Luganda


The Ganda language or Luganda is a Bantu language spoken in the African Great Lakes region. It is one of the major languages in Uganda, spoken by more than eight million Baganda and other people principally in central Uganda, including the capital Kampala of Uganda. It belongs to the Bantu branch of the Niger–Congo language family. Typologically, it is a highly-agglutinating, tonal language with subject–verb–object, word order and nominative–accusative morphosyntactic alignment.
With about six million first-language speakers in the Buganda region and a million others fluent elsewhere, it is the most widely spoken Ugandan language. As a second language, it follows English and precedes Swahili.
Luganda is used in some primary schools in Buganda as pupils begin to learn English, the primary official language of Uganda. Until the 1960s, Luganda was also the official language of instruction in primary schools in Eastern Uganda.

Phonology

A notable feature of Luganda phonology is its geminate consonants and distinctions between long and short vowels. Speakers generally consider consonantal gemination and vowel lengthening to be two manifestations of the same effect, which they call simply "doubling" or "stressing".
Luganda is also a tonal language; the change in the pitch of a syllable can change the meaning of a word. For example, the word kabaka means 'king' if all three syllables are given the same pitch. If the first syllable is high then the meaning changes to 'the little one catches'. This feature makes Luganda a difficult language for speakers of non-tonal languages to learn. A non-native speaker has to learn the variations of pitch by prolonged listening.
Unlike some other Bantu languages, there is no tendency in Luganda for penultimate vowels to become long; in fact they are very frequently short, as in the city name Kampala Kámpalâ, pronounced , in which the second vowel is short in Luganda.

Vowels

All five vowels have two forms: long and short. The distinction is phonemic but can occur only in certain positions. After two consonants, the latter being a semivowel, all vowels are long. The quality of a vowel is not affected by its length.
Long vowels in Luganda are very long, more than twice the length of a short vowel. A vowel before a prenasalised consonant, as in Bugáńda 'Buganda' is also lengthened, although it is not as long as a long vowel; laboratory measurements show that the vowel + nasal takes the same length of time to say as a long vowel. Before a geminate, all vowels are short. A segment such as tugg, where a short vowel is followed by a geminate consonant, is very slightly shorter than tuuk or tung.

Consonants

The table below gives the consonant set of Luganda, grouping voiceless and voiced consonants together in a cell where appropriate, in that order.
LabialAlveolarPalatalVelar
Plosive
Fricative
Nasal
Approximant
Trill

Apart from, all these consonants can be geminated, even at the start of a word: bbiri 'two', kitto 'cold'. The approximants and are geminated as and : eggwanga 'country'; jjenje 'cricket'—from the roots -wanga and -yenje respectively, with the singular noun prefix e- that doubles the following consonant.
Historically, geminated consonants appear to have arisen when a very close between two consonants dropped out; for example -dduka from *-okuduka 'run'.
Apart from, and, all consonants can also be prenasalised. This consonant will be,, or according to the place of articulation of the consonant which follows, and belongs to the same syllable as that consonant.
The liquid becomes when geminated or prenasalised. For example, ndaba 'I see' ; enddagala 'leaf'.
A consonant cannot be both geminated and prenasalised. When morphological processes require this, the gemination is dropped and the syllable is inserted, which can then be prenasalised. For example, when the prefix en- is added to the adjective -ddugavu 'black' the result is enzirugavu.
The nasals,, and can be syllabic at the start of a word: nkima 'monkey', mpa 'I give', nnyinyonnyola or 'I explain'. Note that this last example can be analysed in two ways, reflecting the fact that there is no distinction between prenasalisation and gemination when applied to nasal stops.

Tone

Luganda is a tonal language, with three tones: high, low and falling. There are, however, no syllables in Luganda with rising tone , since these automatically become .
According to one analysis, tones are carried on morae. In Luganda, a short vowel has one mora and a long vowel has two morae. A geminate or prenasalised consonant has one mora. A consonant + semivowel also has one mora. A vowel followed by a prenasalised consonant has two morae including the one belonging to the prenasalised consonant. The initial vowel of words like ekitabo 'book' is considered to have one mora, even though such vowels are often pronounced long. No syllable can have more than two morae.
Falling tones can be heard in syllables which have two morae, e.g. those with a long vowel, those with a short vowel followed by a geminate consonant, those with a vowel followed by a prenasalised consonant, and those following a consonant plus semivowel. They can also be heard on final vowels, e.g. ensî 'country'.
Words in Luganda commonly belong to one of three patterns : toneless, e.g. ekitabo 'book'; with one high tone, e.g. ekibúga 'city'; with two high tones, e.g. Kámpalá which link together to make HHH, i.e. or .
Although words like ekitabo are theoretically toneless, they are generally subject to a tone-raising rule whereby all but the first mora acquire a high tone. Thus ekitabo 'book' is pronounced and ssomero 'school' is pronounced . These tones added to toneless words are called 'phrasal tones'. The tone-raising rule also applies to the toneless syllables at the end of words like eddwâliro 'hospital' and túgenda 'we are going', provided that there is at least one low-toned mora after the lexical tone. When this happens, the high tones which follow the low tone are slightly lower than the one which precedes it.
However, there are certain contexts, such as when a toneless word is used as the subject of a sentence or before a numeral, when this tone-raising rule does not apply: Masindi kibúga 'Masindi is a city'; ebitabo kkúmi 'ten books'.
In a sentence, the lexical tones tend to fall gradually in a series of steps from high to low. For example, in the sentence kye kibúga ekikúlu mu Ugáńda 'it is the chief city in Uganda', the lexical high tones of the syllables bú, kú and gá stand out and gradually descend in pitch, the toneless syllables in between being lower. This phenomenon is called 'downdrift'.
However, there are certain types of phrase, notably those in the form 'noun + of + noun', or 'verb + location', where downdrift does not occur, and instead all the syllables in between the two lexical high tones link together into a 'plateau', in which all the vowels have tones of equal height, for example mu maséréngétá gá Úgáńda 'in the south of Uganda' or kírí mú Úgáńda 'it is in Uganda'. Plateauing also occurs within a word, as in Kámpálâ.
A plateau cannot be formed between a lexical tone and a phrasal tone; so in the sentence kíri mu Bunyóró 'it is in Bunyoro' there is downdrift, since the tones of Bunyóró are phrasal. But a phrasal tone can and frequently does form a plateau with a following high tone of either sort. So in abántú mú Úgáńda 'people in Uganda', there is a plateau from the phrasal tone of tú to the lexical tone of gá, and in túgendá mú lúgúúdó 'we are going into the street', there is a plateau from the phrasal tone of ndá to the phrasal tone of dó. Again there are certain exceptions; for example, there is no plateau before the words ono 'this' or bonnâ 'all': muntú onó 'this person', abántú bonnâ 'all the people'.
Prefixes sometimes change the tones in a word. For example, Baganda 'they are Baganda' has LHHL, but adding the initial vowel a- gives Abaganda 'Baganda people' with LLHLH. rather than with the length mark, to allow for tones to be written on each mora.)
Different verb tenses have different tonal patterns. The tones of verbs are made more complicated by the fact that some verbs have a high tone on the first syllable of the root, while others do not, and also by the fact that the sequence HH generally becomes HL by a rule called Meeussen's rule. Thus asóma means 'he reads', but when the toneless prefix a- 'he/she' is replaced by the high-toned prefix bá- 'they', instead of básóma it becomes básomá 'they read'. The tones of verbs in relative clauses and in negative sentences differ from those in ordinary positive sentences and the addition of an object-marker such as mu 'him' adds further complications.
In addition to lexical tones, phrasal tones, and the tonal patterns of tenses, there are also intonational tones in Luganda, for example, tones of questions. One rather unexpected phenomenon for English speakers is that if a yes-no question ends in a toneless word, instead of a rise, there is a sharp drop in pitch, e.g. lúnó lúgúúdò? 'is this a road?'.

Phonotactics

s can take any of the following forms:
where V = vowel, C = single consonant, G = geminate consonant, N = nasal stop, S = semivowel
These forms are subject to certain phonotactic restrictions:
The net effect of this is that all Luganda words follow the general pattern of alternating consonant clusters and vowels, beginning with either but always ending in a vowel:
where V = vowel, X = consonant cluster, = optional vowel
This is reflected in the syllabification rule that in writing, words are always hyphenated after a vowel. For example, Emmotoka yange ezze 'My car has arrived' would be split into syllables as E‧mmo‧to‧ka ya‧nge e‧zze.

Variant pronunciations

The palatal plosives and may be realised with some affrication — either as and or as postalveolars and respectively.
In speech, word-final vowels are often elided in these conditioning environments:
For example, ekiddugavu 'black' may be pronounced or. Similarly lwaki 'why' may be pronounced, or.
Long vowels before prenasalised fricatives may be nasalised, and the nasal is then often elided. Additionally, when not elided, the usually becomes a labiodental in,. For example:
The liquid has two allophones and, conditioned by the preceding vowel. It is usually realised as a tap or flap after a front unrounded vowel, and as a lateral approximant elsewhere. However, there is considerable variation in this, and using one allophone instead of the other causes no ambiguity. So lwaki 'why' may also be pronounced,, etc.

Alternative analysis

Treating the geminate and prenasalised consonants as separate phonemes yields the expanded consonant set below:
LabialAlveolarPalatalVelar
Simple plosive
Geminate plosive
Prenasalised plosive
Simple fricative
Geminate fricative
Prenasalised fricative
Simple nasal
Geminate nasal
Approximant
Liquid

This simplifies the phonotactic rules so that all syllables are of one of three forms:
where V = vowel, C = consonant, N = nasal stop, S = semivowel.
Vowel length is then only distinctive before simple consonants —not before geminate or nasalised consonants or at the end of a word.

Orthography

Luganda spelling, which has been standardized since 1947, uses a Latin alphabet, augmented with one new letter ŋ and a digraph ny, which is treated as a single letter. It has a very high sound-to-letter correspondence: one letter usually represents one sound and vice versa.
The distinction between simple and geminate consonants is always represented explicitly: simple consonants are written single, and geminates are written double. The distinction between long and short vowels is always made clear from the spelling but not always explicitly: short vowels are always written single; long vowels are written double only if their length cannot be inferred from the context. Stress and tones are not represented in the spelling.
The following phonemes are always represented with the same letter or combination of letters:
The following phonemes can be represented with two letters or combinations of letters, with the alternation predictable from the context:
The following phonemes can be represented with two letters or combinations of letters, with unpredictable alternation between the two:
It is therefore possible to predict the pronunciation of any word from the spelling. It is also usually possible to predict the spelling of a word from the pronunciation. The only words where this is not possible are those that include one of the affricate–vowel combinations discussed above.
Note, however, that some proper names are not spelled as they are pronounced. For example, Uganda is pronounced as though written Yuganda and Teso is pronounced Tteeso.

Vowels

The five vowels in Luganda are spelt with the same letters as in many other languages :
As mentioned above, the distinction between long and short vowels is phonemic and is therefore represented in the alphabet. Long vowels are written as double and short vowels are written single. For example:
In certain contexts, phonotactic constraints mean that a vowel must be long, and in these cases it is not written double:
For example:
But
Vowels at the start or end of the word are not written double, even if they are long. The only exception to this is yee 'yes'.

Consonants

With the exception of ny, each consonant sound in Luganda corresponds to a single letter. The ny combination is treated as a single letter and therefore doesn't have any effect on vowel length.
The following letters are pronounced approximately as in English:
A few letters have unusual values:
The letters l and r represent the same sound in Luganda——but the orthography requires r after e or i, and l elsewhere:
There are also two letters whose pronunciation depends on the following letter:
Compare this to the pronunciation of c and g in many Romance languages. As in the Romance languages the 'softening letter' is not itself pronounced, although in Luganda it does have the effect of lengthening the following vowel.
Finally the sounds and are spelt n before another consonant with the same place of articulation rather than ny and ŋ:
The standard Luganda alphabet is composed of twenty-four letters:
Since the last consonant ŋ does not appear on standard typewriters or computer keyboards, it is often replaced by the combination ng'. In some non-standard orthographies, the apostrophe is not used, which can lead to confusion with the letter combination ng, which is different from ŋ.
In addition, the letter combination ny is treated as a unique consonant. When the letters n and y appear next to each other, they are written as nÿ, with the diaeresis mark to distinguish this combination from ny.
Other letters are not used in the alphabet, but are often used to write loanwords from other languages. Most such loanwords have standardised spellings consistent with Luganda orthography, but these spelling are not often used, particularly for English words.
The full alphabet, including both standard Luganda letters and those used only for loanwords, is as follows:
Like most Bantu languages, Luganda's grammar can be said to be noun-centric, as most words in a sentence agree with a noun. Agreement is by gender and number and is indicated with prefixes attached to the start of word stems.
The following parts of speech agree with nouns in class and number:
NB: In the study of Bantu languages the term noun class is often used to refer to what is called gender in comparative linguistics and in the study of certain other languages. Hereafter, both terms may be used.
There is some disagreement as to how to count Luganda's noun classes. Some authorities count singular and plural forms as two separate noun classes, but others treat the singular-plural pairs as genders. By the former method, there are 17 classes, and by the latter there are 10 since there are two pairs of classes with identical plurals and one class with no singular-plural distinction. The latter method is consistent with the study of non-Bantu languages. Applying the method to Luganda gives ten noun classes, nine of which have separate singular and plural forms. This is the usual way to discuss Luganda but not when discussing Bantu languages, generally. In addition, Luganda has four locative classes, e, ku, mu, and wa.
The following table shows how the ten traditional classes of Luganda map onto the Proto-Bantu noun classes:
Luganda ClassNumberProto-Bantu Class
I Singular1, 1a
I Plural2
II Singular3
II Plural4
III Singular9
III Plural10
IV Singular7
IV Plural8
V Singular5
V Plural6
VI Singular12
VI Plural14
VII Singular11
VII Plural10
VIII Singular20
VIII Plural22
IX Singular15
IX Plural6
X 13

As the table shows, Proto-Bantu's polyplural classes are treated as separate in this article.
As is the case with most languages, the distribution of nouns among the classes is essentially arbitrary, but there are some loose patterns:
The class that a noun belongs to can usually be determined by its prefix:
There are a few cases where prefixes overlap: the singulars of Classes I and II ; the singular of Class III and plurals of Classes III and VII ; and the plurals of Classes V and IX. Genuine ambiguity, however, is rare, since even where the noun prefixes are the same, the other prefixes are often different. For example, there can be no confusion between omuntu 'person' and omuntu 'seat' in the sentences Omuntu ali wano 'The person is here' and Omuntu guli wano 'The seat is here' because the verb prefixes a- and gu- are different, even if the noun prefixes are the same. The same is true with the singular and plural of Class III: Embwa erya 'The dog is eating' vs Embwa zirya 'The dogs are eating'.
In fact, the plurals of Classes III and VII, and those of Classes V and IX, are identical in all their prefixes.
Class V uses its noun prefixes a little differently from the other classes. The singular noun prefix, eri-, is often reduced to e- with an accompanying doubling of the stem's initial consonant. This happens when the stem begins with a single plosive, or a single nasal stop followed by a long vowel, a nasal stop and then a plosive. For example:
Other stems use the full prefix:
There are also some nouns that have no prefix. Their genders must simply be learnt by rote:
Adjectives, verbs, certain adverbs, the possessive and a few special forms of conjunctions are inflected to agree with nouns in Luganda.

Nouns

Nouns are inflected for number and state.
Number is indicated by replacing the singular prefix with the plural prefix. For example, omusajja 'man', abasajja 'men'; ekisanirizo 'comb', ebisanirizo 'combs'. All word classes agree with nouns in number and class.
State is similar to case but applies to verbs and other parts of speech as well as nouns, pronouns and adjectives. There are two states in Luganda, which may be called the base state and the topic state. The base state is unmarked and the topic state is indicated by the presence of the initial vowel.
The topic state is used for nouns in the following conditions:
The base state is used for the following conditions:
As in other Niger–Congo languages, adjectives must agree in gender and number with the noun they qualify. For example:
In these examples the adjective -lungi changes its prefix according to the gender and number of the noun it is qualifying. In some cases the prefix causes the initial l of the stem to change to n or r.
Attributive adjectives agree in state with the noun they qualify, but predicative adjectives never take the initial vowel. Similarly, the subject relative is formed by adding the initial vowel to the verb.

Adverbs

True adverbs in the grammatical sense are far rarer in Luganda than in, say, English, being mostly translated by other parts of speech—for example adjectives or particles.
When the adverb is qualifying a verb, it is usually translated by an adjective, which then agrees with the subject of the verb. For example:
Here, 'badly' is translated with the adjective -bi 'bad, ugly', which is declined to agree with the subject.
Other concepts can be translated by invariant particles. for example the intensifying particle nnyo is attached to an adjective or verb to mean 'very', 'a lot'. For example: Lukwago anywa nnyo 'Lukwago drinks a lot'.
There are also two groups of true adverb in Luganda, both of which agree with the verbal subject or qualified noun, but which are inflected differently. The first group is conjugated in the same way as verbs and contains only a few words: tya 'how', ti 'like this', tyo 'like that':
The adverb ti 'like this' is conjugated as a verb to agree with the subject of the sentence in gender, number and person.
The second group takes a different set of prefixes, based on the pronouns. Adverbs in this group include -nna 'all', -kka 'only', -mbi, -mbiriri 'both' and -nsatule 'all three':
Note how, in the last two examples, the adverb -kka agrees with whichever antecedent it is qualifying — either the implicit nze 'I' or the explicit emmotoka 'the car'.
Note also, in the first two examples, how the placement of nzekka before or after the verb makes the difference between 'only' and 'alone'.

Possessive

The possessive in Luganda is indicated with a different particle for each singular and plural noun class. An alternative way of thinking about the Luganda possessive is as a single word whose initial consonant cluster is altered to agree with the possessed noun in class and number.
Depending on the possessed noun, the possessive takes one of the following forms:
If the possessor is a personal pronoun, the separate possessive form is not used. Instead, the following personal possessives are used:
There are also a few nouns that take special forms when used with a possessive:

Subjects

As in other Bantu languages, every verb must also agree with its subject in gender and number. For example:
Here, the verb nywa changes its prefix according to the gender and number of its subject.
Note, in the third and fourth examples, how the verb agrees with the number of the noun even when the noun doesn't explicitly reflect the number distinction.
The subject prefixes for the personal pronouns are:
For impersonal pronouns the subject prefixes are:
When the verb governs one or more objects, there is also an agreement between the object prefixes and the gender and number of their antecedents:
As with the subject prefix, the third person prefixes also agree with their antecedents in person. The personal object prefixes are:
For the impersonal third person the object prefixes are:
Note the similarity between each subject prefix and the corresponding object prefix: they are the same in all cases except Class I and the singular of Class III. Note also the correspondence between the object prefixes and the noun prefixes : when every m- in the noun prefix is replaced by a g- in the object prefix, the only differences are in Classes I and III.
The direct object prefix is usually inserted directly after the subject prefix:
The indirect object prefix comes after the direct object:
The negative is usually formed by prefixing te- or t- to the subject prefix, or, in the case of the first person singular, replacing the prefix with si-. This results in the following set of personal subject prefixes:
The negative impersonal subject prefixes are:
When used with object relatives or the narrative tense, the negative is formed with the prefix ta-, which is inserted after the subject and object affixes:
To form some tenses, a special form of the verb stem, called the 'modified form', is used. This is formed by making various changes to the final syllable of the stem, usually involving either changing the final syllable to one of the following suffixes:
The modified form of verb stems is the only real source of irregularity in Luganda's verbal system. Monosyllabic verbs, in particular, have unpredictable modified forms:
in Luganda is explicitly marked on the verb, as it is in most other Bantu languages.
Present tense
The present tense is formed by simply adding the subject prefixes to the stem. The negative is formed in the same way but with the negative subject prefixes.
InflectionGlossNegativeGloss
nkola'I do'sikola'I don't do'
okola'you do'tokola'you don't do'
akola'he, she does'takola'he, she doesn't do'
tukola'we do'tetukola'we don't do'
mukola'you do'temukola'you don't do'
bakola'they do'tebakola'they don't do'
gukola'it does'tegukola'it doesn't do'
bikola'they do'tebikola'they don't do'
zikola'they do'tezikola'they don't do'

The present perfect is just the subject prefix plus the modified stem:
The present perfect in Luganda is sometimes slightly weaker in its past meaning than in English. It is often used with intransitive verbs with the sense of being in the state of having done something. For example, baze azze means 'my husband has arrived' means 'I am in the process of sitting myself down'; to say 'I'm sitting down' in the usual sense of 'I'm seated' in standard English, a Muganda would use the present perfect: ntudde.
Past tenses
The near past is formed by inserting the prefix -a- before the modified form of the stem. This prefix, being a vowel, has the effect of changing the form of the subject prefixes:
The near past tense is used for events that have happened in the past 18 hours. The negative is formed in the usual way.
The far past is formed with the same prefix a- as the near past, but using the simple form of the stem:
The far past tense is used for events that happened more than 18 hours ago, and can also be used as a weak pluperfect. This is the tense that's used in novels and storytelling.
Future tenses
The near future is used when describing things that are going to happen within the next 18 hours. It is formed with the prefix naa- on the simple form of the stem:
In the second person singular and the singular of Class III, the prefix becomes noo- and nee- in harmony with the subject prefix.
The negative form of this tense is formed by changing the final -a of the stem to an -e and using vowel-lengthened negative subject prefixes; no tense prefix is used:
The far future is used for events that will take place more than 18 hours in the future. It is formed with the prefix li- on the simple form of the stem:
Note how the l of the tense prefix becomes a d after the n- of the first person singular subject prefix.
Other
The conditional mood is formed with the prefix andi- and the modified form of the stem:
The subjunctive is formed by changing the final -a of the stem to an -e:
The negative is formed either with the auxiliary verb lema plus the infinitive:
or using the same forms as the negative of the near future:
Luganda has some special tenses not found in many other languages. The 'still' tense is used to say that something is still happening. It is formed with the prefix kya-:
In the negative it means 'no longer':
With intransitive verbs, especially verbs of physical attitude, the kya- prefix can also be used with the modified verb stem to give a sense of 'still being in a state'. For example, nkyatudde means 'I'm still seated'.
The 'so far' tense is used when talking about what has happened so far, with the implication that more is to come. It is formed with the prefix aaka-:
This tense is found only in the affirmative.
The 'not yet' tense, on the other hand, is found only in the negative. It is used to talk about things that have not happened yet, and is formed with the prefix nna-:
When describing a series of events that happen sequentially, the narrative form is used for all but the first verb in the sentence. It’s formed by the particle ne followed by the present tense:
The narrative can be used with any tense, as long as the events it describes are in immediate sequence. The negative is formed with the prefix si- placed immediately after the object prefixes :
Compare this with the negative construction used with the object relatives.

Auxiliary verbs

Other tenses can be formed periphrastically, with the use of auxiliary verbs. Some of Luganda's auxiliary verbs can also be used as main verbs; some are always auxiliaries:
The meaning of a verb can be altered in an almost unlimited number of ways by means of modifications to the verb stem. There are only a handful of core derivational modifications, but these can be added to the verb stem in virtually any combination, resulting in hundreds of possible compound modifications.
The passive is produced by replacing the final -a with -wa or -ibwa/-ebwa:
The reflexive is created by adding the prefix e- to the verb stem :
Many verbs are used only in their reflexive form:
Reduplication is formed by doubling the stem, and generally adds the sense of repetition or intensity:
The applied, or prepositional, modification, allows the verb to take an extra object and gives it the meaning 'to do for or with '. It is formed with the suffix ir- inserted before the final -a of the verb:
Adding the applied suffix twice gives the 'augmentative applied' modification, which has an alternative applied sense, usually further removed from the original sense than the simple applied modification:
The causative is formed with various changes applied to the end of the verb, usually involving the final -a changing to -ya, -sa or -za. It gives a verb the sense of 'to cause to do', and can also make an intransitive verb transitive:
Applying two causative modifications results in the 'second causative':
The neuter modification, also known as the stative, is similar to the '-able' suffix in English, except that the result is a verb meaning 'to be x-able' rather than an adjective meaning 'x-able'. It is formed by inserting the suffix -ik/-ek before the verb's final -a:
The intransitive conversive modification reverses the meaning of an intransitive verb and leaves it intransitive, or reverses the meaning of a transitive verb and makes it intransitive, similar to English's 'un-' prefix. It is formed with the prefix uk- inserted before the verb's final -a:
The transitive conversive is similar to the intransitive conversive except that it results in a transitive verb. In other words, it reverses the meaning of an intransitive verb and makes it transitive, or reverses the meaning of a transitive verb and leaves it transitive. It is formed with the suffix ul-:
Two conversive suffixes create the augmentative conversive modification:
The reciprocal modification is formed with the suffix -na or -gana :
The progressive is formed with the suffix -nga. It is used with finite verbs to give the sense of continuousness:
This is not really a modification but a clitic, so it is always applied 'after' any grammatical inflexions.

Combinations of modifications

More than one modification can be made to a single stem:
There are some restrictions that apply to the combinations in which these modifications can be made. For example, the 'applied' modification can't be made to a causative stem; any causative modifications must first be removed, the applied modification made and the causative modifications then reapplied. And since the reflexive is formed with a prefix rather than a suffix, it is impossible to distinguish between, for example, reflexive causative and causative reflexive.

Numbers

The Luganda system of cardinal numbers is quite complicated. The numbers 'one' to 'five' are specialised numerical adjectives that agree with the noun they qualify. The words for 'six' to 'ten' are numerical nouns that don't agree with the qualified noun.
'Twenty' to 'fifty' are expressed as multiples of ten using the cardinal numbers for 'two' to 'five' with the plural of 'ten'. 'Sixty' to 'one hundred' are numerical nouns in their own right, derived from the same roots as the nouns for 'six' to 'ten' but with different class prefixes.
In a similar pattern, 'two hundred' to 'five hundred' are expressed as multiples of a hundred using the cardinal numbers with the plural of 'hundred'. Then 'six hundred' to 'one thousand' are nouns, again derived from the same roots as 'six' to 'ten'. The pattern repeats up to 'ten thousand', then standard nouns are used for 'ten thousand', 'one hundred thousand' and 'one million'.
The words used for this system are:
Numerical adjectives :
Numerical nouns:
Standard nouns:
Digits are specified from left to right, combined with na and mu. For example:
The numerical adjectives agree with the qualified noun:
but
and
The forms emu, bbiri, ssatu, nnya and ttaano are used when counting.
However, a complication arises from the agreement of numerical adjectives with the powers of ten. Since the words for 'ten', 'hundred', 'thousand' and so on belong to different classes, each power of ten can be inferred from the form of the adjective qualifying it, so the plural forms of the powers of ten are usually omitted, as long as this doesn't result in ambiguity.
For example:
Note that
amanda amakumi ana '40 batteries' cannot be shortened to amanda ana because this means "four batteries", and embwa amakumi ana '40 dogs' cannot be shortened to embwa ana because ana is the form of nnya used with embwa, so this actually means 'four dogs'!
Nkumi 'thousands' is also not usually omitted because the form the numerical adjectives take when qualifying it is the same as the counting form, so 3,000 will always be rendered nkumi ssatu.