1. Grammar
What is 'grammar'?
The English word grammar derives ultimately from the Greekfgrämmž meaning a 'letter'. In classical Greek and Latin the word grammatica referred to thegenerai study of Iitej^Iiii^e_ajidJailguage^,When the word grammar came into Englisrtln the medieval period, it was used to refer only to Latin grammar; and it was not until the seventeenth century that the term took on a more general meaning and so had to be prefaced by 'Latin', 'English', 'French', etc.
Two meanings have competed with each other in English since the seventeenth century. In 1605 Francis Bacon wrote: ''Cmice^ning_sjDeech^^^ sideration of them hath produced the science of Grammar". While in 1637 Ben Jo~nsoTwrTtesTrrG^^                          true and well speaking a Language". For
Bacon, grammar is a science, a study of a set of phenomena; but for Jonson, grammar is an art, the skill or technique of speaking well. It is Jonson's definition, extended to include writing, that has predominated until recent times. Typical is the quote from an 1824 "English Grammar" by L Murray: "English grammar is the art of speaking and writing the English Language with propriety". Popularly, this is probably still what is understood by 'grammar'; but for linguists and students of language, it is Bacon's definition that is preferred.
'Grammar' vs 'a grammar'
Even having distinguished the 'science' from the 'art' definitions of 'grammar', the word is still ambiguous. On the one hand, the term 'grammar' can refer to certain features of a language, for example the "speech and words" of the Bacon quotation. That is to say that we may speak of the 'grammar of a language', being those characteristics of a language which we denote by the word 'grammar', rather than, say, the phonetic or semantic characteristics. On the other hand, the term 'grammar', or morejecuxately 'a grammar', is used to refer to a description either of the ljmguagé~~ása whole} (sounds, structures and meanings) or of the grammatical featureTüfXTänguage. A grammar in this sense is^Jheji^-a, linguist's or grammarian's representation of those traits of a languagej^esignatedyby 'grammar' in the first sense.
Description vs prescription
Modern linguists, following Bacon's kind of definition of 'grammar', would claim to be describing the grammar of a language when they write a grammar. They would emphasise the word 'describe', in order to distinguish clearly what they are doing from the 'prescriptive' attitudes implied in the definitions of Jonson and Murray.
1
2          Discovering grammar
Descriptive linguists claim to be giving an account of what people actually do when they speak and write a language, rather than to be telling people what they ought to do when they speak and write their language, as the prescriptive grammarians tended to do. In the event, although different basic attitudes prevail, the distinction is probably not so clearcut as the terms 'descriptive' and 'prescriptive' imply. To be sure /prescriptive grammarians incIudedjuieSuin their grammars, such as "You shoulcr-fterrend a sentence with a preposition"; but in so doing they still had to describe what a 'sentence' and a 'preposition' are. And a descriptive linguist producing a grammar of modern English, for example, has to make a choice ßf-wrtich English usage he is going to describe; and he would usually select the itandarai variety, perhaps even 'standard educated usage', and by so doing he wourd-bave indulged in an implicit prescription.
(Good' and f bad j grammar
There still remains a usage of the term fenmima)' that we have not mentioned, in instances such as: "She wrote me a nice letter, but her grammar was awful". Here the speaker is passing a judgement on someone's use of language^-HejifiYaluating a piece of English usage against some supposed norm, called ^gpj^dJEnglisi)' or 'ran££LEflgKsh\ The norm is usually what is called 's^ajid^nlJEnglish' or 'educated English usage'. In other words, it is a particularly prestigious variety of English, which has achieved normative status becausejt is^widely used in public life, and in its written form in mosf kinds "čTjxíblic communication (media, government, arts). Because it is a nationally used public variety of English does not automatically confer on it the epithet 'good', by which all alternative varieties are judged 'bad'. The use of double negation ("They didn't have none"), for example, is not intrinsically 'bad grammar'; it merely represents a 'non-standardl-English usage. Linguists would prefer to replace the^evaluative terms 'good' and 'bad' with the more neutral terms 'appropriate' andonappropriate'. That is to say, double nega-
fion may not be appropriate in the EngTrsh^c^xiblic communication, but in more asual or intimate circumstances among certain groups of English speakers it may well constitute a normal feature of English grammar.
Clearly, the extension of this argument is that no language can be regarded as a homog^neous_entity. The term, 'the English language' subsumes a whole variety of 'Englishes', differentiated regionally and socially, one of which is the prestigious 'standard' variety, which is the variety of public communication, of the education system, and that is taught to foreign learners. When we talk of the grammar of a language, therefore, we need to specify which the variety of the language is whose grammar we are describing. Even then, we have to allow for variatioruwithm^a variety; for example, in the previous sentence I used the relative pronouhwhose i refer to the non-human noun language, which not all users of thejstandard' variety of English would do.
The scope of grammar
In his definition of the "science of Grammar", Bacon described the concerns of grammar as "speech and words". Certainly, words fall into the province of gram-
Grammar          3
mar but it is not clear what Bacon means by/speech,'. Although the term 'grammar' was at one time used to denote the study bUlLaápects of a language, linguists generally restrict the scope of the term these days. The linguistic study of a Ian-          ^J
guageJ_ox3iara£ivof a language, is^#ten said to comprise three components: rA*   ^ pRongcJp^onoWy, g^mar, sfaantics. The three aspects of linguistic study^^V , are, riowřvěTTňot independent of eäcrrother.                                                           a^
Phonetics and phonology are both concerned with the pronunciation of language, how language sounds, the transmission of utterances through the medium of sound. The province of phonetics is the general study of speech sounds, their articulation,, acoustic qualities and range; while phonology concentrates on the use particular languages make of the available human speech sounds. Phonology .deals, for example, with the nu^r and types of distinctive speech sounds employedb^a lanffual^lhlm7]^^^                                                       the accompanying
felífíiíelofltirělslr^^                             where phonology intersects with gram-
mar is in the consideration of the way in which grammatical categories are manifested in speech sound: note, for example, the different pronunciations of the English past tense ending (-ed) in the verbs walked, filled, loaded.
Semantics is the study of meaning. In a way, nearly all of language study is concerned with how language means. It is arguable, indeed, that there is no study of meaning separate from the study of grammar, phonology and the like. And some linguists would replace semantics with a branch of linguistics called lexicology, which comprises the study of words, their meanings and their interrelationships. Semantics is often conceived as being wider in scope than lexicology, concerned not only with the meaning of wojd^JmLaisiLwiilLth.e^eaning of sentences and with meaning relations between words and between sentences^
Language is sometimes viewed as the means by which meanings are transmitted in sound via the organising principle of grammar. Grammar is thus concerned with the counters of meaning—words, parts of words, grammatical categories—and their combination into meaningful strings—sentences, texts—prior to vocalisation through the sound system of a language. The study of grammar is often subdivided into syntax and morphology, the former dealing with the structure of sentences, and the latter with the structure of words.
Syntax
The English word^yÄjderives from a similar Greek word meaning 'arrangement' and came to refer spTfclfically to the arrangement of words. Syntax, then, is about the ways in which words may combine with one another, with the arrangements or patterns or orders of words that are possible in a language, and with the differences in meaning that different orders of words may brhigjibouL
Let us illustrate these points from English. The string "Every good boy deserves favours" is a possible arrangement of words in English. The following are not permissible ordenngs (Note: '*' is used to mark an ungrammatical or unacceptable
string,'?' to mark a doubtful string):
4          Discovering grammar
* "Deserves favours every good boy"
* "Good every boy favours deserves" *"Boy good every deserves favours"
etc,
while "Favours deserves every good boy" could possibly be acceptable in an appropriate context, though this is perhaps doubtful. The distinction between possible and imposs.ble, acceptable and unacceptable, is not always cle^rcuA native speakers of a language may vary in their judgements about the acce^bnitv of an //    arrangement. (See further, Chapter23.)                                        r\X^-J——-
ŕ /       Consider now the string "Harry is tickling Susan". This is an acceptable arrangement m English, as is the following using the same words: "Susan is tickling Harry  . But the two sentences do not mean the same: in the first, the person per-/ forming the action of tickling is Harry, while Susan is the one undergoing the ^      action; in the second, their roles are reversed. Clearly, then, an alteration in the ^            arrangement of words may bring about a change in meaning. In our investigations
syntax  therefore, we need to be able to identify elements hTs^nto^s that are sTucíře        alSO determine th-^2^^et4oi^he^^
Finally, look at the  following string:  "Very boy is because  being atrocious naughty . This is obviously an unacceptable string in English, but it could not be t^A       made acceptable by a rearrangement of the words that it comprises. The problem yť              lies m the incompatibility of a word like very with a word like boy  of a word like
because with words like is or being, and of a word like atrocious with a word like naughty-, although the last oi these combinations could be made acceptable by »i convertingWa^j into atrocious^, giV]ng the sequence "atrociously naughty"^ What these^xamf les illustrate is that you cannot combine a word with just any other word in a language. In part, this arises from the lexical or semantic facts of a language; eg we talk about "naughty boys", but we do not usually talk about naughty buildings". But the incompatibilities we have been discussing do not arise particularly from the incongruities of meaning produced; rather from the grammatical, or more accurately, s^actjcJaci^iLüieJaQgua^ To put what we have been saying into the more precise linguistic terms we shall be using later intensifying adverbs (like very) do not go together with nnnn« flffc-fc-j.) conjunctions (hje brause) do notjoTogether withverbsjlike^ and being), adjectives (like atrocious) do not ^to^tjierjyjthother adjectiv^slukě^jgSryj Part of the task of a syntactic description is, then, to "specify the possible combinations of words (more accurately, word classes-^g noun, adverb, conjunction, adjective) in a language.                                                                                                           '
Morphology               •
The word morphology derives from two Greek words which together meanJihe Study-aUorms'. In botany, morphology^ the study of plant forms. In linguistics. mo^hoIogyjlÄincsni^jvitfoE^Äát^ structure oTwords; that is to sa\T not_ withthesognd^ letters {ox^^^^^^^o^m^^^^^— but witrüjönsütuerlif^rwoHsl^ich have a grammatical function or meaning~For
ixt« i
ft     ^      íl                                                                              Grammar         5
/    /     // example, the/English word redefinitions may be analysed into four constituents:
re-define-tion-s. Each of these constituents occurs as a constituent in the structure of other words in the language, with the same function or meaning (eg define, re-inspect-{t)ion-s). As in syntax, so in morphology, the permissible combinations of constituents are limited: the study of morphology must, therefore, identify the possible types of word constituent in a language, investigate the patterns and arrangements of the constituents in words, and determine the function(s) or meaning^) realised by each constituent.
The study of the morphology of a language comprises two parts. On the one hand, morphology deals with the realisation of grjyrimatic-aLcategories, systems, or mean-ings-(such as number—singular/plural, tense—present/past) by means of inflections. The(ís_at the end of the English examples in the previous paragraph would be considered arrijíH©co&n, realising the category of plural number in English. Sometimes an infleelion^will realise monTthan~öne grammatical category; for example, the -s' of boys' as in "the boys' bicycles" realises both plural number and possessive case, and it is pronounced in exactly the same way as -s realising only plural number in boys and -'s realising only possessive case in boy's.
On the other hand, morphology deals with the combining ofjword constituents as the means, nf dgrisdngjieat wqtHs  One of the prinp^a^a^sin which a language
may add new words to its vocabulary is to mató neologisms /Fom existing word constituents by making combinations that have n^v^r--fjegn~made_before in the-. language. For example, the word redefinition is (wášTmade by adding the consti-ttrenT'-ŕíC»« to the constituent (in fact a potentially independent word) define, and the constituent re- to the combination definition; or alternatively by adding re- to define, and then -tion to redefine. It is not clear in this case which order of combination might be regarded as more plausible. Sometimes, more than two constituents may be necessary in order to make a new combination^-for example, unparalleled in English is a combination of un-parallel-ed. \Unparallel is not a possibTe^aTrange-ment in English, and ?paralleled is probably^t-besťTnarginal; so that we must regard unparalleled as a combination directly of three constituents.
Exercise 1
Say which of the following strings are grammatically unacceptable sentences of English, and explain why they are unacceptable:
1.  All the birds have been migrating to warmer climates.
2.  The dog big is intimidating the puppy small.
3.  The baby eats not his dinner.
4.  The violent wardrobe flew voraciously into the singing washing machine.
5.  Every beautifully dancer receiving many into bouquet.
6.  Some unhelpful decisionsmay not have been taken by wise men.
Exercise 2
Make as many acceptable English words as you can from the following word constituents:
1.   re-                                    5. -ment
2.  -ness                                 6. large
3.  state                                 7. -ly
4.  en-                                   8. treat
2, Word classes

ŕ/ y
'Vjt^í:
Classification
From our earliest years we are taught to sort objects into groups on the basis of their similarityJn_one_vv£Xili^^                                 seems' almostlo be a human
orj^nisaBünaTurge to group similar things together in all kinds of areas of lifeTThis is no less true in scientific study, where classification of phenomena forms an important part of description. In the study of language, then, we shall be grouping linguistic items into classes or types or categories on the basis of appropriate criteria ofsimilari
One set of Jjnguistic items that has traditionally been the subject of classification is that of/words; In older grammars, word classes are often referred to as 'parts of speech^anďíhe consideration of them forms the basis and main part of the grammar's content. In grammars such as these, deriving from a tradition based on Latin and Greekížlght parts of speecTpare recognised: Noun, Adjective, Pronoun, Verb, Adverb, PrepôsTtTon;T5ň junction, Interjection.
There are two problems with such a classification. One is that a classification worked out for one language may not be appropriate for another language; for instance, it is not clear where the English articles (a, the) fit into this Latin and Greek derived classification, since the classical languages did not have articles. The other problem concerns the criteria for the classification: on what basis^ssuhe, words of a languag.e_gEQ,upecj together into classes? One principle"which emerges from this is that each language must be cónsideredon-řtstíwnleřms.
Criteria of classification
/T\
21.
One basis of classification that is applied uses semantic or^notional'triteria, beginning from the question: What kinds ofjnqtions or concepts do the"words refer to? It was this kind of criterioTTthat gave rise to the traditional definitions of nouns as 'names of persons, places or things' and of verbs as 'doing words'. The problem with such definitions, even if they could be refined a great deal, is that they tend to remain vague and difficult to apply in any rigorous way. Such definitions are, however, useful as general characterising devices,..and can serve as 'rules of thumb' in the recognition of words in an unfamiliar language.
t
A second basis of classification, that of_ morphology, is more easy to apply in a
rigorous way, sinee- it focuses on the form or structure of words; and this is not
subject to the^vagaries of intuitive or introspective judgements. A classification of
words in Engľkij_baséd on morphological criteria would, for example, note that one
clj^£J3L^Y°rillIüöUßs)^n^
(girls, g/r/^jvhUd another class (verbs) inflects for past tense, present and past
Word classes          7
participle (showed, showing, shown). Additionally, it might be noted that words containing the constituents -tion, -ness, -ment, -ance belong to the noun class {celebration, keenness, retirement, remembrance). However, derivational criteria of this kind would apply to only a small part of the membership of a class. Indeed, inflectional criteria are not necessarily universal in their application and are therefore not absolutely reliable: for example, nouns like mud or information in English do not form a plural, and the verbs must and ought to do not form a past tense or participles. Morphological criteria are not, moreover, universally applicable in the sense that some languages do not have words composed of more than one constituent, or alternatively do not have inflections regularly applying to all or most members of a class. The morphological basis of classification is, therefore, not totally reliable, though more reliable than the notional basis, and more reliably applied.
Exercise 3
In the following English data group the words into classes on the basis of their morphological characteristics:
1.  big
2.  bigger
3.  biggest
4.  read
5.  reads
6.  reader
7.  sell
8.  sells
9.  seller
10.  fast
11.  faster
12.  fastest
13.  grow
14.  grows
15.  grower
16.  tall
17.   taller
18.  tallest
Exercise 4
Do the same for the following Isthmus Zapotec (Mexico) data (SIL 1980: Dl):
1.	taburete	'chair'
2.	/taburetebe	'his chair'
3.	Jtaburetedu	'our chair'
4.	pan	'bread'
5.	Jpanbe	'his bread'
6.	fpandu	'our bread'
7.	rukaadu	'we write'
8.	rukaabe	'he writes'
9	zukaabe	'he will write
10.	bikaabe	'he wrote'
11.	kukaabe	'he is writing1
12.	ru30oftedu	'we run'
13.	ru30oňebe	'he runs'
14.	zu30oňebe	'he will run'
15.	bi300fiebe	■■   'heran'
16.	ku30oňebe	'he is running
Syntactic criteria
A third basis for classification uses syntactic or 'functional' criteria. Words are grouped into classes according to their operation in syntactic structure: this covers
8        Discovering grammar
both position in relation to other words and function within linguistic units. For example, we could establish a class of words in English which fitted into the frame
"the ------- man",  and  it  would  include  items  like good,  bad,  big,  small,
fat, thin, etc. Generalising, we would establish the class of words that could follow a definite article (the) and precede a noun. It would correspond to the traditional class of adjectives, but it would be established on the basis of a clear and more-or-less absolute criterion. In English, the same set of words would also be found in the
frame   "'Noun'   is  _____"   (eg   "the   man   is   good/bad/big/small,   etc").   In
fact, this is not entirely correct, since there is a small set of adjectives in English that can function only in the pre-noun position (eg mere in "a mere boy"), and a small set that functions only after verbs like be (eg asleep in "the dog is asleep"). Apart from these few exceptions, the class of adjectives in English can be defined as the set of words that functions attributively (or as a modifier) before nouns and as a complement after be and similar verbs.
Consider the following Portuguese data (SIL 1980: D3)
1.   A  meninapequenachorapouco         The little girl cries a little' the girl       little       cries a little
2.   O  meninopequenochora                  'The little boy cries' the boy      little       cries
3.  A  porcapequenacomepouco            The little sow eats a little' the sow    little       eats   a little
4.   O  porcopequenocomemuito            The little pig eats a lot' the pig     little       eats   a lot
5.   O  porco gordo come mais                  The fat pig eats more' the pig     fat     eats   more
6.  A  meninagorda chora muito              The fat girl cries a lot' the girl       fat     cries a lot
From this data it is possible to establish five word classes in Portuguese, on the basis of syntactic function:
1.  A class of articles (a, o) functioning attributively before the noun.
2.  A class of nouns functioning as head of the phrase functioning as subject of the sentence (menina, menino, porca, porco).
3.   A class of adjectives functioning attributively after the noun (pequena, pequeno, gorda, gordo).
4.   A class of verbs functioning as nucleus of the sentence (chora, come).
5.   A class of adverbs functioning as attributive after the verb (pouco, muito, mais).
How many word classes?
The number of word classes that one identifies for a language will depend on the nature of thegrammar of the language itself, as well as on the judgements of the linguistic analyst. Some languages are thought to have as few as three word classes As we have already noted, a traditional 'notional' classification for English recog-
Word classes         9
J
nises eight. A more rigorous 'functional' classification might also identify eight —Noun, Verb, Adjective, Adverb, Pronoun, Determiner, Preposition, Conjunction—with various subclasses. It would seem that all languages probably have a class approximating to that of Nouns and one approximating to that of Verbs in European languages, corresponding notionaUy to the fnlrip^öf experience onlhe one hand, and to the 'events' of experience on the other.               *4UAf?/#ior ^^ ci^řĚÍ;
If we establish word classes in a language on a syntactic, functional basis, it will 0fte7r~birffie"ča^riIfäTl^                                   oflhFdasš^r ovMäppíngľ'For
example-,-the-'words' cut, bottle, net ~Jeel,'^                                                    and
the Verb clä^sTnTingfisliľ In such a case, where we are dealing with a small set of overlapping members, we can say either that these words have multiple class membership, or that there are two 'words' cut, etc, in English, one a Noun and the other a Verb. You will find that dictionaries will sometimes deal with such a word under one headword and sometimes make twcffieadwordsjfeig cut1 N, cut2 V). If, on the other hand, two functionally establišHěTcIalsSš are completely (or almost / completely) overlaprMnj*J^^ furicfioIřísTT^šlsTnVcase in English with the Adjective word'class, which functions both attributively before Nouns and predicatively after 'copula' verbs (eg "the tall tree"—"the tree is tall"). Qnty a small number of adjectives is restricted in each case to attributive or predicative position (eg "a mere youth", "the baby is asleep").
Exercise 5
Set up word classes for the following Apinaye (Brazil) data (SIL 1980: D2):
1.   kukrěkokoi                                         'The monkey eats' eats    monkey
2.   kukrěkra                                             'The child eats' eats    child
3.  ape     kra                                             The child works' works child
4.   kukrěkokoi    ratf                                'The big monkey eats' eats    monkey big
5.  ape     kra   met/                                   'The good child works' works child good
6.  ape     mctf kra                                     'The child works well' works well child
7.  ape    rat J mi   met J                              'The good man works a lot' works a lot man good
8.   kukrératf kokoi    punui                       'The had monkey eats a lot' eats    a lot monkey bad
9.  ape    punui mi   pirjetj"                         'The old man works badly' works badly man old
10. ape       pinetj     mi                              'The man works a long time'
works a long time man
Types of word class
It is often the case that some word classes in a language have a more readily extendable membership than others. The class membership is, in the first place, large and     V(t *
10/      Discovering grammar / .
, ,ť.p^°nfrnj±?]l;^ucj^^                                                                       ones faj| out of
V ^ f      fasRion or ojherwiseje
European languages include táouns, Verbs,.....Adjectives, and (Adverbs^ As new
substances are invented or disWéredSréw thoughts and idealTčIévdoped, new .sensations or qualities or ways of doing things and talking about things come into ' being, so new words are added to these word classes in the languages concerned.
These word classes are called/openj or 'lexical' word classes: the term 'open' refers to the type of membership, wMLfUhe term 'lexical' refers to the typical function of words in these classes. It is the members of these classes that carry the main burden of lexical meaning, referring to entities outside of language, in the structure of sentences. For example, note the open class words underlined in the following sentence:
"The motorist came round the corner and has knocked the child over".
\       It is predominantly the lexical words "that survive in newspaper headlines and in !\J      telegrams.                                          _  \              -"         ■-----—-—■—-------—------
The other type of word classjsjhe fclosed'/or igrammatical^ora^ass..Examp-les of these wprd-eiasses int^^r^opeán-kfřguages are(^onoun,(6eterminef^Prev positionfConjunction^lembership of these word classeSisTTot reärfHy-expandable and chkngeroTuyslo^ overjime; moreover, the membership is limited and can easily be enumerated. For example, the subclass of Personal Pronouns contains the following members in modern English: /, me, we, us, you, he, him, she, her, it, they, them.
The 'grammatical' label for these word classes refers to their mainly intra-linguistic function; they constitute the cement, holding the lexical building bricks together. Determiners, like a, the or this, have a function within a text; similarly many pronouns, like he or they. It would be incorrect, however, to assert that members of closed classes cannot refer at all; for example, prepositions like into, below or along clearly refer to specific kinds of describable spatial relationships. What we have here, as so often in language, is a continuum of more-or-less rather than the discrete classifications that scientists tend to make: at the 'most lexical' end of the continuum come nouns and verbs (though not all verbs), at the 'most grammatical' (ie 'least lexical') end come determiners and some pronouns; adjectives and adverbs are more lexical than grammatical, while prepositions and conjunctions are more grammatical than lexical. Only by examining the functions of the members of each word class can this categorisation be made for a particular language.    «      lv-

3. Noun
Reference
Nouns refer toothings), in the broadest sense of that term. When we choose to refer to some entitylryTneans ofräTnoun,. we choose to view it as a 'thing'. Some entities can be referred to only by \tmxitftree, window, van.XU other instances a choice can be made. Compare the following: "The bombers destroyed the village"—"the destruction of the village by the bombers". In the first, destroy is a verb and the action is viewed as an 'event'; in the second, destruction is a noun and the action is viewed as a 'thing', about which something further may then be predicated: "The destruction of the village by the bombers was unprovoked".
The kinds of 'things' that .nouns-refer to are varied. If a noun refers to a 'unique' thing, it is said to be a;Proper Noun, such as personal names (Wojfgang Amadeus Mozart) or names of places or institutions (Berlin, Home Office); other nouns are 'common' nouns, which refer to classes of things or to instances of things. A traditional classification of common nouns is made into 'concrete' nouns and 'abstract' nouns. 'Concrete' nouns refer to tangible, observable things, that can be defined ostensively (ie by pointing to them and saying "That is a ..."). In some dictionaries the definition of a concrete noun includes a photograph or a line drawing; such nouns are, arguably, best defined in this way. 'Abstract' nouns, on the other hand, cannot be defined ostensively or by means of illustrations; they refer to unobserv-able things like imagination or truth or feminism. They can be defined either by giving instances or by comparing their meaning with that of other related words in the vocabulary.
A further traditional classification of nouns is into 'animate' and 'inanimate'. Animate nouns refer to living 'things', ie people, animals, birds, fish, insects, etc. Sometimes, within this group, the distinction between 'person' (ie human) and non-person is grammatically important; for example, it underlies the choice between who and which as relative pronouns in English. Inanimate nouns refer to non-living things and thus include all the abstract nouns; alternatively, one might regard the animate/inanimate distinction as being a further subclassification of concrete nouns.
Gender
We turn now from semantic classifications of nouns to grammatical ones, which have been traditionally called 'gender', since in the European languages at least such classifications are based on natural gender or sex, ie the distinction between masculine, feminine and neuter. This is not the universal basis for a grammatical classification of nouns, however. In the Algonquian languages of North America,
11
12          Discovering grammar
for example, the gender classification is based on the animate/inanimate distinction. In Swahili there are six gender classes, the first containing mostly animate nouns, the second inanimate nouns, and the third plants. No gender system is completely consistent with its meaning base; there is always arbitrary assignment of words to genders from a semantic point of view, the overriding criterion has to be grammatical.
In German, for example, there are three genders: masculine, feminine and neuter. Virtually all nouns denoting male beings belong to the masculine gender, and those denoting female beings to the feminine gender, although Weib 'woman' and Mädchen 'girl' belong to the neuter gender. The neuter gender contains predominantly inanimate nouns, but these are also found in considerable numbers in the other two genders, eg Tisch 'table', Baum 'tree' are masculine, while Blume 'flower', Wahrheit 'truth' and many other abstract nouns are feminine.
There are no universal grammatical markers of gender; languages vary in the grammatical means used to signal the gender of a noun. Sometimes gender is marked in the noun itself, eg in Swahili by a set of prefixes. In French, nouns ending in -tion are usually feminine; in German, nouns ending in -e are usually feminine, as are abstract nouns in -heit or -keit. But such rules for the European languages do not account for a great number of words; here the markers of gender are outside the noun itself, usually in the forms of accompanying words like determiners or adjectives. In French, for example, the gender of a noun is identified by the form of the article accompanying it (un, une 'a'; le, la 'the' for masculine and feminine singular respectively), or by the form of a demonstrative determiner (ce, cette 'this'/'that'), or by the form of an accompanying adjective (eg "le garcon heureux", 'the happy boy'; "la fille heureuse", 'the happy girl').
A similar picture emerges for German. The singular definite article varies in form according to the gender of the noun: der masculine, die feminine, das neuter. There is also variation in the adjective, especially with the indefinite article: "ein grosser Mann" 'a big man', "eine grosse Frau" 'a big woman', "ein grosses Haus" 'a big house'. The pattern of agreement is, in fact, more complex than this, as there is a tendency to mark gender once only in any noun phrase.
Some languages do not make a gender classification of their nouns, for example Quechuan languages. It is arguable, moreover, that English does not exhibit grammatical gender: there is no mark of gender in the noun, apart from one or two endings that are restricted to nouns referring to females (eg tigress, usherette); and there is no mark in accompanying words (articles, adjectives, etc) either. The only argument for gender in English would be on the basis of the substitutability by a third person singular pronoun (ie he, she or it): this usage corresponds almost exactly with natural gender, except that she is sometimes extended in usage to refer to ships, cars, mountains, and there is variability of usage when referring to animals.
Exercise 6
Arrange the nouns in the following"Chontal of Tabasco (Mexico) data (SIL 1980: C7) into genders, indicating how each gender is marked:
Noun          13
1.	unk?e hun	'one paper'	13.	untu winik	'one man'
2.	ufk?e hun	'three papers'	14.	uftu winik	'three men'
3.	wsk?e hun	'six papers'	15.	w'tu winik	'six men'
4.	untu tsitam	'one pig'	16.	unk?e wah	'one tortilla'
5.	ujtu tsitam	'three pigs'	17.	ujk?e wah	'three tortillas'
6.	wstu tsitam	'six pigs'	18.	wsk3e wah	'six tortillas'
7.	unts^it ahin	'one alligator'	19.	untu mut	'one bird'
8.	ujts?it shin	'three alligators'	20.	uftu mut	'three birds'
9.	wsts'it ahin	'six alligators'	21.	watu mut	'six birds'
10.	unk'e jopo	'one leaf	22.	unts?it tsan	'one snake"
11.	u/k?e jopo	'three leaves'	23.	ujts'it tsan	'three snakes'
12.	wak?e jopo	'six leaves'	24.	wats'it tsan	'six snakes'
Cc	mutability				
Many nouns refer to objects that are countable. The noun may be preceded or followed by a numeral indicating how many of the objects are being referred to, eg six boxes. Very often, the fact that more than one is being referred to is also marked in the noun (boxes rather than box). Besides numerals, quantity may also be indicated by 'indefinite' quantifiers, which do not specify an exact number, eg several boxes, a lot of boxes, a few boxes. Nouns, whose quantity may be specified in these ways, are called 'countable' nouns.
Some nouns refer to things that may not be counted. For example, it is not possible in English to speak of *six waters. A plural form of the noun does not occur, and the quantifier some occurs before the singular form of the noun to indicate indefinite quantity: some water (compare some boxes—plural, but not *some box). Such nouns are called 'uncountable' or 'mass' nouns. With some mass nouns in English a numeral may occur, but the meaning is not 'plural' but 'kinds of, eg six wines, six cheeses.
Number
Number is a grammatical category applicable to nouns when they are countable. Such nouns are often marked in languages for 'plural' number. That is to say, we may distinguish 'singular' number (referring to one) from 'plural' number (referring to more than one). Singular nouns do not usually have a positive mark for number: they are said to be 'unmarked'; it is plural number that is usually marked in nouns. Not all languages mark their nouns for number, and some only mark number when the noun is not accompanied by any other indication of quantity in the form of a numeral, indefinite quantifier, etc. Some languages, in addition to plural number, also mark a separate 'dual' number, when a noun refers to two objects.
Number is usually marked in nouns by means of suffixes (endings) or prefixes (added to the front). In European languages suffixation is the most common method. For example, in English, the regularmark of a plural is the (written) suffix -(e)s, eg apples, apricots, peaches. It might be noted that the pronunciation of the plural suffix is different for each of these three words, a point that will be taken up in Chapter 8. It might also be noted that, although suffixation by -(e)s is the regular
14          Discovering grammar
way of marking plurai number in nouns in English, it is also marked in several other ways: compare, for example, man—men,mouse—mice, ox—oxen.
In German there are five different kinds of suffix to mark plural, three of which may in addition be accompanied by a change in the vowel of the noun (called 'umlaut', indicated in writing by '"' above the vowel concerned). The first suffix is -e, with a noun like Monat—Monate 'months', and with umlaut in Sohn—Söhne 'sons'. The second suffix is -{e)n, with a noun like Schmerz—Schmerzen 'pains' or Biene—Bienen 'bees'; this suffix does not occur with umlaut. The third suffix is a 'zero'; that is, there is no ending, the plural is the same as the singular, as in a noun like Daumen—Daumen 'thumbs', or just with umlaut as in Mantel—Mäntel 'coats'. The fourth suffix is -er, with a noun like Bild—Bilder 'pictures', and with umlaut in Rad—Räder 'wheels'. The fifth and least frequent plural suffix, which is restricted mainly to words borrowed from other languages, especially English and French, is -s, as in a noun like Park—Parks; it does not occur with umlaut.
Number and gender
In some languages the marking of number and gender intersects. This is true to a limited extent for German: the -e suffix tends to be associated with masculine nouns, the -{e)n suffix with feminine nouns, and the -er suffix with neuter nouns. In Latin the intersection is more systematic: for example, corvus—corvi 'crow/s' represents a masculine plural, while puella—puellae 'girl/s' represents a feminine plural, and bellum—bella 'war/s' a neuter plural.
Consider the following data (SIL 1980: C8) from Konkomba (Ghana):
1.	libil libaa	'one year'	7.	kid3uk kibaa	'one knife'
2.	bubib bugmu	'five years'	8.	nd3um mulee	'two knives'
3.	lilal libaa	'one axe'	9.	kipipeek kibaa	'one-place'
4.	bulab bulee	'two axes'	10.	npipeem mugmu	'five places'
5.	lid30ol libaa	'one hill'	11.	kigbalik kibaa	'one spear'
6.	bud30ob buniin	'eight hills'	12.	ngbalim muniin	'eight spears'
The nouns in this data are: -bi- 'year', -la- 'axe', -d^oo- 'hill', -d^u- 'knife', -pipee-'place', -gbali- 'spear'. The first three of these (in Nos 1-6) belong to one gender, as marked by one set of prefixes and suffixes; the remaining three belong to a second gender, being marked by another set of prefixes and suffixes. The numeral {-baa 'one', -lee 'two', -gmu 'five', -niin 'eight') is also marked (by prefixes) according to the gender of the noun. The gender markers are, however, different in singular and plural. The markers of number thus intersect with the markers of gender, so that any prefix/suffix marks at the same time both a number (singular or plural) and a gender (1 or 2), according to the following scheme:
Noun
Numeral
Gender 1	1      sing	ii-	-1	li-
eg -bi- 'year'	I       Pi	bu-	-b	fo u-
Gender 2	1     sing	ki-	-k	ki-
eg -d^u- 'knife'	}       Pi	n-	-m	mu-
Noun        15
Case
Case is a further grammatical category that is marked in the noun and additionally (or alternatively) in words that accompany nouns (eg articles, adjectives). The function of case is to signal the grammatical/semantic relationships between nouns in a larger syntactic structure (eg clause or sentence). Languages vary in the extent to which case marking is used to signal these relationships. For example, in German the article (not the noun) is sometimes marked to signal whether the noun is functioning as Subject or Object in a sentence (see Chapter 11 for discussion of 'Subject', 'Object' and similar terms). In the following German sentence: "Der Hund hat den Mann gebissen"—'The dog bit the man', the noun Hund 'dog' is marked as Subject of the sentence by the form of the definite article der 'the', while Mann 'man' is marked as Object by the form of the.definite article den. In English, on the other hand, the grammatical relationships 'Subject-of' and 'Object-of are signalled merely by the relative ordering of the elements in the sentence. So, "The man bit the dog" has a different Subject and Object from "The dog bit the man"; while "Den Mann hat der Hund gebissen" has the same Subject and Object as "Der Hund hat den Mann gebissen".
There is a great variation in the number and kinds of case marking found in different languages. Some languages (eg Chinese) do not mark case at all. In English case marking in the noun is limited to a 'possessive' (or 'genitive') case, marked in writing by -'s in the singular noun and by -s' in the plural noun; it links two nouns in a semantic relationship of 'belonging' or 'possession' (eg "the cow's tail", "the girl's name") or in some kind of largely grammatical relationship (eg "the world's end", "the enquiry's conclusions"). In the pronoun in English, however, some distinctions are made between subject and object forms, eg, I—me, she—her, (hey—them.
Four cases are distinguished in German, marked most consistently in the article: nominative, accusative, dative, genitive. The nominative and accusative are the Subject and Object cases respectively, mentioned earlier. The dative marks the Indirect Object in a sentence, eg seiner Frau '(to) his wife' in the following sentence: "Der Mann hat seiner Frau einen Brief geschickt", 'The man sent his wife a letter'. The genitive in German functions like the genitive/possessive in English. In addition, all the cases except the nominative are governed (see Chapter 16) by particular prepositions, eg für 'for' is followed by noun phrases in the accusative case, von 'by'/'from' by those in the dative, and während 'during' by those in the genitive. Latin has a further case, additional to the four found in German: the ablative, which usually translates into English by means of the prepositions by, with ox from.
In Punjabi a 'direct' and an 'oblique' case are distinguished. The oblique case is used before postpositions '(in common with other North-Indian languages, Punjabi has postpositions, coming after the noun referred to, rather than prepositions). The direct case is used for subjects and objects, although under certain circumstances these are realised by a noun in the oblique case followed by a particular postposition. In addition to the direct and oblique, Punjabi marks a 'vocative' case on nouns used as forms of address, an 'instrumental' case (in the plural only) to signal the
16          Discovering grammar
object 'with' which an action is performed, an 'ablative' case to signal movement 'from' a place, and a 'locative' case to signal 'ať a place.
In some languages, known as 'ergative' languages, the case of the Object of a transitive verb (for explanation of this term, see Chapter 10) is the same as that of the Subject of an intransitive verb. Such a language is Alawa, spoken in Australia. Among other cases, Alawa has a 'nominative', which is unmarked, and an 'operative', which is marked by a suffix. The nominative is used for the Subject of intransitive verbs and the Objects of transitive verbs; while the operative is used for the Subject of transitive verbs, as well as the instrument or location involved in an action. Consider the following examples:
lilmi            na-wutala gutaru                  The man came to a hill'
man (nom) he-went    hill (nom)
lilmi-ri    yarjka-natan-nadakiribu                  parakal-ta
man (op) hit he-did-it           kangaroo (nom) spear (op)
'The man killed the kangaroo with a spear'
Exercise 7
For the following Basari (Ghana) data (SIL 1980: C2), identify what the genders are and how they are marked for singular and plural in the noun and the demonstrative:
1.  uni umbini
2.   uboti umbini
3.   diyin dimbini
4.  dibil dimbini
5.   kusaau kumbini
6.   kukabuu kumbini
'this person' 'this chief 'this name' 'this seed' 'this farm' 'this basket'
biniib bimbini bibotiib bimbini ayin rjimbini abil rjimbini tisaati timbini tikabuti timbini
'these persons' 'these chiefs' 'these names' 'these seeds' 'these farms' 'these baskets'
4. Verb
Reference
Verbs refer to 'events'. Included under the heading of 'event' are: actions, eg spill in "The cat has spilled the milk"; 'process', eg spill in "The milk has spilled"; 'happenings', eg occur in "The accident occurred at lunch-time". One may also need to extend the notion of 'event' to include 'states', expressed in English by verbs like seem, look ("She seems/looks tired"), and above all be ("She is happy"). In the case of be in particular, however, two points might be noted. Firstly, many languages would not have the 'copula' verb be in sentences like "She is happy"; ie it would be equivalently "She happy" (cf Punjabi "ik patla" = 'One (is) thin'). Secondly, it has been argued that for languages like English one should regard the 'verb' in sentences like "She is happy" as 'be happy'. The effect of this suggestion would be to vastly increase the number of verbs in a language like English, by including the majority of adjectives, and possibly many nouns as well (cf "She is a doctor"). One may well decide that the implementation of such a suggestion would not result in a very economical description of the language, since one would still need, for example, to account for the attributive use of adjectives in English (eg "the happy girl").
From the examples above ("The cat has spilled the milk"—"The milk has spilled") we can see that there may exist a relationship between actions and processes. Indeed, the use of spill in the first of these examples is sometimes referred to as an 'action-process', to distinguish it from action verbs that do not imply a simultaneous process, such as laugh in "Everybody laughed", or kick in "The referee has kicked the ball into the goal". Processes do not necessarily p|esuppose some actionJn order to trigger them off fäll *Hať is implied Is söme^hange^taking place in the 'thing' undergoing the process, eg "The sky has clearecT^The leaves are turning yellow".
Tense
One piece of information that languages often encode with relation to events is the time when they take place, specifically either before the time of speaking, at the time of speaking, or after the time of speaking. These real-world time differences are encoded in language by means of the grammatical category of 'tense'. We thus often find verbs marked for 'past', 'present' and 'future' tenses, for example in French: "ilparlait" 'he talked', "il parle" 'he talks', "ilparlera" 'he will talk'.
In French, then, tense is marked by suffixes in the verb. Not all languages mark tense in this way; the English translations given for the French examples show that English marks present and past tense by means of suffixes (-s for present, and -ed
17
18           Discovering grammar
for past), but the future tense is marked periphrastically, by using another, 'auxiliary', verb. In fact, it is disputable whether a future tense can be clearly recognised in English, since it is possible to refer to future time in a number of ways, eg "he is going to talk", "he is to talk". Some languages do not mark tense in the verb by means of prefixes or suffixes at all, but by the use of auxiliary verbs or particles (see Chapter 9). In the Yurok language (North America), for example, tense is indicated by a series of particles placed before the verb, eg "ho s?egok" = 'past they-box', ie "they boxed". Other languages do not mark tense at all, but instead have an elaborate system of 'aspects' (see below); this seems to be the case in Classical Arabic.
We have suggested so far that tense, encoding real-world time, distinguishes past, present and future. These are not, however, the only time relations that may be encoded in language; they are time-relations viewed from the perspective of the time of speaking, the present. Some languages encode time relations that might be referred to as 'past-in-the-pasť and 'past-in-the-future', eg in English "He had finished his breakfast, when .. .", "He will have completed the report by Thursday". The first of these sets one event in a past previous to another past event, while the second views an event as being past from a future perspective. These tenses are sometimes referred to as 'pluperfect' and 'future perfect' respectively.
Additionally, in some languages we find that there is no single past tense form, for example. Such is the case in French, where there are three past tenses, called the 'imperfect', 'preterite' and 'perfect'. A distinction between preterite and perfect is maintained in written French: the perfect implies that a past event has relationship to the present (eg "Jean m'a écrit deux fois cette semaine"—'John has written to me twice this week'), while the preterite places an event unequivocally in the past (eg "Jean lui écrivit ä Noěl"—'John wrote to him at Christmas'). In spoken French, however, the perfect has taken over the functions of the preterite, so that the latter is now restricted to the written form of the language. The imperfect should perhaps be considered under aspect (see below) rather than tense: it indicates 'past with duration' (eg "II écrivait au moment oů le telephone sonna"—'He was writing when the telephone rang'), or 'past event repeated' (eg "Je lui écrivais chaque jour"—'I wrote to him every day'). But the imperfect in French is also used for past description ("Elle avait les cheuveux blonds"—'She had blond hair') and may be interchangeable with the preterite or perfect (eg "II décidait soudain de se lever"—'He suddenly decided to get up").
Tense and time
Our discussion of tense up to now has centred on its relationship with time. But tense, although it clearly relates to real-world time, is nevertheless independent of it, in the sense that on occasions tense distinctions may have nothing to do with time. This independence of tense and time may be illustrated by the fact that they may not be in a one-to-one relationship in a language. Consider the 'present' tense in English: it may refer to present time, eg "I see no ships". But it may also be timeless in its reference, eg "The sun rises in the east", "Oil floats on water". It may be used with future reference, eg "Father comes home tomorrow"; or with
Verb        19
reference to habitual events, eg "We go shopping on Fridays". And in a few instances in English a present tense verb may refer to a past event, eg "I hear that you've changed your job", when the hearing was presumably before the time of speaking. A further illustration of this lack of one-to-one correspondence between tense and time is the perfect tense in German: on the one hand it is used like the perfect in French; on the other it may have a future-perfect reference, eg "Bald hat er es geschafft"—'He will have soon done it'.
The independence of tense and time is also evident in the fact that tenses may be used for other than time reference. The only difference in English between "May I use your telephone?" (present) and "Might I use your telephone?" (past) is one of tentativeness or politeness; the might-imm is interpreted as a more polite request than the may-form. Similarly in conditional clauses in English, present tense implies a 'real' condition while past tense implies a 'hypothetical' condition, and both refer to future time: compare "If you visit the museum tomorrow . . .", "If you visited the museum tomorrow . . .". One might also point to the use of the present tense in past narratives (the so-calied 'historic present') in order to mark a particular episode as climactic or to bring it into particular focus for some reason; spo ken narratives in some styles of speaking are entirely in the present tense.
Tense and person
While dealing with tense we need to note that the marking of tense in the verb by means of suffixes (or prefixes) sometimes intersects with the marking of person/ number. The category of person is dealt with in detail in the next chapter; let us note for the moment the distinction between the speaker (1st person singular/ plural—I/we), the addressee (2nd person singular/plural—you for both in English-), and the thing/person talked about (3rd person singular/plural—he, she, it/they). In the past tense in English, there is no mark of person along with that of tense: -ed signals past tense for all persons/numbers ("I talked", "you talked", "he talked", "we talked", "they talked"). In the present tense, there is a mark for the 3rd person singular, but for none of the other person/numbers ("he talks" but "I talk", "you talk", etc).
In French, on the other hand, each tense is marked by a different set of person-/number suffixes, so that each suffix is at the same time a mark of tense and person-/number, with very little overlap between forms. Consider the following table of forms for the verb parier 'speak', 'talk':
Imperfect                 Present                     Future
1st sing ye 2nd sing tu 3rd sing il/elle 1st pi nous 2nd pi vous 3rd pi ilsjelles
The relative absence of overlap between forms applies primarily to the written form of the language; there is far more overlap in the spoken language, eg the 1st, 2nd
parlais	parle	parlerai
parlais	paries	parleras
párlait	parle	parlera
parlions	parlons	parlerons
parliez	parlez	parle rez
parlaient	parlent	parleront
20           Discovering grammar
and 3rd person singular and the 3rd person plural forms of the imperfect are identical in pronunciation, as are the same forms of the present. Spokejn French has diverged quite radically from the written form, which represents an oldeT~prt> nunclatkm;-------------________■•-------—■—---------------....................-......-............= -.............-
^Consider how the following data (SIL 1980: A3) from Mbembe (Eastern Nigeria):
1.   nwaogwo                      'the child drinks'
2.   nwa oci                          'the child eats'
3.   nwa oci eten                  'the child eats fish'
4.   aci eten                         'you eat fish'
5.  oci eten                         'he eats fish'
6.   nwa mogwo                   'the child will drink'
7.   maci eten                      'you will eat fish'
8.   kogwo                           'he does not drink'
9.  kagwo                           'you do not drink'.
By comparing Nos 1 and 2 we can conclude that nwa means 'child', and ogwo and oci 'drinks' and 'eats' respectively. No 3 adds the further vocabulary item eten 'fish'. Nos 4 and 5 show us that the person is marked in the verb by means of a prefix, o- for 3rd person singular, a- for 2nd person: the verbs 'drink' and 'eat' are in fact then -gwo, -ci. Nos 6 and 7 show us that future tense is also marked in the verb with a prefix, independently of person and prior to it in order: m-. We conclude that present tense is ««marked. Nos 8 and 9 show us additionally that the negative is marked in the verb by means of a prefix (k-), independently of person, though our data does not allow us to draw conclusions about its relationship to tense.
Exercise 8
Identify the constituents of the verbs in the following Congo Swahili data (SIL 1980: A2), and list them with their meanings:
1.  ninasem                        'I speak'
2.  wunasem                       'you speak'
3.  anasem                         'he speaks'
4.  wanasem                       'they speak'
5.  ninaon                          'I see'
6.  nilion                            'I saw'
7.  ninawaon                       'I see them'
8.  niliwuon                        'I saw you'
9.  ananion                         'he sees me'
10.  wutakanion                    'you will see me'
Now translate 'they saw him' into Congo Swahili.
Aspect
In addition to encoding in the verb the tíme at which an event takes place, many languages encode various ways of viewing the event, particularly from the point of view of the extension of the event in time. Ways of viewing the event are encoded in language by means of the grammatical category of 'aspect'. One aspect that is frequently found is the one we have already noted in the French imperfect: the
Verb         21
'durative', also termed 'continuous' and 'progressive'. The durative encodes a view of the event that regards it as having duration, continuing through a period of time, or being in progress. The English 'progressive' is an aspect with a similar meaning: compare "He slept all day" with "He was sleeping ali day". The first views the sleeping as a once-and-for-all past event, the second views it as an event that has duration: the event is the same, the way of viewing it differs.
Another aspect that is often marked in languages is the 'completive' or 'perfective', which views an event as completed or finished rather than not yet completed. The difference in meaning may be illustrated from English with the following examples: "I was reading a book last, nighty—"I ,r&ad.-a«book .lastnighť^The first, in the past progressive form," implies an uncompleted event, while the second, in the simple past form, implies a completed event—the book was finished. A perfective aspect is found commonly in Slavonic languages among others. For example, in Polish one finds a set of perfective verbs contrastive with an equivalent set of imperfective verbs, eg mrugnač (perfective) 'to blink once'—mrugac (imperfective) 'to blink/ several times', zabolet (perfective) 'to begin aching'—boleč (imperfective) 'tó ache', domówič (perfective) 'to finish speaking'—mówič (imperfective) 'to speak'.
As a counterpart to the continuous aspect, some languages positively mark £ 'punctiliar' aspect: the event is viewed as taking place at a point in time. This i? done in Alawa (Australia); compare the following forms: nawiňa'l will go' (future,\ no aspect marking), rjadiňa 'I can go' (present, punctiliar), rjadiři T can go'\ (present, continuous).
As with tense, aspect may be marked in the verb by means of affixes (as in Alawa); or periphrastically with auxiliary verbs, as in the English progressive, which is formed with be as an auxiliary followed by a present participle {is going, was talking); or by means of particles.
Consider the following Ekpeye (Eastern Nigeria) data (SIL 1980: A6); there is no marker for 'future':
1.   edi                                'he will eat'
2.   edikpo                          'he will finish eating'
3.   edile                             'he has eaten'
4.   adikpole                       'we have finished eating'
5.   edikpohwo                    'he will eventually finish eating'
6.   adigbale                        'we have eaten again'
7.   edikpohwole                 'he has eventually finished eating'
8.   emegba                        'he will make again'
9.   amekpogbale                'we have finished making again' 10.  amegbahwo                  'we will eventually make again'
From the English glosses it is clear that a number of aspects are marked in the verb in Ekpeye, translated by English 'have', 'finish', 'eventually', 'again'. They are marked by the suffixes -le, -kpo, -hwo, -gba respectively; and they occur in the order -kpo-gba-hwo-le. The aspects might be termed 'perfect' (-le), 'completive' (-kpo), 'culminative' {-hwo), 'iterative' {-gba). To complete the analysis, -di- 'eat' and -me- 'make' are the verbs, while e- 'he' and a- 'we' are person prefixes.
22         Discovering grammar
Exercise 9
Identify the constituents of the verb words in the following Inga (Colombia) data (SIL 1980: A7), and suggest terms for the aspects that are marked (Note: 3rd person singular is unmarked):
1. chayarca
2. chayacurca
3. chayarcani
4. chayacurcanimi
5. chayacumi
6.  chayamucunimi
7.  chayamurcachar
8.   chayamucurcanguichar
9.   puňucuchar
'he arrived'
'he was arriving'
'I arrived'
'I was definitely arriving'
'he is definitely arriving'
'I am definitely arriving here'
'he probably arrived here'
'you were probably arriving here'
'he is probably sleeping'
5. Pronoun
Types ofpronoun
The term 'pronoun' implies an item which stands for or instead of a noun. In general, pronouns can be used in grammar where nouns would normally function, as Subject of a sentence, Object of a sentence, and so on. The class of pronouns, however, is not homogeneous; there are several types of pronoun.
Perhaps the words most commonly thought of as pronouns are those represented by the English set l/me, you, he/him, shejher, it, we/us, they/them. These are the 'personal' pronouns. It is to be expected that a similar set of pronouns would be more or less universal, although not necessarily with all the distinctions found in the English set.
Allied to the personal pronouns one may also find a set of 'reflexive' pronouns, eg in English myself, yourself, himself, herself, itself, ourselves, yourselves, themselves. In English these have two uses: when the Object of a sentence is the same as the Subject (eg "she has hurt herself"); and as an emphatic pronoun (eg "I did it myself"). Many languages do not have such an elaborate system of reflexive or emphatic pronouns; one form may well serve for all persons, as in North-Indian languages. In German the reflexive pronoun is the same as the accusative case form of the personal pronoun for 1st and 2nd persons, with a special form for 3rd person (singular and plural) - sich, eg "Ich habe mich verletzt" T have hurt myself, "Sie hat sich verlatzt" 'she has hurt herself. The emphatic pronoun in German is an invariant form—selbst, eg "Ich habe es selbst gemacht" T did it myself.
Also related to the personal pronouns one may find a set of 'possessive' pronouns, eg in English mine, yours, his, hers, its, ours, theirs, used in sentences like "This coat is yours" (ie '... belongs to you'), or "Mine is the brown one" (ie '(the coat) belonging to me ...').
Other types of pronoun found in languages include: 'relative' pronouns, used to introduce relative clauses (see Chapter 13), eg who, which, whose, that in English; 'interrogative' pronouns, used in questions, eg who?, what? in English; 'indefinite' pronouns, used to refer to a non-specified person or thing, eg someone, anything, everybody in English; 'demonstrative' pronouns, with a distinguishing or pointing reference, this/these, that/those in English, eg in "That is a good idea", "This looks like the one we want".
Person
We noted in the previous chapter that probably all languages have a grammatical category of person, distinguishing a 1st, 2nd and 3rd person, referring to the speaker, the addressee, and the person/thing being talked about, respectively.
OG-B
23
24          Discovering grammar
Unless, unusually, the reference of a 3rd person pronoun is clear in the situational context, a 3rd person pronoun will have a textual function: it will be standing for a noun occurring previously in the text; eg "Once upon a time there was an emperor. He ruled the country harshly". The reference of 1st and 2nd person pronouns is clear in the situational context, ie the person talking and the person being talked to, respectively.
Many languages have more than one 2nd person pronoun; indeed, English appears to be unusual in having only one. In French, for example, we find tu and vous. Although vous is the plural counterpart to singular tu, it is also used to speak to a singular addressee. A distinction is made in terms of familiarity or politeness: tu is used when talking to family or friends, vous with strangers. But the distinction is also connected with relative social status: a social superior may speak to a social inferior (eg adult to child, master to servant) with tu, expecting to be addressed with vous in return. This use of 2nd person pronouns to signal social asymmetry has, however, been gradually breaking down in the increasing egaUtarianism of post-war society. Similarly in German: du, which has its own plural form Ihr, is the 'familiar' 2nd person pronoun, and Sie the 'polite' one, used for both singular and plural. A further level of politeness or deference may be added by using a 3rd person form of address; this may also be found in English, in deferential butler-talk, eg "My lord has requested the car. Would he like arug?"
In Javanese there are three 2nd person pronouns, whose use depends on whether one is addressing a superior, an equal, or an inferior in social status; although how many of these an individual may use will depend on his own social status, and in Javanese it is not only 2nd person pronouns that are marked for social status but other vocabulary items as well (eg there are three forms for 'now', two for 'rice', etc). Similarly, in Hindi and Urdu there are three 2nd person pronouns, one for addressing inferiors, one for addressing intimates, and one for expressing politeness or respect. In Bengali there are also 'ordinary' and 'honorific' forms of the 3rd person pronoun.
Person and number/gender/case
We have already mentioned that number intersects with person in the pronoun system; that is to say, pronouns occur in singular and plural forms for each person. In the case of 2nd and 3rd person pronouns the reference is to a plurality of addressees or things under discussion, eg Ihr, they. But with 1st person pronouns we do not have a straightforward case of plurality; unless the speaker is a chorus, 1st person plural must refer to 1st person (/) plus others. Now 'others' could include either the addressee (you) or persons other than the addressee (they) or both together. Some languages encode these possible distinctions by having more than one 1st person plural pronoun: an 'exclusive' 1st person plural pronoun (ie / + they, but not you), and an 'inclusive' 1st person plural pronoun (ie 1 + you, and also possibly they). Alternatively, a distinction may be made, as in Alawa (see below), between I + you but not they, and / + you jthey.
Usually only in the 3rd person, person also intersects with gender, often reflecting the gender distinctions made in the noun class. In French, for example, a mas-
Pronoun        25
culine/feminine distinction is made in both singular and plural 3rd person pronouns: il/elle, ils/elles. In German the masculine/feminine/neuter gender distinction is found only in the 3rd person singular pronoun (er, sie, es), the plural pronoun being common for all genders (sie). A similar pattern is found in English, where the 3rd person singular pronoun is the only place in the language that a gender distinction is maintained (he, she, it). In Punjabi we find the reverse case: the masculine/feminine gender distinction of nouns is not replicated in the 3rd person pronoun.
A third intersection of grammatical categories may occur in the pronoun between person and case, affecting potentially all person/numbers. English makes a case distinction (subject/object) in the personal pronoun system, though no longer anywhere else in the language, eg i/me, she/her. In fact, the 'subject'/'object' labels represent a simplification: the Morms are restricted to subject position, though native speakers argue about whether one says "It is I" or "It's me" (cf the less likely "It's I", "It is me"); but the me-forms are used everywhere else in the structure of sentences, as object ("They saw me"), after a preposition ("Don't wait for me"), as an emphatic ("Who's coming?"—"Me!").
Pronoun systems
We will now diagram two personal pronoun systems, to show how the categories we have been discussing intersect to produce an array of pronoun forms.
German:
	Nominative	Accusative	Dative
sing 1	ich	mich	mir
2 familiar	du	dich	dir
3m	er	ihn	ihm
3f	sie	sie	ihr
3n	es	es	ihm
pi    1	wir	uns	uns
2 familiar	ihr	euch	euch
2 sing/pi polite	Sie	Sie	Ihnen
3	sie	sie	ihnen
It will be noted that the 2nd person 'polite' forms are identical in pronunciation with the 3rd person plural, although situational context will always disambiguate them. In writing, it is conventional to capitalize the initial letter of the 2nd person forms.
Alawa (Australia):
Direct
Indirect
sing 12		ňanu
1		nma
2		nagana
3m		nula
3f		nadula
non-sing	12	ňalu
ňaka napa ňaba nipa natu ňalarja
26         Discovering grammar
Alawa (Australia) (cont.)
Direct                             Indirect
dual 1                             rjařu                               nařarja
2                                wuřu                               wuřurja
3                                yiřula                              yiřuna pl 1                                 rjalu                                nalarja
2                                    wulu                               wulurja
3                                    yilula                              yilurja
In Alawa a case distinction is made between 'direct' and 'indirect': the direct pronoun is used for the Subject of most clauses and wherever a pronoun is placed at the beginning of a clause for emphasis; the indirect pronoun is used elsewhere in the clause and for possession. The usual three persons are distinguished, together with an additional 1st person, marked '12': in the singular, this person refers to the speaker plus one hearer, in the non-singular to the speaker plus two or more hearers. Three numbers are distinguished for 1st, 2nd and 3rd persons singular, dual and plural. In the 3rd person singular a gender distinction (masculine/feminine) operates. There are no overlaps among the twenty-four forms of the personal pronoun.
Bound and free pronouns
The pronouns that we have discussed so far in this chapter all have the status of free and independent words. In the previous chapter we noted that person/number is also frequently marked in the verb by means of a prefix or suffix, eg in French or Congo Swahili. However, these two languages differ in the way in which they mark person in the verb: in the case of French the category of person intersects with that of tense, so that a particular suffix signals both person and tense simultaneously, eg "je parle", "je parlais", "je parlerai"—where -e, -ais and -erai mark both 1st person and present, imperfect and future tenses respectively. In the case of Swahili (Exercise 8) the category of person is marked independently of the category of tense: ni-, wu-, a-, and wa- are person prefixes, while na-, li- and taka- are tense prefixes. When person is marked in this way in the verb, independently offense and aspect in particular (as categories that do not intersect with person in free pronouns), then we are justified in talking about a set of bound pronoun forms. Many languages exhibit parallel sets of free and bound pronouns.
Consider the system of personal pronouns in Bimoba (Ghana):
Emphatic            Non-emphatic       Bound            Bound
free                       free             subject            object
sing 1                              min                    n
2                                 fin                      a
3 animate                  ngoo                   u pl 1                                   timm                  te
2                                 yimm.                i
3 animate                  ngamm               be
sing/pl 3 inanimate           ngann                 le
m-/ma-	-n
f-/fa-	-a
w-/wu-	-u/-o
t-	-et
y-/yi-	-i
b-	-eb
1-	-er
Pronoun          27
Bimoba has two sets of free pronouns, one used in emphatic contexts, the other when the pronoun is not being emphasised. There are also two sets of bound pronouns, distinguished for case: a subject set, and an object set. The pronouns are distinguished for the usual three persons, and for singular and plural number. In the 3rd person a gender distinction is made between 'animate' and 'inanimate', and the 3rd person inanimate pronoun shows no distinction of number.
Exercise 10
In the following Zulu data (SIL 1980: Ell), identify the personal pronouns and suggest what functions the different pronouns have (Note: -ya- in the verb word is an aspect marker which appears in the present tense when no complement follows the verb):
1.   Umfanau-fana   ukudla boy        he-want food
2.   U-fana  ukudla he-want food
3.   uSipho yena u-fana   ukudla Sipho   he    he-want food
4.   Yena u-fana   ukudla he     he-want food
5.   Yena uSipho u-fana   ukudla he      Sipho   he-want food
6.   U-shayaimbongolo he-hit    donkey
7.   U-shayayona he-hit     it
8.   U-ya-yi-shaya he-   -it-hit
9.   U-shayayonaimbongolo he-hit     it      donkey
10.   U-ya-yi-shaya yona he-   -it-hit     it
11.   U-ya-yi-shaya imbongolo he-    -it-hit     donkey
12.   U-ya-yi-shaya yona imbongolo he-    -it-hit     it      donkey
The boy wants food' He wants food'
_____but Sipho (himself) wants food'
_____but he (himself) wants food'
.___ but he, Sipho (himself), wants food'
He hits a donkey'
He hits it'
He hits it'
He hits it, the donkey'
_____but he hits if
_____but he hits it, the donkey'
_____but he hits it, the donkey'
6. Word structure
Morpheme
In considering data in previous chapters we have on a number of occasions been involved in identifying constituents of words (eg in Exercises 7-9). The implication of this procedure is that words may have an identifiable structure and be composed of smaiier recurrent parts. There is a sense in which words are composed of sounds or letters, eg cat is composed of c + a + f; but such structure is not grammatical structure. We are concerned with the analysisof wordžinto recurrent parts with an identifiable constant -grammatical TülffcYion v^
are^ITeď^Jňorphemíeíi^Wé may, then, define a morpheme as "the minimal unit of "grTmmar", and impose the condition that a morpheme must have a discernible meaning or function, which is more or less constant for all its occurrences in the structure of the words of a language.
Morphemes can be identified by comparing the words of a language which have a similar form. Recurrent, sequencesJ^o|_soun,ds_or Jetters) -with -the.,s                     or
meaning are recognised as morphemes. Consider the following English examples:
refuse	dismiss	revive	refer
refus-al	dismiss-aí	reviv-al	refer-al
refus-al-s	dismiss-al-s	reviv-al-s	refer-al-s
A comparison of these forms enables us to identify the following morphemes: refuse, dismiss, revive, refer with their usual dictionary meanings; -al, which has the function of deriving a noun from a verb; -s, which signals the grammatical category 'plural number'.
Now consider the following data (SIL 1980: Al) from Jebero (Peru):
1.	nurjja	'a little canoe'
2.	nunjawak	'my little canoe'
3.	wilaľa	'a little child'
4.	tulapsn	'your leg'
5.	piĚteknan	'his house'
6.	piÖaklusa?	'houses'
7.	piöakparjlusa?	'your houses'
By comparing Nos 1 and 3, we can recognise morphemes nur;- meaning 'canoe', wila- 'child', and -fa 'little' or 'diminutive' (cf -let in English piglet, notelet). Returning to No 2, we conclude that -wsk is a morpheme meaning 'my', ie '1st person singular possessive'. Looking now at Nos 5 and 6, we can identify a morpheme piöak- meaning 'house', and conclude that -nay is a morpheme meaning 'his' or '3rd person singular possessive', and -lusa?is a morpheme meaning 'plural'.
28
Word structure        29
From No 7 we recognise a further morpheme -pay meaning 'your' or '2nd person possessive'. Returning finally to No 4 we conclude that it is composed of the morpheme tula- meaning 'leg' and the '2nd person possessive' morpheme -psy. A comparison of -nay and -pay might suggest a further subdivision into -dtj 'possessive' and -n/-p '3rd person'/'2nd person'; but this analysis is contradicted by -wak, which cannot be analysed in this way: on the basis of the data given we are not justified in proposing a further analysis of -n?y and -psy.
Summarising, the Jebero data contains the following morphemes:
Nouns: nuy- 'canoe',piöak- 'house', tula- 'leg', wila- 'child'.
Possessives:   -wdk '1st person singular',  -psy  '2nd person',  -my  '3rd person
singular'.
Number: -lusa?'plural'.
Derivative: -fa 'diminutive'.
Root and affix
If you look back at the examples we have discussed, it is evident that a word is composed of a central morpheme, to which other morphemes are added. Sem-antically, the central morpheme carries the main lexical reference of the worJjancl the o"ffier~morpEeměs modify that reference in various'ways. So, for example, the / English word refusals has refuse as its central morpheme, and the Jebero word / nuyfawak has nuy as its central morpheme. The central morpheme is called the 'root' of the word, and the peripheral morphemes That are äťtäčhéďto'theroot are    j
called 'affixes'.                                     '"....."■ "■i-"™'-™--~—-............-......- -.-, .,„..                       I
__...........................                                                                                                    I
Affixes are always 'bound' morphemes: they cannot occur alone as words, they  1 occur only in combination with other morphemes. Thus in the English examples, -al J and -s cannot be independent words, they are always bound to another morpheme. I You will notice that in writing these morphemes, as indeed in writing many of the | other morphemes we have identified in this and previous chapters, in fact all the f bound morphemes, we have written them with a hyphen on the side or sides where! they are bound. In the case of -al and -s, they are bound on the left, similarly all the; bound affixes in the Jebero data; in Exercise 8, however, the Congo Swahili verb-word has affix morphemes bound on the right.
Root morphemes may be bound or 'free', ie able to stand alone as an independent word. In the English data above, the roots refuse, dismiss, etc, are free mor- \t phemes, since they constitute words by themselves. MjOjs^oiits..in-E&glish are free.    / In the Jebero data it appears that the roots are bound"; there is no example where a \ root stands alone, but it is not inconceivable, for example, that nuy might consti- / tute an independent word" in Jebero, with the meaning 'canoe', similarly wila 'child', tula 'leg', piÖ3k 'house'. The limited data that we have does not allow us to definitely conclude this, although it could represent a reasonable hypothesis awaiting confirmation from further data. In the Congo Swahili data of Exercise 8, it seems fairly certain that the verb roots are bound, since they are preceded by at least a tense morpheme and a person morpheme functioning as subject.
30
Discovering grammar
Exercise 11
Identify the morphemes in the following Isthmus Zapotec (Mexico) data (SIL 1980: A5); list the morphemes with their meanings, and mark the bound morphemes with hyphen's as appropriate:
1. ňee	'foot'
2. kanee	'ťeeť
3, ňeebe	'his foot'
4. ňeelu3	'your foot'
5. kaneetu	'your(pl) feet
6. kaňeedu	'our feet'
?• 3igi	'chin*
8. ka3igi	'chins'
9. 3igibe	'his chin'
Types of affix	
10.  3igilu?
11.  ka3igitu
12.  ka3igidu
13.  3ike
14.  3ikebe
15.   ka3ikelu?
16.  diaga
17.  kadiagatu
18.  kadiagadu
'your chin' 'your(pl) chins' 'our chins' 'shoulder' 'his shoulder' 'your shoulders' 'ear'
'your(pl) ears' 'our ears'
In the examples we have been concerned with so far, we have come across two types of affix: those bound to the right (eg ka- 'plural' in the previous exercise); and those bound to the left (eg -be '3rd person singular masculine possessive' in the previous exercise). An affix which is bound to the right is called a 'prefix', and an affix which is bound to the left is called a 'suffix'. We introduced these terms informally in Chapter 3. Prefixes and suffixes are the most commonly occurring types of affix found in the world's languages, and in European languages they are probably the only regularly occurring affixes. But there are two further types of affix that are found less commonly: the 'infix' and the 'suprafix'.
An infix is an affix which interrupts the root; it is inserted into the root, instead of being joined at the side as with a prefix or suffix. Consider the following data (SIL 1980: A15) from Chontal of Oaxaca (Mexico):
1.   tsetse
2.   tuwa
3.   te?a
4.   meko?
'squirrel' 'foreigner' 'elder' 'spoon'
tseitse tuiwa tePa metko?
'squirrels' 'foreigners' 'elders' 'spoons'
The 'plural' morpheme is an infix -ŕ-, placed between the two syllables of the root morpheme.
Look now at the following Kamhmu? (Laos) data (SIL 1980: A16):
1.   kap
2.   poot
3.   sal
4.   hiip
'to grasp with tongs'           krnap
'to walk on'                        prnoot
'to place in ear-lobe'          srnal
'to eat with a spoon'           hrniip
'tongs'
'platform round a house'
'ear ornament'
'spoon'
The morpheme -rn- is an infix, placed after the initial consonant of the root, with the function of deriving a noun from a verb; it might be called a 'nominaliser' morpheme.
The final affix type, the suprafix, is so called because it accompanies the root, and is usually written above the root. Suprafixes are, therefore, features of tone or stress
Word structure          31
or nasalisation, which have an identifiable meaning or grammatical function. This may be illustrated from English: the word import has two stress patternings associated with a difference of word-class membership. If import receives its main or primary stress on the first syllable, it belongs to the noun class; but if it receives its main stress on the second syllable—import—it belongs to the verb class. Arguably, therefore, the shift in stress patterning may be regarded as a morpheme (suprafix), since it has the same function as, for example, the suffix -ment in statement (verb —* noun) or the prefix be- in befriend (noun —» verb), depending on whether one takes import (noun) or import (verb) as the basic (root) morpheme.
Consider the following data (SIL 1980: A17) from Bekwarra (Nigeria). The language is tonal (ie pitch change on a word may signal a lexical/meaning difference), having a system of three tones: high tone is marked '; low tone is markedv; mid tone is unmarked. Compare: abe éfää 'they grind', abe éfaä 'they learn', abe éfaa 'they roast'.
1.  abe éfää                         'they grind'
2.  abeefää                         'they ground'
3.   abeefää                        'they should grind'
4.   abeéhärä                      'they answer'
5.   abeéhärä                      'theyanswered'
6.  abeéhärä                      'they should answer'
The tone on the initial syllable of the root is a suprafix with a number of distinct meanings: high tone signals 'present tense'; mid tone signals 'past tense'; low tone signals'obligation'.
Word structure
We have described the structure of words in terms of the constituent morphemes that compose words. For each word there is a root morpheme, to which may be added one or more affixes. A root morpheme is the minimal form of a word, and some words consist only of a root morpheme, eg art, circle, difficult in English. In some languages (eg Chinese or Vietnamese) the vast majority of words are of this kind; ie there are few if any affixes in these languages. In languages like English, words consisting of a single root morpheme occur freely, as well as words composed of a root and one or more affixes, eg gives, un-decide-d, stigmat-is{e)-ing, general-is(e)-ation-s, re-institut-ion-al-is(e)-at\on. In a language like Swahili probably the majority of words contain affixes attachec to a root.
When i multiple affixationVcurs, it is usually the case that the affixes occur in a specific^orde^ ,so that k'fs important in the description of word structure to state affix orders if this is appropriate. We have done this, for example, in the solutions to Exercises 9 and 11. Consider now the following data (SIL 1980: All) from Sierra Popoluca (Mexico):
1.   anAkpa                          'I go'
2.  tanAkgakpa                   'You and I go again'
3.  minAkta?mgakum          'You all went again'
32          Discovering grammar
i
4. 5. 6.
7.
nAkjahpa
nAknejahum
anAkneta'mgakum
nAkjahgakpa
nAkpa
'They go'
'They have gone'
'We (exclusive) have gone again'
'They go again'
'He goes'
I    Let us first of all identify the morphemes, and then establish the order of affixes.
|   From its recurrence in all the eight examples, it would seem clear that n\k- is the
l   root morpheme meaning 'go'; it appears to be bound on the right (cf Nos 4, 5, 7, 8).
^     |   Comparing Nos 1 and 4, which differ only in person, we conclude that a- is the '1st
■»        |   person  singular'  morpheme,  and  -jah the '3rd person plural' morpheme.  A
.         |   comparison of Nos 4 and 8 reveals that -gak means 'again' or 'iterative'. From No 2
|   we can now conclude that ta~ means 'you and ľ or Tst/2nd person singular'. What
z    Nos 1, 2, 4, 7 and 8 have in common by contrast with the remainder is that they are
I    present tense, which appears to be marked by -pa. If this is so, we conclude from
j     No 8 that '3rd person singular' is unmarked. Nos 3, 5 and 6 contain the suffix -urn,
which appears to mean 'past tense'. Comparing Nos 3 and 6, we can conclude that
-taPm means 'plural (for lst/2nd person)', while mi- is the morpheme for '2nd
person'; we must revise our assessment of a- to '1st person', being singular in No 1
without -ta?m, but becoming plural in No 6 only by the presence of -td?m. The
remaining morpheme to be identified is -ne in Nos 5 and 6, which corresponds to
have in the English gloss, ie it has the meaning 'perfective'.
Summarising, we have identified the following morphemes in the verb word in Sierra Popoluca:
Verb Root: riAk- 'go'.
Person Affix: a- '1st person', mi- '2nd person', ta- 'lst/2nd person'
plural'. Tense Affix: -pa 'present', -urn 'past'. Aspect Affix: -gak 'iterative', -ne 'perfective'. Number Affix: -ta^m 'plural'.
The order of morphemes is as follows:
Person (lst-2nd)—Root—Perfective—Plural/3rd person plural—Iterative—Tense.
-jah '3rd person
Root and stem
We have identified the root as the part of a word to which affixes are added. Alternatively, the root may be viewed as the part of a word that remains after all the affixes have been removed. What we need in addition to the terms 'root' and 'affix' is a term to designate a structure (word) composed of root plus affix(es), but which may itself be further affixed. The term used is 'stem'. A stem is a word which may stiH have_ additional affixes attached to it. The term 'steirflhus includes "'root': a root is a minimal stem. LolSk^^^                              example:   ■   ......     --• ■■-—-.
re-align-ment-s
ROOT -STEM------STEM—
Word structure          33
Align is the root. It is prefixed by re-, and realign is a stem, since it may be further affixed, eg by -ment. Realignment is also a stem, since it may be affixed by the 'plural' morpheme -s. However, realignments is no longer a stem, since it may not take any further affixes.
Exercise 12
Identify the morphemes in the following data (SIL 1980: A19) from Chatino (Mexico), and list them with their meanings (Notes: treat ngu- as a single morpheme; ~ marks nasalisation):
1.	ngudaó	'we(inclusive) gave'
2.	ndaö	'we(inclusive) give'
3.	ngudaba	'we(exclusive) gave'
4.	ngudä	'I gave'
5.	nguda	'You/he gave'
6.	nda	'You/he give'
7.	ndja	'You/he pay' (ie are caused to give)
8.	ngudja	You/he paid'
9.	kuda	You/he will give'
10.	kusi?ju	You/he will cut'
11.	nsPjü	'1 cut (present)'
12.	nsjr'ju	You/ he are caused to cut'
13.	ngusPjuô	'We(inclusive) cut (past)'
14.	ngusPjuwö	You(plural) cut (past)'
15.	ngusji'ju	You/he were caused to cut'
7. Combining morphemes
Derivation and inflection
In the previous chapter we considered the decomposition of words into their constituent parts, ie morphemes. That is to say, we took an analytical aproach to the consideration of word structure. In this chapter we are going to take a 'synthetic' approach, attempting to answer the questions: "How do morphemes combine together to form words?", "What kinds of processes are at work?"
We need first of all to notice that two general processes are at work in combining morphemes into words: 'derivation' and 'inflection'. Derivation is a lexical process which forms a new word out of an existing word by the addition of a morpheme. For example, in English the suffix -ment may be added to the verb state to derive the noun statement, which is a different word, an additional vocabulary item in English. Similarly in the Kamhmu? examples cited in the previous chapter, the addition of the infix -rn- to, say, the verb poot 'to walk on' derives a noun prnoot 'platform around a house', an additional lexical item in this language. Strictly speaking, the term 'derivation' refers to the creation of a new word by means of the addition of an affix to a stem; but we shall also be considering other word-formation processes in this chapter, which have a similar, lexical, function.
The other general process at work in combining morphemes, inflection, is a grammatical process. For example, the plural morpheme in English is an inflectional morpheme. Inflections represent alternative grammatical forms of a word. The plural form boys, for instance, does not represent a different lexical item from the singular form boy, but rather merely the addition of the morphemic realisation of the grammatical category of plural number. An inflectional morpheme, then, does not derive a different vocabulary item; it realises alternative grammatical forms of the same word.
The distinction between derivation and inflection is a morphological one. It should not be concluded that a particular meaning (eg 'plural') is universally represented either inflectionally or derivationally. Plural number, for example, may be marked equally well by quantifiers (eg many, several) or numerals, ie lexically, as by an inflectional affix. Similarly, 'person' may be realised by an affix in the verb word (ie inflectionally) or alternatively by a separate pronoun word (ie lexically); compare the distinction made in Chapter 5 between bound and free pronouns. Or consider the Sierra Popolucä data in the previous chapter: here the meaning 'iterative' is realised grammatically by the inflectional morpheme -gak, whereas in English this meaning is realised lexically by the separate word again. The term 'derivational' and 'inflectional' are, then, a means of distinguishing a morphological process that creates new vocabulary items on the one hand, and on the other a morphological process that realises alternative grammatical forms of the same word.
34
Combining morphemes        35
Compounding
We look first under the broad heading of derivation at a word-formation process known as 'compounding'. Comt^mding involves the combinationof two roots (Root + Root), to form a new word_or_stem. Foj example, in English the word dnßvb^TiTma^lj^ofcwoloo^, the verb drift and the noun wood, both of which may be independent words in their own right. Similarly in German, the word Lebensabend 'old age' is a compound of the nouns Leben 'life' and Abend 'evening'.
In compounding, two roots combine in such a way that each loses its independent word status and the new combination takes on the characteristics of a single word. These single-word characteristics may include the semantic dimension, eg a compound may have a unitary meaning not immediately derivable from its elements, as in German Augenblick 'moment', made up of Auge 'eye' and Blick 'look/glance'. Or they may include the grammatical dimension, eg a compound will inflect like a single word, as English babysit. This has all the usual verb inflections of English, eg babysits, babysitting, babysat; but the baby part of the compound does not inflect, eg for 'plural', however many babies may be involved. Additionally, single-word characteristics may include the phonological dimension, eg a compound will take on the stress features of single words, or regular phonological changes may occur when roots are compounded. In English, for example, single words have characteristically one primary (or main), stress, so that in a compound one of the roots normally loses its primary stress; compare "a black board" and "a blackboard". In German, a number of roots forming the first element of a compound regularly take an additional -s, eg Universitätsprofessor 'university professor', Universitätsgebäude 'university building', Universitätsstudium 'study at a university'.
Consider now the following data (SIL 1980:18) from Igede (Nigeria):
1.   iyo        'meat'                enyi           'water'               iyenyi        'fish'
2.   imi        'hunger'             enyi           'water'               imenyi       'thirst'
3.   ube       'room'               utoji          'medicine'          ubutoji      'clinic'.
It will be noticed that all the nouns in this data begin with a vowel and end with a vowel. However, it appears that two vowels cannot occur together in a word, so that on compounding the final vowel of the first element is omitted. Note that in the compounds in this data it is the first element that is semantically primary, ie iyenyi 'fish' is a kind of meat (iyo), imenyi 'thirst' is a kind of 'hunger' (imi), and ubutoji 'clinic' is a kind of 'room' (ube). In our English and German examples, representative of these languages generally, the order is reversed: driftwood is a kind of 'wood', Universitätsgebäude is a kind of Gebäude 'building'.
Compounding is not always a matter of the combination of two roots. Take the cómpouná'waičhmdícérIn English, representative of a common type of English compound. This is a combination of the noun root watch and the derived noun maker, which has a verb root make. But there is no *watchmake in English, so that the second element of watchmaker is not, strictly speaking, a root, but a stem. Compounds themselves should also be regarded as stems, since they may be
36          Discovering grammar
affixed; eg watchmakers by the 'plural' -s suffix, overtaker by the 'nominaliser' -er suffix. Indeed, compounds may be elements in further compounds, eg in German Mitunterschrift 'joint signature', made up of mit 'with' plus Unterschrift 'signature', which in turn is a compound made up of unter 'under' plus Schrift 'writing'.
Exercise 13
Examine  the  following  data  (SIL  1980;   19)  from Mbembe  (Nigeria)  and  say what phonological processes occur as a result of compounding:
1.  ana
2.  obsik oso:m eci
eciraqa eci oda eba
'oil'
'arm'
'house'
'tree'
'medicine'
'tree'
'sleep'
'breast'
e-pya
edorp
emma
eraga
3so:m
ikwuma
icen
asi
'market'
'throat'
'mouth'
'root'
'house'
'half
'eye'
'liquid'
enapya
eboidorp
£so:mma
ecirana
ocirarjasoim
icikwuma
idacen
abasi
'selling of oil'
'wrist'
'threshold'
'medicine'
'hospital'
'log of wood'
'dream'
'milk'
Derivation
We look secondly at derivatioa^inapej^Jü^oj^^                                              by
means of affixation. All four types of affix may be used derivationally. We meÄtiorretf~a"äenvational suffix and a derivational infix at the beginning of the chapter. A derivational prefix can be illustrated from English, eg be- in befriend, deriving a verb from the noun friend. The stress change on the English word import, deriving a noun from a verb or vice versa, could be regarded as a derivational suprafix. Derivation involves the addition of an affix to a root or stem to form a different lexical item (word or new stem).
The majority of derivational affixes effect a change in the word class membership of the stem to which they are added, ie the derived word (or stem) has a different word class membership from that of the stem being affixed. For example, in German the addition of the suffix -heit to the adjective wahr 'true' derives the noun Wahrheit 'truth', in English the addition of the prefix en- to the adjective large derives the verb enlarge. A derivational affix may not in some cases effect a change in the word class membership of the stem to which it is added, eg in English the negative prefix un- when added to the adjective true derives another adjective untrue, and when added to the verb tie derives another verb untie. Intermediate between these two types of derivational affix is a third type which does not effect a change of word class, but of sub-class within a class, eg in English there are several derivational suffixes forming abstract nouns from concrete nouns: machine-ery, boy-hood, dictator-ship.
Consider now the following data (SIL 1980:12) from Mbembe (Nigeria):
1.   běrja
2.  -ebr
3.   dôrj
'to grow' 'to be pale' 'to be deep'
áběrjíjí ab bríjí
ädôijíjí
'growth'
'paleness'
'depth'
Combining morphemes        37
4. göro	'to be proud'	ägôríjí	'pride'
5. bán	'to be dirty'	ábárjíjí	'dirtiness'
6. came	'to be small'	ácámíjí	'smallness'
In this data, the derivation of a noun from a verb involves a simultaneous prefix-ation and suffixation. Prefixation is by a- which 'agrees' in tone with the root, low tone before low tone (Nos 1-4), high tone before high tone (Nos 5 and 6). Suffixation is by -íjí. If the verb root ends in a vowel (Nos 1, 4 and 6), this is deleted when -í/í is affixed.
Even more true of derivation than of compounding is the possibility of a word resulting from the application of several derivational processes. We noted this point in the previous chapter. For example, the English word intermarriage may be viewed as the prefixation of the verb root marry by inter- (which does not change the word class), and then suffixation by -age to derive a noun. Similarly, the German word misstonig is first of all the prefixation of the noun root Ton 'sound' by miss- to derive Mission 'dissonance' (still a noun), and the suffixation of this stem by -ig accompanied by umlaut of the root vowel to derive the adjective misstonig 'dissonant'.
Exercise 14
Examine the following Lisu (North Thailand) data (SIL 1980: 13) and identify the derivational affixes, giving each a 'meaning'.
1.  syi
2.  syidwu
3.  syiswu
4.  tshye
5.  tshyegwu
6.  tya
7.  tyaswu
8.  tyagwu
'to sweep' 'broom' 'sweeper' 'to hide' 'hiding place' 'to live' 'inhabitant' 'house'
9. tshi
10.  tshiswu
11.  tshigwu
12.  tshidwu
13.  pú
14.  púdwu
15.  püswu
'to wash'
'washerwoman'
'washing place'
'thing for washing clothes'
'to write'
'pencil'
'writer'
Other word-formation processes
Apart from the non-morphological ways of adding new vocabulary to a language such as borrowing'words from other languages (cf detente in English from French, or le weekend in French from English) or creating words from roots in classical Latin and Greek (eg many scientific and medical terms), there are a few more strictly morphological ways of forming new words, but of a minor nature by comparison with compounding and derivation proper.
Perhaps the moji widespread of these minor word-formation processes is one known as/'conversiori\ which could be described as derivation without affixation. A word (orstern) belonging to one word class is transferred to another word class without any change of form by means of affixation. For example, the word bottle in English is a noun, but it is also used as a verb without any change of form, except that as a verb it is subject to the usual inflectional affixation of verbs (he/she bottles, bottling, bottled). Compare also: peel (noun and verb), race (verb and
38          Discovering grammar
noun), dirty (adjective and verb), escape (verb and noun), picture (noun and verb). Similarly in other languages, eg French travail/travailler 'work', lutte/lutter 'fight' (noun and verb, the -er being the verb inflectional suffix for the infinitive, ie citation form), innocent 'innocent' (adjective and noun) together with the verb innocenter 'to declare innocent'; also German Blick/blicken 'look, glance', Arbeit/arbeiten 'work', Rauch/rauchen 'smoke' (all noun and verb, with -en the inflectional suffix for the infinitive form of the verb).
A second minor word-formation process is one known as 'back-formation', which is in a way the converse of derivation: a word is coined by the removal of a (supposed) affix. For example, the verb babysit in English was derived from the noun babysitter by the removal of the -er, which was associated with the 'agentive' suffix of words like walker, seller, flyer, etc. The verb edit was derived from the noun editor in a similar way. Unlike derivation proper, however, the process of back-formation is not obvious to the modern native-speaker without access to historical linguistic information. Nevertheless, back-formation is a process by which new words may yet be coined in a contemporary language.
A last minor word-formation process that we might mention is the coining of new words by formingji^ranj/rnSa ie taking the initial letters of the words in a phrase and forming them into a new word. For example, the word radar was formed in this way, standing for 'radio detecting and ranging'; also laser, 'lightwave amplification by stimulated emission of radiation'. But this word-formation process is popular with international organisations, cf NATO 'North Atlantic Treaty Organisation', UNESCO 'United Nations Educational, Scientific and Cultural Organisation'.
Inflection
We come finally in this chapter to inflectional morphemes, those which add grammatical meaning to a word. Inflectional morphemes cannot change the word class membership of a stem to which they are affixed. They realise grammatical categories and produce different grammatical forms of the same word. Take the verb in English: as a rule, it has potentially five different forms—base form/present tense (eg show), 3rd person singular present tense (eg shows), past tense (eg showed), present participle (egshowing), past participle (egshown). In fact, for the majority of verbs the past tense and the past participle have the same form (eg walked). The inflectional morphemes here realise categories like person/tense (-s), past tense (-ed), and participles {-ing, -n); and the resulting verb word forms are used appropriately in specific lexical and syntactic contexts, eg the present participle is used with be to realise 'progressive aspect' ("they are showing") or after certain other verbs ("they like reading stories").
In English, inflectional morphemes are always suffixes (cf also 'plural' and 'possessive' in the noun—girls, girl's); but any of the four types of affix may be inflectional. In the Congo Swahili data of Exercise 8, the inflections for person and tense are both prefixes, eg ninasem 'I speak' (ni- '1st person singular', na- 'present tense'). In the previous chapter, one of the illustrations of an infix was the inflectional morpheme 'plural' in Chontal of Oaxaca, eg tuiwa 'foreigners' {-i- 'plural'). And in the
M
Combining morphemes         39
Chatino data of Exercise 12, the inflectional morpheme '1st person singular' is a suprafix of nasalisation, eg ngudä T gave' (ngu- 'past tense', -da 'give', ~ '1st person singular').
We have noted on a number of occasions that grammatical categories intersect and find a realisation by a common morpheme; for example, number and person intersect in pronouns; tense, person and number intersect in verbs. That is to say, one inflectional affix may 'carry' a number of grammatical meanings, and it is not possible to identify a separate segment of the word for each meaning. For example, the -s inflection in the English verb (eg shows) 'carries' the meanings '3rd person', 'singular number', and 'present tense'. This point can perhaps be illustrated best from the conjugational and declensional paradigms of Latin or Greek grammar. Consider, for instance, the verb form amo T love' in Latin. It is composed of the root am- 'to love' and the inflectional suffix -o, meaning '1st person singular present active indicative'; ie it 'carries' five meanings. It is '1st person' by contrast with, say, '2nd person', eg amas 'you love'. It is 'singular' in number by contrast with 'plural', eg amamus 'we love'. It is 'present' in tense by contrast with, say, 'past', eg amabam T loved'. It is 'active' in voice by contrast with 'passive', eg amor T am loved'. And it is 'indicative' in mood by contrast with, say, 'subjunctive', eg amarem T should love'. Thus, there is no necessary one-to-one relationship between inflectional affix and grammatical category or meaning.
Exercise 15
Identify the morphemes in the following Sierra Aztec (Mexico) data (SIL 1980: A9) and list them with their meanings. Say if the affixes are inflectional or derivational. Chart the order of morphemes.
1.	nimitsita	'I see you'
2.	nikita	'I see him'
3.	nikmaka	'I give it to him'
4.	tikmaka	'You give it to him'
5.	tinetjita	'You see me'
6.	nannet/maka	'You(pl) give it to me'
7.	tikonmaka	You give it to him, Sir'
8.	tikonitatihtsinoh	You see him, most honoured Sir'
9.	tinetjonita	You see me, Sir'
.0.	tinetjonmakatihtsinoh	You give it to me, most honoured Sir'
.1.	nannetjonmakatsikah	'You(pl) give it to me, honoured Sirs'
.2.	nannetjonitatihtsinoh	'You(pl) see me, most honoured Sirs'
.3.	tinetfonitatsikah	'You see me, honoured Sir'
14. nannetjonmakatihtsinoh   'You(pl) give it to me, most honoured Sirs'
8. Morpheme variants
Some knowledge of phonetics is assumed in this chapter.
AHomorphs
In Chapter 3 the reader was requested to note the different pronunciations of the 'plural' suffix in the English words apples, apricots and peaches. We come now to consider the nature of such differences. What differences like this mean is that a particular morpheme (eg 'plural number') may not be always pronounced (or spelt) in the same way wherever it occurs as a constituent of a word. Morphemes, that is to say, may have variant realisations (pronunciations or spellings), depending on their context of occurrence. These variant realisations are known as 'allomorphs'. From our examples above, the English 'plural' morpheme has the allomorphs/z/, /s/ and/iz/ respectively. Strictly speaking, then, the 'morpheme' is an abstract unit, whose concrete realisation in sound or spelling is by means of a 'morph' or—if there are variant realisations—'allomorphs'. We can equate the term 'morpheme' with the 'meaning' or 'function', and the term 'morph' with sounds or letters.
Some morphemes have only a single realisation; they always appear with the same pronunciation/spelling in a particular variety of the language (eg dialect or accent). For example, the 'present participle' morpheme in standard English has the single realisation -iWg/irj/in all its contexts of occurrence, ie whichever verb root it is affixed to (cf sowing, sleeping, finding). Most inflectional morphemes in English, however, do have alternative realisations (allomorphs), such as the 'plural' inflection. But it is not only inflectional morphemes that may have allomorphs; derivational morphemes and root morphemes may also have variant forms. For example, the 'negative' prefix un- has variant forms in unpleasant/Am/, unkind/At}/ and unchanged/An/; the root resign has variant forms in resí'g/u/rizaín/and resignation)rezign/.
There are two kinds_QLde&ctipt.ive-.explanatiônTor morpheme variants. On the one hand, the pronunciation of a morpheme may vary bécaUsěTof its phonological context, ie the nature of the sounds in the accompanying morpheme(s): such allomorphs are said to be 'phonologically conditioned'. On the other hand, the realisation of a morpheme may vary quite arbitrarily when combined with certain other morphemes: such variation is said to'Be^mörpKöIögicaliy conditioned', ie the variation depends purely on Which morpheme accompahieTthit with variant forms, but not on its phonological form.
Exercise 16
Identify the variant forms (allomorphs) of the 'past tense' morpheme in the following English data. Be sure to consider the pronunciation;
40
Morpheme variants        41
1.	believe	believed	6.	miss	missed
2.	float	floated	7.	redeem	redeemed
3.	snatch	snatched	8.	give	gave
4.	allow	allowed	9.	catch	caught
5.	need	needed	10.	put	put
Phonological conditioning
The allomorphs of a morpheme are said to be phonologically conditioned when the variation in pronunciation can be explained from the phonological context',le the sounds composing the adjoining moŕph(s); Take the allomorphs of the 'plural' morpheme-in English, with which we began this chapter: /z/, /s/, /iz/. These allomorphs are phonologically conditioned: which one of them occurs depends on the nature of the final sound of the noun stem to which the 'plural' morpheme is affixed. If the final sound of the noun stem is a sibilant (ie/s, z, J, 3, tj\ &■$!), then the 'plural' morpheme is realised by the allomorph /iz/; if the stem-final sound is a voiceless consonant (other than a sibilant), the allomorph is /s/; if the stem-final sound is a voiced sound, ie a vowel or a voiced consonant (other than a sibilant), then the allomorph of the 'plural' is /z/.
This conditioning just described may be expressed in the following formula:
{plural}
/iz/    after stem-final sibilants /s/    after-stem-final voiceless consonants /z/    elsewhere
The morpheme is conventionally written in curly brackets (braces), and the allomorphs in the slashed brackets of phonemic transcription. The arrow represents the statement "is realised by the allomorphs", and the contexts of conditioning are put after each allomorph. Note that the allomorph with the narrowest distribution (ie occurring after the least number of sounds) is put first, and that with the widest distribution last. Each successive conditioning statement excludes the one(s) that have preceded, so that the last one is appropriately expressed by an "elsewhere" statement, since it refers to every phonological environment not previously mentioned. A similar statement can be made for the 'past tense' suffix identified in Exercise 16:
/id/    after stem-final alveolar plosives {past tense}-*      /t/    after stem-final voiceless consonants
/d/    elsewhere Phonological conditioning, thus, implies providing a plausible phonological explanation from surrounding context for allomorphic variation. Consider now the following data (SIL 1980: H10) from Ilocano (Philippines):
1.	tugawko	'my chair'	7.	bagasko	'my rice'
2.	tugawmo	'your chair'	8.	bagasmo	'your rice'
3.	tugawna	'his chair'	9.	bagasna	'his rice'
4.	sabak	'my banana'	10.	sidak	'my food'
5.	sabam	'your banana'	11.	sidam	'your food'
6.	sabana	'his banana'	12.	sidana	'his food'
42           Discovering grammar
The nouns in this data are composed of a root followed by a 'possessive' suffix. Three possessive morphemes are represented in this data: '1st person singular', '2nd person', '3rd person singular masculine'. The '3rd person singular masculine possessive* morpheme is realised by a single morph /-na/ (Nos 3, 6, 9, 12). The other two possessive morphemes are each realised by two allomorphs, the '1st person singular possessive' by /-ko/ and /-k/, and the '2nd person possessive' by /-mo/ and /-m/. The allomorphs of both morphemes are phonologically conditioned in the same way: /-ko/ and /-mo/ occur after a root-final consonant, while /-k/ and /-m/ occur after a root-final vowel. A statement for the '1st person singular possessive' would be as follows:
f       1st person       ""i                  /k/
\ singular possessive f               /ko/
And similarly for the '2nd person possessive':
f 2nd person "i                            /m/    after root-final vowel
i  possessive j                          /mo/    elsewhere
after root-final vowel elsewhere
Exercise 17
Identify the allomorphs of the 'plural' and '2nd person possessive' morphemes in the following Turkish data (SIL 1980: H12), and state the phonological conditioning. (Note: the answer lies in the vowels.)
		plural	'your' -l- sing noun	'your' + pi noun
1. mum	'candle'	mumlar	mumun	mumlarin
2. kibrit	'match'	kibritler	kibritin	kibritlerin
3. yzym	'grape'	yzymler	yzymyn	yzymlerin
4. sinif	'class'	sinifiar	sinifin	siniflarin
5. ders	'lesson'	dersler	dersin	derslerin
6. satJ	'hair'	satjlar	sat/in	satjlarin
7. g0Z	'eye'	g0zler	g0zyn	g0zlerin
8. top	'gun'	toplar	topun	toplarin
9. kuj	'bird'	kuflar	kufun	kuflarin
10. dij-	'tooth'	di/ler	dijin	dijlerin
11. ok	'arrow'	oklar	okun	oklartn
12. kak	'root'	k0kler	kakyn	k0klerin
Morphological conditioning
Allomorphs of a morpheme, whose variation cannot be accounted for phonologically, are; said to be 'morphologically conditioned'. That is to say, a particular allomorph of a morpheme occurs with an arbitrary set of other morphemes; the presence of one of that set determines or conditions the occurrence of the particular allomorph. For example, the allomorphs of the 'plural' morpheme in German, which we discussed in Chapter 3, are morphologically conditioned allomorphs: -e/a/ occurs with noun morphemes like Tag 'day', Mond 'moon', Schuh 'shoe', etc, -er /sr/ occurs with noun morphemes like Ei 'egg', Kind 'child', Licht
Morpheme variants        43
'light', etc, and so on. Learners of German know that for each noun learnt as a vocabulary item they must also learn which plural allomorph it is associated with; apart from a few rules of thumb (eg feminine nouns ending in e usually take plural in -n), there is no general predictability, and certainly not of a phonological nature.
A similar kind of statement of allomorphic variation can be made for morphologically conditioned allomorphs as for phonologically conditioned ones, except that in the case of morphological conditioning a list of possible accompanying morphemes must be provided for each allomorph, rather than a general statement of conditioning. A statement for the 'plural' morpheme in German would have the following kind of form:
{plural}^*                /s/   with the following noun stems:
Detail 'detail', Hotel 'hotel', Team 'team', etc j-3rl    with the following noun stems:
Brett 'board', Kleid 'dress', Leib 'body', etc jssj   with the following noun stems:
Haken 'hook', Koffer'case', Fenster 'window', etc /(a)n/   with the following noun stems:
Biene 'bee', Frau 'woman', Vetter 'cousin', etc /a/    elsewhere, ie with the following noun stems: Erfolg 'success', Hund 'dog', Dieb 'thief, etc
To avoid over-complicating this statement, no account has been taken of the operation of 'umlaut' on stem vowels with some of these suffixes in respect of specific noun stems.
Consider now the following data (SIL 1980: H2) from Hixkaryana (Brazil):
Verb stem_________Adjective           ______
1.                  tawa-                    tawanje               'dark'
2.                  tak-                      takdje                'warm'
3.                  bata-                     bataje               'rotten'
4.                   tak-                      takrje                 'wet'
5.                 tawas-                   tawasdje               'light'
6.                 tut fu-                   tutjurje                'red'
7.               tatjenot-               tatfenotdje_______'cold'
In this data, the 'adjectiviser' morpheme has four allomorphs, which are morphologically conditioned as follows:
{adjectiviser}^              /nje/    with the following verb stem:
tawa-/Je/   with the following verb stem:
bata-/rje/    with the following verb stems:
tak-, tutfu-/dje/    elsewhere, ie with the following verb stems: tak-, tatjenot-, tawas-
44          Discovering grammar
Exercise 18
Identify the morphemes in the following Isthmus Zapotec (Mexico) data (SIL 1980: H5), and describe the allomorphs of the tense morphemes.
1.	rukaadu	'we write'	14.	ruyubibe	'he looks for'
2.	rukaabe	'he writes'	15.	zuyubibe	'he will look for'
3.	zukaabe	'he will write'	16.	biyubibe	'he looked for'
4.	bikaabe	'he wrote'	17.	kuyubibe	'he is looking for'
5,	kukaabe	'he is writing'	18.	rireebe	'he goes out'
6.	ru30onebe	'he runs'	19.	zareebe	'he will go out'
7.	zu3ooňebe	'he will run'	20.	bireebe	'he went out'
8.	bÍ30oňebe	'heran'	21.	kareebe	'he is going out'
9.	ku30oňebe	'he is running'	22.	ribanibe	'he wakes up'
10.	rid3elabe	'he finds'	23.	zabanibe	'he will wake up'
11.	zad3elabe	'he will find7	24.	bibanibe	'he woke up'
12.	bid3elabe	'he found'	25.	kabanibe	'he is waking up'
13.	kad3elabe	'he is finding'	26.	kabanidu	'we are waking up
Mixed conditioning
Sometimes the allomorphs of a morpheme cannot be accounted for completely by either phonological conditioning or morphological conditioning alone; some of the allomorphs are found to be phonologically conditioned and some morphologically conditioned. This is, in fact, the case with the allomorphs of the 'plural' morpheme and of the 'past tense' morpheme in English. In our previous discussions of these morphemes we have considered only the 'regular' forms, ie /s/, /z/, /iz/ for 'plural' and /t/, jáj, /id/ for 'past tense' (but cf Exercise 16). Consider, however, the following plurals of English nouns: deer, mice, oxen, teeth, criteria. The allomorph of the 'plural' in deer is zero /o/, ie there is no visible or audible mark of plurality; in mice the aliomorph is the stem vowel change /at) —> al/; in oxen it is the suffix /an/; in teeth it is the stem vowel change /u -» i) and in criteria it is the change taken over from Latin from /an/ {criterion /kraltlarian/) to /a/ (/kraltlsrla/), ie in effect the loss of final /n/. These allomorphs are all morphologically conditioned; they apply to only a small number of noun stems, and there are more of a similar kind. Clearly, then, the allomorphs of the 'plural' morpheme in English represent an instance of mixed conditioning.
In making a statement of mixed conditioning, since the general principle' is that allomorphs with the most restricted distribution are listed first and in mixed conditioning the phonologically conditioned allomorphs generally have the wider distribution, those that are morphologically conditioned are stated first. For example, a statement of the variants of the 'plural' morpheme in English would have the following kind of form:
{plural}^»           /an/    with the following noun stems: ox, child
/an/    with the following noun stems: deer, sheep, etc
/au-^al/    with the following noun stems: mouse, louse, etc
/u—»i/    with the following noun stems: tooth, goose, etc
/an-H-o/    with the following noun stems: criterion,phenomenon, etc
Morpheme variants        45
/íz/    after stem-final sibilants /s/    after stem-final voiceless consonants /z/    elsewhere
Now consider the following Northern Tepehuan (Mexico) data (SIL 1980: H19):
Singular
Plural
1.   'rabbit'
2.   'man'
3.   'foreigner'
4.   'tree'
5.   'son'
6.  'stone'
7.  'friend'
8.  'arrow'
9.   'turkey'
10.   'older brother'
11.   'species of bird'
12.   'needle'
13.   'younger brother'
14.   'species of fish'
15.   'rat'
16.   'water jar'
toji kAli obai
u/i
mara
obai aduni
uyi tova
j"ÍAgi
adatomaii
oyi
sukuli
aafi dAgi
ay i
toto/i
kAkAli
obai
uji
mamara
oxodai
aaduni
uxuyi
totova
JiJiAgi
aadatomali
oxoyi
susukuli
aafi
dAdAgi
axayi
There are three allomorphs of the 'plural' morpheme in this data. The allomorph with the widest distribution is the reduplication of the initial syllable (consisting of a Consonant + Vowel or just of a Vowel, ie/(C)V/), Nos 1, 2,5,7, 9,10,11, 13, 15: this allomorph can be regarded as phonologically conditioned. The one with the next widest distribution is the reduplication of the initial syllable consisting of just a Vowel with the addition of the consonant /x/, ie /Vx/: this must be regarded as morphologically conditioned, since the same initial vowel occurs as in the regular allomorph, viz /a/, cf Nos 7/14 with 16. The third allomorph is jaj, Nos 3, 4 and 14, which must also be regarded as morphologically conditioned. Our statement for the 'plural' morpheme in this data might, then, be as follows:
with the following noun stems: obai 'foreigner', u$i
'tree', aaji 'species of fish'
with the following noun stems: odai 'stone', uyi
'arrow', oyi 'needle', ayi 'water jar'
elsewhere
Make a statement for the '3rd person singular masculine possessive' (his) and the '3rd person singular masculine reflexive possessive' (his own) morphemes in the following data (SIL 1980: H20) based on Kaiwa (Brazil).
{plural} —»	jaj
	/Vx/
	/(QV)/
Exercise 19	
1.	iňäka	'his head'	oäka	'his own head'
2.	inambi	'his ear'	onambi	'his own ear'
3.	ipo	'his hand'	opo	'his own hand
4.	ijavati	'his corn'	oavati	'his own corn'
5.	i jib*	'his land'	o-ibi	'his own land'
46	Discovering grammar			
6.	iňäpeku	'his tongue'	oapeku	'his own tongue'
7.	h Pi	'his water'	o*i	'his own water'
8.	hi^abí	'his hair'	o?abt	'his own hair'
9.	iňämba	'his place'	oämba	'his own place'
10.	ijape	'his shell'	oape	'his own shell'
11.	ijagwa	'his dog'	ojagwa	'his own dog'
12.	ijape	'his track'	gwape	'his own track'
13.	ijupa	'his lying place'	gwupa	'his own lying place'
14.	íňäta	'his home'	nwáta	'his own home'
15.	iňémbPu	'his food'	nwembi-'u	'his own food' 'his own younger brother'   \
16.	ijíviri	'his younger brother'	gwiviri	
17.	hParjgwe	'his shadow'	gw?arjgwe	'his own shadow'
9. What is a word?
'Word' an ambiguous term
Within linguistics, as indeed within ordinary language, the term 'word' is multiply ambiguous. We use it to refer to different kinds of unit.'TöFexample, I might say that this sérírerrce'"confäins eleven * words ľ'B'y''word' here I would mean blocks of letters separated by spaces. A dictionary will claim to contain "15,000 words", and the term has changed its reference: for example, the items sing, sings, singing, sang and sung will count as only one of the 15,000 'words' in the dictionary. Or consider the case of 'homonyms', either spelt the same (ie 'homographs') or pronounced the same (ie 'homophones'); for example, long (opposite of short) and long ('desire strongly') are homonyms (both homographs and homophones), bow (as in "bow and arrow") and bow ('inclination of the head') are homographs but pronounced differently, while bow ('inclination of the head') and bough are homophones but not homographs. In each of these cases, we may ask whether they are the same word, and the answer will depend on the point of view that we take.
If we take the point of view of the written language, regarding words as combinations of letters (the 'orthographic' point of view), then row (ie r+o. + w) represents one word, however many 'meanings' it might have. Similarly, from the point of view of pronunciation ('phonological'), regarding words as combinations of sounds (phonemes), then /bau/ (ie /b/ + /a/ + /u/) represents one word, however many spellings or meanings it might have. From the point of view of the dictionary, however,.every, item .with, a 'different meaning' (though-iKaf is open to interpTetatlon) is a different word; so that bow (as in "bow and arrow"), bow ('inclination oťí^é'^Qaä^ánSBough are all separate words, as indeed are bow ('inclination of the head'—noun) and bow ('incline the head'—verb). However, as we noted earlier, bows, bowing and bowed would not be separate words in the dictionary, but regarded as 'forms' of the word bow (verb). Clearly, then, we need some new terminology to enable us to make the necessary distinctions that we have been discussing.
Word-forms
Following P H Matthews (Morphology, CUP 1979), we.will refer to orthographic and phonological words as 'word-forms'. These are words viewed as combinations of letters or combinations of sounds, respectively. In languages that are well-established in the written medium, the question of what constitutes an orthographic word-form poses few problems: the convention of writing 'words' with spaces either side provides the answer, except in the cgse of emergent compounds. For example, is bird bath spelt thus as two orthographic word-forms, or hyphenated bird-bath, or together birdbath, the last two counting each as one word-form. Decisions of this
47
48          Discovering grammar
nature at the orthographic level are generally dependent on prior lexical or phonological decisions about word status.
/Phonologically, the question of what is a word is considerably more /fraught/. In i some languages, there are clear phonological indicators of word bodReafieS;_In languages with a fixed stress v where the main (pi.pnmarx)L^tr^^aHsjgn the sjjge syllabj£m^ě^^~wořdT"a clear wor"d^tt^rnjenigigesI---Another such indicator, /whicITmäycön^                                           the syllable structure in a language.
Some languages have a very simple syllable structure, which does not allow combinations of consonants within a syllable; in those which do allow consonant clusters, there are specifiable constraints on combinations. For example, in the sequence /boldtri/ "bald tree" in English, the boundary would have to be established between jáj and /t/, because the only permissible consonant clusters are /Id/ syllable-finally and /tr/ syllable-initially. A few languages exhibit a further phonological feature that contributes to the establishing of phonological word-forms, a feature termed 'vowel harmony'. This feature, found for example in Turkish (see Exercise 17), means that within a given word all the vowels will be of a given type (eg back rounded), so that a new syllable with a vowel not of this type will indicate the beginning of a new word.
In spite of these kinds of phonological indicators, it is extremely difficult in some languages to establish word forms phonologically. French is a notorious example, with stress distributed at utterance rather than word level and with extensive contraction of 'grammatical' words. Consider the following sequence: /33nle3amevy/ "je ne ľai jamais vu" ('I have never seen him/her'), where it is difficult to apply any of the phonological criteria we have discussed.
Lexemes-
The term that has been coined to refer to 'words' from the point of view of the dictionary (or lexicon) is 'lexeme', on the pattern of 'phoneme' and 'morpheme'. 'Words' in this sense correspond to the headwords of dictionary entries. So far we have assumed that a lexeme comprises a single word-form, but this is by no means always the case. One problematical area is that of compound words (cf Chapter 7). Many compounds are identifiable as a single word-form because they adopt phonological characteristics of ordinary words, eg a single primary stress in English or German: driftwood, Dampfschifffahrt 'steamboat journey'. But other root + or combinations, that one would want to recognise as compounds semantically root grammatically, do not adopt these phonological characteristics, eg sélf-séeker, whose compound status is recognised orthographically by the hyphen. Other compounds, whose single-word status is confirmed phonologically are, nevertheless, written as two word-forms, eg cattle grid, carrier pigeon. Another group of lexemes in English that has single-word status phonologically but not orthographically are the prepositional and phrasal verbs, eg look after, look up (ie 'visit'), look up to—to which we shall return later.
Leaving compounds aside, there are other combinations of word-forms that constitute a single item lexically: these are 'fixed phrases', such as metaphors or idioms,
What is a word?         49
eg on the straight and narrow, hammer and tongs, get the wrong end of the stick. Characteristic of such expressions, which occur as items in many dictionaries and are thus to be treated as single lexemes, is that their meaning cannot be read off from the meaning of their constituents and that they have become institutionalised as 'fixed phrases' with idiosyncratic meanings. There is, however, in English at least, no clear boundary between sequences of words, whose meaning derives from the amalgamation of the meaning of each constituent word, and those fixed phrases whose meaning is not derivable in this way. For example, there are sequences of words that regularly occur together (collocations), of which one member only does not have its usual meaning, eg "a happy chance". Happy is normally predicated of people, and here the meaning is slightly different from its normal usage: do we then count "happy chance" as one lexeme or two? For any particular language there may well not be any clear-cut answers; lexicographers of English have varied in their judgements of what are to be counted as lexemes in the language. Clearly, though, there may be no direct one-to-one correspondence between lexeme and word-form in a language.
Words
We reserve the term 'word' for the grammatical level of analysis. Words are units of grammar, whose structure is described in terms of morphemes, and which enter into the structure of units at higher levels of grammatical patterning, eg phrases and clauses. The different forms of lexemes (eg sing, sings, singing, sang, sung as forms of SING) thus constitute separate words grammatically. Equally, homonyms like skins (plural of noun skin) and skins (3rd person singular present tense of verb to skin) are different grammatical words. So are the plural, the possessive, and the plural possessive, all represented by the pronunciation /gelz/ (but spelt girls, girl's, girls' respectively).
One classic definition of the 'word' regards it as "the minimal free form", ie the smallest unit that may occur on its own, eg as a one-word response to a question. Clearly, under this definition an inflectional morpheme would be excluded from word status, since it may only occur bound to a root morpheme or stem. However, such a definition also calls into question the status of many so-called 'grammatical' or 'form'-words, for example the articles (a and the in English). It would be unusual to hear the as a one-word response, except as a reply to a request for clarification ("Did you say a tiger or the tiger?"—"The"); but such replies may also comprise bound morphemes ("Did you say disbelief or unbelief?"—"Dis").
The status of the as a word, as of 'form'-words generally, depends not on their ability to function as minimal responses, but on their detachability from other words and their relative "freedom of occurrence. For example, the may occur directly before nouns ("the tiger"), before adjectives ("the ferocious tiger"), before quantifiers and numerals ("the many/seven ferocious tigers"), etc. If the is not to be regarded as free, and thus grammatically a word, it would be necessary to find some other class of word or morpheme to which it would be affixed: from our examples it is clear that this is not the case. An alternative solution would be to say
50          Discovering grammar
that the is neither a word nor a bound morpheme, but has some indeterminate status, eg as a 'particle'—of which more below.
We consider first of all whether the 'word' or the 'morpheme' is to be considered the smallest unit of grammar. The traditional division of grammar into morphology and syntax regards the word as pivotal in both: in morphology it is the unit whose structure is described in terms of morphemes; and in syntax it is the smallest unit in the analysis of sentences. However, different types of language will lead to alternative evaluations of the relative status of word and morpheme. In an 'isolating' language like Vietnamese, where there are no inflectional or derivational affixes, it may be questionable whether a distinction between word and morpheme is strictly necessary. In an 'agglutinating' language like Turkish, where each morph realises only one morpheme, so that every word can be segmented into units, each of which has a single 'meaning', it would seem sensible not to insist on a division between morphology and syntax, but to regard the morpheme as the smallest unit in the syntactic hierarchy. In a 'synthetic' language like Latin, however, where a segment of a word (a morph) may realise a number of 'meanings' (eg the -a of puella 'girl' realises 'nominative' case, 'singular' number, 'feminine' gender), the distinction between morphology (as the description of the grammar of the word) and syntax (the description of the grammar of the sentence) is essential. We have described the three classic types of language: few languages conform exclusively in their structure to one only of these types. Linguistic analysts, therefore, need to make judgements on the relative status of word and morpheme in the description of a particular language on the basis of the nature of that individual language.
Consider now the following data (SIL 1980: Kl) from Kaiwa (Brazil). Word boundaries are not marked!
1.  iwi'tuma                       'The wind is already blowing (lit. 'It's winding')
2.  iwi'tupo'rá                    'lovely wind'
3.   ipo'räiwi'tu                   'The wind is lovely'
4.   ipi'täha'gwe                  'Its feathers are red'
5.   ha'gwepi'täpo'rä           'Its feathers are a beautiful red'
6.   iho'viha'gwe                  'Its feathers are green'
7.  ha'gweho'vipo'ra          'Its feathers are a beautiful green'
8.  ha'gwema                      'It already has feathers' (lit. 'It's feathering')
9.  ipo'räha'gwe                 'Its feathers are beautiful'
In this data it is clear that some items occur in the same position with respect to other (kinds of) items, while for others their relative position is variable. In particular: -ma, which is a 'verbaliser' (deriving a verb from a noun), and í-, also a 'verbaliser' (but deriving a verb from an adjective), must be regarded as bound morphemes. The other items, however, which we establish as items on the basis of their non-interruptibility, must be regarded as words, since they are mobile items: ha'gwe 'feathers', iwi'tu 'wind'; ho'vi 'green', pi'tä 'red',po'rä 'lovely/beautiful'.
Particles
We now take up the question of items which are clearly not morphemes, because of their relative freedom of occurrence, but which do not seem to attain to full word
What is a word?          51
status. We earlier cited the question of the in English. A further example from English would be the so-called 'adverb particles', which combine with lexical verbs to form 'phrasal verbs', eg look up (- 'visit'), slow down, give inJcnockjmL bring on (- 'induce'), take off (= 'imitate'). These are regarded as single lexemes (eg in the Longman Dictionary of Contemporary English), but they are~not compounds, because the adverb particle may be separated from the lexical verb (eg "the boxer knocked him out"), neither is the adverb particle an affix morpheme, for the same reason. But it is not a free word, since it is clearly part of the verb lexeme; and like the article, an adverb particle may not normally be the locus of intonational prominence in speech.
Another interesting case in English is that of the items more and most, which along A with the suffixes -er and -est are 'allomorphsjaf the icomparatiyeLand..^siipexlative' morphemes respectively. They are'not affix morph(eme)s, however, since they may occur before words other than adjectives, eg nouns (cf "more difficult", "more money"; "most decisive", "most respect"). Should we, then, perhaps regard them as particles, since they are like affix morph(eme)s in respect of the inflections of the adjective, but like words in their general behaviour in the sentence?
In German, a set of word-forms called 'modal particles' occurs, including doch, ja, wohl, (ein)mal. They never receive any intonational prominence, have a generally 'emphatic' function, and are frequently not translatable (into English, at least!). For example: "Sie wissen doch, wie das so ist" ('You know how "it is, don't you?'), "Das kann wohl mal vorkommen" ('That might well happen').
If we define particles as items that do not have bound morpheme status, but are not clearly words because they are generally unable to function as minimal responses, we ought probably to include prepositions and conjunctions within the category of particle. Prepositions, after all, are often functionally equivalent to inflectional affixes of synthetic languages, cf Latin urbe 'from the city', Old High German swertu 'with a sword'.
ČBlics-~-____
^ Particles are items that have ajmarginal status as wor Clitics are items that have a margináTstatus as morphemes^. That is to say, they clearíyform paTTor another wordTeg on phonological grounds, but they are moveable and do not regularly form part of one particular class of words. A candidate for the status of clitic in English is the 'possessive' suffix, which we generally regard as being an inflectional suffix of the noun. It does not, however, always occur on the noun: if the noun is postmodified (see Chapter 13), the 'possessive' clitic is attached to the last item in the noun phrase; cf "the president's proposals", "the president of the United States of America's proposals", "the president who addressed the nation yesterday's proposals". The last of these examples is perhaps less likely to occur; in such cases the alternative formulation with of is probably preferred ("the proposals of the president, who addressed the nation yesterday").
Also a candidate for clitic status is the Latin item Quer and', which only occurs as part of another word, but which may be affixed Walmost any word, eg "pueri puellaeque" 'boys and girls', "virmulierque" 'a man and a woman'.
52          Discovering grammar
Consider now the following data (SIL 1980: Kil) from Choctaw (USA):
1.   alatik-mattalowa                                       'the girl sings' girl           sings
2.   alatik osi-mat talowa                                  'the little girl sings' girl    little     sings
3.   alatik osi   taSapi-mat talowa                    'the four little girls sing' girl    little four           sing
4.   hatak-matabilini-mapilatok                      'the man threw the chair' man         chair        threw
5.   hatak-matabilinichite-mapilatok             'the man threw the big chair' man          chair  big          threw
6.   hatak-mat abilini-ma ayimpa-pito pilatok man          chair        table-on      threw
'the man threw the chair on the table'
7.   hatak-mat abilini-ma alatik osi-pito pilatok man          chair        girl    little-to threw
'the man threw the chair at the little girl'
Here the items -mat, -ma and -pito are to be regarded as clitics; they are affixed to stems, but not always to the same stem. The item -mat is a subject-marker and it occurs on the last item of the noun phrase functioning as subject, whether that is a noun (No 1), an adjective (No 2), or a numeral (No 3). The item -ma is an object marker and occurs similarly on a noun (No 4), or an adjective (No 5). The item -pito is a postposition, occurring on the last item of the noun phrase governed by it, a noun (No 6) or an adjective (No 7).
We must conclude, therefore, that the descriptive status of words, morphemes and any other intermediate unit (particles and clitics) must depend on the nature of the individual language and on the considered judgement of the linguistic analyst describing that language.
Exercise 20
Consider which items in the following Tlingit (Alaska) data (SIL 1980: K9) might be regarded as particles or clitics.
1.   ax hidee-t           oowagut                       'He went to (and arrived at) my house' my house-poss-to he-went
2.  ax hi dee         tlen-t oowagut                  'He went to my big house' my house-poss big-to he went
3.  ax hidee-dei      «woogoot                       'He went to (towards) my house' my house-poss-to he-went
4.  ax hidee         tten-dei woogoot               'He went to my big house' my house-poss big-to   he-went
5.   dooaat-gaa woogoot                              'He went for (to fetch) his aunt' his aunt-for he-went
6.   doo aat   has-gaa woogoot                        'He went to fetch his aunts' his  aunt pi.-for he-went
What is a word?        53
7.  ax aat   nak       ax hide-dei woogoot      'He went to my house without my aunt' my aunt without my house-to he-went
8.   dooaat   has nak       woogoot                   'He went without his aunts' his aunt pi. without he-went
9.   tleildooaat   teenwoogoot                      'He didn't go with his aunt' neg. his aunt with he-went
10.  dooaat   has teen woo.aat                         'He went with his aunts' his aunt pi. with he-went
10. Sentences (1)
Nouns and verbs
Sentences may be viewed essentially as the combinations of nouns and verbs,-More precisely, a sentence may be viewed as the combination of a verb^vith one or more nouns. We mentioned earlier that nouns refer to the^thiňgsr~of our experience (persons, objects, ideas, feelings, etc), and that verbs refer to 'ejvej^'-^induding actions, processes and states). Sentences, then, are concerned with thé^nyriadWyŠN ^xTIiTwhich ^he^thlngs' of experience are involved in the 'events' of expefieľiceT-oT' rather with the myriad ways in which we are able to talk about them.
Consider the following sentences of English:
i                 1.  Horses eat hay.
'!■- y"                   2.  The.committee has awarded the novelist first prize.
i>   3.  Thejheckiera)rgued with the speaker about his ideas.
4.  The children played in the garden after tea.
The first of these sentences contains just one verb and two nouns: the verb refers to an 'action', and the nouns to the 'things' involved in the action. In the second sentence, the structure still essentially comprises a verb and nouns (viz award, committee, novelist, prize): the additional words are depeňtfěňTlMf or modify these main items. Similarly in the third sentence, where the verb and nouns are argue, heckler, speaker, ideas:, except that here, besides 'modifying' words like the and his, there are a|soTeIatiorialSivords (prepositions) like with and about. The fourth sentence (in wnich~the-~verb and nouns are play, children, garden, tea) also has prepositions (in^ after) in addition to modifiers, but here the prepositions have a spatial and (temporal reference respectively, and are substitutable. The function of the prepositions in No 4 is thus different from that o'f the"prepösitions in No 3: you can only argue with somebody about something; but you can play in, on, outside, etc somewhere, after, before, at, etc, some point in time. But the basic point remains: sentences are essentially structures comprising a verb and a number of nouns.                                   <~      "              ———~.
//
ThsJferirisreentral
A useful way-qf approaching the analysis of sentence structure is to regard the verb as the central elementVhich determines how many (and which kind of) nouns may be present hi"--^-Sentence,- including any non-substitutab]e_ relational items (eg prepositions and case inflections). This function of the verb as the determining element in a sentence derives from a 'dependency'view of sentence structure (see Chapter 22). Which other elements a verb "determines is attributable to the 'meaning' of the verb. For example, we implied earlier that the verb argue
fyt,.,,   w ■                                        ■
I
ľ
Sentences (1)          55
determines the presence of three nouns: someone^argue with someone about something. In fact, our English specification allows lis to indicate additionally that two of the nouns refer to persons and one to a non-person. Clearly, this kind of statement has a good measure of generality, since "otfrgrkíríds of sentence structure occur with argue that make some refinements of this statement necessary; eg ''The boys are arguing", "He is arguing with his sister", "They are arguing about the rules of the !~gärneJiTTwo furtKěrpointsueexltoiíělnoticed about argue}: firstly, it may be used to refer generally to the action, when only the participants in the argument are mentioned', Bürnot the^sorrrethihg' argued about; secondly, the participanťšTň the ,;*., argument^£yj;b:c^IecĚ^4ogethě-r- as; one; (plural) noun (th^boy^piéý)"with or without mention of the'something' argued about.                                                           r>
Consider another exampie:.bwy ./This; verb could be said to determine the following nouns: someone buy something from someone/somewhere for some amount, eg "Bill bought the radio from the corner shop for £10". We might add: at someJime, eg "the day before yesterday"; though we might regard this as not being strictly determined by the verb buy, since almost any event may be specified for point of time._A refinement of this general statement for buy would have to indicate that not all of the nouns need to be specified, ie that the someone/somewhere from which the thing is bought and the amount for which it is.bought may be omitted (eg "Jack has_bmigliLajisÄJcar,'4..''Bill bought a radio Tor £10?, "Harry buys his newspapers from the corner.shop''),_*                                   K,^
English does not have case marking for sentence elements. We will now consider a couple of examples from German, which does mark case (though-as we have seen-—Chapter 3—not in the noun itself). Our first example is geben 'give', which determines as follows: -someone (nominative case) geben jpmeone (dative case) isöifißthittg (accusative case), eg "Der Mann hat seiner.Trau einen Diamanten gegeben/ The man gave his wife a diamond'. Our second example from German is -^hklagen 'accuse', which determines as follows {wegen is a preposition often translated By_jb_e£ause of in English): someone (nominative case) anklagen someone (accusative case) wegen.«wzer/n>jg (genitive case), eg "Die Polizei hat den Gefangenen wegen des Diebstahls angeklagt" 'The police accused the prisönerof thefť:-------
Exercise 21
Make 'determining' statements for the following English verbs, adding any 'refining' details you think necessary:
1- write                                2.  win
3. spend                               4.  wait
5- invite                                6.  clean
7. fight                                 8.  report
Sentence types
From the limited number of examples that we have looked at so far in this chapter, it has probably not been apparent that for any language a limited number of patterns of sentence structure occurs. That is to say, it is possible to make generalis-
'?/
In
v
r :
-rtfS j
1/
56
Discovering grammar
V"
ttY>
ations about the types of sentence patterns jthat occur in a language, many of which are applicable across languages.                                      ...-------             /"    ^
Traditio^all^Tlišfinction has been made between) 'transitiyy' and 'intransitiv^' sentences. Intransitive sentences have only one noun_'cH^rh/involveaI inJJTeeyeňt: the 'doej^ofjanjLctioS^or the 'undergoeT)of a process; eg^The^hlMreiraTeTaugh-ing'V'The tree is falling". In transitive sentences, more than one noun indirectly involved in the event/though usually not more than three. The seconZňjímn usually refers to the 'undergoes in an action, eg "The'goalkeeper caughUne"bail". The third noun, in what is sometimes called a 'ditransitiv^ sentence, usually refers to the 'recipisjLťiir3.benefi'ciajy;involved in theTaetionyeg "The shopkeeper sent the customer a bill". In German, and other languages with case systems,These different nouns are identifiable by their case marking: in German, the first noun (ie the one occurring alone in intransitive sentences, and the 'doer' in transitive ones) is in the nominative case, the second noun is in the accusative case, and the third noun is in the dative case.
Two further sentence types commonly found are'stative' and 'equate' sentences, both of which are characterised by the presence of the verb tejn English (though this is not the only verb which may occur), eg
Stative;                        "The leaves are brown".
r'f            Equative:                     "Harry is the ship's doctor".
In the case of stative sentences, the second (non-verbal) element is often—as in our example—ah adjective rather than a noun; though we could equally well say "The leaves are a brown colour", where the second element is now a noun (colour), modified by an adjective (brown). In stative sentences the adjective or second noun describes (the state of) the first noun. In equative>.sentences, as the term suggests, there is an equationaLrelationship betvveen'the two nouns^-thěÍ!ř~ardej.Js.,often reyeisiUe, eg we could have said for our example."The_ship's_doctor is Harrys, though this would have a different contextual meaning. In languages that have case inflections (eg Latin, German), the second element is often in the same case as the .ftfsTnoun (eg nominative case), cf German "Dieser Mann ist der Schiffsarzt" 'This man is the ship's doctor'.
Our list of sentence types is by no means exhaustive. Each language must be considered on its merits, on the basis of the possible generalisations from occurring sentence patterns. Consider now the following data from Kolami (India):
'he slept'
amd mangten he    slept addellangseddin she house went
*
amd aambal tinden he    rice       ate amd anung pustok siyten he   me     book   gave ramak poc[am andean Ram   tall       is
'she went to the house'
'he ate rice'
'he gave me a book'
'Ram is tall'
Sentences (1)         57
Here we can clearly jeeognise^njntransitive sentence (No 1), with just one noun and an action verb;(a transitive sentencfejjJNô' 3),with twoí nouns and an action verb; and a ditransitive seMencfi (No tyrvmh three nouns. No 5 falls into the stative (or 'descriptive') type, with a noun and an adjective describing the noun. No 2 is spme-what^of a problem, since it looks like a transitive sentence, with two nouns; but the action verb go does not involve the second noun (ellang 'house)} in the same way that ändert 'ate' Involves aambal 'rice', ie the house does not have something done to itHRithěr, thTlSó^jí^I^S or tne location, towards which the action is directed^S.uch sentence types, with a location as the second nouri but structured tire a transitive sentence, are often called 'semi-transitive' sentences; they occur typically in North-Indian languages.                          ...............________________
Exercise 22
identify the sentence types in the following Engenni (Nigeria) data (SIL 198Q: E7):
1.	edei ne adide man the rich-man		'The man is a rich man'
2.	Ade mise n'eki slept in-market		'Ade slept in the market'
3.	Ade doriya is-tall		'Ade is tall'
4.	edei ne moni ozhi avu	n'ukwo	'The man saw a thief at the farm'
	man the saw   thief one	at-farm	
5.	Ade do     eseni stole fish		'Ade stole a fish'
6.	eseni kuya fish    is-scarce		'Fish is scarce'
7.	edeine fina   n'oko man the enter in-canoe		'The man entered the canoe'
8.	adide      ne wu rich-man the died		'The rich man died'
9.	edei dori ne Ade man tall the		'The tall man is Ade'
S<	;ntences without verbs	\ftfl	
In the previous exercise, two of the examples were sentences without averb (Nos 1 and 9), which would seem to contradict our earlier statement that a sentence was constituted by a verb and^one or more nouns. In fact, that statement does not have general validity because, of-one other kind of sentence: one. that contains only a verb, eg Latin "Pluit" 'It is raining). But the more interesting cases of the statement not being valid are thosejyh_ere-there is no verb. In the Engenni data (Exercise 22), the sentences without verbs are equative sentences. And we fiŕídth-at it is not unusual for equative and stative/des^Fiptwe-sentences to be/verbless (where in English the verb be would be used).
Consider the following examples from Western Desert (Australia):
1.  watimirpanpa                                    'The man is angry' man angry
2.  wati ngalyayala                                  'The man is a doctor'
doctor
58         Discovering grammar
Similarly in Hixkaryana (Brazil) for equative sentences, eg
1.   Hoňko woto                                         'The game is/was wild pig' wild-pig game
2.   Towankaŕhemi anaro komo                 'Others were courageous' courageous-one other group
And in Tlingit (Alaska) there is a locative sentence that is verbless, eg
1.   ee    xooneeněilu                                  'Your friend is at home' your friend   inside-LM
2.   aatlěindáanaadoojěewu                      'He has lots of money' lots     money his hand-LM
(LM = locative marker, ie an inflection signalling location).
Exercise 23
Identify the sentence types of the examples in the following data (SIL 1980: E8) based on Pocomchi (Guatemala):
1.   huan fokik     pan ripat                            'John entered his house' John entered in    his-house
2.   re huan Jilwatj"       rasbes                       'John recognised his older brother'
recognised older-brother
3.   re ribahil       maria huan                         'John is Mary's husband' her-husband Mary John
4.   rijq'un          na   rik'uTum tulul              'Her daughter will eat bananas' her-daughter will eat           bananas
5.   re   tulul       firix     pat                           'The bananas are behind the house' the bananas behind house
6.   re        kinaq' wiki                                   'The beans are there'
beans  there
7.   re        maria re rijq'un                           'Her daughter is Mary'
Mary      her-daughter
8.   re be    inox pan tinamit                      'The road goes into town'
road goes in   town
9.   ripat         Jinx     k'aybal                        'Her house is behind the market' her-house behind market
10. maria na   ri'ayem tulu        pan tinamit   'Mary will sell bananas in town' Mary will sell        bananas in    town