sign in sign up
<<
Home , Grammatical category, Inessive case, Inflection, Noun class

4.1

Within a -based theory of , the difference between

derivation and inflection is very simple. Derivation gives you new , and

inflection gives you the forms of a lexeme that are determined by syntactic

environment (cf. 2.1.2). But what exactly does this mean? Is there really a need for

such a distinction? This section explores the answers to these questions, and in the

process, goes deeper into the relation between morphology and .

4.1.1 Inflection vs. derivation

The first question we can ask about the distinction between inflection and

derivation is whether there is any formal basis for distinguishing the two: can we tell

them apart because they do different things to ? One generalization that has

been made is that derivational tend to occur closer to the or stem than

inflectional affixes. For example, (1) shows that the English third person singular

present inflectional -s occurs outside of derivational like the deadjectival -ize, and the ending -s follows derivational affixes including the

deverbal -al:

(1) a. popular-ize-s

commercial-ize-s

b. upheav-al-s

arriv-al-s

Similarly, Japanese derivational suffixes like passive -rare or -sase precede

inflectional suffixes marking tense and aspect:1

(2) a. tabe-ru tabe-ta

eat- IMP eat- PERF INFLECTION 113

‘eats’ ‘ate’

b. tabe-rare- ru tabe-rare- ta

eat - PASS-IMP eat- PASS-PERF

‘is eaten’ ‘was eaten’

c. tabe-sase- ru tabe-sase- ta

eat- CAUS-IMP eat- CAUS-PERF

‘makes eat’ ‘made eat’

It is also the case that inflectional morphology does not change the meaning or of the that it applies to. A with a plural suffix attached to it is still a noun; slurp means the same thing whether it is past or present; and so on. Derivational morphology may or may not the grammatical category of a word it applies to, and it typically changes its meaning. Glory is a noun, and glorious is an . And while their meanings are related, they cannot be said to mean the same thing.

A final generalization we can make is that inflectional morphology tends to be more productive than derivational morphology. Inflectional morphology can apply to words of a given category with relative freedom. Virtually any noun in English can be made plural with the addition of [z] or one of its two allomorphs. The only exceptions are those that have irregular , such as children or phenomena, and those that logically do not allow a plural form: mass like rice and abstract nouns like intelligence generally fall into this category.

Despite these three generalizations, it turns out that in terms of actual morphs, it is hard to see what difference, if any, there is between inflection and derivation.

Crosslinguistically, the forms of inflectional morphology and the forms of derivational morphology are usually not very different from one another. Both can be 114 CHAPTER FOUR

expressed through prefixal, suffixal, or non-segmental means. For this reason, the

difference between inflection and derivation is not so much a difference in form as a

difference in function — what they do and what they tell us.

The word inflection comes from traditional . It represents the

idea that speakers bend the shape of a word so it will fit in a particular position within

a sentence. Its root flect-, which we see in the English word flex, means ‘bend’. We

really like to think of it in this way: every sentence is a syntactic frame with positions

for a series of lexemes. In order to fill one of those positions, you take a lexeme and

bend it to fit.

What are the ways in which speakers can bend lexemes to make them fit into a

certain syntactic slot? In like English or German, ‘bending’ generally

means adding affixes, although in some cases affixation is supplemented or replaced

by , or vowel alternations, as shown below for English (3) and the Bernese dialect of Swiss German (4):

(3) a. sing, sang, sung

b. drive, drove, driven

(4) a. suuffe [su…f´] ‘drink (inf.)’

gsoffe [g(sOf…´] ‘drunk (past part.)’

b. schwimme [S√Im…´] ‘swim (inf.)’

gschwomme [g(S√Um…´] ‘swum’

c. pfyffe [pfi…f´] ‘whistle (inf.)’

pfiffe [pfIf…´] ‘whistled (past part.)’

Another term for apophony is internal change. INFLECTION 115

When referring to English and other , the terms ablaut and umlaut often arise. These terms describe particular types of internal change that must be understood in a historical context. We use the term ablaut for apophony in paradigms, as in (3) and (4). Umlaut, on the other hand, is used to describe the apophony found in pairs like goose~geese or foot~feet. Umlaut differs from ablaut in that at one time it was a phonologically-conditioned alternation, with the high vowel in geese or feet resulting from with a high vowel in the plural suffix, which has since disappeared.

In the Semitic family, ‘bending’ often involves internal variations in a vocalic pattern, while the consonantal frame stays fairly stable. This is illustrated in

(5) for the expression of number in certain nouns (examples from McCarthy and Prince 1990: 212, 217). The inclusion of the loanwords ‘film’ and ‘bank’ is to show that this particular way of forming the plural (referred to as the ‘broken plural’) is robust and productive. What all of the plural forms in (5) have in common is that they begin with the pattern CVCVV+:

(5) Root Singular Plural Gloss

jndb jundub janaadib ‘locust’

slt¢n sult¤aan salaa?iin ‘sultan’

?nb ?inab ?anaab ‘grape’

flm film /aflaam ‘film’

nfs nafs nufuus ‘soul’

bnk bank bunuuk ‘bank’

The last type of ‘bending’ that we will mention here is . Suppletion is the morphological process that replaces one form with a completely different one in order to signal a grammatical . In English, the paradigm for the verb be is 116 CHAPTER FOUR

characterized by suppletion. Am, are, is, was, were, and be have completely different

phonological shapes. We also find suppletion with ; compare and me or she and her. Suppletion is most likely to be found in the paradigms of high-frequency words, as seen in the following table:

(6) Suppletion in some languages of Europe

Language

French aller ‘to go’, être ‘to be’ vais ‘go (1sg)’, suis ‘am (1sg)’

Spanish ir ‘to go’, ser ‘to be’ fue ‘went (1sg)’, fue ‘was (1sg)’

Finnish hyvä ‘good (nom. sg.)’ parempi ‘better’, paras ‘best’

Greek [Enas] ‘a, one (m.nom.sg.) [mja] ‘a, one (f.nom.sg.)’

Swedish ett ‘one’, två ‘two’ första ‘first’, andra ‘second’

We can look to historical linguistics for an explanation of why suppletive forms arise.

Take, for example, the paradigm of the verb ‘to go’ in French. It comes from three

different Latin sources. The , aller, and the first person and second person plural forms in the present, allons ‘we go’ and allez ‘you (pl.) go’, come from Latin ambula¤re ‘to walk, to walk along’. The stem of future and conditional forms, such as

irai ‘will go (1sg.)’, has evolved from the Latin verb ire ‘to go’. Finally, forms like

vais ‘go (1sg.)’ or vont ‘go (3pl.)’ come from Latin vadere ‘to go, to walk’. Thus we

see that the idiosyncracies of languages today can often be explained by looking at the

languages of yesterday.

In certain cases, such as with catch~caught or think~thought and other

like them in English, it is most convenient to use the term partial suppletion. In these

cases, the initial or of the word remain the same, but there is both INFLECTION 117

internal change and change to the end of the word (loss of segments and addition of a

indicator [t]).

To summarize this section, ‘bending’ a lexeme here simply means ‘changing

shape’. That’s all it means. Any change in form will count as inflection, whether it

involves affixation or not.

4.1.2 Morphological vs. syntactic inflection

The presence of inflectional morphology in a language depends on the

existence of multiple forms of a lexeme. From a morphological point of view, if a

lexeme has only one form, then you can’t get inflection. Take Chinese, a famous case

of an uninflected language. Chinese lexemes have only one form, abstracting away

from phonologically determined alternations (mostly changes in ). While Chinese

has a few or particles, including one that expresses past tense, these are

generally not considered to be affixes. The same is true of Vietnamese, though the

two languages are unrelated. For the morphologist, therefore, these two languages

have no inflection. Traditionally, they are described as isolating.

From a syntactician’s point of view, whether or not Chinese and Vietnamese have inflection is an entirely different matter. Even if a language does not express a particular notion such as number or case, it is typically assumed to be present in the syntax. So a syntactician may argue that noun in a particular language raise to check features, even though nominative case is not realized overtly.

But from a morphologist’s point of view, if a particular notion is never realized overtly in a given language, then it cannot be called inflection in that language.

A second condition for inflection is that the realization of a given category must be obligatory. What do we mean by that? Let’s look at a language that seems to 118 CHAPTER FOUR

pay attention to a particular category but does not inflect for it. Indonesian expresses

plural by , as shown in (7) (examples from Sneddon 1996:16). It

certainly appears as if this language has an inflectional plural for nouns:

(7) kuda-kuda ‘horses’

rumah-rumah ‘houses’

singkatan-singkatan ‘abbreviations’

perubahan - perubahan ‘changes’

There is a catch: The forms in (7) are not obligatory. Speakers of Indonesian have the

option of using the unreduplicated form to refer to either singular or plural. So kuda

not only means ‘horse’, but also ‘horses’; rumah can refer to one house or more than

one, and so on. The reduplicated plural is most likely to be used when the number of

the noun is not clear from the context, as in the examples below (Sneddon 1996:17):2

(8) a. Rumah-nya dekat pohon-pohon mangga itu

house- 3SG.POSS near tree- REDUP mango that

‘His house is near those mango trees’

b. Pada pinggang-nya terikat bumbung- bumbung kosong

LOC waist- 3SG.POSS PASS.tie water.container-REDUP empty

‘At his waist are tied empty bamboo water containers’

Without reduplication of pohon, (8a) would be ambiguous between ‘His house is near

that mango tree’ and ‘His house is near those mango trees’; likewise, if bumbung

were not reduplicated in (8b), the sentence could have the interpretation, ‘At his waist

is tied an empty bamboo water container’, as well as the one given above.3

It is possible to say that Indonesian has the category plural in the syntax, but

because speakers are under no obligation to express it morphologically, it is not an INFLECTION 119 inflectional of the language. We only talk about inflection when speakers of a language are obliged to express that category morphologically. In English, plural is an inflectional category because its expression is obligatory. Even when it is obvious that a noun is plural, such as when it is preceded by a plural word, the noun must have a plural form. We must say I own two dogs, never *I own two dog.

So far we have argued that for something to be classified as morphological inflection, it has to be overt and obligatory. There is a third condition. Since we have characterized inflection as bending the shape of a single lexeme, we talk about inflection only when dealing with bound forms. Let’s go back to English. English has a of modals like can, may, must, so called because they express moods. Should we consider these modals to be inflection? The answer is no. In order to be classified as morphological inflection, a syntactic category must be expressed through bound forms. In the case of the English modals, we are not dealing with bound forms, but rather with separate words. Again, syntacticians differ from morphologists on this point. Many treat auxiliaries as part of the inflectional system of a language.

4.1.3 A Disease Model of morphology

It is reasonable to look at morphology as a disease. A few languages escape it entirely; some languages have minor cases of the disease; in others it is pervasive. It’s when dealing with the third group that we can talk of languages ‘living with morphology’. For example, it turns out that the languages of the Athapaskan family, such as Navajo, Apache, Hupa, Chipewyan, or Sarcee, have been living with very bad cases of morphology for thousands of years. Yet somehow they’ve managed to survive. 120 CHAPTER FOUR

So what do we mean by some languages having morphology worse than

others? Let’s look at English inflection. And let’s ask how much inflection English

really has. Nouns have only two forms. They have a bare form, like umbrella, and a

suffixed form, like umbrellas. Some people will tell you that English nouns have three forms, singular, plural, and , but don’t be fooled: the plural, the possessive,

and the plural possessive are identical in speech (though in writing, we use an

, which is an innovation dating from Shakespeare’s time, to distinguish the

possessive form). We call this lack of morphological distinction among syntactically

differentiated forms inflectional homophony. Almost every English verb has four

forms: X, X-s, X-ed, and X-ing. A few, like break, have five forms because the preterite (in this case, non-affixal) and the past have distinct forms (broke, broken). A few very common verbs have an irregular third person singular form (has, does, says).4 And one, be, has eight (be, am, is, are, was, were, been,

being). English pronouns have four forms, for example, I, me, my, mine. All in all,

English has a fairly trivial case of inflection.

An intermediate case might be German. An average German noun has from

two to four forms, while have a few more. German verbs may have up to

about fifteen forms. Latin has a much worse case of morphology, with rich

inflectional paradigms for nouns, pronouns, adjectives, demonstratives, and verbs. A

typical Latin verb, depending on how you count, might have about 125 forms. Your

average Navajo verb, with inflectional marking for , , as well as tense,

mood, or aspect, has approximately 4,000 forms. All that we are trying to point out is

that some languages have rich inflection, and others, like English, have rather poor INFLECTION 121

inflection, from a morphologist’s point of view. Again, from a syntactic point of

view, it’s a whole different story.

Recall that we said that lexemes bend to fit syntactic frames. What kinds of things do they express? In general we speak of lexemes expressing morphosyntactic information. This includes the abstract syntactic categories of tense, aspect, number, and case. Specific values for these categories, such as past, imperfective, plural, or genitive, are generally referred to as morphosyntactic features or morphosyntactic properties, the latter a term from Matthews (1991). Syntacticians sometimes call these things , even when not overtly realized. For example, a syntactician might refer to the present in Chinese. Here, however, we will stick to our terminology and call them properties or features. Inflection expresses these properties or features by changing the shape of the lexeme.

4.1.4 Inflection as morphological realization

Linguists often talk about inflection as the phonological realization of

morphosyntactic properties. In some sense, inflection makes these properties

detectable by giving them a physical manifestation — it turns them into actual

acoustic signals.

To return to English, mood — expressed by modals like can, may, and must

— is an abstract syntactic category all right; it is being realized, and it is obligatory.

So we are not talking about Chinese here, or Indonesian. But note that it is not being realized morphologically, through the bending of a lexeme. From the morphology’s point of view, this is not inflection. From the syntax’s point of view, it is.

We will just mention in passing some interesting problem cases. One involves

languages where inflection is obligatory in some contexts, but impossible in others. In 122 CHAPTER FOUR

Jóola we will see (cf. 4.2.2) that subject is obligatorily expressed, except in the past subordinate and positive imperative forms of the verb. In some related languages, like Balanta, however, verbs agree with their subject in some contexts, but not in all. In Balanta, verbs may be marked for subject agreement, but generally only in the absence of a subject noun (9a). When a subject is expressed, a subject on the verb indicates that the subject is focused (9b) (data from N’Diaye-Corréard 1970: 30):

(9) a. b´- ≠a¤aN be¤nte

CL2-people come

‘The people came’

b. b´- ≠a¤aN b´- do¤olo! b´- beeTa ma

CL2-people CL2-few CL2-see 3SG.OBJ

‘A FEW PEOPLE saw him.’

Another interesting case is the Central Khoisan language //Ani, which displays an interesting system of object agreement (Vossen 1985).5 Finite verbs (except in the imperative) bear affixes that agree with a pronominal object in person, , and number (10a) or with a object for number and gender (10b):

(10) a. ti$ tsa! mu)-tiŸ- teŸ

me you see-1SG-PRES

‘You see me’

b. gúénì=kÓòè-//ùà ?xóà- mà- /à !xòè-!xòè - `m- tè

hunter- M.PL elephant-M.SG-OBJ run- REDUP -M.SG.OBJ-PRES

‘The hunters make the elephant run’ INFLECTION 123

The catch is this: if a nominal object is not marked for gender and number, object

agreement does not appear on the verb:

(11) gúénì=kÓòè-//ùà ?xóà !xòè-!xòè- tè

hunter- M.PL elephant run- REDUP -PRES

‘The hunters make the elephant run’

Here, in contrast to (10b), ‘elephant’ does not bear a gender-number suffix, and object agreement morphology fails to appear on the verb.

Finally, a well-known case is that of Arabic. Here the basic generalization is that marking for subject agreement appears on the verb when the is SV

(subject-verb) (12a) but not when it is VS (12b) (data from Ouhalla 1994: 43):

(12) a. l- tullaab- u wasal- uu

the-students-NOM arrived-3PL

‘The students have arrived’

b. wasal- a l- tullaab- u

arrived-3SG the-students-NOM

‘The students have arrived’

In (12b) third plural subject agreement is blocked, and instead we get default third

person singular agreement.

How should we deal with cases like these? It would be misleading to call

Balanta or Arabic subject agreement and //Ani object agreement ‘optional’ as we did

for the expression of plural in Indonesian, since here realization of agreement is either

obligatorily present or obligatorily absent, depending on context. Because there are a

number of possible analyses of cases like these, we will not propose solutions here.

We simply acknowledge that tricky cases like these exist. 124 CHAPTER FOUR

One question we can ask is whether inflectional categories are universal. For

syntacticians working in the Chomskian framework, all languages are theorized to be

identical at some level of analysis, and so the push is for inflectional categories to be

seen as universal. At some abstract level, then, all languages have tense (the same

tenses) and all languages have aspects (the same aspects). Only the realization differs.

This is stated explicitly by Chomsky (1995: 169-70), conjectures that “beyond

PF options and lexical arbitrariness . . . [linguistic] variation is limited to

nonsubstantive parts of the and general properties of lexical items.” This is

an extreme hypothesis, as Chomsky himself acknowledges (p. 170), and it probably

does not work.

Let’s take case. Do all languages have case? Abstract Case, maybe. But how

about morphological case? Let’s look at Jóola. Have we seen any case markers

running around? No — there are none. We might infer one of two things from this:

(13) a. Kujamaat Jóola has no case.

b. Kujamaat Jóola has case, but only secretly.

(In other words, (13b) suggests that case is present, but somehow it is not realized

morphologically.)

Now let’s compare Kujamaat Jóola to English. Does your average English

noun have case forms? We might distinguish between a possessive case (e.g.,

linguist’s) and a default case (e.g., linguist). But it turns out that English pronouns

give evidence for one more case, which is the object case:

(14) nominative: I

possessive: my

object: me INFLECTION 125

The point here is that because English pronouns inflect for case, you can plausibly argue that nouns are also marked for nominative or objective case underlyingly, although the surface realization is the same for both. In Kujamaat Jóola, however, such an analysis could only be justified on theory-internal grounds, since there is never any morphological evidence in the language for it. How could a child learning

Kujamaat Jóola learn the case distinctions in the absence of evidence?

Maintaining the view that all languages are the same underlyingly gets more complicated if we look beyond English. Some languages, like German, have the . Do we want to say that English does, too? We could say that English has the dative case, but that it is always equivalent to the object case. Their homonymy is an accident in English. Let’s look beyond German, at the . The

Slavic languages have an . Does English have an instrumental case?

“Well, yes,” we might say. It’s just that the instrumental case is equivalent to the dative case which is equivalent to the object case. Well, what about the , which we find in Latin? And what about the , or any one of the

(depending on who you talk to) twenty-three to twenty-seven cases in Finnish? Do we want to say that all languages have all of the case forms we find in Finnish — we just don’t see them?

The same is true with tenses and aspects. Morphologically, English has two tenses, which we might loosely refer to as the present and the past: break(s), broke.

We will see in 4.2 that Jóola, on the other hand, has five morphologically distinguishable tenses: the (recent past or present), future, simple past, remote past, and past hypothetical. Do we want to say that English has all of these, too, but only at some abstract level? 126 CHAPTER FOUR

For now we are going to put this question aside. From a purely selfish morphological point of view, the only abstract syntactic categories that matter are those that are realized morphologically. So we can assume that may be inflected for only two tenses.

We would like to return to a topic we discussed earlier, one that is highly problematic from a universalist point of view — gender (cf. 2.2). We will see in the second half of this chapter that gender is manifested in agreement classes for nouns and therefore that the gender specification of the noun itself is an abstract notion. We might refer to a Kujamaat Jóola noun as belonging to class 5, although in reality, people almost certainly don’t run around with little numbers in their heads.

Let’s look at a language that has three , masculine, feminine, and neuter. German is such a language. Examples of nouns belonging to each gender are given in (15):

(15) Masculine Feminine Neuter

Mensch ‘man’ Frau ‘woman’ Parlament ‘parliament’

Tag ‘day’ Lüge ‘lie’ Messer ‘knife’

Zuschlag ‘surcharge’ Erde ‘earth’ Mädchen ‘girl’

Masculine, feminine, and neuter are obligatory inflectional categories of German.

What does that mean? It means that every noun in the language, including borrowings like Parlament ‘parliament’, must belong to a gender. A noun cannot be genderless. Furthermore, gender is obligatory in that a noun cannot simply carry it around: its gender category must be expressed through agreement. INFLECTION 127

Now let’s compare German to Ojibwa, an Algonquian language discussed by

Corbett (1991: 20-22). Ojibwa has two genders, animate and inanimate (but note that some inanimate objects have grammatically animate forms):

(16) Animate Inanimate

enini ‘man’ essin ‘stone’

enim ‘dog’ peka…n ‘nut’

mettikumi…šš ‘oak’ pekkwe…šekan ‘bread’

a…kim ‘snowshoe’ wa…wan ‘egg’

a…sso…kka…n ‘sacred story’

meskomin ‘raspberry’

These function in the same way as the German genders in that they are obligatory and must be expressed. Let’s compare German and Ojibwa to Kujamaat Jóola. Kujamaat

Jóola has nineteen noun classes. Every noun must belong to a in all of its occurrences, and this class must be expressed. (The noun class system of Kujamaat

Jóola was presented in detail in chapter 2.)

The question we should now ask is how the noun classification systems of

German, Ojibwa, and Kujamaat Jóola are related to one another. The obvious answer is that they are not related at all. The only thing they have in common is that for all of them gender is an obligatory inflectional category. Every noun must have a gender, and that gender must be expressed in the morphology.

The lack of correspondence between many of the world’s gender systems poses a problem for universal syntax. But if gender categories are not universal, what can we do with them? Under the disease model of morphology, we can look at them as a kind of cancer that grows on otherwise “normal” languages.6 This gives a bleak picture because it seems to say that there is no hope of ever relating inflection types in 128 CHAPTER FOUR

one language to those of others. But there is hope. If you look at lots and lots of languages, you discover that certain types of inflectional categories keep appearing over and over again. For example, when you look at nouns and ask what nouns may be inflected for, the answer is case, number, and gender. This is pretty much all you will find. Nouns in different languages will not necessarily inflect for the same cases

— some languages will have two and others twenty; nor will they inflect for the same genders. For verbs, the picture is similarly limited. Verbs might inflect for tense, aspect, mood, , or agreement (subject-object or ergative-absolutive), but you generally will not find languages where verbs inflect for other categories. Whether these categories are rooted in a universalist-type grammar or whether they derive from functional notions is an interesting question, but one that we will not speculate about here. Our point is clear: the same, or very similar, categories pop up over and over again in many of the world’s languages, just as the same diseases do in various groups of people. We will on questions like the following: Are these categories realized in the morphology of a given language? Are they systematic? Are they obligatory?

4.1.5 The mechanism of inflection

Let’s step back for a moment and situate inflection in the grammar. Assuming that the syntax provides some sort of abstract morphosyntactic representation, the job of the morphology is to get from there to the actual phonological realization:

(17) Morphosyntactic Morphology Phonological representation representation

INFLECTION 129

This diagram portrays the relationship between the syntax, morphology, and as derivational, but it is equally possible to model a non-derivational, parallel relationship. Either way, however, a diagram like (17) is bound to be deceptively simple. We are still left asking precisely how words become inflected. A lot of work in morphology and syntax has been devoted to this question. Here we are going to give a very simple but useful typology. Traditionally people talk about two different reasons for inflection: government and concord.

First we want to distinguish between inherent and assigned inflection. Let’s take gender. For nouns, gender is inherent. That means that nouns are marked as having a particular gender in the speaker’s mental lexicon. For any other lexical category (e.g., adjectives, verbs) that happens to agree with nouns for gender, gender cannot be inherent. It must be assigned. What about number? Is it is generally not inherent, although there exist some words, like scissors, with inherent number. In some languages, there are even verbs that occur only in the singular or in the plural.

An example of assigned inflection is case. Nouns in the lexicon do not have case.

They obtain case by virtue of their position in the sentence: for example, nouns in object position will surface with an objective case.

Once we talk about the difference between inherent and assigned, we can address the question of how categories may be assigned, which is generally in one of two ways: government or concord. What’s the difference? Concord occurs when a particular word in a sentence says to another, “Hey, why not be like me?” Another word for concord is agreement. Kujamaat Jóola nouns trigger concord — they each

‘tell’ adjectives that modify them to be like them in gender or noun class. We therefore describe adjectives as agreeing with nouns for gender. Similarly, verbs in

Kujamaat Jóola agree with their subject. In formal theories, concord or agreement is 130 CHAPTER FOUR

sometimes attributed to feature percolation or feature inheritance (see, e.g.,

Anderson 1992: 106-11). One element has a particular feature, which then percolates through the structure of the sentence to other elements that stand in a certain relation to the first element; these other elements thereby acquire the feature.

The other way in which a word can acquire a category is government, which is more or less what it sounds like: one word tells another how to behave. (The traditional notion of government gave rise to the Chomskian notion of government, but the two are somewhat distinct.) Case assignment by verbs is usually thought of in this way. It cannot be described as agreement because verbs themselves don’t have case. The same holds for prepositions. Instead, verbs and prepositions “tell” their objects to surface bearing a certain type of case-marking.

Note that we cannot talk about the morphosyntactic features themselves as being ‘government features’ or ‘concord features’. It might seem, for instance, that case should be described as a ‘government feature’ because nouns receive case under government by a verb or preposition. In (18), the noun object of the verb is in the , because the verb sehen demands that its direct object be accusative:

(18) Wir haben [NPden jungen Piloten gesehen (German)

we have [NPthe.M.ACC young.ACC pilot.ACC] seen

Accusative

‘We saw the young pilot’

The problem is that the definite den and the adjective jungen are usually thought to acquire this same case via concord with the noun. If this is the case, then the mechanism of inflection is independent of inflectional features.7 INFLECTION 131

We should briefly point out that there are other ways of assigning inflectional features besides concord and government. For example, let’s look at tense. How does a verb get tense? The simple answer in formal syntax is that it is governed by the tense phrase (TP), or something like that. But if we look at actual languages, this answer is not so satisfying because, from a morphological point of view, TP is just an abstract notion. Tense seems to be something that verbs sometimes just have; it is not assigned to them by either government or concord, by some other concrete element.

4.1.6 Exponence

Exponence, a term coined by Peter Matthews, refers to morphosyntactic realization. In the word seas, [z] is the exponent of plural, and in sailed, [d] is the exponent of past tense or past participle (Matthews 1991:175). In both cases there is a one-to-one relationship between form and meaning, a situation that Matthews calls simple exponence.

When we go beyond simple exponence, we get into some very interesting data that has been central to modern theories of morphology. One type is what Matthews first called cumulative exponence. These are cases where more than one morphosyntactic feature maps onto a single formative. The most famous case of this is one of Matthews’ original examples, Latin verbal . In the Latin first person singular present indicative active form, five features (person, number, tense, mood, voice) are spelled out with a single morph, -o@:

(19) cant-o@

sing-1SG.PRES.IND.ACT

‘I sing’ 132 CHAPTER FOUR

Another example is subject-object agreement in Cherokee, an Iroquoian language.

Verbs in Cherokee bear that agree with their subject and object in person,

number, and . What is most interesting in the present context is that some

prefixes, including those listed in (20), indicate both subject and object (Scancarelli

1987: 71):

(20) ski-, skw- 2SG.SUB/1SG.OBJ

sti… 2DU.SUB/3SG.INAN.OBJ

kaci…y 1SG.SUB/3PL.AN.OBJ

ci…y 1SG.SUB/3SG.AN.OBJ

Examples of subject/object prefixes in context are given below (tones are not marked;

/v/ is a nasalized central vowel). Following Scancarelli, verbs are given in both their

surface and phonemic forms, with the phonemic forms aligned with the glosses:

(21) a. sv…kthv kaci…ne…lv…/i (Scancarelli 1987: 68)

/kaci…y- /ne…lv…/i/

apple 1SG.SUB/3PL.AN.OBJ-give.PERF

‘I gave them an apple’

b. ci…ko…wthiha (Scancarelli 1987: 74)

/ci…y- ko…/wthiha/

1SG.SUB/3SG.AN.OBJ-see.PRES

‘I see him’

Lastly, inflection for case, number, and gender in many Indo-European languages involves cumulative exponence. The -os ending of the Modern Greek adjective kalós

‘good’ indicates that it is masculine, nominative, and singular. The -á of Russian stolá

‘table’ denotes both genitive and singular. INFLECTION 133

Related to cumulative exponence is the notion of a portmanteau (a term

coined by Lewis Carroll). In a portmanteau, what were historically two or more

distinct markers (and what may still be, in some contexts) are fused together. The

most famous case is in French. The French definite occurs in two forms, le

(masculine) and la (feminine). The feminine form can be preceded by the prepositions

à ‘in, to’ or de ‘of, from’, as shown below:

(22) à la plage ‘to/at the beach’

de la plage ‘from/of the beach’

The masculine form, however, may not follow either of these prepositions. Instead,

we get a portmanteau:

(23) au [o] marché ‘to/at the market’ (*à le marché)

du [dy] marché ‘from/of the market’ (*de le marché)

The existence of cumulative exponence is very important to a proper characterization

of the morphology-syntax interface. We see that fairly complex syntactic structures

may get reduced morphologically. Even more interesting, however, is that we also

find the opposite.

In extended exponence, a single morphological feature is realized

simultaneously on more than one form. Matthews’ most famous case is probably that

of the perfective. The verb elely!kete ‘you had unfastened’ (stem -ly-) is marked as perfective by reduplication (le-), -k- infixation, and the presence of a

special stem (-ly- versus -ly…-). We cannot single out any one of them — they mark the

together. In Kujamaat Jóola, deverbal nouns can be formed from some

by changing the noun class and tensing the vowels (24). One or the other

isn’t sufficient. This is another example of extended exponence: 134 CHAPTER FOUR

(24) a. E-ka…y ‘to divorce’ (a man by a woman)

bu-k´…y ‘a divorce’ (man by woman)

b. ka-kO≠En ‘to send a message’

ku-kO≠En ‘a message’

c. E-lOk ‘to cry (of an animal), bark’

bu-lok ‘an animal cry, bark’

The most complicated cases are those where we get a combination of cumulative and extended exponence. In Latin, the notion perfect is realized by having a special verb stem in addition to a special set of suffixes that encode person, number, and mood (we might also want to list voice, although the perfect is realized periphrastically in the passive):

(25) a. re¤x- ist¸¤

rule.PERF-2SG.ACT. PERF

‘you ruled’

b. rex-¤ e¤runt

rule.PERF-3PL.ACT.PERF

‘they ruled’

Here we are getting simultaneous cumulative and extended exponence: the mapping from the syntax to the phonology is both many-to-one and one-to-many.

4.1.7 Context-free vs. context-sensitive inflection

The last topic we are going to address in this chapter is the distinction between context-free and context-sensitive inflection. We refer to context-free inflection when there is a simple directional mapping between a morphosyntactic feature and a INFLECTION 135

particular phonological string. Imagine that English has a feature [PRESENT

PARTICIPLE] or [PROGRESSIVE]. Because this feature is always realized as /-ing/, we

refer to context-free inflection: all present in English bear the same

suffix.8 In context-sensitive inflection, the realization of a morphosyntactic feature

varies. For example, the feature [PAST] in English corresponds to several possible phonological realizations:

(26) a. Ablaut: ran, sat, won, drank, shone . . .

b. Suppletion was, went, thought . . .

c. Ø hit, cut, put . . .

d. /-t/ sent, lent . . .

e. /-d/ helped [-t], shrugged [-d], wanted [-´d]. . .

We even find combinations of the mechanisms in (26). Slept involves both ablaut

(26a) and /-d/ suffixation (26e). So inflection for past tense in English is context-

sensitive in the sense that the feature [PAST] is realized as many things depending on the lexeme it attaches to, with /-d/ suffixation being the default case, or the elsewhere rule. As you continue to look at morphological data from a variety of languages, you will discover that context-sensitive inflection is much more common than context-free inflection.

4.2 Agreement in Kujamaat Jóola

It is virtually impossible to discuss noun class without also discussing agreement.

While noun class, or gender, is a property of nouns, it can be reliably detected only by

looking at those words with which the noun enters into an agreement relation. We got a

hint of this in 2.2 where we learned that certain Kujamaat Jóola words, such as mbur

‘bread’ and dakar ‘Dakar’, bear no noun class prefix, and others, such as (e)jimukor

‘lion’ and (E)bEkan ‘bicycle’, bear one only optionally. Yet these nouns always have

gender.

The set of words with which a noun enters into an agreement relation varies from

language to language. In Kujamaat Jóola this set includes the categories in (1):

(1) a. Definite articles

b. Pronouns

c. Nominal modifiers (demonstratives, cardinal and ordinal numbers,

particularizers, adjectives)

d. Verbs

e. Relative pronouns

f. Genitive markers

In this section we present some of the forms that agreement in Kujamaat Jóola may take.

Definite article

The Kujamaat Jóola definite article has the form -aC, where C is identical to the

consonant of the class prefix. Three noun classes have prefixes that consist of a single INFLECTION 137 vowel: class 1 (a-), class 3 (E-) and class 8 (u-). For these, the definite article has the

form -aw, -Ey, and -aw, respectively. Examples from Sapir (1965: 68) are given in (2):

(2) Cl. 1 a-≠il-aw 'the child'

Cl. 2 ku-≠il-ak 'the children'

Cl. 3 E-yEn-Ey 'the dog'

Cl. 4 si-yEn-as 'the dogs'

Cl. 5 fu-ti…k-af 'the war'

Cl. 6 ku-ti…k-ak 'the wars'

Cl. 7 ka-si…n-ak 'the horn'

Cl. 8 u-si…n-aw 'the horns'

Cl. 9 bu-b´…r-´b 'the tree'

Cl. 10 ji-sEk-aj 'the small woman'

Cl. 11 mu-sEk-am 'the small women'

Cl. 12 ≠i-sEk-a≠ 'the large women'

Cl. 13 ba-≠il-ab 'the many small children'

Cl. 14 fa-kOr-af 'the smoke'

Cl. 15 ma-kuk-am 'the brains'

Pronouns

The form of first and second person pronouns obviously do not depend on noun

class. Third person pronouns do, however. Both subject and object pronouns have the

shape CO when free (as opposed to bound), with C corresponding to the consonant of the

noun class prefix. The forms of third person pronouns for classes 1-15 are listed in (3): 138 CHAPTER FOUR

(3) Cl. 1: O- Cl. 8: wO-

Cl. 2: kO- Cl. 9: bO-

Cl. 3: yO- Cl. 10: jO-

Cl. 4: sO- Cl. 11: mO-

Cl. 5: fO- Cl. 12: ≠O-

Cl. 6: kO- Cl. 13: bO-

Cl. 7: kO- Cl. 14: fO-

Cl. 15: mO-

Nominal modifiers

While Kujamaat Jóola has a variety of nominal modifiers, including demonstratives, numbers, and particularizers (Sapir 1965: 27-8), there is no well-defined category that corresponds to Indo-European adjectives. This is not unusual among the world’s languages. In fact, throughout Africa, what might be considered canonical adjectives by speakers of Indo-European languages are rare (Welmers 1973). Instead, adjectival concepts like ‘small’, ‘beautiful’, or ‘angry’ are generally expressed by a verbal element. In Kujamaat Jóola, what at first glance appear to be adjectives are often formed with a verbal stem prefixed with a relativizer that agrees in noun class with the noun:

(4) a. E- yEn- Ey ya- gOn- O- E

3CL- dog- DEF3 3REL- be.mad- STAT- HAB

‘the mad dog’ (lit. ‘the dog which is mad’) INFLECTION 139 b. si- b´…- s sa- jak- as

4CL- cow- DEF4 4REL- be.good- DEF4

‘the good cows’ (lit. the cows which are good’)

The relativizer takes the form Ca- when modifying a subject, and Can- when modifying an object. We see it again in the forms in (5), where the nature of the verb makes it more obvious that we are dealing with relative . Note in (5b) that the object relativizer may stand on its own:

(5) a. ku- ≠il- ak ka- ri≠- ulOm

2CL- child- DEF2 2REL- arrive.from-SUBORD

‘the children who arrive’

b. E- lO…l- Ey yan ndaw a- sEk- O ´- pur- en

3CL- chicken- DEF3 3REL Ndaw 1CL- woman- POSS 1CL- leave-CAUS

‘The chicken that Ndaw’s wife brought out’

A second way to form what appear to be adjectives is to attach a prefix of the form (C)V- to what Sapir calls a ‘neutral theme’ — that is, a stem that can be used as a noun, verb, adjective, or depending on context. Again, the initial consonant of the prefix corresponds to the consonant of the noun class prefix. We see examples of neutral themes being used as adjectives in (6).

(6) a. si- jamEn- as si- lulum- ´s

4CL- goat- DEF4 4CL- European- DEF4

'the European goats' 140 CHAPTER FOUR

b. E- yEn e- lulum

3CL- dog 3CL- European

'a European dog'

c. e- be E- narE

3CL- cow 3CL-woman

‘a female cow’

Prefixes of the form Ca- (7 ka-; 13 ba-; 14 fa-; 15 ma-) take the form Cu- in this construction, perhaps to avoid confusion with the relative pronoun presented immediately above.

Demonstratives, which have the form uC(E), and particularizers, which have the

forms CV-kE(n) (indefinite) and CV-kila (definite), are illustrated in (7a-c). As we saw

with neutral themes, prefixes of the form Ca- surface as Cu-:

(7) a. si- jamEn- as use

4CL- goat- DEF4 4DEM

‘these goats’

b. ka- rEg ku- kEn

7CL- story 7CL- PARTIC

‘a certain story’ INFLECTION 141

c. e- b´- y E- kila

3CL- cow- DEF3 3CL- PARTIC

‘the cow (of which I am talking)’

Cardinal numbers up to ‘four’ and ordinals up to ‘fifth’ also agree with a head

noun (8-9):

(8) a. E- yEn y´- kon

3CL- dog 3CL- one

‘one dog’

b. u- b´…r u- ba…kir

8CL- tree 8CL- four

‘four trees’

(9) a. E- yEn- Ey E- tONOndEy

3CL- dog- DEF3 3CL- first

‘the first dog’

b. si- ≠ara…- s si- tOkEn- as

4CL- monkey- DEF4 4CL-fifth- DEF4

‘the fifth monkeys’

Higher numbers do not agree with a head noun (examples from Sapir 1970):

(10) butinken ‘fifteen’ (class 9 invariable)

kabanan ‘twenty’ (class 7 invariable)

ceme ‘one hundred’ (invariable) < Mdk. keme

In languages where numerals agree with a head noun, it is typical for agreement marking to be limited to the lower numerals. 142 CHAPTER FOUR

Verb Agreement

The last type of agreement we present here is that which holds between a verb and

its subject. As we saw with neutral themes, numerals, and particularizers, class prefixes

of the form Ca- surface as Cu-:

(11) a. e- munguno E- jum bo

3CL- hyena 3AGR- stop there

‘a hyena stopped there’

b. fu- gOl- af fu- lOlO tentam

5CL- stick- DEF5 5AGR- fall ground

‘a stick fell to the ground’

c. ba- suwa…-b bu- iit

13CL- bird- DEF13 13AGR- fly

‘the birds flew off’

A subject noun phrase does not have to be present in order for an agreement prefix to

appear on the verb. Verbs may agree with implied subjects, as the sentences in (11)

illustrate. In (12a), the verb jOl ‘come’ agrees with the class 4 subject si-jamEn-as ‘the goats’, as shown by the si- prefix. The verb appears with this prefix even when the

implied subject ‘goats’ is left unstated:

(12) a. si-jamEn-as si-jOl ‘the goats came’

b. si-gaba si-jOl ‘two came’

c. s-O si-jOl ‘they came’ INFLECTION 143

d. si-nifan-as si-jOl ‘the old ones came’

e. s-Eti alasan si-jOl ‘Alasanne’s came’

Even this brief sketch of agreement in Kujamaat Jóola makes it clear that agreement is a pervasive part of its grammar. It is impossible to ignore. As the sentences in (13) illustrate, the repetition of agreement markers gives the impression of alliteration:

(13) a. fu- gOl- af f- umb´ f- O- f- E

5CL- stick- DEF5 5AGR- 1POSS 5CL- here- DEF5- EMPH

‘my stick is here ~ this is my stick’

b. si- jamEn- as s- umb´ s- O- s- E

4CL- goat- DEF4 4AGR- 1POSS 4CL- here- DEF4- EMPH

‘these are my goats’

c. bu- b´…r- ´b b´- m´k- ´b bu- lOlO

9CL- tree- DEF9 9REL- be.big- DEF9 9AGR- fall

‘the big tree fell’ CHAPTER FOUR: EXERCISES

1. Identify the morphological process at work in each set of words, and think of at least one more English example to add to each set.

a) report/reported, grovel/groveled, purr/purred, saddle/saddled

b) goose/geese, foot/feet, louse/lice, eat/ate, run/ran

c) go/went, good/better, I/me, am/was

2. Using an etymological dictionary, investigate the reasons for the following suppletive pairs in English: (a) go/went; (b) good/better.

3. The following words do not have plurals or, if they do, their plurals have a special meaning. Describe the meaning of the plural forms of each word, if it exists. Then try to come up with an account of the behavior of these nouns as a group.

a) water

b) rice

c) fish

d) air

e) laughter

f) courage

g) hate INFLECTION 145

h) heat (i.e., hot temperature)

i) humidity

j) intelligence

10. Examine the following data from Tzeltal, a language of (Nida 1965: 100).

For each pair, indicate whether the formation is derivational or inflectional. Then list the characteristics that form the basis of your decision.

a. h-čamel ‘sick person’ čamel ‘illness’

b. šiwel ‘fright’ šiw ‘to be afraid’

c. lumal ‘land’ lum ‘earth’

d. mahk’il ‘lid’ mahk’ ‘to close’

e. awinam ‘your wife’ /inam ‘wife’

f. čenk’ultik ‘bean patches’ čenk’ul ‘bean patch’

g. h-/u/el ‘influencial person’ /u/el ‘power’

h. č’uunel ‘offering’ č’uun ‘to believe, obey’

i. k’abal ‘custody’ k’ab ‘hand’

5. Italian verbs

First, analyze the following data and identify all of the morphemes. Group suffixes denoting person into one set. Then determine whether each suffix or set of suffixes exemplifies simple or cumulative exponence.

146 CHAPTER FOUR

parlo ‘I speak’ parlavo ‘I spoke’

parli ‘you (sg) speak’ parlavi ‘you (sg) spoke’

parla ‘he speaks’ parlava ‘he spoke’

parliamo ‘we speak’ parlavamo ‘we spoke’

parlate ‘you (pl) speak’ parlavate ‘you (pl) spoke’

parlano ‘they speak’ parlavano ‘they spoke’

6. French adjectives

The adjectives in the first column are masculine, while those in the second are feminine. How are masculine and feminine adjectives differentiated in French? (You may ignore changes in vowel quality.) Outline a possible analysis.

a) mauvais [movE] mauvaise [movEz] ‘bad’

b) bon [bo‡] bonne [bOn] ‘good’

c) grand [gÂA‡] grande [gÂA‡d] ‘big’

d) froid [fÂwa] froide [fÂwad] ‘cold’

e) petit [p´ti] petite [p´tit] ‘small’

f) gros [gÂo] grosse [gÂos] ‘fat’

g) sot [so] sotte [sOt] ‘foolish’

h) premier [pÂ{mje] première [pÂ{mjEÂ] ‘first’

i) entier [A‡tje] entière [A‡tjEÂ] ‘entire’

j) frais [fÂE] fraîche [fÂES] ‘fresh’ INFLECTION 147

k) long [lo‡] longue [lo‡g] ‘long’

l) heureux [PÂP] heureuse [PÂPz] ‘happy’

m) gentil [ZA‡ti] gentille [ZA‡tij] ‘kind’

n) soûl [su] soûle [sul] ‘drunk’

7. Zapotec of the Isthmus (Mexico) (Nida 1965: 38)

Identify all the morphemes in the following set of data. List all morphemes that have allomorphs. Describe the distribution of all allomorphs having phonologically definable positions of occurrence. Finally, discuss the data in terms of exponence

(simple, cumulative, or extended).

a) geta ‘corncake’ sketabe ‘his corncake’ sketalu/ ‘your corncake’

b) bere ‘chicken’ sperebe ‘his chicken’ sperelu/ ‘your chicken’

c) do/o ‘rope’ sto/obe ‘his rope’ sto/olu/ ‘your rope’

d) yaga ‘wood’ syagabe ‘his wood’ syagalu/ ‘your wood’

e) di/idZa ‘word’ sti/idZabe ‘his word’ sti/idZalu/ ‘your word’

f) palu ‘stick’ spalube ‘his stick’ spalulu/ ‘your stick’

g) kuba ‘dough’ skubabe ‘his dough’ skubalu/ ‘your dough’

h) tapa ‘four’ stapabe ‘his four’ stapalu/ ‘your four’

148 CHAPTER FOUR

8. Spanish (adapted from Nida 1965: 124-5)

Describe how the third person singular present indicative of –er verbs in Spanish is formed from the basic stem on the basis of the following forms. You should describe the types of allomorphy that you encounter.

3sg present indicative Basic stem

a) pide /pide/ ‘asks for’ ped- /ped-/

b) sirve /sirbe/ ‘serves’ serv- /serv-/

c) vende /bende/ ‘sells’ vend- /bend-/

d) barre /bare/ ‘sweeps’ barr- /bar-/

e) come /kome/ ‘eats’ com- /kom-/

f) aprende /ap|ende/ ‘learns’ aprend- /aprend-/

g) decide /deside/ ‘decides’ decid- /desid-/

h) siente /siente/ ‘feels’ sent- /sent-/

i) miente /miente/ ‘lies’ ment- /ment-/

j) duerme /due|me/ ‘sleeps’ dorm- /dorm-/

k) pierde /pie|de/ ‘loses’ perd- /perd-/

l) vuelve /buelbe/ ‘returns’ volv- /bolb-/

m) mueve /muebe/ ‘moves’ mov- /mob-/

INFLECTION 149

9. Sudan Colloquial Arabic (Nida 1965: 41)

Identify as many morphs as the data indicate. List all morphs that have allomorphs. Describe the distribution of all allomorphs that have phonologically definable positions of occurrence. Although the nouns in the following problem consist of a vocalic pattern superimposed on a template (as was illustrated in the chapter), you should treat the stems as single morphemic units for the purposes of this problem.

a) kitabi ‘my book’ axuy ‘my brother’ b) kitabak ‘your (m.sg.) book’ axuk ‘your (m.sg.) brother’ c) kitabik ‘your (f.sg.) book’ axuki ‘your (f.sg.) brother’ d) kitabu ‘his book’ axuhu ‘his brother’ e) bitaba ‘her book’ axuha ‘her brother’ f) kitabna ‘our book’ axuna ‘our brother’ g) kitabkum ‘your (m.pl.) book’ axukum ‘your (m.pl.) brother’ h) kitabkan ‘your (f.pl.) book’ axukan ‘your (f.pl.) brother’ i) kitabum ‘their (m.) book’ axuhum ‘their (m.) brother’ j) kitabin ‘their (f.) book’ axuhin ‘their (f.) brother’

10. Hebrew

Describe how the following paradigms are formed. Use CV notation, where C stands for a consonant, and V for a vowel.

150 CHAPTER FOUR

Past a) yashavti ‘I sat’ katavti ‘I wrote’ b) yashavta ‘you (m.sg.) sat’ katavta ‘you (m.sg.) wrote’ c) yashavt ‘you (f.sg.) sat’ katavt ‘you (f.sg.) wrote’ d) yashav ‘he sat’ katav ‘he wrote’ e) yashva ‘she sat’ katva ‘she wrote’ f) yashavnu ‘we sat’ katavnu ‘we wrote’ g) yashavtem ‘you (m.pl.) sat’ katavtem ‘you (m.pl.) wrote’ h) yashavten ‘you (f.pl.) sat’ katavten ‘you (f.pl.) wrote’ i) yasvu ‘they sat’ katvu ‘they wrote’

Present a) yoshev ‘sit (m.sg.)’ kotev ‘write (m.sg.)’ b) yoshevet ‘sit (f.sg.)’ kotevet ‘write (f.sg.)’ c) yoshvim ‘sit (m.pl.)’ kotvim ‘write (m.pl.)’ d) yoshvot ‘sit (f.pl.) kotvot ‘write (f.pl.)

11. Now add the following Hebrew data to those provided in (8). How can we express the formation of the future? How does its formation differ from that of the past and present? (Note that x alternates with k in Hebrew depending on position. You should INFLECTION 151 assume that x in the forms below is the same underlying segment as k in the second column of example (8), but you do not have to account for its distribution.)

a) /extov ‘I will write’

b) tixtov ‘you (m.sg.) will write’

c) tixtvi ‘you (f.sg.) will write’

d) yixtov ‘he will write’

e) tixtov ‘she will write’

f) nixtov ‘we will write’

g) tixtvu ‘you (m.pl.) will write’

h) tixtovna ‘you (f.pl.) will write’

i) yixtvu ‘they (m) will write’

j) tixtovna ‘they (f) will write’

12. Chiluba, Luba-Lulua dialect (Congo) (Nida 1965: 39)

Identify all the morphemes in the following set of data. List all morphemes that have

allomorphs. Finally, describe the distribution of all allomorphs that have phonologically

defined positions of occurrence.

Supplementary information:

• Forms (a-c) are participial forms. 152 CHAPTER FOUR

• The allomorphic form of the first person pronoun in the second word of item (h)

is not characteristic of prevowel position, but should be treated as such in terms of

these limited data.

• The final –u of form (k) may be disregarded.

• The second forms in items (l) and (m) are past participles. The prefix mú- is a

person indicator, employed with all singular persons.

• The vowel /u/ can become [w] in certain situations, as in (i) and (j).

• The tonal part of a morph can spread to some contiguous segments.

a) n`sùmá ‘I, biting’ b) úsùmá ‘you (sg), biting’ c) ùsùmá ‘he, biting’ d) n`dí nsùmá` ‘I am biting’ e) údí úsùmá ‘you (sg) are biting’ f) ùdí ùsùmá ‘he is biting’ g) n`tSìdí n`sùmá ‘I am now biting’ h) n`tú Násùmá ‘I am always biting’ i) útú wásùmá ‘you (sg) are always biting’ j) ùtú wàsùmá ‘he is always biting’ k) m`básùmú ‘I bit (indefinite past)’ l) n`dí músùmé ‘I have bitten’ m) údí músùmé ‘you (sg) have bitten’ INFLECTION 153

13. Mongbandi (Congo) (Nida 1965: 63)

Determine the allomorphs of the morph that consists solely of suprasegmental features. Note that all forms are in the completive aspect and that the forms in the first column represent the basic forms of the roots. The tones on the forms in the first column are a fundamental part of the root.

Forms with singular subjects Forms with plural subjects a) Ngbò Ngbó ‘swam’ b) gwè gwé ‘went’ c) ma¤ má ‘heard’ d) kpe¤ kpé ‘fled’ e) yó yó ‘carried’ f) yé yé ‘agreed’ g) bàtà ba¤tá ‘guarded’ h) hùlù hulú¤ ‘jumped’ i) ha¤kà ha¤ká ‘taught’ j) d¸¤rì d¸¤rí ‘answered’ k) ko¤ló koló¤ ‘pierced’ l) s¸¤gí s¸¤gí ‘went out’ m) díkò d¸¤kó ‘read’ n) gbíNgà gb¸¤Ngá ‘translated’ 154 CHAPTER FOUR

1 Regarding the morphemic breakdowns of example sentences, we have consistently put

hyphens at the right edge of the leftmost element, rather than putting them at the right

edge of prefixes and left edge of suffixes. Therefore, hyphenation in this book is not intended to provide information about direction of attachment of affixes.

2 Word-for-word glosses for Sneddon’s examples were provided by Niken Adisasmito-

Smith.

3 In Indonesian the picture is slightly more complicated. Expression of the category

plural is ungrammatical in contexts like the following where plurality of the noun is

explicit, in (i) because of the presence of the number tiga ‘three’, and in (ii) because

shoes typically come in pairs (Sneddon 1996:16-17):

(i) Menteri mengunjungi tiga negeri (*-negeri) asing

minister visit.ACTIVE three country (*-REDUP) foreign

‘The minister visited three foreign countries’

(ii) Saya harus membeli sepatu (*-sepatu) baru

I must buy.ACTIVE shoe (*-REDUP) new

‘I must buy new shoes’

4 Says is spelled regularly, but the vowel represented by is pronounced differently

from the vowel in other forms of the same verb. Some British speakers do not have this

distinction.

5 // represents a lateral click, ! a palatal retroflex click, and ? an alveolar click. INFLECTION 155

6 We should point out here that there are some languages — Navajo is one — that do not make morphological gender distinctions at all. We might be tempted to say that English is a language of this type, but it is not: it distinguishes three genders in the third person singular pronouns (cf. he, she, it; him, her, it). Navajo does not even do this.

7 An alternative is to say that the entire DP receives accusative case under government and that the accusative case feature is distributed over all of its members.

8 Note that the mapping is directional. We cannot work backwards and pair up every instance of /-ing/ with the feature [PRESENT PARTICIPLE] or [PROGRESSIVE], because there is more than one type of ‘-ing’ in English (e.g., babysitting can be a present participle or a noun).