Building words: morphology

In the two previous chapters we have considered the basic building blocks of language (phonemes) and the ways in which they can be combined in speech. Our focus in this chapter and the next will be on grammar, which linguists have traditionally seen as comprising the two sub-disciplines of syntax, the arrangement of words in a sentence, which we explore in Chapter 6, and morphology, the internal structure of words, which we examine here. We begin by introducing the concept of the morpheme, which often proves a more helpful analytical tool than the word, a satisfactory definition of which proves elusive. We then review the traditional division between derivational morphology (or word formation) and inflectional morphology (the marking of grammatical categories), and look at attempts to classify languages on the basis of their morphological structure. We close the chapter by looking more closely at grammatical categories such as number and gender and their values, which prove extremely variable cross-linguistically and often differ greatly from those which are familiar to us as English speakers.

Words and morphemes

Languages differ more in morphology than in syntax. The variety is so great that no simple scheme will classify languages as to their morphology. One such scheme distinguishes analytic languages, which use few bound forms, from synthetic, which use many. At one extreme is a completely analytic language, like modern Chinese, where each word is a one-syllable morpheme or compound word or phrase-word; at the other, a highly synthetic language like Eskimo, which unites long strings into single words (…) This distinction, however, except for cases at the former extreme, is relative; any one language may be in some respects more analytic, but in other respects more synthetic, than some other language.’(Bloomfield 1933: 207)

Key idea: Morphology: the study of word structure

Derivational morphology is concerned with the creation of new words.

Inflectional morphology involves the marking of grammatical categories (for example number, tense or gender).

The distinction between morphology and syntax as sketched above would seem to presuppose a clear definition of the term word but, as we shall see, ‘words’ prove problematical on a number of levels. If asked ‘how many words there on this page?’, we’d probably count the number of items with blank spaces either side (would contractions like that ‘we’d’ in the previous clause count as one word or two?) and ignore the fact that some words (e.g. ‘the’) are used more than once. But if asked instead how many words there are in the Oxford English Dictionary, or how many words of Spanish we know, our criteria would change: ‘words’ in this case would imply different words as cited independently in the dictionary, and would not include inflected forms predictable by rule, i.e. we would treat dog/dogs or read/reads/reading as the same ‘word’ in each case. i.e. dog+s dogs, or read+ing reading.

For word in this second sense, linguists generally use the term lexeme, and the scope of lexemes includes items made up of more than one word (e.g. set up; windscreen wiper), and even idioms like to penny pinch, to keep tabs on, where the meaning cannot be broken down into the component parts. Where the distinction is important, by convention linguists use small capitals to refer to lexemes, so READ refers to the verb to read and all its inflected forms.

To add yet a further complication, words can have more than one function: in a sentence like ‘He is thinking about Mary’ the word thinking is a verbal participle, indicating an action which is ongoing; in ‘Bill was overly fond of thinking’ the same form is a gerund, i.e. it functions as a noun (we could substitute, for example, ‘football’ or ‘jam’ for ‘thinking’ and the sentence would remain grammatical). Here we need to distinguish a third sense of word, i.e. a grammatical or morphosyntactic word.

Even allowing for these qualifications, deciding what counts as a ‘word’ and what does not proves surprisingly tricky. The definition we routinely apply in everyday life, namely calling something a ‘word’ if it is separated by orthographical spaces on the page, is unsatisfactory on a number of counts. Firstly, this seems an arbitrary basis for definition in the absence of pre-existing criteria for separation: these ‘gaps’ do not after all correspond to pauses or breaks that are actually made in speech (if they did, there would be no need for the comma I’ve just used to indicate that a pause should be made, nor for the full stop with which this sentence will end). Secondly, as we saw in Chapter 1, only a minority of present and past languages have a writing system in any case, so in most instances we have no orthographic conventions to help us.

One criterion offered by Bloomfield is that a word should be a minimal free form: John, houses, riding, hopeless, for example, all qualify as ‘words’ because they could occur as one-word answers to a question (‘What are you doing?’ – ‘Riding’; ‘How’s your arithmetic?’ – ‘Hopeless!’). But this poses problems because some items which we would probably like to think of as words fail this criterion. In English, these would include ‘functional’ items such as the articles a and the, or the subject pronouns:

• Who’s there?

\- Me.- \*I

More awkward still from this point of view is French, which requires nouns to have articles, so for example ‘I want bread’ is ‘Je veux du pain’: on the minimal free-form criterion not only pain (bread) but almost all nouns would be ruled out. A more promising criterion is separability: the dog should be seen as a sequence of two words because adjectives, for example, can be interspersed between them, e.g. the great big lovable old dog. But this criterion proves no more watertight: broad beans, for example, looks like two words because broad and beans can both occur independently in other contexts (it’s as broad as it’s long; ‘Mum! I managed to sell the cow for some magic beans!’), but we cannot separate the two elements (\*broad big beans;? how broad are your beans?^^), suggesting that they form a single lexical unit.

Furthermore, in informal language, we do encounter interspersed elements at points where there is clearly no word boundary (abso-bloody-lutely!). A final criterion might be stress: blackbird, for example, is one word rather than two because, unlike black bird, it carries only one main stress. This will work for a whole range of items, including thorough, achieve, resist, but would rule out a number of items which would not normally bear stress, for example a, the, he, it, and so on, and in any case, not all languages have word-level stress: stress in French for example is borne by the last syllable of the rhythm group, which may consist of several ‘words’ on other criteria.

The criterion of ‘potential pause’ has even been advanced, unconvincingly, to align linguistic and orthographical word criteria: while we do not normally pause between words, we could – potentially – do so (as I’ve attempted to indicate here by dashes). But this criterion soon proved to be an impostor: how can we distinguish a hesitation in mid-word from a pause at a word boundary? Only, of course, by knowing in advance where the word boundaries are, in which case the definition becomes circular.

Spotlight: Human lexeme-inators?

Derivational morphology is part of a native speaker’s linguistic knowledge. An understanding of its functioning allows speakers to be extraordinarily creative in coining new terms. Evil Dr Heinz Doofenshmirtz, from Disney Channel’s Phineas and Ferb is not only a prolific inventor of (woefully ineffective) devices, he’s also a great coiner of new (and very short-lived) nouns using the suffix -inator (cf. terminator), e.g. audience controlinator, drillinator, media erasinator, giant baking-soda volcanoinator. Children’s ability to recognize ‘inator’ in this context as a nominalizing suffix with the meaning ‘device used to achieve a specified aim’ makes these unfamiliar words readily comprehensible even though they’re unlikely to make it into any dictionary, or indeed survive longer than a single episode.

By whatever criteria we apply, then, some meaningful linguistic items look more like ‘words’ than others: for this reason it is often more productive to look at meaning-bearing elements or morphemes. A word like internationalization, for example, seems naturally divisible into five elements:

inter+nation+al+iz+ation

The second, [naʃ], derives from a free morpheme, namely the noun nation [nεIʃ] which can occur independently (a powerful nation, etc.). The rest are bound morphemes, which can only occur as parts of bigger units and not on their own: inter- is a prefix conveying the notion of ‘between’ in a range of adjectives (interactive, interpersonal, interplanetary), verbs (interpose, interact) and nouns (interpol, interface); -al is a grammatical suffix frequently used to derive adjectives from nouns (structural, financial, orbital); -ize/ise is a verbal suffix used to derive verbs, while -ation is an abstract noun suffix (rationalization, penetration, realization). Morphemes, then, are minimal meaning-bearing units, uniting an arbitrary form and meaning or grammatical function. As we have seen, a distinction is usually made between inflectional morphemes and derivational morphemes.

Derivational morphology

All living languages need constantly to renew and update their lexical stock. They may do so in two different ways: the first, lexical borrowing, involves taking words from another language and assimilating them according to the phonological and morphological rules of the ‘borrower’ language. All of these English words, for example, have been borrowed from other languages: robot (Czech); shampoo (Hindi); kangaroo (Guugu Yimidhirr: North Queensland, Australia); entrepreneur (French); rucksack (German). They are, for the most part, now so well assimilated that we no longer notice that they are borrowings: indeed, some estimates suggest that around 30 per cent of English words are ultimately of French or Norman French origin, the vast majority of which pass unnoticed.

Key ideas: Derivation

•  Languages may either borrow new lexical items from other languages, or create them from existing resources (derivation).•  Derivation involving bound morphemes is called affixation; word creation using free morphemes is known as compounding.

Alternatively, new words can be created or derived using the language’s existing lexical resources, whether this involves bound morphemes (affixation), or free ones (compounding); the lexical resources deployed are known as derivational morphemes. The morphemes most commonly used in affixation are prefixes and suffixes. English is rich in both:

Table 6.1: Some English affixes

Spotlight: Gates and burgers

The creative potential of derivational resources is such that affixes may emerge from the most unlikely sources. At the time of writing the UK government was embroiled in a difficulty labelled ‘Plebgate’ in the popular press, in which the suffix -gate was attached to the derogatory term ‘pleb’ (plebeian). This followed Irangate, Sachsgate, Bigotgate, Debategate, Dianagate and numerous other ‘-gates’ in which a scandal, and the subsequent alleged cover-up, have made headline news.

There’s no connection, obviously, between ‘gate’ meaning ‘opening or door’ and the notion of scandal or cover-up, but the association was cemented by the Watergate bugging scandal which eventually brought down US President Richard Nixon in 1973, since when the suffix ‘gate’ has been applied to any number of scandals. Plebgate involved accusations that a minister called a police officer a pleb in Downing Street when told he was not allowed to take his bicycle through the Downing Street gate, which led to the affair being dubbed, perhaps inevitably, Gategate by some commentators. (For an extensive list of -gate scandals, see http://en.wikipedia.org/wiki/List\_of\_scandals\_with\_%22-gate%22\_suffix.)

The suffix burger has a longer history, and is based on a false derivation ham+burger. In fact, as any schoolchild knows, hamburgers contain beef, not ham, and the original derivation in German was Hamburg+er, ‘native of Hamburg’, and by extension the hot snack associated with that city. But the misderivation has spawned a range of new lexemes, including cheeseburger, veggieburger, chickenburger and beanburger.

English generally does not use infixes, inserted within words, but there are some informal expletive or emphatic uses, e.g. a-whole-nother story, abso-bloody-lutely. In Russian, the verbal infix -vy- carries the nuance that an action happens on a regular basis, e.g. arestovac (to arrest); arestovyvac (to arrest repeatedly). Note that these affixes are subject to selectional restrictions: un-, for example (see Spotlight on previous page), can be used with adjectives (unclear, unreasonable) and verbs (unfasten, undo), but not with nouns (barring one or two marginal exceptions, such as unconcern). The comparative and superlative suffixes -er and -est can normally only be used with gradable antonyms (e.g. warm-warmer-warmest, but not pregnant-\*pregnanter-\*pregnantest), and not with adjectives of more than two syllables (\*marvellouser, \*incrediblest).

Prefixes in English, unlike suffixes, almost never change word class: there are a few, generally unproductive, exceptions such as a- which derives the adjectives ablaze, awash and abuzz from the nouns blaze, wash and buzz, and the verbal prefix en- in enrage, enamour, entangle. Not all languages restrict the functions of prefixes in this way. In Zulu, for example, the prefix um-, coupled with a change of final vowel, is used to derive nouns from verbs:

Table 6.2: Affixal nominalization in Zulu

Spotlight: Un-stoppable?

George Orwell’s famous novel Nineteen Eighty-Four presents a world in which the state conditions thought through control of language. The hero Winston Smith is charged with rewriting documents in ‘Newspeak’, a version of English in which the words to express opposition to the all-powerful Big Brother simply do not exist. Part of this involves removal of antonyms; thus ‘bad’ becomes ‘ungood’ and heretical statements beyond ‘Big Brother is ungood’ become all but impossible:

‘After all, what justification is there for a word which is simply the opposite of some other word? A word contains its opposite in itself. Take "good", for instance. If you have a word like "good", what need is there for a word like "bad"? "Ungood" will do just as well — better, because it’s an exact opposite, which the other is not.’

Fortunately, language is far too complex, and human beings far too creative in its use, for top-down control ever to be possible. Ungood may never have caught on, but uncool has, and unfriend, while ruled out as a noun, has emerged as a verb with the advent of Facebook. What Orwell appears to have grasped is the flexibility with which English speakers apply this prefix to create antonyms:

•  But I’ve already booked a table!- Well, unbook it!

I have even heard a speaker, known for his fondness for British understatement (litotes), observe that: ‘this actor doesn’t unremind me of a young Robert Redford’!

Our derivational inventiveness stems in part from the fact that we associate morphemes with a particular meaning, and learn those meanings in the same way we do those of full lexemes. But there is an interesting subset of derivational morphemes with no apparent meaning outside the isolated lexemes in which they occur. These have become known as cranberry morphemes after their most celebrated example:

• straw+berry

• black+berry

• goose+berry

• blue+berry

• cran+berry

All of these cases appear to involve compounding of free morphemes, but cran has no independent meaning or function outside of the word cranberry. A slightly more marginal case is lukewarm, in which the first element luke- appears to qualify warm and is thought to derive from a Middle English word meaning ‘tepid’, but has no such meaning in any other lexeme.

Children learn not just a list of derivational morphemes, which enables them to understand new words like giant-baking-soda-volcano-inator (see the Spotlight box earlier in this chapter), but also the rules by which they may be combined. A child needs to know not just that prefixes can only be placed at the beginning of the word and suffixes at the end, but also that affixes attach to particular kinds of word-class.

The word uncontrollableness, for example, can be divided into four morphemes un+control+able+ness, but these morphemes have an internal constituent structure and are not simply juxtaposed. Control needs to combine first with able to form controllable, because the prefix un- can only attach to adjectives (unwary) and verbs (undo), but not nouns, ruling out \*uncontrol. The same restriction applies to the abstract noun controllableness, which suggests that, in spite of the fact that controllableness does exist, the proper constituent structure is uncontrollable+ness rather than un+controllableness. This can be illustrated by the following tree diagram.

Figure 6.1: Derivational structure of uncontrollableness

In the above example, morphologists would distinguish between root and stem morphemes: the root noun (in this case) around which uncontrollableness is built is control, which is also the stem of controllable. But controllable itself is the stem of uncontrollable, and likewise uncontrollable is the stem that yields uncontrollableness.

While many aspects of derivational morphology reveal regular patterns, much has to be learned on an item-by-item basis. In the example above, the meaning of the -ness suffix used for coining abstract nouns is broadly synonymous with that of -ity, and many speakers prefer uncontrollability to uncontrollableness (prescriptive dictionaries allow both). A quick Google search gave around 25,000 hits for uncontrollableness, but 241,000 for uncontrollability, but unfathomableness gave 41,600 hits as opposed to only 10,700 for unfathomability. The same highly unscientific test suggested a preference for unremarkableness over unremarkability but a strong preference the other way for predictability over predictableness. Similarly, there is no obvious reason why the antonyms of complete and capable are incomplete and incapable while those of conscious and comfortable are unconscious and uncomfortable: this is simply an arbitrary fact about present-day English.

Derivational morphology reveals many grey areas in which form or meaning can vary and change. The suffix -phobia, for example, has acquired a generally pejorative meaning in xenophobia and homophobia which it lacks in claustrophobia or agoraphobia, and while some speakers insist on the dictionary distinction between disinterested and uninterested, for many others the two words are now synonymous.

Inflectional morphology

We saw in the definition of lexemes above that in some cases different words are related by rule, and perceived to be forms of the same word. A dictionary would not, for example, list book and books separately, because the latter can be formed from the other. Furthermore, the meaning is entirely predictable: if you know what book is then you know what the plural form books means.

This kind of morphology, in which words are modified to express grammatical categories, is known as inflectional morphology. It does not involve the creation of new words, and the markings involved are subject to grammatical rule. Nouns, or more precisely count nouns, in English are marked for number, having generally a singular and a plural form, even if some of these forms are irregular (child, for example, has an irregular plural form: one child, many children). English verbs may be marked for tense, aspect and person: for the verb decide, for instance, we can identify four separate forms:

1. decide (infinitive; all present tenses except the third person)

2. decides (third person, present tense)

3. deciding (present progressive/present participle/gerund)

4. decided (past tense/past participle/passive participle)

5. The morpheme (marked in bold) which marks the particular grammatical function in question is often referred to as the exponent, thus -ed above is the exponent of \<past> in English for decide and many other regular verbs.

6. Languages differ considerably in the richness of their inflectional morphology. Isolating languages, for example Mandarin or Vietnamese, have little or no inflectional morphology: the concept of ‘plural’ in Mandarin for example has to be deduced from context (one dog, two dog, many dog and so on) and is not marked on the noun itself. Russian or Latin, by contrast, are examples of highly inflecting languages: both Latin and its daughter language, Portuguese, for example, have full verbal paradigms in which all persons in all tenses are marked by a suffix (compare English, which marks only third person singular in the present tense). In both Latin and Russian, nouns are additionally marked for case, indicating by means of a suffix their function within a sentence. English, which has lost most of its case marking except in pronouns (compare she as a subject or nominative form, and her as an accusative or object form), achieves this through word order (subjects tend to precede verbs, objects follow them), or by prepositions. In Russian, these endings vary according to the gender of the noun, and there is a separate plural form.

Table 6.3: Nominal inflection in Russian

Note that the genitive plural form for a regular feminine noun like kniga contrasts with the rest by virtue of adding nothing to the stem knig-. Because it contrasts with realized suffixes in all the other cases, this ‘nothing’ is actually meaningful: in cases like this, it makes sense to talk of a zero morpheme. Similarly, for plurals such as sheep, fish or deer in English, there is a strong case, however counter-intuitive it might appear, for arguing that these are all in fact stem+Ø (zero) sequences: this enables us to maintain our generalization that plurals in English are generally formed by addition of a suffix.

In practice, the classification of languages into ‘inflecting’ and ‘isolating’ types should not be thought of in absolute terms, but rather as a continuum in which English is rather less inflecting than, say, Russian, Latin or Basque, but more inflecting than Vietnamese or Tok Pisin. A third type of language, known as agglutinating, is, however, exemplified by Turkish, Hungarian, Aleut and Finnish. In agglutinating languages, words are built up from morphemic blocks, each of which has a single meaning or grammatical function.

Consider the verb ‘to make’ in Turkish:

• yap verb stem/imperative

• yapmak infinitive

• yapiyor present tense

• yapiyorsun ‘you make/are making’

• yapiyorsunuz ‘you (plural) make/are making’

The one-form to one-meaning relationship in Turkish is clearly exemplified here: yap provides the verb stem, iyor marks present tense, sun second person and finally uz marks plural.

In an inflecting language, affixes often have more than one function, as illustrated by the past tense paradigm of the Spanish verb hablar (to speak):

• 1

• pers. sg. yo hablé

• 2

• pers. sg. tu hablaste

• 3

• pers. sg. el/ella habló

• 1

• pers. pl. nosotros hablamos

• 2

• pers. pl. vosotros hablasteis

• 3

• pers. pl. ellos/ellas hablaron

In each case a single suffix marks the categories of person, tense and number: thus é in hablé is the exponent of \<1st person>, \<singular> and \<past>, and we cannot, as in Turkish, find a specific marker for each of these properties. In this case, the relationship between form and grammatical properties is one to many, but the reverse relationship, in which a single grammatical property is marked more than once, is also common. A good example is negation in Turkish:

• geliyorum [geli'jɔrum] I’m coming

• gelmiyorum ['gelmijɔrum] I’m not coming

• yapıyorsunuz [japə'jɔrsunuz] you are making

• yapmıyorsunuz ['japməjɔrsunuz] you are not making

• görüyor [gøry'jɔr] he/she is seeing

• görmüyor ['gørmyjɔr] he/she is not seeing

In each of these examples, negation is realized by an infix -m-, but there is also a change of stress position, so the negation is doubly marked.

Key idea: Classifying languages

Linguists have informally classified languages on the basis of their morphological structure:

•  Isolating languages (e.g. Chinese) have no bound morphemes.•  Agglutinating languages (e.g. Turkish) have words built from morphemes with a single meaning or grammatical function.•  Inflecting languages (e.g. Russian) use bound morphemes to mark several grammatical categories.•  Polysynthetic languages (e.g. Inuit) build ‘sentence-words’ out of bound morphemes rather than constructing sentences from free ones, as for example in English.

As with the analytic/synthetic distinction (see the Bloomfield quotation earlier in the chapter), differences between languages are relative rather than absolute, and languages may combine elements of more than category.

ALLOMORPHY

As we saw in Chapter 5, the same phoneme may be realized in different environments by more than one allophone. Interestingly, a single morpheme may likewise have several allomorphs. A good example is provided by regular plurals in English:

• bat+s ‘bats’

• dog+s ‘dogs’

• place+s ‘places’

At first sight, this looks like a highly regular pattern of inflection, in which plurals are formed by adding -s to the singular noun. But the orthography disguises the fact that the three endings here are different: /S/ for the first, /Z/ for the second and /IZ/ for the last. The three allomorphs are, furthermore, in complementary distribution: /S/ is used after voiceless consonants (e.g. caps, bets, bricks, coughs), /Z/ after voiced ones (beds, ribs, logs, lathes) and /IZ/ after the sibilants or ‘hissing sounds’ /S/, /Z/, /ʃ/, /ʒ/, /tʃ/ and /dʒ/ (mazes, wishes, matches). We can describe /S/, /Z/ and /IZ/ as three allomorphs of the morpheme \<PLURAL>, although we should note here that we are using ‘morpheme’ here in a slightly more abstract sense than hitherto: to avoid terminological confusion where the distinction is important some linguists reserve the term ‘morpheme’ for the abstract representation of a particular grammatical value (e.g. \<PLURAL> for the category of ‘number’), and use the term morph (or allomorph) for its actual realization.

More serious problems do, however, beset the morpheme concept in the case of what are termed discontinuous morphemes. Consider the following irregular plurals in English:

• mouse – mice

• foot – feet

• tooth – teeth

• man – men

• louse – lice

Though this fact is somewhat disguised by the orthography in the case of louse/lice and mouse/mice, all these examples involve monosyllabic items in which a vowel change occurs in the nucleus position in the plural, but the onset and coda are left unchanged. One might therefore wish to posit a discontinuous root morpheme in each case, i.e. /m_S/ for mouse and /f_t/ for foot. Though this might appear to be stretching a point, it is not in fact implausible: in Semitic languages, for example, a number of related words share a common root in which the vowels change, as illustrated by the root k\t\b in Arabic:

• kitāb ‘book’

• kutub ‘books’

• kātib ‘writer’

• kuttāb ‘writers’

• kataba ‘he wrote’

Similarly, in German regular past participle forms involve a prefix ge- and a suffix -t (e.g. gelernt, from the verb lernen in ‘Ich habe gelernt’ – ‘I have learned’). But even if one is prepared to extend the morpheme concept to discontinuous elements, a problem remains with our English plurals, namely how are we to analyse \<PLURAL> in these words?

A first, obvious difficulty is that there is no meaningful sense in which these forms can be analysed as a noun+plural marker sequence: the vowel change which marks plurality is not a suffix. Secondly, whereas with regular plurals a plural morpheme is added (we can include zero plurals here), these plural forms involve a change rather than an addition. One solution might be to describe the \<PLURAL> allomorph in mice as a process /aɷ→aI/, but this seems to bend the original concept of the morpheme as ‘minimal meaning bearing unit’ beyond all recognition without any compensatory gains in terms of descriptive power or elegance. It seems better simply to view these plurals as a non-productive subset of English nouns – in fact a vestige of the umlaut process which survives in German, for example in der Vogel (sg. ‘the bird’) – die Vögel (‘the birds’).

The morpheme-to-meaning relationship also breaks down in the Celtic languages, where gender agreement is marked by mutation of initial consonants, as in the following examples from Breton:

• ar paotr bras ‘the big lad’

• an daol vras ‘the big table’

• ul levr kozh ‘ an old book’

• un gador gozh ‘an old chair’

…the idea of morphemes with constant phonetic and semantic identity is fully appropriate only to the agglutinative languages. Morpheme analysis, therefore, is essentially an attempt to mould all languages (including those that are inflectional) into the form of the agglutinative ones.(Palmer (1971: 112)

A mutation known as lenition affects some (but not all) initial consonants of singular feminine nouns after articles (thus taol ‘table’ becomes _d_aol; and kador ‘chair’ becomes _g_ador, but the masculines paotr ‘lad’ and levr ‘book’ are unchanged), and the initial consonant of an adjective qualifying a feminine noun (contrast bras/vras ‘big’, and kozh/gozh ‘old’). Mutation in Breton exemplifies what is known as non-concatenative morphology in that, in contrast to the Turkish examples, nothing is actually added to a stem (calling -aol a ‘stem’ is an unsatisfactory solution because many other Breton words have initial consonants which are unaffected by mutation) and, as was the case for the exponents of \<PLURAL> in the forms feet, mice, etc. in English, we cannot identify a morpheme which marks gender.

Key idea: The morpheme/allomorph distinction

The morpheme/allomorph distinction parallels that of phoneme and allophone at the phonological level. Allomorphs may, like allophones, be in complementary distribution, each occurring in a particular environment: this is the case, for example, for regular plural noun suffixes in English, the distribution of which is determined by the final consonant of the noun.

Grammatical categories

As we have seen in the examples above, inflection represents a morphological marking on items according to grammatical categories (for example tense, number or gender), which have a number of different values (e.g. masculine or feminine for gender). Categories relevant to English can have very different values and inflectional systems in other languages, as the brief survey of number and gender below will demonstrate.

Categories that are not manifested or that are marginal in English often play a significant role in the inflectional systems of other languages. Animacy, for example, is important in Navajo, in Basque and in Spanish, where animate direct objects are inflected with the preposition a:

Key idea: Number and gender values

Inflectional morphology marks grammatical categories (e.g. tense, number or gender), the values of which vary according to language.

•  Number in English has two values: singular and plural; many languages have more complex, multivalued systems.•  Gender in French has two values (‘genders’), usually termed masculine and feminine; German has three (masculine, feminine and neuter), while Dyirbal has four.

Fact-check

1. [1](answers.mdx#rfn6-1)  Which of these is not a regular plural allomorph in English?

1. [Ən]

1. [IZ]

1. [S]

1. None of them is a regular plural allomorph

1. [2](answers.mdx#rfn6-2)  How is gender motivated in Italian?

1. Formally

1. Largely formally, but partly semantically

1. Largely semantically, but partly formally

1. As an inherent property of articles and adjectives

1. [3](answers.mdx#rfn6-3)  What characterizes derivational morphology?

1. Concern with marking of grammatical categories

1. It involves only affixes

1. The creation of new words

1. It marks agreement

1. [4](answers.mdx#rfn6-4)  What characterizes isolating languages?

1. They are highly inflected

1. They have little or no inflectional morphology

1. They have only short words

1. They build up words from morphemic blocks, each having a separate meaning or grammatical function

1. [5](answers.mdx#rfn6-5)  What is the role of the morpheme ii in the following examples from Kurdish?

1. aaqil ‘wise’      aaqilii ‘forethought, wisdom’

1. garm ‘warm’   garmii ‘warmth’

1. draiž ‘long’      draižii ‘length’

1. A nominal suffix

1. An adjectival suffix

1. A nominal prefix

1. An agreement marker

1. [6](answers.mdx#rfn6-6)  How do inflecting languages differ from agglutinative ones?

1. They have no derivational morphology

1. Inflectional morphemes often mark more than one grammatical category

1. Grammatical categories are often marked by more than one inflectional morpheme

1. They all have grammatical gender

1. [7](answers.mdx#rfn6-7)  Which of these is a regular past-tense allomorph of English?

1. [Id]

1. [d]

1. [t]

1. All of the above

1. [8](answers.mdx#rfn6-8)  What characterizes prefixes in English?

1. They rarely change word class

1. They are not used with nouns

1. They generally mark grammatical categories

1. They are entirely rule-governed in their distribution

1. [9](answers.mdx#rfn6-9)  Which grammatical categories are inflected in ‘The girls were sitting on the table’?

1. number, gender and tense

1. number, tense and case

1. gender and tense

1. number and tense

1. [10](answers.mdx#rfn6-10)  When do linguists sometimes posit ‘zero morphemes’?

1. When a language has no inflection

1. When a language does not mark plural

1. When the absence of marked inflection is meaningful

1. When they wish to be deliberately contrary and confuse people

Dig deeper

B. J. Blake, All About Language (Oxford University Press, 2008), Chapters 2 & 3

G. Corbett, Gender (Cambridge University Press, 1991)

G. Corbett, Number (Cambridge University Press, 2000)

F. Palmer, Grammar (Penguin, 1971), Chapter 3

A. Radford, M. Atkinson, D. Britain, H. Clahsen & A. Spenser, Linguistics: An Introduction (2nd Edition) (Cambridge University Press, 2009), Chapters 8–11

Online source

For a critique of Lakoff’s ‘Women Fire and Dangerous Things’ classification, see K. Plaster, K. & M. Polinksy ‘Women are not dangerous things: Gender and categorization’, Harvard Working Papers in Linguistics 12 (2007). Available online: http://dash.harvard.edu/bitstream/handle/1/3209556/Women%20are%20not%20dangerous%20things%20-%20Pol,%20M.pdf?sequence=2