It is not uncommon to hear people say (usually derogatorily) of a language ‘It doesn’t have any grammar.’ To appreciate the absurdity of this statement, it is helpful to specify what ‘grammar’ is. For linguists, the ‘grammar’ of a language is the set of rules which the speakers of the language follow when they speak. It encompasses rules about the possible forms of words (shplernk is not a possible word in English), rules about the way bits of words can be put together (you can’t make plurals in English by putting the -s first), rules about the way words are put together to make longer units (in English you have to say This is an interesting book, not A book interesting is this) and rules about the way meanings are encoded by speakers. For some non-linguists, ‘grammar’ refers only to the second and third of these types of rules. Even on that narrower definition, it is easy to demonstrate that all languages have grammar.
For argument’s sake, let us discuss the proposition ‘Spelitzian has no grammar.’ I shall demonstrate that this cannot be true by considering what Spelitzian would be like if it were true.
If Spelitzian had no grammar, it would be impossible to make a mistake when speaking Spelitzian. Saying that a sentence is wrong in Spelitzian is the same as saying that it breaks a rule or rules of Spelitzian. If Spelitzian has rules, then it has grammar.
If Spelitzian had no grammar, there could be no difference between nouns and verbs or other word classes. There could be no pronouns, because they – by definition – stand for nouns, not verbs, and thus imply a distinction between the classes. If it is possible to distinguish word classes in Spelitzian, Spelitzian has grammar. All known human languages distinguish at least nouns and verbs.
If Spelitzian had no grammar, there could be no rules for the placement of words in sentences. Every order would be possible in every case. To say ‘John said Pip hit the fence’ you could say the equivalent of ‘Pip say john hit fence,’ ‘Fence say john hit pip’ ‘Hit say fence pip john,’ ‘Say fence john pip hit,’ or any other of the 120 possible word orders! Each of these would, of course, also be able to mean ‘Pip said John hit the fence,’ and other things like ‘John, say “The fence hit Pip.” ‘ Clearly, the listener would not know what was intended. Such a ‘language’ would only allow people to communicate extremely simple messages, probably only one-word messages.
If Spelitzian had no grammar, it could not have prefixes or suffixes, for example. These would imply that Spelitzian had rules for forming words, and thus that it had grammar. (If English had no rules for the placement of such forms, then pots could equally well be replaced by spot, psot or post on any occasion. Imagine what would happen if you tried to say ‘Spot the pots on the post.’)
If Spelitzian had no grammar, it could not have ‘little words’ (particles) to mark grammatical functions. Suppose it had a conjunction like and. If it could not be fixed in position, you would not know what it joined, so it would be possible to say ‘Pip and Pat like John’ by the Spelitzian equivalent for ‘pip and john like pat’ or ‘pat pip john and like.’ Particles work because they occur in specified positions in relation to other words.
If Spelitzian had no grammar, it would not be possible to mark the differences between sentences by changing the ‘tune’ or intonation of the sentence. Thus it would not be possible to mark differences between statements, questions and commands, between Pip hit Pat, Pip hit Pat? (or Did Pip hit Pat?) and Pip, hit Pat. If Spelitzian used intonation to mark these changes, the intonation would have a grammatical function, and thus Spelitzian would have grammar. In all known human languages these differences are conveyed by some combination of word order, modification of word forms (e.g. adding a suffix), function-marking particles and intonation.
To sum up, if Spelitzian distinguishes different word classes it has grammar. If Spelitzian has rules about word order, it has grammar. If Spelitzian has rules about the addition of prefixes, suffixes, etc., it has grammar. If Spelitzian has particles which ‘go with’ particular types of words (such as prepositions like to, in, by), it has grammar. If Spelitzian uses different ‘tunes’ which change the meaning of what the speaker of Spelitzian says, it has grammar.
If Spelitzian had none of these, then when two speakers of this ‘language’ were talking, the listener would not know what the speaker intended. At best the listener would guess. With such an imprecise system, language would be of very little use. Such a system resembles what we know of very simple animal communication systems. No human languages we know behave like this. All allow the precise communication of complex messages, and this requires grammar.
Next I will consider whether it is true that some languages don’t have very much grammar, or that some languages have more grammar than others. Latin is often taken by non-linguists as the ‘standard’ against which other languages are measured. Is it possible for Spelitzian to have less grammar than Latin?
There are several classes of nouns in Latin, and each class has a special pattern of endings, different in singular and plural, which mark the functions of nouns in sentences. I will illustrate with the first two classes (or declensions, as they are usually called). Puella ‘girl’ is 1st declension (feminine), and dominus, ‘lord’ is 2nd declension (masculine):
| 1st Declension | Singular | Plural |
| Nominative | puell-a | puell-ae |
| Vocative | puell-a | puell-ae |
| Accusative | puell-am | puell-ās |
| Genitive | puell-ae | puell-ārum |
| Dative | puell-ae | puell-īs |
| Ablative | puell-ā | puell-īs |
| 2nd Declension | Singular | Plural |
| Nominative | domin-us | domin-ī |
| Vocative | domin-e | domin-ī |
| Accusative | domin-um | domin-ōs |
| Genitive | domin-ī | domin-ōrum |
| Dative | domin-ō | domin-īs |
| Ablative | domin-ō | domin-īs |
In basic terms, the nominative is used for the subject of a sentence (the person or thing that performs the action), the vocative is used to mark the person or thing addressed, the accusative is used for the direct object of a sentence (the person or thing which is affected by the action), the genitive is used for possessors of things, the dative is used for the goal of an action (to phrases) and the ablative is used for by, with and from phrases. Thus to say ‘The girl saw the lord’ in Latin, girl is translated with the form puella, the nominative, and lord is translated with dominum, the accusative. To say ‘O girls, bow to your lord!’, girls would be translated with the vocative plural form puellae, lord would be translated with the dative singular domino, and your would require the genitive form of the pronoun.
Latin verbs also have patterns of endings which mark person, number and tenses, as in the following list of endings for one type of verb (first conjugation) in present and perfect tenses:
| Present | Perfect | |||
| Singular | Plural | Singular | Plural | |
| 1st (I, we) | am-ō | amā-mus | amā-vī | amā-vimus |
| 2nd (you) | amā-s | amā-tis | amā-visti | amā-vistis |
| 3rd (he/she) | ama-t | ama-nt | amā-vit | amā-vērunt |
Because Latin words carry on their ends markers which show their function, Latin has relatively free word order. Thus ‘The girl saw the lord’ can be translated by any of the following: Puella vidit dominum, Vidit puella dominum, Dominum vidit puella, Puella dominum vidit, Vidit dominum puella, Dominum puella vidit. All possible orders are in theory acceptable, although some are much more usual than others. (However, it should be noted that Latin word order is not entirely free: such forms as conjunctions and some adverbs have specific places in relation to the sentence as a whole or to other word classes.)
The way Latin works has led to the perception that ‘grammar’ means sets of endings. The fact that Latin also has rules for word order is often ignored. You cannot jumble completely freely the words in the Latin sentence equivalent to ‘John said the girl saw the lord,’ although there are several possible orders.
Many languages do not have sets of endings on nouns and verbs like Latin. Modern English has very few in comparison with Latin (and in comparison with Old English). Chinese has none at all. However, we cannot conclude that English has less grammar than Latin and Chinese none at all. English has replaced a Latin-like system of endings on nouns and free word order with a system of few endings and fixed word order. The fixed word order has the effect of marking the function of words just as clearly as the use of endings. Consider (a) The girl protected the lord and (b) The lord protected the girl The performer of the action always comes before the verb protected, and the recipient of the action always comes after the verb. In (a) we know that the girl performed the action with just as much certainty as in the equivalent Latin sentence. Thus fixed word order in English does the same job as the marking of nominative and accusative on nouns in Latin. Chinese also uses word order to mark these functions.
Yet other languages use particles to mark these functions. Maori has a particle i, which occurs before the recipient of the action and contrasts with no particle before the performer of the action. Thus in Maori ‘the girl’ is te kōtiro, te ariki is ‘the lord’ and ‘protected’ is ka tiaki, so Ka tiaki te kōtiro i te ariki is ‘The girl protected the lord’ and Ka tiaki te ariki i te kōtiro is ‘The lord protected the girl.’ While it is usual for the actor to precede the recipient, as in these sentences, it is possible under some circumstances to reverse the order, as in Ka tiaki i te kōtiro te ariki, which still means ‘The lord protected the girl’ and cannot mean ‘The girl protected the lord.’ Thus particles are another grammatical device parallel to endings and word order for marking such grammatical functions.
Since all three means are equivalent, a language which uses word order or particles has a grammatical system equivalent to one which uses endings. Latin, which makes extensive use of endings (or inflections), Chinese, which uses word order, Maori, which uses particles and relatively fixed word order, and English, which uses fixed word order and some inflections all have equivalent grammatical systems in this regard. Once we accept that languages can make similar distinctions using different sorts of grammatical devices, it becomes clear that it is very difficult to quantify how much ‘grammar’ a language has, and thus statements like ‘Spelitzian has hardly any grammar’ or ‘Latin has more grammar than Spelitzian’ cannot readily be supported. You have to consider the kinds of grammatical distinctions a language makes, not how it makes them.
While all known human languages mark basic distinctions such as the differences between the actor and the recipient of an action or between statements, questions and commands, there are still many other distinctions which are marked in some languages, but not in others. It is not true that Latin (or any other language) marks all the possible grammatical distinctions, and other languages mark only a subset of them. Here I can only hint at some of the wealth of differences in what languages mark.
English marks a difference between ‘definite’ noun phrases and ‘indefinite’ noun phrases by (amongst other things) the use of a and the. Latin does not mark this distinction at all. Maori makes a distinction somewhat similar to that in English, also through articles roughly comparable to a and the, while Chinese makes a somewhat similar distinction through word order. In this field, then, Latin makes fewest grammatical distinctions.
There are American-Indian languages which distinguish between things which happened recently in the past and things which happened a long time ago, and between things which speakers know from their own experience and things which they’ve been told. So in these languages, I would require a different form of the verb lose to say in 1996 ‘England lost to Germany in the semi-finals’ and to say in 1996 ‘The English lost the battle of Hastings in 1066.’ And both of these (which I only know because I have been told so) would require a different verb from’ ’I lost my game of Scrabble this afternoon/ which I know from my own experience.
Some languages, like Maori, have different pronouns for we when it means ‘you the listener and me the speaker’ and when it means ‘me the speaker and someone else, like my mother’. And Fijian, in addition, has not just singular and plural pronouns, but singular, dual (for two people), trial (for three people) and plural pronouns. European languages have relatively simple pronoun systems, although they distinguish gender in the third person, which Maori and Fijian do not.
Examples like these show that language A may have more complex systems than language Β in one area and less complex systems in other areas. We cannot sensibly quantify the amount of grammar a language has. All languages have immensely complex grammatical systems.
Sometimes when people assert that a language has no grammar, what they really mean is that there is no grammar book for that particular language. But the rules of a language exist in the heads of speakers of that language. We know the rules are there because of the way the speakers behave. They use similar structures for similar events. If you hear the Spelitzian sentence for ‘Give me some water,’ you can be pretty sure you can use the same pattern for ‘Give me some food,’ even if there are other possible patterns as well. If there weren’t any patterns, people wouldn’t be able to communicate because they would have no way of knowing what other people meant to say. You can understand what I say only because you know the same rules I do. That is what it means to speak the same language. Speakers can tell you whether a particular string of words is an acceptable sentence of their language. You, as a speaker of English, know that ‘Spelitzian isn’t a real language’ is a possible sentence of English, even though I feel confident that you have never met it before. And you know, equally well, that ‘Spelitzian real language isn’t’ is not a possible sentence of English. You do not need a grammar book to tell you this. Your own internal grammar tells you.
A grammar book of a language contains rules which mirror the rules speakers use when they speak that language. You can work out the rules which must be in the heads of speakers of Spelitzian by observing what they say. Suppose the following were sentences of Spelitzian:
Mashak Spelitziask op Pat
‘The Spelitzian saw Pat.’
Mashak Pat op Speltiziask
‘Pat saw the Spelitzian.’
Trakak Spelitziask op Pat
‘The Spelitzian greeted Pat’
Trakak Pat op Spelitziask
‘Pat greeted the Spelitzian.’
On the basis of these sentences, we might deduce the following patterns: the verb comes first, followed by the performer, followed by the recipient. The recipient always has op in front of it. We could start to write a grammar book for Spelitzian by writing down these rules, although they would almost certainly need some refinement as we accounted for other sentences. Every time we noticed a new pattern which Spelitzians follow, that would tell us about a new rule of Spelitzian, and we could add it to the grammar book. That is how grammar books are written. Thus the existence of a grammar book is irrelevant to the question of whether the language has grammar. A grammar book can be written for any language, because every language has grammar.
If Spelitzian is a language, it has a highly complex grammatical system, involving some combination of devices like word order, inflections, particles and intonation. A language without any grammar is a contradiction in terms.