% | very common | WALS, 207 coded |
| Is there a decimal numeral system? | Decimal counting | 70.8% | very common | Grambank, 1743 coded |
| Is the adnominal possessive construction different for alienable and inalienable nouns? | One possessive construction | 52.7% | very common | Grambank, 2031 coded |
| Are there numeral classifiers? | No numeral classifiers | 76.1% | very common | Grambank, 2175 coded |
| Is there grammatical marking of indirect evidence (hearsay, inference, etc.)? | Indirect evidence not marked | 70% | very common | Grambank, 1429 coded |
| Is there grammatical marking of direct evidence (perceived with the senses)? | Direct evidence not marked | 84.6% | very common | Grambank, 1423 coded |
| Is there a politeness distinction in 2nd person forms? | No politeness distinction in 2nd person | 86.1% | very common | Grambank, 1739 coded |
| Is there productive morphological plural marking on nouns? | Productive plural marking | 55.7% | very common | Grambank, 2302 coded |
| Is there any ergative alignment of flagging? | No ergative flagging | 78% | very common | Grambank, 1769 coded |
| Are there morphological cases for non-pronominal core arguments (i.e. S/A/P)? | No morphological case | 65.6% | very common | Grambank, 2262 coded |
| What is the order of numeral and noun in the NP? | Numeral after noun | 49.4% | very common | Grambank, 2199 coded |
| What is the order of adnominal demonstrative and noun? | Demonstrative after noun | 47.2% | very common | Grambank, 2259 coded |
| What is the pragmatically unmarked order of adnominal possessor noun and possessed noun? | Possessor after noun | 37.9% | common | Grambank, 2322 coded |
| What is the order of adnominal property word and noun? | Property word after noun | 58.4% | very common | Grambank, 2117 coded |
| Are there prepositions? | Prepositions | 54.9% | very common | Grambank, 2268 coded |
| Are there postpositions? | No postpositions | 51% | very common | Grambank, 2152 coded |
| Are there definite or specific articles? | No definite articles | 62.5% | very common | Grambank, 2198 coded |
| Do indefinite nominals commonly have indefinite articles? | No indefinite articles | 86.4% | very common | Grambank, 2221 coded |
| Is there a gender/noun class system where sex is a factor in class assignment? | Sex is not a factor in class assignment | 81.4% | very common | Grambank, 2206 coded |
| Is there a gender/noun class system where shape is a factor in class assignment? | Class assigned partly by shape | 7% | rare | Grambank, 2228 coded |
| Do verbs have prefixes/proclitics, other than those that only mark A, S or P (do include portmanteau: A & S + TAM)? | Verb prefixes or proclitics | 67.6% | very common | Grambank, 2173 coded |
| Do verbs have suffixes/enclitics, other than those that only mark A, S or P (do include portmanteau: A & S + TAM)? | No verb suffixes | 16.4% | uncommon | Grambank, 2265 coded |
| Is there overt morphological marking of present tense on verbs? | Present marked on the verb | 34.2% | common | Grambank, 2253 coded |
| Is there overt morphological marking on the verb dedicated to past tense? | Past marked on the verb | 52.4% | very common | Grambank, 2269 coded |
| Is there overt morphological marking on the verb dedicated to future tense? | Future marked on the verb | 45.3% | very common | Grambank, 2273 coded |
| Can the S argument be indexed by a suffix/enclitic on the verb in the simple main clause? | Subject not indexed by a suffix | 59.6% | very common | Grambank, 2315 coded |
| Can the S argument be indexed by a prefix/proclitic on the verb in the simple main clause? | Subject indexed by a prefix | 41.8% | very common | Grambank, 2324 coded |
| Can standard negation be marked by an affix, clitic or modification of the verb? | Negation marked on the verb | 46.8% | very common | Grambank, 2301 coded |
| What is the pragmatically unmarked order of S and V in intransitive clauses? | Subject before verb | 78.8% | very common | Grambank, 2320 coded |
| Can polar interrogation be indicated by a special word order? | Polar questions not by word order | 97% | very common | Grambank, 1758 coded |
| Is there a clause-initial polar interrogative particle? | Clause-initial question particle | 14.8% | uncommon | Grambank, 1776 coded |
| Is there a clause-final polar interrogative particle? | No clause-final question particle | 60.3% | very common | Grambank, 1765 coded |
| 'Want' Complement Subjects | Subject is expressed overtly | 25.4% | common | WALS, 283 coded |
| Purpose Clauses | Deranked | 60% | very common | WALS, 170 coded |
| Reason Clauses | Balanced | 53.3% | very common | WALS, 169 coded |
| Case Syncretism | No case marking | 62.1% | very common | WALS, 198 coded |
| Exponence of Selected Inflectional Formatives | No case | 46.3% | very common | WALS, 162 coded |
| Inflectional Synthesis of the Verb | 6-7 categories per word (5 marked) | 21.4% | common | WALS, 145 coded |
| Locus of Marking in Possessive Noun Phrases | Double marking | 9.3% | uncommon | WALS, 236 coded |
| Locus of Marking in the Clause | Head marking | 30.1% | common | WALS, 236 coded |
| Locus of Marking: Whole-language Typology | Clause head, phrase double marking | 51.3% | very common | WALS, 236 coded |
| Reduplication | No productive reduplication | 14.9% | uncommon | WALS, 368 coded |
| Syncretism in Verbal Person/Number Marking | Not syncretic | 40.9% | very common | WALS, 198 coded |
| Comitatives and Instrumentals | Differentiation | 66.1% | very common | WALS, 322 coded |
| Distance Contrasts in Demonstratives | Three-way contrast | 37.6% | common | WALS, 234 coded |
| Gender Distinctions in Independent Personal Pronouns | 3rd person singular only | 16.1% | uncommon | WALS, 378 coded |
| Inclusive/Exclusive Distinction in Independent Pronouns | No inclusive/exclusive | 60% | very common | WALS, 200 coded |
| Indefinite Articles | No definite or indefinite article | 37.1% | common | WALS, 534 coded |
| Indefinite Pronouns | Generic-noun-based | 26.1% | common | WALS, 326 coded |
| Number of Genders | Five or more | 9.3% | uncommon | WALS, 257 coded |
| Occurrence of Nominal Plurality | All nouns, always obligatory | 45.7% | very common | WALS, 291 coded |
| Person Marking on Adpositions | No person marking | 55.3% | very common | WALS, 378 coded |
| Position of Pronominal Possessive Affixes | Possessive suffixes | 39.4% | common | WALS, 902 coded |
| Sex-based and Non-sex-based Gender Systems | Non-sex-based | 10.9% | uncommon | WALS, 257 coded |
| Action Nominal Constructions | Possessive-Accusative | 17.3% | uncommon | WALS, 168 coded |
| Genitives, Adjectives and Relative Clauses | Gen after, Adj after, Rel after | 10.9% | uncommon | WALS, 138 coded |
| Nominal and Verbal Conjunction | Identity | 53.5% | very common | WALS, 301 coded |
| Noun Phrase Conjunction | 'And' different from 'with' | 56% | very common | WALS, 234 coded |
| Absence of Common Consonants | All present | 88.7% | very common | WALS, 567 coded |
| Consonant Inventories | Average (22) | 35.7% | common | WALS, 563 coded |
| Consonant-Vowel Ratio | Average (4.4) | 41.5% | very common | WALS, 564 coded |
| Fixed Stress Locations | Penultimate | 21.9% | common | WALS, 502 coded |
| Front Rounded Vowels | None | 93.4% | very common | WALS, 562 coded |
| Glottalized Consonants | No glottalized consonants | 72.1% | very common | WALS, 567 coded |
| Lateral Consonants | /l/, no obstruent laterals | 68.4% | very common | WALS, 567 coded |
| Rhythm Types | No rhythmic stress | 30.3% | common | WALS, 323 coded |
| Syllable Structure | Complex | 31.1% | common | WALS, 486 coded |
| The Velar Nasal | No initial velar nasal | 18.6% | uncommon | WALS, 469 coded |
| Tone | No tones | 58.3% | very common | WALS, 527 coded |
| Uvular Consonants | None | 82.9% | very common | WALS, 567 coded |
| Voicing and Gaps in Plosive Systems | None missing in /p t k b d g/ | 45% | very common | WALS, 567 coded |
| Voicing in Plosives and Fricatives | In both plosives and fricatives | 27.9% | common | WALS, 567 coded |
| Vowel Nasalization | Contrast absent | 73.8% | very common | WALS, 244 coded |
| Vowel Quality Inventories | Average (5-6) (5) | 50.9% | very common | WALS, 564 coded |
| Weight-Sensitive Stress | Not weight-sensitive | 5.2% | rare | WALS, 500 coded |
| Alignment of Verbal Person Marking | Accusative | 55.8% | very common | WALS, 380 coded |
| Antipassive Constructions | No antipassive | 75.3% | very common | WALS, 194 coded |
| Applicative Constructions | Benefactive and other; both bases | 26.8% | common | WALS, 183 coded |
| Comparative Constructions | Exceed | 19.8% | uncommon | WALS, 167 coded |
| Ditransitive Constructions: The Verb 'Give' | Double-object construction | 22.2% | common | WALS, 378 coded |
| Negative Indefinite Pronouns and Predicate Negation | Predicate negation also present | 82.5% | very common | WALS, 206 coded |
| Nominal and Locational Predication | Identical | 30.3% | common | WALS, 386 coded |
| Nonperiphrastic Causative Constructions | Morphological but no compound | 81.9% | very common | WALS, 310 coded |
| Order of Person Markers on the Verb | A precedes P | 25.3% | common | WALS, 379 coded |
| Passive Constructions | Present | 43.4% | very common | WALS, 373 coded |
| Periphrastic Causative Constructions | Purposive but no sequential | 57.6% | very common | WALS, 118 coded |
| Predicative Adjectives | Mixed | 26.7% | common | WALS, 386 coded |
| Predicative Possession | Conjunctional | 24.6% | common | WALS, 240 coded |
| Reciprocal Constructions | Distinct from reflexive | 56.6% | very common | WALS, 175 coded |
| Symmetric and Asymmetric Standard Negation | Asymmetric | 17.8% | uncommon | WALS, 297 coded |
| Third Person Zero of Verbal Person Marking | No zero realization | 47.6% | very common | WALS, 380 coded |
| Verbal Person Marking | Both the A and P arguments | 51.1% | very common | WALS, 378 coded |
| Imperative-Hortative Systems | Minimal system | 5.3% | rare | WALS, 375 coded |
| Perfective/Imperfective Aspect | Grammatical marking | 45.5% | very common | WALS, 222 coded |
| Suppletion According to Tense and Aspect | None | 63.7% | very common | WALS, 193 coded |
| The Future Tense | Inflectional future exists | 49.5% | very common | WALS, 222 coded |
| The Past Tense | Present, no remoteness distinctions | 42.3% | very common | WALS, 222 coded |
| The Perfect | Other perfect | 36% | common | WALS, 222 coded |
| The Prohibitive | Normal imperative + special negative | 36.7% | common | WALS, 496 coded |
| Order of Adverbial Subordinator and Clause | Initial subordinator word | 60.4% | very common | WALS, 659 coded |
| Order of Degree Word and Adjective | Adjective-Degree word | 39.9% | common | WALS, 481 coded |
| Order of Demonstrative and Noun | Noun-Demonstrative | 45.9% | very common | WALS, 1225 coded |
| Order of Object and Verb | VO | 46.4% | very common | WALS, 1518 coded |
| Order of Object, Oblique, and Verb | VOX | 42% | very common | WALS, 500 coded |
| Order of Subject and Verb | SV | 79.7% | very common | WALS, 1496 coded |
| Position of Interrogative Phrases in Content Questions | Not initial interrogative phrase | 68.2% | very common | WALS, 902 coded |
| Position of Polar Question Particles | Initial | 14.6% | uncommon | WALS, 884 coded |
| Relationship between the Order of Object and Verb and the Order of Adjective and Noun | VO and NAdj | 34.7% | common | WALS, 1316 coded |
| Relationship between the Order of Object and Verb and the Order of Adposition and Noun Phrase | VO and Prepositions | 39.9% | common | WALS, 1142 coded |
| Relationship between the Order of Object and Verb and the Order of Relative Clause and Noun | VO and NRel | 47.3% | very common | WALS, 879 coded |

*WALS and Grambank are read side by side and never averaged: WALS asks what a language's dominant pattern is, Grambank mostly asks whether a language can do something at all, so a mean of the two would mean neither thing.*

### Combinations

No single answer above can be right or wrong. Combinations can be rare, and these are the ones worth remarking on.

**Word order and adpositions**  
Verb-initial and SVO languages overwhelmingly use prepositions.  
301 of 1028 languages coded for both share this, among them Arabic (Egyptian), Arapesh (Mountain), Bagirmi, English, French, Hausa.

**Adpositions and possessors**  
Prepositional languages usually put the possessor after the noun; postpositional languages usually put it before.  
350 of 1004 languages coded for both share this, among them Arabic (Egyptian), Bagirmi, Berber (Middle Atlas), Chamorro, Fijian, French.

**Word order and relative clauses**  
Relative clauses go before the noun almost only in verb-final languages. Mandarin is the well-known exception.  
293 of 751 languages coded for both share this, among them Apurinã, Arabic (Egyptian), Bagirmi, English, Finnish, French.

**Possessors and relative clauses**  
Both modify a noun, and languages are fairly consistent about which side modifiers go.  
328 of 755 languages coded for both share this, among them Arabic (Egyptian), Bagirmi, Berber (Middle Atlas), Chamorro, Fijian, French.

**Word order and adjectives**  
Adjective order correlates only weakly with verb position, so this combination is freer than the others.  
346 of 1182 languages coded for both share this, among them Apurinã, Arabic (Egyptian), Bagirmi, French, Grebo, Guaraní.

### Nearest real languages

The languages answering the most features the same way. Resemblance here is typological only: it says nothing about relatedness, and two languages can match closely without sharing a word.

| Language | Family | Features shared |
|---|---|---|
| Kinga | Niger-Congo | 13 of 13 |
| Langi | Niger-Congo | 12 of 12 |
| Amara | Austronesian | 11 of 11 |
| Hehe | Niger-Congo | 11 of 11 |
| Ayiwo | Austronesian | 10 of 10 |
| Bankon | Niger-Congo | 10 of 10 |
| Hemba | Niger-Congo | 10 of 10 |
| Kasong | Austro-Asiatic | 10 of 10 |

*WALS Online v2020.5, ed. Matthew S. Dryer & Martin Haspelmath (2013), Max Planck Institute for Evolutionary Anthropology. wals.info, CC BY 4.0.*

## 10. Descent

Nothing to record here: this language is not derived from anything. It is the root of its own family, which is exactly what a proto-language is.

A language gains this chapter by being evolved from another one, borrowed from, or reconstructed — at which point the parent, the sound changes that applied and a set of worked derivations are recorded below.

### How it could change from here

Four mechanisms, and only one of them makes anything new. Nothing below has happened; this is what each would do to the language as it stands.

**Analogy.** Sound change makes a language irregular and nothing in it pushes back. Of your 27 written words, 16 sit in frequency bands that analogy reaches and 11 are common enough to resist it. The frequency effect runs OPPOSITE to sound change here: a common word leads a reductive change and resists an analogical one, which is why English still has "went" and "children" while every rare strong verb has quietly gone weak. Real languages settle at 0.881 distinct forms per paradigm cell, so a fully regular paradigm is not the destination either.

**Reanalysis.** Reanalysis changes nothing you can hear. A listener assigns a different structure to exactly the string the speaker produced, and the language carries both readings with no evidence either way — sometimes for centuries — until somebody says something the old structure could not have generated. That utterance is the first moment anyone, including this app, can tell it happened.

- "I am going to eat" was parsed as a motion verb with a purpose clause: going, in order to eat, and is now parsed as a future auxiliary with its main verb: will eat. Nothing changed on the surface. What gives it away is "I'm going to stay here" — motion and staying cannot both be meant, so the old parse is gone.  
- "a stem ending in the same sound as a common suffix" was parsed as part of the root, and is now parsed as the suffix itself, leaving a shorter root behind. Nothing changed on the surface. What gives it away is the new shorter root turning up with a DIFFERENT suffix, which proves the speaker thinks the boundary moved.  

**Borrowing.** Regular contact moves nouns, verbs, adjectives. Everything below that stays put — the hierarchy is an ordering, so a language that has borrowed affixes has certainly borrowed nouns, and one that has borrowed only nouns tells you the contact was shallow. Measured against the Leipzig-Jakarta ranking, the words of yours most likely to survive it are "go", "I", "big", "child". On the sound side, SegBo's most-travelled segments your language lacks are /d̠ʒ/ (74 languages), /t̠ʃ/ (68 languages), /ʒ/ (32 languages).

**Or a shift, which runs the other way.** Did your speakers keep their language and take words in, or did they give up a language of their own and carry its habits into this one? The two produce nearly opposite evidence from the same contact, and only you know which happened.

Your speakers gave up a language of their own and carried its habits into this one. **Lexicon.** FEW loanwords — conspicuously few — the shifting speakers are learning the new vocabulary, not contributing theirs. **Phonology.** A foreign accent: the old sound system showing through the new one — phonology is the hardest thing to acquire and the first thing to leak. **Morphology.** Simplification, and categories the old language lacked going missing — imperfect learning drops distinctions the learners had no equivalent for. **Syntax.** Substantial transfer, early — syntactic habits survive a change of vocabulary almost intact. Loanwords are the tell, and they run the wrong way round. Heavy borrowing leaves many; a shift leaves almost none, because nobody was borrowing — they were failing to acquire. A language with a foreign accent, simplified morphology and a native word-stock has been shifted into, not borrowed from.

**Pattern without matter.** A language can be reshaped to the bone by a neighbour and show almost no loanwords, because what travelled was the structure and not the form. Calque: A foreign compound rebuilt word for word out of native parts. German "Fernsehen" for television, from "far" and "see"; English "loanword" itself, a calque of German "Lehnwort". Replicated word order: The neighbour's order adopted while every word stays native. The Balkan languages converging on a postposed article and losing the infinitive, each with its own vocabulary throughout.

*For scale: Glottolog counts 7,674 spoken languages in 421 families. The median one sits 6 branchings below its family's root, and the deepest — in Atlantic-Congo — sits 26. 183 of those families are a single language with no known relative, so a family of one is the commonest shape there is rather than an unfinished one.*

## 11. Examples

```
Kwa ndi- paka-ku- sana kubwa-le m- -a- -ni a--na-li-ruka-i- quickly katikakwa nyasi-ku- sana kijani-le -ni.
TOP FOC.POS cat-INFIX DEG big-SUBST CL AGR.CL =HOST 3SG-PFV-PST-jump-APPL quickly into-INSTR grass-INFIX DEG green-SUBST =HOST
kwa ndi- paka-ku- sana kubwa-le m- -a- -ni a--na-li-ruka-i- katikakwa ɲasi-ku- sana -le -ni
The big cat quickly jumped into the green grass.
```

```
Kwa ndi- ni--na-na-enda nyumbani.
TOP FOC.POS 1SG-IPFV-PRS-go home
kwa ndi- ni--na-na-enda ɲumbani
I am going home.
```

```
Kwa ndi- hata ha--m-a--na-li-ona wa-ndege-ku- mbili m- -a- kila -a kwanza -ni.
TOP FOC.POS NEG.CONC NEG-3PL.O-3SG-PFV-PST-see PL-bird-INFIX two CL AGR.CL DISTR.RED ORD =HOST
kwa ndi- hata ha--m-a--na-li-ona wa-ndeɡe-ku- mbili m- -a- kila -a kwanza -ni
She did not see the two birds.
```

```
Kwa ndi- je mwanamke-ku- sana mzee-le m- -a- -ni -m-a--na-li-pa mtoto-ku- m- -a- -ni kitabu-ku- m- -a- -ni.
TOP FOC.POS Q woman-INFIX DEG old-SUBST CL AGR.CL =HOST 3SG.O-3SG-PFV-PST-give child-INFIX CL AGR.CL =HOST book-INFIX CL AGR.CL =HOST
kwa ndi- mwanamke-ku- sana mzee-le m- -a- -ni -m-a--na-li-pa mtoto-ku- m- -a- -ni kitabu-ku- m- -a- -ni
Did the old woman give the child a book?
```

```
Ni na kwa ndi- kaka-ku- my -a -angu m- -a- -ni -m-a--na-na-be mkulima-ku- m- -a- -ni.
PRED.MIX POSS.WITH TOP FOC.POS brother-INFIX My POSS.HEAD POSS CL AGR.CL =HOST 3SG.O-3SG-PFV-PRS-be farmer-INFIX CL AGR.CL =HOST
ni na kwa ndi- kaka-ku- -a -aŋu m- -a- -ni -m-a--na-na- mkulima-ku- m- -a- -ni
My brother is a farmer.
```

```
Kwa ndi- mwanamume-ku- m- -a- -ni -m-a--na-li-ona me-ku- m- -a- -ni a--na-li-ondoka.
TOP FOC.POS man-INFIX CL AGR.CL =HOST 3SG.O-3SG-PFV-PST-see me-INFIX CL AGR.CL =HOST 3SG-PFV-PST-leave
kwa ndi- mwanamume-ku- m- -a- -ni -m-a--na-li-ona -ku- m- -a- -ni a--na-li-ondoka
The man who saw me left.
```

```
Kwa ndi- ni--na-na-jua kwamba a--na-li-ondoka.
TOP FOC.POS 1SG-PFV-PRS-know COMP 3SG-PFV-PST-leave
kwa ndi- ni--na-na- kwamba a--na-li-ondoka
I know that she left.
```

## 12. Lexicon

27 words.

| Word | IPA | Part of speech | Means |
|---|---|---|---|
| kubwa | /kubwa/ | adjective | big |
| ndege | /ndeɡe/ | noun | bird |
| kitabu | /kitabu/ | noun | book |
| kaka | /kaka/ | noun | brother |
| paka | /paka/ | noun | cat |
| mtoto | /mtoto/ | noun | child |
| mkulima | /mkulima/ | noun | farmer |
| pa | /pa/ | verb | give |
| enda | /enda/ | verb | go |
| nyasi | /ɲasi/ | noun | grass |
| kijani |  | adjective | green |
| nyumbani | /ɲumbani/ | noun | home |
| mimi | /mimi/ | pronoun | I |
| katika | /katika/ | adposition | in |
| katika | /katika/ | adposition | into |
| ruka | /ruka/ | verb | jump |
| jua |  | verb | know |
| ondoka | /ondoka/ | verb | leave |
| mwanamume | /mwanamume/ | noun | man |
| si | /si/ | adverb | not |
| mzee | /mzee/ | adjective | old |
| haraka | /haraka/ | adverb | quick |
| ona | /ona/ | verb | see |
| yeye | /jeje/ | pronoun | she |
| kwa | /kwa/ | adposition | to |
| mbili | /mbili/ | determiner | two |
| mwanamke | /mwanamke/ | noun | woman |

## Still to do

- 6 grammatical markers have no form yet, so they are not expressed.
- The writing system has no glyphs.