an affix is a bound morpheme that is joined before, after or within a root or stem.
Types of affixes
Prefix: this is added before the root or stem of the word. E.g. unpredictable, unresponsive, revitalize, etc.
Suffix: a suffix is an affix that is attached to the end of a root or stem. E.g. the plural formatives, -s, -en, -ing, -d, -er, -est, and –less, -ment, -ion, etc.
Affix types
Infix: An infix is an affix that is inserted within a root or stem. It is added in the base form of a word to create a new word or intensify meaning. E.g.
In Philippines (Tagalog)
buli root, ‘buy’
-um- infix, ‘agentive’
Bumili ‘ bought’
In Indonesian:
Cerlang ‘luminous’, cemerlang ‘ brilliant’
Circumfixes: Consists of two parts- a prefix and a suffix that together create a new word. The prefix and suffix are not considered as separate. E.g. In Dutch berg ‘mountain’à ge-berg-te ‘mountain chain’.
Simulfix: A simulfix is a change or replacement of vowels or consonants (usually vowels) which changes the meaning of a word: E.g. eatà ate, tooth à teeth etc.
In English derivational morphology, suffixes heavily outnumber prefixes. In Chapter 6 we saw that most compounds are headed, with the head on the right. Superficially these two facts are unconnected. Consider, however, the role played by the head house of a compound such as greenhouse. As head, house determines the compound’s 71 syntactic status (as a noun), and also its meaning, inasmuch as a greenhouse is a kind of house for plants. This is very like the role played by the suffix -er in the derived word teacher: it determines that teacher is a noun, unlike its base, the verb teach, and it contributes the meaning ‘someone who Xs’, where the semantic blank X is here filled in by teach. Many (though not all) linguists therefore treat -er as the head of teacher in just the same way as house is the head of greenhouse. This is relevant to the distinction between helpful and *-ful-help. In helpful, the affix is what determines that the whole word is an adjective, and so counts as its head. Accordingly, *-ful-help violates English expectations not just because the affix is on the wrong side, but also because the rightmost element is not the head. In the derived words teacher and helpful, therefore, the two components are not equal contributors, so to speak; rather, the righthand element (as in most compounds) has a special status. Superficially, this view of affixes as heads leads us to expect that prefixed words should be as rare in English as left-headed compounds are (items such as attorney general). Yet prefixes, though fewer than suffixes, include some that are of very common occurrence, such as un- ‘not’ and re- ‘again’. Is our expectation disappointed, then? Not really, despite first appearances. Consider the relationship between helpful and unhelpful. In helpful, -ful has a clearly wordclass-determining role because it changes a noun, help, into an adjective. In unhelpful, however, un- has no such role; rather, it leaves the wordclass of helpful unchanged (see Section 5.6). This characteristic of un- is not restricted to adjectives, moreover. Verbs to which un- is prefixed remain verbs (e.g. untie, unfasten, unclasp), and those few nouns to which un- is prefixed remain nouns (unease, unrest). This strongly suggests that the head of of all these words is not un- but the base to which un- is attached (helpful, tie, ease etc.) – and which is the righthand element. Similar arguments apply to re-: rearrange, repaint and re-educate are verbs, just as arrange, paint and educate are. These prefixed verbs, therefore, are right-headed also. The only prefixes that are unequivocally heads are those that change wordclass, such as de- in delouse (deriving verbs from nouns) and en- in enfeeble and enslave (deriving verbs from nouns and adjectives) (see Section 5.9). So, while left-headed derived words do exist, just as left-headed compounds do, they are also not so numerous as may at first appear.
More elaborate word forms: multiple affixation Many derived words contain more than one affix. Examples are unhelpfulness and helplessness. Imagine now that the structure of these words ientirely ‘flat’: that is, that they each consist of merely a string of affixes plus a root, no portions of the string being grouped together as a substring or smaller constituent within the word. An unfortunate consequence of that analysis is that it would complicate considerably what needs to be said about the behaviour of the suffixes -ful and -less. In Chapter 5 these were straightforwardly treated as suffixes that attach to nouns to form adjectives. However, if the nouns unhelpfulness and helplessness are flat-structured, we must also allow -ful and -less to appear internally in a string that constitutes a noun – but not just anywhere in such a string, because (for example) the imaginary nouns *sadlessness and *meanlessingness, though they contain -less, are nevertheless not words, and (one feels) could never be words. The flat-structure approach misses a crucial observation. Unhelpfulness contains the suffix -ful only by virtue of the fact that it contains (in some sense) the adjective helpful. Likewise, helplessness contains -less by virtue of the fact that it contains helpless. Once that is recognised, the apparent need to make special provision for -ful and -less when they appear inside complex words, rather than as their rightmost element, disappears. In fact, both these words can be seen as built up from the root help by successive processes of affixation (with N, V and A standing for noun, verb and adjective respectively):
(1) helpN + -ful →helpfulA
un- + helpful →unhelpfulA
unhelpful + -ness →unhelpfulnessN
(2) helpN + -less →helplessA
helpless + -ness →helplessnessN
Another way of representing this information is in terms of a branching tree diagram, as in (3) and (4), which also represent the fact that the noun help is formed by conversion from the verb:(3) A WORD AND ITS STRUCTURE 73 un help ful ness
N
A
N
V
A
(The term ‘tree diagram’ is odd, because the ‘branches’ point downwards, more like roots than branches! However, this topsy-turvy usage has become well established in linguistic discussions.) The points in a tree diagram from which branches sprout are called nodes. The nodes in (3) and (4) are all labelled, to indicate the wordclass of the string (that is, of the part of the whole word) that is dominated by the node in question. For example, the second-to-top node in (3) is labelled ‘A’ to indicate that the string unhelpful that it dominates is an adjective, while the topmost node is labelled ‘N’ because the whole word is a noun. The information about structure contained in tree diagrams such as (3) and (4) can also be conveyed in a labelled bracketing, where one pair of brackets corresponds to each node in the tree: [[un-[[helpV]N-ful]A]A-ness]N, [[[helpV]N-less]A-ness]N. One thing stands out about all the nodes in (3) and (4): each has no more than two branches sprouting downwards from it. This reflects the fact that, in English, derivational processes operate by adding no more than one affix to a base – unlike languages where material may be added simultaneously at both ends, constituting what is sometimes called a circumfix. English possesses no uncontroversial examples of circumfixes, and branching within word-structure tree diagrams is never more than binary (i.e. with two branches). (The only plausible candidate for a circumfix in English is the en-…-en combination that forms enliven and embolden from live and bold; but en- and -en each appears on its own too, e.g. in enfeeble and redden, so an alternative analysis as a combination of a prefix and a suffix seems preferable.) The single branch connecting N to V above help in (3) and (4) reflects the fact that the noun help is derived from the verb help by conversion, with no affix. At (5) and (6) are two more word tree diagrams, incorporating an adverbial (Adv) node and also illustrating both affixal and non-affixal heads, each italicised element being the head of the constituent dominated by the node immediately above it: help less ness
N
A
N
V
Some complex words contain elements about which one may reasonably argue whether they are complex or not. For example, the word reflection is clearly divisible into a base reflect and a suffix -ion; but does reflect itself consist of one morpheme or two? This kind of uncertainty was discussed in Chapter 2. But, if we put it on one side, then any complex word form consisting of a free root and affixes turns out to be readily analysable in the simple fashion illustrated here, with binary branching and with either the affix or the base as the head. (I say ‘free root’ rather than ‘root’ only because some bound roots are hard to assign to a wordclass: for example, matern- in maternal and maternity.) Another salient point in all of (3)–(6) is that more than one node in a tree diagram may carry the same wordclass label (N, V, A). At first sight, this may not seem particularly remarkable. However, it has considerable implications for the size of the class of all possible words in English. Linguists are fond of pointing out that there is no such thing as the longest sentence of English (or of any language), because any candidate for longest-sentence status can be lengthened by embedding it in a context such as Sharon says that ___ . One cannot so easily demonstrate that there is no such thing as the longest word in English; but it is not necessary to do so in order to demonstrate the versatility and vigour of English word-formation processes. Given that we can find nouns inside
(5)
Adv
A
V
A
un assert ive ly
(6)
V
re de class ify
V
V
N
nouns, verbs inside verbs, and so on, it is hardly surprising that (as was shown in Chapter 2) the vocabulary of English, or of any individual speaker, is not a closed, finite list.