last updated 26 April 2004
This page is part of the Encyclopedia of English Grammar and Word Grammar
|
Notation: `J' `Adjective' is a word-class, not a grammatical function. One of the two main uses of adjectives is `attributive', i.e. modifying a common noun (e.g. big book), but this does not mean that any word which modifies a common noun must be an adjective. For example, in big linguistics book, linguistics modifies book, and has the same grammatical function as big (namely, adjunct), but belongs to a different word-class: noun. This is easy to prove because linguistics may in turn be modified by an adjective, such as theoretical: thus in theoretical linguistics book, theoretical is an adjective modifying the noun linguistics, which in turn modifies book. Had linguistics been an adjective, it could not have taken another adjective as its modifier. For example, although nice big book is possible, with two adjectives, these both modify book; if we use them after BE only one is possible (*It is nice big), which shows that nice does not modify big (unlike, say, extremely in extremely big). Most adjectives can be used in two ways, either attributively (as pre-adjuncts of common nouns) or predicatively (as sharer of a verb such as BE): (1) That big book is heavy. The structures concerned are shown here: |
|
However, some (apparent) adjectives can only be used attributively (e.g. SOLE) and others only predicatively (e.g. ALONE): (2) The sole/*alone survivor was Pat. (3) Pat was alone/*sole. It is not at all clear how best to handle these restrictions, and further
research is certainly needed. Are SOLE and ALONE really adjectives? For
the time being I assume that they are, but this may well be wrong. Notation: `A'. `Adverb' is the name of one of the main word-classes, but it is undoubtedly the least convincing word-class because it has so few distinctive characteristics. Here is a list of distinctive characteristics that are shared by words which are clear adverb:
(1) They (have) frequently/often walked to work.
(2) They acted quickly/*quick. (3) Their quick/*quickly action saved the day.
(4) extremely/*extreme quick/quickly In general, if a word can be replaced by an ly-adverb (i.e. an adverb such as QUICKLY or EXTREMELY derived from an adjective), it is probably an adverb; putting it another way, if a word can be replaced by an ly-adverb, it must be being used in a way which is permitted for adverbs, so you can safely call it an adverb unless you have good reasons for thinking otherwise. (For example, this morning can often be replaced by recently, but we have excellent reasons for believing that both this and morning are nouns.) See also: sentence adverb.
See determiner.
This is a possible name for the difference between the three kinds of non-finite verb: `infinitive', `perfect' and `participle'. The name `aspect' is not widely used to refer to this contrast, but it is the best available term, and does at least fit the feature `perfect' (which is one of the recognised `aspects'). The three values are distinguished as follows: Infinitive: a verb whose form is just like an imperative, e.g. talk, be. Used after TO and a range of special verbs. (See modal verb.) (1) I will be late. Perfect: a verb whose form is just like a passive, i.e. normally ending in -ed (e.g. walked), but with -en in many irregular verbs (e.g. taken). The perfect only has one use: as complement of HAVEperfect. (2) I have been late. Most verbs have the same form in the perfect and in the past tense, but it is very easy to distinguish the two: look for HAVE among the preceding words, as in (2). If there's a HAVE, it's almost bound to be perfect; otherwise it's bound to be passive. Traditionally the perfect is called the `past participle'. Participle: This is further subdivided by the feature `voice',
which distinguishes active and passive participles.
There are three verb inflections which are available only for auxiliary verbs. reduced forms like's and 'll in she's and we'll; negative verbs containing the suffix -n't, which are a particular kind of reduced verb; `inverted' verbs whose subjects are inverted. These are not strictly inflections because they are not distinguished morphologically from their uninverted counterparts, but they can be recognised as a distinct word class because they combine distinctive syntax (the position of the subject) with the distinctive semantics of interrogatives. These three inflections combine freely with each other and with past tense as shown in Figure 2, to give the 12 possibilities shown in the table below for HAVE with it as subject. |
| it has | ||||
| it's | reduced | |||
| it hasn't | reduced | negative | ||
| has it | inverted | |||
| 's it | reduced | inverted | ||
| hasn't it | reduced | negative | inverted | |
| it had | past | |||
| it'd | reduced | past | ||
| it hadn't | reduced | negative | past | |
| had it | inverted | past | ||
| 'd it | reduced | inverted | past | |
| hadn't it | reduced | negative | inverted | past |
|
Notation: `v' This is the only sub-class of `verb' that is recognised (as a word-class) in WG, apart from its own sub-class `modal verb' and (possibly) the remaining verbs, which we can call `full verbs'. Its members are:
It is a beautiful word-class because it is so clearly defined by so many characteristics (most of which, incidentally, have developed since the 14th century - our auxiliary verbs are much more clearly defined than Shakespeare's! See Hudson 1990). Here are its distinctive characteristics.
(1) Are/*get they tired?
(2) They are/*get not tired.
(3) They aren't/*getn't tired.
(4) They're/*Theyg't tired. (5) They may've been tired.
(6) I don't know whether he likes grammar, but he may/tries. (full ellipsis) (7) I don't know whether he likes grammar, but he may/*tries phonology. (partial ellipsis, meaning `he may like phonology' or `he tries to like phonology') Both subject-auxiliary inversion and the addition of -n't are handled as inflections of the auxiliary verb. The name `auxiliary verb' is highly inappropriate to this word-class, because only one of these characteristics (the last one) has anything to do with the presence or absence of another verb which this one `helps' (Latin AUXILIUM, `help'). Indeed, some auxiliary verbs (as defined by these characteristics) may be used without any other verb, overt or ellipsed. (8) Is she here? (9) Has she time? In both these examples the single verb (is, has) has an inverted subject, so it must be an auxiliary verb. My own preferred term is `polarity verb', but it seems perverse not to use the established term, especially since these verbs are most typically used to support other verbs. Please bear this in mind in deciding how to classify a word - the only relevant question is whether it allows subject-auxiliary inversion, accepts -n't, etc. Another characteristic shared by most auxiliary verbs is that they have very little `referential' meaning, i.e. meaning you could define in terms of characteristics of situations in the world, which helps you to know what the speaker is talking about; e.g. in (10) the main distinctive load is carried by wash, not by may. (10) Pat may wash it. In some cases the semantic relation between the auxiliary and the next verb is co-reference; e.g. in will come or did come the two verbs both refer to the same event (with the auxiliary defining its time, like a tense inflection). In spite of this, the auxiliary verb is syntactically the parent of the next verb, which is its complement. (This is perfectly normal for a parent-dependent pattern, as explained in dependency.) I.e. the syntactic structure of (10) is just the same as that of (11) and (12), where the first verb is not an auxiliary verb. (11) Pat helps wash it. (12) Pat keeps washing it. These examples illustrate the irrelevance of the notion `auxiliary' which I grumbled about in the previous paragraph: in each case the first verb `helps' the second one, by supporting it syntactically, so there is nothing unusual about so-called auxiliary verbs from this point of view. All such verbs take a non-finite verb as their `sharer', which shares their subject. The figure illustrates the pattern. |
Auxiliary verbs include the class of modal verbs, as shown in the next figure.
|
In my opinion English does not have any case at all. In languages such as Latin, German or Russian, various kinds of nouns and adjectives have inflectional `case' (see inflectional feature); e.g. in Latin, a noun such as URB- `city', has five distinct cases in singular and plural: |
| singular | plural | |
| nominative | urb-s | urb-es |
| accusative | urb-em | urb-es |
| genitive | urb-is | urb-ium |
| dative | urb-i | urb-ibus |
| ablative | urb-e | urb-ibus |
|
A noun's case is mostly determined (`governed') by the word it depends on, and by its function in relation to this word; e.g. the preposition per, `through', governs an accusative complement noun whereas ab, `from', governs an ablative one. Old English was like Latin, but the word-endings which distinguished cases had disappeared well before Shakespeare's time (late Middle English). Nothing like Latin is found in English nouns, though it is often claimed that pronoun-forms (e.g. me versus I) are examples of case. See pronoun-form. See word-class
Not all words can be assigned to a word-class,
because word-classes must be defined by general characteristics that are
true of their members but not of other classes. A word such as HELLO does
not seem to share the characteristics of any word-class, so we leave it
unclassified. It isa just `word'. Another
class-less word is the 'complementiser' that.
A clitic is a word which is treated, as far as its position and phonology are concerned, as though it was an affix. For example, all the reduced forms of auxiliary verbs ('re, 'll, 's, etc) are clitics, as is the possessive 'S. The phonology treats them as part of the preceding word (e.g. possessive 'S has the same range of alternative pronunciations as the suffix -s in cats, dogs and horses), though their positions are the same as if they had been full words. In some languages there are clitic pronouns which are positioned differently from ordinary nouns; e.g. French object pronouns are attached just before the verb, whereas ordinary objects follow the verb: (1) Paul mange le fromage. `Paul eats the cheese.' (but not: *Paul le fromage mange. ) (2) Paul le mange. `Paul it eats.' (but not *Paul mange le.) The position of a clitic may cause serious structural oddities, as well as reversing the normal order. It can even lead to tangling in the surface structure, as in (4). (3) Paul mange beaucoup de fromage. `Paul eats lots of cheese.' (4) Paul en mange beaucoup. `Paul of-it eats lots.' The figure below shows the structure for examples (3) and (4), with the tangling above mange in (4). However, the No-tangling principle explicitly excludes clitics from its scope, on the grounds that clitics are positioned by rules of morphology which always take priority over rules of syntax. |
| The No-tangling principle has now been replaced by Order concord but this too makes special provision for clitics by defining a word's possible positions in relation to its landmark. For a clitic the landmark is the stem of the hostword that contains it - for example, in (4) the hostword is en mange, whose stem is the verb mange. The next diagram shows the structure of this sentence including the hostword. |
|
For more discussion of clitics see Hudson 2001 and (especially) Camdzic and Hudson 2002. The analysis just presented isn't quite the same as that theory, because it takes account of subsequent changes to the WG theory of morphology. `Common noun' is one of the main sub-classes of noun. Unlike a pronoun,
a common noun can occur freely with a determiner
(compare the book with *the she), and unlike a proper
noun, it must combine with a determiner if it is singular and countable
(compare I saw Pat/*student and I saw the student/*Pat).
The vast majority of nouns in a dictionary are common nouns; for many
people, `noun' means `common noun'. Some adjectives and adverbs have inflections which are traditionally called 'comparative', and which are typically marked by the suffix -er: bigger, faster, sooner. Their syntactic characteristic is that they allow than as a complement: bigger than Mary; sooner than he expected. See also superlative. Like 'specifier', the term `complementiser'
is an invention of the 1970s, and includes a small group of words whose
function is merely to introduce a clause. The main members are THATsub,
IF and WHETHER,
but it may also include TO and used to include
`POSS ING', responsible for gerunds. WG does not
recognise `complementiser' as a word class because every one of these
words is different. WHETHER may be a pronoun (see interrogative
pronoun), and THATsub, IF and TO have nothing in common
other than that they each link a verb to some higher word. In a WG grammar
it can be argued
there are no rules which apply to all these words and to noo others, so
the category `complementiser' would do no work at all, and is not recognised
in WG. A compound pronoun is made up (at least historically) of a determiner plus one of the nouns BODY, ONE, THING or WHERE:
In WG they are treated as single words, but the network analysis in principle
allows the close semantic and morphological relations to the relevant
determiners and nouns to be revealed; but this analysis remains to be
done.
See coordinator, subordinating
conjunction `Coordinator' is the WG name of one of the major word-classes (traditionally called `coordinating conjunction'), although it is a very small class whose most obvious members are AND and OR. (For other members see below.) Their primary function is to signal a coordination, and unlike other words they do not have any dependency links to other words. For example, consider (1): (1) I drink {[white coffee] [and tea with milk]}. Here both coffee and tea have dependency links to drink, and all the other dependency links are as in any sentence, but the coordinator and is outside the dependency system. (For examples see the diagrams in coordination and conjunct.) Why do the brackets in these examples treat the coordinator as part of the next conjunct? Why `{[...] [C...]}', rather than `{[...] C [...]}'? It is actually quite hard to choose between these two analyses, and I myself used the second for some years, but I finally switched to the first for two reasons, neither of which is particularly persuasive. The first is that every other linguist seemed to take the second one for granted, so why not follow the crowd? (I already had quite enough disagreements that I cared much more about!) The second reason was slightly more respectable: both intonation and punctuation locate conjunct boundaries before the coordinator rather than after it: white coffee, and tea with milk is a reasonable way to punctuate the example above, in contrast with the completely impossible *white coffee and, tea with milk. (See correlative conjunction for the positioning of BOTH or EITHER.) What other coordinators are there? Since coordinators are by definition used to signal coordinations, we must first decide how to recognise coordination on the one hand in contrast with subordination and on the other hand in contrast with loose juxtaposition. The crucial characteristic of coordination is the Dependency-in-Coordination Principle which allows conjunct roots to share dependencies. The easiest kind of shared dependency to recognise is a shared subject dependency as in (2). (2) Pat {[works hard] [and plays squash]} It is easy to be sure that both works and plays share Pat as their subject because without such sharing plays would have no subject; but we know that, being a tensed verb, it must have a subject. There is no syntactic construction other than coordination which allows tensed verbs to share a subject, so if we find a word which can replace AND in a sentence like (2), it must be a coordinator. By this criterion, the other coordinators are BUT, NOR and (somewhat surprisingly) THEN. Sentence (3) illustrates the use of BUT and THEN. (3) Pat {[works hard] [but/then plays squash]}. Sentence (4) is different because NOR has to be used in combination with a correlative conjunction NEITHER: (4) Pat {[neither works hard] [nor plays squash]}. THEN is an odd member of the class because it can also be used in combination with AND, in which case it is presumably not a coordinator but just an adverb: (5) Pat {[works hard] [and then plays squash]}.
A correlative conjunction is a special type of coordinator which is only used in the first conjunct of a coordination, and which is always paired with one particular coordinator; for example, the correlative conjunction BOTH is always paired with AND as in {[both black coffee] [and tea with milk]}. The correlative conjunctions and the ordinary coordinators with which they are paired are the following:
Correlative conjunctions are useful tools for deciding where a coordination starts. For example, take an example like (1) (for further discussion of examples like this, see conjunct). (1) Pat drinks coffee in the morning and tea in the afternoon. One possible place for BOTH, the counterpart of AND, is before coffee, which shows that this must be the first word in the coordination of (1): (2) Pat drinks {[both coffee in the morning] [and tea in the afternoon]}. Unfortunately the correlative conjunctions can also be used as ordinary adverbs which are pre-adjuncts of the first verb. This explains the pattern in (3), which is more or less acceptable - at any rate it is certainly widely used. (3) Pat both drinks {[coffee in the morning] [and tea in the afternoon]}. Sentence (3) does not prove that drinks is the first word of the
coordination, because this is impossible - drinks is shared by
both conjuncts so it must be outside the coordination. The position of
a correlative conjunction thus works as a test for coordination boundaries
only when it cannot be explained by the usual rules for positioning adverbs
as pre-adjuncts of verbs. Traditional grammar classified `articles' (the determiners THE and A) as definite or indefinite, but this distinction applies to all the other pronouns as well (see the lists in pronoun). Furthermore, proper nouns are all inherently definite, while common nouns are indefinite unless combined with a definite determiner (so books is indefinite in Books were lying all over the place). Although definiteness is clearly predictable (in English) from the syntactic
structure, it is not a syntactic category (there being no syntactic rules
that mention it) so in WG it is not treated as a syntactic distinction.
It is represented in the semantic structure, however, in terms of whether
or not the noun's referent is already known to the addressee.
The referent is known if the noun is definite, but not if it is indefinite;
for example, if I say the book to you, I am inviting you to find
the book concerned in your memory, among all the books you know; but if
I say a book I am advising you not to bother to look for it, as
you may not already know it.
See pronoun. Notation: `n', just as for `pronoun'. 'Determiner' is a word-class which is not part of traditional grammar but was invented earlier this century. Typical examples are SOME, ANY, MY and THIS, which were traditionally classified as adjectives because they combine with common nouns. However, the most common determiners are A(N) and THE, which were traditionally called `articles' (from the Latin for `joint', as in articulated). The modern word-class `determiner' therefore combines the traditional articles with some of the traditional adjectives - a major reorganisation of the word-class system. Other modern theories take `determiner' as an additional basic word-class, distinct from all the others. After all the upheaval, however, the WG position is that determiners do not in fact constitute a separate word-class, but are all pronouns (NB certainly not adjectives as in traditional grammar). See below for evidence. Determiners are controlled by two general rules, which between them set rigid minimum and maximum requirements for determiners: Rule 1. Any singular countable common noun must combine with a determiner. (More precisely, the noun must depend on the determiner, as explained below.) (1) I found *(a) book. (i.e. a cannot be omitted) (An apparent but explicable exception is the common noun one.) Rule 2. A determiner cannot combine with another determiner. (2) I found a/my book. (3) *I found a my book. (Other languages have very different rules for the use of determiners, if indeed they have such words at all.) These two rules distinguish determiners very clearly from adjectives, which have neither minimum nor maximum requirements; for example, adding good to (1) and (2) makes no difference to their grammaticality. It is quite certain that English determiners are not adjectives. A much more relevant comparison is with pronouns. Most determiners can be used either with or without a following common noun. For example: (4) I found some (books). (I.e. book may be omitted.) (5) I like this (book). (6) Which (book) did you read? Though traditional grammar classifies these words as adjectives when they combine with a common noun, it also calls them pronouns when used without a common noun - a classification which fits well with the characteristics of pronouns. If determiners are pronouns when used without common nouns, why not call them pronouns when there is a common noun as well? After all, we don't say that EAT belongs to different major word classes according to whether it combines with an object (eat at 5 versus eat sandwiches at 5), so why treat SOME differently according to the presence or absence of a common noun? This argument rests, of course, on the assumption that the relation of some to books is similar to that between eat and sandwiches, but this in turn can be justified: Just as verbs vary according to whether an object is obligatory or optional, so do determiners. After DEVOUR the object is obligatory, whereas it is optional after EAT; and similarly the common noun is obligatory after THE but optional after THIS: (7) I ate/devoured a sandwich. (8) I ate/*devoured at five o'clock. (9) I ate this/the sandwich. (10) I ate this/*the. When either a verb's object or a determiner's common noun is omitted, its meaning may nevertheless still be present in the semantic structure - it is missing through ellipsis; and in both cases the same range of strategies is available for reconstructing this meaning. When EAT is used without an object noun the event referred to still has to involve some kind of typical food (usually a meal); similarly, when THIS is used without a common noun it has to refer to a thing (not a person). Example (11) is very odd without woman. (11) Do you take this !(woman) to be your wife? In both these cases, the meaning is reconstructed as an example of some predefined general concept (food, thing). In other cases, again found in both verbs and determiners, the meaning is reconstructed on the basis of the preceding context (typically anaphorically): (12) He switched on the TV but he only watched [understood: TV] for 10 minutes. (13) She needed some money but she hadn't got any [understood: money]. There is a subclass of verbs, auxiliary verbs, which typically combine with a following verb (e.g. will eat). The relationship between an auxiliary verb and its accompanying verb (its sharer) is very similar both in syntax and in meaning to that between a determiner and its accompanying common noun (e.g. this book). In both cases:
These similarities between verb-object patterns and determiner-common
noun patterns show they need similar analyses. We know that a verb's object
or sharer depends on it; therefore we can conclude that the determiner's
accompanying common noun also depends on it. More precisely, the verb's
object is a complement of the verb,
so the determiner's common noun is also its complement. The analysis is
illustrated in Figure 10. |
|
Somewhat surprisingly, this means that in a noun-phrase like this book there are actually two nouns, since this is not only a pronoun but also a noun; and the noun which is the head of the phrase is not the one traditionally recognised as such, book, but this. This analysis has a number of advantages in addition to those listed above. One is that it removes the problem of `zero determiners', as found in sentences like Boys like girls. If we require subjects and objects to be noun phrases, and if we require noun-phrases to have determiners as their heads, we must assume some kind of invisible determiner before boys and girls. But according to the WG analysis, subjects and objects are nouns, so common nouns and pronouns (including determiners) are equally good candidates. What, then, is a determiner? According to WG, this term has no official status in the grammar. It is just a loose way of talking about pronouns that have common nouns as their complements (or, equally loosely, about pronouns that may have such complements). All the grammar has to do is to specify which pronouns have (or may have) complement nouns. The table below contains a complete list of all determiners, classified according to the general scheme for pronouns given in pronoun. Notice that some pronouns have alternative forms according to whether or not their complement is present, or according to their complement's number. Problems: This analysis is controversial, and may be wrong (though I think any alternative faces even bigger problems). The idea that the common noun depends on the determiner is also found in the `DP' (Determiner Phrase) analysis which is more or less standard in Chomskyan GB/MP theory, but even there the analysis is fundamentally different from the WG one because `Determiner' is taken as a class on its own (possibly including pronouns), and certainly not as a subtype of noun. The analysis suggested here has been criticised within the WG assumptions by Langendonck, who points out some factual problems. Perhaps the most worrying involves the use of noun phrases as adjuncts, as in (1). (1) He got up at 7 [each/this/every/one morning/*occasion]. What decides whether a noun phrase can be used in this way is its common noun, not its determiner. This is not as predicted by the WG analysis, but it may be possible to explain it in semantic terms, given that the determiner and the common noun share their meanings through co-reference. The table below gives a complete list of English determiners. See also: 'S. |
| pronoun class | lexeme | number | complement obligatory? | alternating forms | example |
| definite | THE | s/p | yes | the book | |
| THIS, THAT | s/p | no* | these, those | this book | |
| YOU, US | p | no | you students | ||
| WHOSE | s/p | no? | whose book | ||
| 'S, HIS, ITS | s/p | no | his book
Pat's book |
||
| MY, YOUR, OUR, HER,
THEIR |
s/p | no | mine, yours,
ours, hers, theirs |
my book
mine |
|
| indefinite | A(N) | s | yes | a book | |
| SOME, ANY | s/p | no | some book | ||
| SOME [sm] | s*/p | no? | some [sm] money | ||
| NO | s/p | no | none | no money, none | |
| EITHER, NEITHER | s | no | neither book | ||
| EVERY | s | yes | every book | ||
| WHICH | s/p | no | which book | ||
| WHAT | s/p | no* | what book | ||
| WHOSE | s/p | no | whose book |
|
Key: s*/p = must be mass when singular (e.g. some money/*thing). no* = the complement is not obligatory, but the pronoun has a specially
restricted meaning when the complement is omitted. See transitivity. A word that simply `fills out' a syntactic position without adding much, or any, meaning of its own. In linguistics, a feature is a variable, or 'attribute', that has a (small and fixed) number of possible values. A typical example from syntax is number, a feature of nouns which has the possible values Singular and Plural. The status of features is one of the most distinctive characteristics of WG, because in other theories of grammar features are the basis for all classification, but in WG their use is very limited indeed. In WG, classification is generally based on the isa relationship (e.g. DOG isa common-noun, which isa noun, which isa word), which does not involve features; for example, there is no feature 'common-ness' or 'pronominality' which distinguishes pronouns, common nouns and proper nouns, or 'noun-ness' distinguishing nouns from other word classes. Features are not recognised in such cases simply because they are not needed - they would do no work at all in the grammar. The one point in a grammar where features really are necessary is where there is agreement. By definition agreement always involves one or more features, for which two words have to have the same value. For example, a determiner and its complement must have the same (value for the feature) number: this book but these books. Without the feature 'number' it would be impossible to state this rule. Consequently WG recognises features only when they are mentioned in some agreement rule. In English, this means that only two features are recognised: number and the closely related feature 'agreement', which is used in subject-verb agreement. One consequence of the WG approach to features is that they involve inflections
rather than word-classes because otherwise the
words concerned could not co-vary.
The difference between finite and non-finite verbs is that normally only a finite verb is allowed to be used without a parent. For example, in (1) was is a finite verb and has no parent, but talking is non-finite and must depend on some other word - in this case, on was. Take away was, and the sentence is ungrammatical. (1) Pat was talking. (2) *Pat talking. This is not to deny that finite verbs may depend on other words - for example, in (3) was does have a parent, that. (3) I know that Pat was talking. The main point is that finite verbs don't have to have a parent, a fact which distinguishes them not only from non-finite verbs but also from all other kinds of word. (The word `normally' is needed here, because this is only true of the 'complete sentences' that we expect in essays, encyclopedias of grammar and the like; almost any kind of word can occur without a parent if it is the answer to a question, part of a list, or a linguistic example.) An easy test for finiteness in English is the possibility of a preceding dependent not: if one is possible, the verb in question is non-finite. For example, take (1): (1) Students| studying grammar | will | enjoy | using an encyclopedia | written for them because the explanations | try | to | be clear. It is possible to add not at any of the points marked by | except just before will and try, the two finite verbs. In contrast, you can be sure that the highlighted verbs are non-finite.'Finite verb' covers not only tensed verbs but also imperatives. (Notice how these too reject not: the negative of Come in! is Don't come in!, not *Not come in!) In English, `non-finite verb' covers infinitives and participles, as well as perfects. `Finite' is a very traditional term, based on the Latin word finis,
`field limit'; finite verbs are limited (in time). For example, finite
saw can normally only refer to an event of seeing that took place
at some time in the past, but non-finite seeing can apply to seeing
at any imaginable time. See reduced verb. Some other theories have followed Chomsky in recognising a basic classification of word-classes into two groups: functional and lexical categories. Functional categories include word classes such as determiner and complementiser, in contrast with noun, verb, adjective and (perhaps) adverb and preposition, which are the lexical categories. This is an extension of the traditional distinction between 'function words' and 'lexical words'. WG rejects the notions 'functional category' and 'lexical category', while accepting (with reservations) the traditional distinction between function words and lexical words. The basis for the rejection is simple: none of the supposedly straightforward examples of functional categories are genuine word classes:
Words like DOG are obviously different from words like THE, a distinction which has been recognised since linguists started to distinguish dictionaries from grammars. DOG has a sense which is a real-world concept (Dog), but THE does not; THE plays an important and unique part in the grammar of definiteness, but DOG is grammatically indistinguishable from CAT (and from thousands of other common nouns). This distinction is often recognised by describing DOG as a 'content word' (or 'lexical word'), and THE as a 'function word' (or 'grammatical word'). WG accepts that such differences exist - how could it deny them? - but they play no part in the grammar. In other words, WG does not recognise the function/content classification within the grammar as part of the system of word-classes. The reasons are as follows:
Moreover, the function/content distinction applied to individual words does not justify a similar contrast applied to whole word classes, as in the distinction between functional categories and lexical categories. The reason for this is that the distinction applied to words cuts across the standard word classes. For example, some prepositions are clearly at the function end of the spectrum (e.g. OF, BY), while others (e.g. BETWEEN, BEHIND) are much nearer to the content end. Similarly for verbs, which range from pure function words like WILL through intermediate cases like GET and START to pure content words like WALK. Even common nouns may be function words - as with the word ONE (a big one), which has no sense of its own. Notation: V:N A gerund is a verb with the suffix -ing, which combines the characteristics of both nouns and verbs in a particularly interesting way. As a dependent, it is a noun; but as a phrase-head, it is a verb. (1) illustrates this possibility: (1) I enjoy not buying books often. The object of ENJOY is normally a noun, and (apart from gerunds) there is no reason to believe any other kind of complement is possible. (2) I enjoy dancing/*to dance/*that I'm dancing. Therefore buying must be a noun in (1). On the other hand, there are equally good reasons why buying must be a verb - the object noun books, the preceding not and the adverb often are all typical dependents of a verb, and not of a noun. None of them is possible with a clear noun such as purchase in (3). (3) I enjoy the (*not) purchase *(of) books (*often). However, the noun and verb qualities apply to distinct parts of its use: noun qualities to its use as a dependent, and verb qualities to its use as head of its phrase. The present WG analysis (which is different from the one in Hudson 1990, pp 316-26) recognises the gerund - buying in (1) - as both noun and verb. More specifically, it recognises an inflection Gerund which isa Noun and which also isa Verb. This classification distinguishes its qualities as dependent and as parent in the desired way just because dependents may be specified as simply 'noun' but never as 'verb', and conversely parents may be specified as simply 'verb' but never simply as 'noun'.
|
|
See also Mixed categories.
This is a class of finite verbs which contrasts (in mood) with `tensed'. Just like infinitives, imperative verbs have no ending (e.g. come, be), but unlike infinitives they use don't as their marker of negation. (1) Be late! (2) Don't be late! (3) Pat may be late. (4) Pat may not/*don't be late. There are two kinds of imperative: the `normal' kind, whose understood subject is `you', and a special kind formed with let's, whose understood subject is `you and me'. Once again negation involves don't. (5) Let's go together. (6) a Don't let's go together.
The meaning of an imperative verb is distinctive. Its referent is an event in the future, but this event is the purpose of the word itself. This analysis rests on the WG assumption that every word is an action with a purpose, but one of the attractions of this assumption is this analysis of imperatives: if a word is an imperative verb this purpose is the verb's referent. So if I say to you Hurry! my purpose is that you should hurry at some time in the (immediate) future. (See LET'S.) The rules for imperatives are as follows: [1] The whole of an imperative verb = its stem. [2] The referent of an imperative verb = its purpose. [3] An imperative verb has an optional subject. [4] The referent of an imperative verb's subject = its addressee. See also: subjunctive. See definiteness. One of the classes which contrast in `aspect', the others being `perfect' and `participle'. For example, the infinitive of BE is be. Please remember that the infinitive is just a single word, the verb, and does not include to - hence the irrelevance of the ban on `splitting infinitives'. Infinitives do occur after to, and do so very frequently, but this isn't their only use. Here is a range of different patterns; in each case the infinitive is make. (1) Pat tried to make it. (2) Pat will make it. (3) Pat helped me make it. (4) I saw Pat make it. (5) Rather than make it myself, I bought it. (6) I want to just make it, without icing it. See also subjunctive. Notation: inflections are written as a single letter following a colon, which is added to the word-class of the word concerned; e.g. a past-tense full verb is 'V:a', and a singular common noun is 'N:s'. An inflection is a general type of word identified by its inflectional morphology, in contrast with lexemes, which ignore inflectional morphology. For example, Singular and Plural are inflections, whereas DOG and CAT are lexemes, giving a two-way classification of every noun in terms of its inflection and its lexeme: |
| inflection |
|
|||
| DOG | CAT | MOUSE | PERSON | |
| singular | dog | cat | mouse | person |
| plural | dogs | cats | mice | people |
|
Similarly for verbs and other word classes which allow inflectional morphology.
In general, a word's lexeme determines its word
class (e.g. DOG is a noun, GO is a verb, and so on),
while inflections make its syntactic and semantic characteristics more
precise. English makes fewer inflectional distinctions than 'highly inflected'
languages like Latin and its descendants, but it has quite a rich set
of verb inflections. See feature. A kind of wh-pronoun used to introduce a wh-interrogative clause. The interrogative pronouns are:
Borderline members are how come and whether. Notation: written in capital letters throughout. A 'dictionary-word', which may comprise several different inflections; e.g. DOG is the lexeme which covers both dog and dogs. Where there is no inflectional variation (e.g. for prepositions, adverbs and the like) it makes no difference whether you write them as lexemes (e.g. OF) or as simple word-forms (e.g. of). In general each lexeme belongs to just one word-class (or combination of word-classes - see mixed category), so a lexeme does not necessarily bring together all words which we feel to be closely related - e.g. the verb and noun walk as in (1) belong to different lexemes, WALKv and WALKn. (1) a We walk everywhere.
This distinction is forced on us by the need to distinguish verbs and nouns. The links between similar lexemes are shown in the network of the grammar by links such as `nominalization of' which handle `word-formation' rules. The general principle, then, is that we must recognise distinct lexemes whenever two (or more) characteristics co-vary - e.g. if meaning and morphology co-vary. This general principle allows us to group lexemes into clusters, with a `super-lexeme' and various `sub-lexemes' which inherit most of the super-lexemes characteristics but differ in specific ways. For example, we can recognise a super-lexeme STANDv, which isa verb and has the irregular ed-form and en-form stood; it is different from STANDn, the noun in examples like (2). (2) We erected a stand for the spectators. Beneath STANDv we recognise two sub-lexemes, the intransitive STANDvi and the transitive STANDvt found in (3a) and (3b). (3) a We stand on our feet.
Each of these has a distinct syntactic valency (object or no object) paired with a distinct meaning (`be upright' versus `tolerate'). One special kind of lexeme is used for idioms (e.g.
`hot dog', `kick the bucket'). Most idioms are built round one word, whose
dependents change its normal meaning. This can be handled by recognising
the parent word as a special sub-lexeme of the
lexeme concerned: DOGsausage or KICKdie, which have
the idiomatic meaning and require the dependent which forces this meaning
- e.g. DOGsausage means `sausage-filled roll' and requires
hot as its preadjunct.
Lexical relations are relations between pairs of lexemes which are sufficiently common to constitute a general pattern. For example:
The similarities handled in terms of lexical relations are primarily matters of word-class and meaning rather than morphology, but derivational morphology may 'support' these relations, as in the second set of examples above. Lexical relations always define one lexeme in relation to the other, so they work 'in one direction' only; for example, the QUICK - QUICKLY pattern defines an 'Adverb' relation which links an adjective to its corresponding adverb. In other words, adverbs are derived from adjectives, not the other way round. In most cases it is quite clear which member is derived from which. For example,
In conclusion, lexical relations are relations from one lexeme, or class of lexemes, to another, on the basis of which the latter's meaning, grammar and possibly morphology may be predicted on the basis of the former's. For example:
A 'mixed category' is a word class which combines two or more of the main word classes. The clearest (and most discussed) example of this is the gerund, which is both a verb and a noun. Other examples in English are:
Notation: `v', as for `auxiliary verb'. `Modal verb' is a subclass of `auxiliary verb'. Its most straightforward members are the lexemes:
Each of these has two forms (present/past): will/would, shall/should, can/could, may/might. A number of characteristics distinguish these modal verbs (and others) from other auxiliary verbs: In standard English English they are always tensed - i.e. either present or past. For example, can is available as present tense, but never as an infinitive (e.g. after WILL); this is a purely grammatical constraint on the words themselves, and not a limitation due to their meanings, as can be seen by comparing CAN with its synonym (BE) ABLE. (1) I can swim now. (2) *I will can swim when I'm older. (cf. I will be able to swim ...) Similarly for all other imaginable uses which require a non-tensed verb: (3) *Can swim by the time I get back! (cf. Be able to swim ...!) (4) *It's important to can swim. (cf. It's important to be able to swim.) (5) *I've could swim for years. (cf. I've been able to swim..) (6) *Canning swim is important. (cf. Being able to swim is important.) The facts are different for Scottish speakers, who allow modal verbs to have infinitives and other non-finite forms. (7) You have to can drive a car to get that job. (8) Ah (= I) would uh (= 've) could uh done it. (9) They might could be working in the shop. The modal verbs have no -s - i.e. no subject-verb agreement - where all other verbs do. (10) She/they can swim. The meaning of the past tense in modal verbs is usually conditional (i.e. it describes a hypothetical situation which depends on some other condition being satisfied), even when the modal verb is independent. In contrast, the past tense of any other kind of verb normally has to be taken as referring to an actual event in the past - a temporal meaning, not a conditional one. The difference can be seen clearly by comparing modal could with its non-modal synonym was able. The modal can have the usual temporal meaning. (11) I could/was able to swim when I was six. But it is more often used conditionally, in which case it cannot be replaced by was able: (12) I could/*was able to swim if only I had arm-bands. This is the only kind of meaning available to most past-tense modals: could, should, might, would. Incidentally, you may wonder why these forms have to be classified as past tense. The reason is that the `sequence of tenses' rule applies to them. A sentence such as Today is Tuesday can be reported, sensibly, as that today was Tuesday provided the verb it depends on is past tense (e.g. I thought that today was Tuesday). This is true if this verb is a modal such as could so such forms must be past tense: (13) Nobody could/should/might/would forget that today was Tuesday. These modal verbs all take a plain infinitive as their sharer: (14) I can swim/*to swim/*swimming/*swum. In addition to these straightforward modals, there are others: MUST. This is straightforward except that it has no past-tense. It must be present tense because the sequence of tenses rule does not apply to it: (15) I must remember that today is/*was Tuesday. DOaux. E.g. Do you like it? I don't like it. This is straightforward except that it has no effect on the meaning of whichever tense it carries, which makes it available when an auxiliary verb is needed but the meaning does not call for any other auxiliary. BEto. E.g. You're to go to bed at once. Like all other uses of BE this is an auxiliary verb, but it counts as a modal auxiliary verb because it has to be tensed (unlike HAVEto, which is semantically similar): (16) I have/am to see Pat this afternoon. (17) Having/*being to see Pat this afternoon is a nuisance. (18) I shall have/*be to see Pat this afternoon. However, it is unlike other modal verbs in having a distinct s-form controlled by subject-verb agreement, and also in taking TO rather than a bare infinitive. OUGHT. This only has a past-tense form (historically its present tense was owe), and takes TO as its sharer. For many people it has ceased to be an auxiliary verb at all, as it uses DOaux for subject-aux inversion and negation. (19) We didn't ought to do that. USED. This is like USED, and seems to be ceasing to be an auxiliary verb (at least for people who say things like didn't used to ..). DARE, NEED. These are both modal verbs which are used only in interrogative or negative sentences - i.e. not in ordinary declarative sentences. They both have an alternative classification as a full verb, in which case they take TO rather than a bare infinitive. (20) Dare/*dares she (*to) do it? (21) She dares/*dare *(to) do it. Very strange verbs indeed! This is the informal WG name for the contrast between two kinds of finite verb: `tensed' and `imperative'. A tensed verb is (obviously) one that has a tense (past or present), whereas an imperative verb is one which is typically used for issuing commands, requests, invitations, and so on. The verbs in (1) are tensed, but the one in (2) is imperative. (1) Pat is/was quiet. (2) Be quiet! Apart from the difference in meaning, they also make different demands on their subjects. A tensed verb has to have a subject, whereas an imperative normally does not have one. (3) *Is/was quiet. Tensed verbs are widely used in subordinate clauses, but imperatives are rarely so used. (4) The child who is/was/*be quiet longest won the prize. See feature. See proper noun. A noun formed out of a verb by means of word
formation is called its nominalization; e.g. ARRIVAL is the nominalization
of ARRIVE. See finiteness.
Notation: M (! a sort of NN, contrasting with n, N and nN; if you prefer, think of 'noMinal', or think of M as the letter before N.) Most people can recognise straightforward nouns and know that `noun' is the name of a word-class; but what they are recognising are generally two subclasses of nouns: common nouns (N) and proper nouns (nN). The WG analysis recognises two other kinds of noun: pronouns (n). These are a kind of noun because they can be used as dependents almost everywhere that common or proper nouns can; the exception is that they cannot be used as complements of determiners (e.g. the book/*him). gerund in examples like Analys-ing sentences is fun. the verb has been 'turned into a noun' by the addition of -ing. See also: definiteness, countability.
null word Number is a feature, whose alternative values are `singular' and `plural'; so dogs and dog contrast in number. Every noun has a number, even if it is not signalled by inflectional morphology, because subject-verb agreement is sensitive to number; for example, they is plural, because it takes the same verb-form as dogs (e.g. They/dogs bark.) Number is also involved in the agreement between a determiner and its complement (e.g. this book but these books). Number is primarily syntactic, though it is highly relevant to both semantics and morphology. In semantics, the referent of a plural noun is a set, and in morphology a plural noun typically contains the suffix -s; so students is a typical plural noun which refers to a set of students and which contains -s. However there are cases where these relations do not hold:
(1) Pat's family are/*is buying presents for each other. (2) The staff are/*is all away. And here are some cases where number does not match the morphology:
(3) This linguistics is/*are fun.
(4) These sheep are mine. Numerals are a special subclass of common noun which may combine in the usual way with determiners (e.g. these three) and even with a handful of adjectives (e.g. a good three - though the use of a here remains unexplained). Their meaning is 'a set of n objects', where n varies with the numeral concerned; so three means 'a set of 3 objects'. They are grammatically special in two respects:
Notice the singular is here, contrasting with plural are in the following: (2) Three (bottles) are in the fridge. In this case they cannot combine with a following common noun, and are in fact regular proper nouns. It is still unclear how to handle this dual membership (as either common nouns or proper nouns), but one possibility is to use a word-formation rule which derives the proper noun from the common noun. Notes:
An old-fashioned name for 'word-class'. It
is a very unhelpful term and is best forgotten. (It is a Latin mistranslation
of a Greek term meaning `part of sentence', but this sounds more like
a grammatical function than
a word-class!) Unfortunately it has spawned another traditional term which
is widely used in linguistics as well, 'parsing' (from the Latin pars
orationis, 'part of speech'). A participle is traditionally an adjective-like form of a verb. For example, take (1). (1) Pat is keen on linguistics and studying it at UCL. The word studying must be a verb, like the other forms of STUDYv, because only verbs take objects such as it; this is not possible after adjectives, as witness the need for a preposition on after keen (compare *She is keen linguistics). Nevertheless, studying is also similar to keen, to the extent that they can be coordinated. This is because studying is a participle. There are two clear examples of participles in English: the active participle (also known as a `present participle'), already illustrated by studying, and the passive participle, illustrated by studied in (2). These differ in `voice', and together they contrast in `aspect' with infinitives and perfects. (2) Linguistics is only studied by the brightest students. Notation: `e' (from the final letter in `particle'; `p' is needed for `prepositional') `Particle' is a grammatical function, and more specifically, it is a kind of verb complement. A typical example is (1). (1) I looked up the word. Like prepositionals, a particle must be a preposition, but it must not have a complement. And like prepositionals, it is often selected by its verb, though this is not always the case. In the combination LOOK UP, no other adverb is possible without a radical change in the verb's meaning, but in PULL UP others are possible (DOWN, ALONG, etc). The distinguishing characteristic of particles is that they may separate a verb from its object, as in (1); in general verbs and their objects cannot be separated in this way. This is why we have to analyse the direction adverbs with PULL as complements, although they look in other respects like adjuncts. Most adjuncts, even short ones, cannot separate the verb and its object. Unfortunately this distinguishing characteristic only works where there is an object. With intransitive verbs it is hard to distinguish particles from ordinary adjuncts; maybe the best way to make the distinction is according to whether the adverb is selected by the verb. If the verb + adverb pair is an idiom, treat the adverb as a particle. For example, UP must be the particle of GIVE in GIVE UP because intransitive GIVE requires UP, and does not contain the ordinary meaning of UP. (Similarly for THROW UP, PASS OUT, HAND OVER, STAY OVER, etc.) A sequence verb + preposition + noun could be analysed in either of two ways, so pay attention! The preposition could be the verb's particle, with the noun as the verb's object; or it could be the verb's prepositional, or even the verb's adjunct, with the noun as its own complement. The contrast is standardly illustrated by the following pair: (2) He ran up big bills. (up is the particle of ran, and bills is the verb's object) (3) He ran up big hills. (up is the adjunct of ran, and hills is its complement) The two analyses are easy to distinguish. If it is possible to reverse the order of the preposition and the noun, the noun must depend on the verb so the preposition must be the verb's particle. (2') He ran big bills up. (3') *He ran big hills up. (If the noun depended on the preposition it would have to follow it, because complements always follow their parents.) The two analyses are illustrated here: |
|
Notation: `n', because some irregular passive participles have the suffix en - e.g. eaten, taken, broken. This distinguishes them clearly from regular past tenses. A passive participle is a verb-form such as cooked and eaten in (1). (1) The meal was cooked and eaten in half an hour. Its syntactic and semantic characteristic is passivization.
Notation: `:a' `Past' (or `past-tense') is one of the two tenses, so it is identified mainly by the morphology of the verb concerned. Past tense verbs are not always used for events (etc) that happened in the past, though this is their most typical use. For example, was typically refers to a situation that obtained in the past, as in (1). (1) I was in the library. But it does not have to be linked to past time in this way. It could be used in an if-clause, referring to a situation that is only imaginary as in (2); (2) If he was in the library, his office light would be off. Or it could even refer to a situation which is currently true, resulting from the 'sequence of tenses' rule. (3) I forgot that today was my birthday. (Cf. Today is/*was my birthday.) See also: modal verb. Notation: 'r' for 'perfect'. A perfect verb is one which has the same form as a passive
participle but which is used after the auxiliary HAVE,
as in has seen. Such verbs are often called 'perfect participles'
but in English they have nothing in common with participles other than
their morphological form.
Traditionally three grammatical `persons' (not 'people'!!) are recognised:
the first person, which is or includes the speaker
(ME, US); the second person, which is or includes the addressee
but not the speaker (YOU); and the third person, which includes every
other noun or pronoun. I reject this classification on the grounds that
if we want to pick out ME, we can do so much more easily by calling it
ME than by calling it `first-person singular'; and I don't know of any
generalisations that apply only to all first-person (or second-person)
words. The WG analysis of subject-verb
agreement makes no mention of person. The most common pronoun, which may be first or
second person (ME, US, YOU) or third person (HIM,
HER, IT, THEM - and for some speakers ONE).
A third person pronoun is most often used anaphorically,
i.e. so that it is co-referential
with some earlier noun or pronoun, its antecedent.
see prepositional See number. `Polarity' is the contrast between `positive' and `negative'. It applies to tensed auxiliary verbs, e.g. is (positive) versus isn't (negative) - hence the name 'polarity verb' which I once used for auxiliary verbs. It is also used to cover any positive/negative contrast. Some polarity contrasts are relevant to semantics but not to syntax - e.g. PERSUADE versus DISSUADE (`persuade not to..'); EVER versus NEVER (`not ever'). See also: negation
See auxiliary verb Pronouns such as my and mine are called possessive pronouns. (In traditional grammar this name is reserved for mine, since this can be used on its own like a pronoun - e.g. I'll bring mine. The form my was called a possessive adjective, but this is wrong as they are clearly determiners, which in WG means that they must also be pronouns.) Two complications have to be addressed:
Since most determiners can occur either with or without a complement, the simplest analysis is to accept that the same is true for possessive pronouns, but that in their case the presence or absence of a complement affects their form (their whole). This can be done in WG by recognising sub-lexemes which have distinct forms and also distinct complement patterns; so if we assume that mine is the default pattern, we can take MY as a sub-lexeme of MINE which (exceptionally) has a complement and is spelt my.
In this example mine clearly refers to my book, since this is what is combined semantically with bring to give the sense 'Bringing my book'. However it also refers to me, just as John in John's book refers to John. Moreover the facts of binding show that this referent is not a co-dependent of the object, so it can be co-referential with a personal pronoun but not with a reflexive pronoun:
The solution to these problems lies in a fusion analysis, in which ME + 'S = mine (or my). This explains the badness of *me's, but it also gives me the same status as John, which explains why their respective referents have the same binding characteristics. See: 'S. `Predicative adjective' is a traditional name for an adjective used as
sharer of a verb, e.g. is nice.
Notation: `P' `Preposition' is the name of one of the major word-classes. Typical prepositions are OF, IN, AT, BY, which are traditionally classified as prepositions. Prepositions (in this traditional sense) have four main distinguishing characteristics:
(1) Of whom are you thinking? (2) In which city do you live?
WG follows several other theories in expanding the traditional range of prepositions in two directions. A. First, it allows prepositions to have an optional complement, just like many verbs and determiners. For example, INSIDE can be used either with or without a noun complement, as in (3): (3) I looked inside (the box). Of course, this is really ellipsis. As with other optional complements the complement is `understood' even when there is no actual complement word - i.e. I looked inside is understood as meaning `I looked inside the box' (or the drawer or whatever is being discussed). In traditional grammar prepositions are defined strictly so that the complement is part of the definition, which means that INSIDE can't be a preposition when it is used without a complement; but its meaning remains the same, and it can be used in the same range of positions, so it would be much better to give it the same classification whether it has a complement or not - especially since the same option is available for a lot of other prepositions. The consequence, then, is that a preposition must be able to have a complement, but need not actually exploit this option in order to be a preposition. To take another example, UP is a preposition whether it has a complement (up the road) or not. This allows us to classify particles (see particle) in examples like look up the word as prepositions without complements. B. The second expansion of the class of prepositions allows the complement to be something other than a noun. One possibility is that it may be another preposition; for example, after BECAUSE we cannot use a bare noun (*because the weather), but we can use the preposition OF which in turn allows a noun (because of the weather). Such examples are traditionally classified as `compound prepositions', but it is better to recognise them as two separate words linked by a dependency. The figure illustrates the structures. A second possibility is that the complement may be a verb, as in (4). (4) We stayed inside because it was raining. Here the preposition because has a tensed verb, was, as its complement; in other words, it introduces a clause. In traditional terminology this makes it a `subordinating conjunction', but it seems wrong to put BECAUSE into a completely different word-class according to whether its complement is a noun or a verb. The objections are the same as with the choice between `adverb' and `preposition' above: the meaning of BECAUSE is the same in both cases, and a substantial number of other words show just the same alternation between two types of complement. A much simpler analysis is to allow a single lexeme, BECAUSE, to have two different types of complement without any change in its word-class. In summary, `preposition' is used for any word which may have either a noun or a verb as its complement and which has the other characteristics listed above. Two of these characteristics are especially important (as Pied-Piping only applies to traditional prepositions):
This definition means that many words qualify as prepositions which traditional grammar would have classified as adverbs or subordinating conjunctions. An attractive consequence of this decision is that it allows a unified analysis of some words which would otherwise have belonged to different word-classes according to what kind of complement they had. For example, BEFORE and AFTER may have no complement, a noun or a verb: (5) I'd seen the film before. (6) I saw them before Christmas. (7) I saw them before they saw me. In our analysis each example of before belongs to the same lexeme BEFORE, which in each case is a preposition, as illustrated below. |
|
Notation: `:p' or `:s' (explained below) `Present' (or `present-tense') is one of the two tenses. In Standard English and most regional dialects a present-tense verb has two forms, one with and the other without -s, controlled by subject-verb agreement. (1) They like ice-cream. (2) He likes ice-cream. These forms signal a feature-difference: [plural] versus [singular], notated `p' (for `present' or `plural'!) and `s' (`singular' or `s-form'). Present-tense verbs generally refer to situations that exist at the moment of speaking - either continuous states (such as liking), or habits: (3) Pat gets up at 7.30. However they can also be used with other meanings:
(4) Pat goes to Edinburgh tomorrow.
(5) Everyone dies.
(6) So he comes up to me and looks me in the eye. (This one is particularly interesting because in some non-standard varieties an -s is used even after I just in narrative, suggesting a third tense: I came/come/comes! Research needed urgently ...) See also: modal verbs Notation: `n' Pronouns are different from both common nouns and proper nouns in not allowing either determiners or adjectives before them. Traditionally they were called pronouns because they were said to be used `instead of' (Latin pro-) a noun; for example, in (1) she is supposed to replace some noun (presumably Mary). (1) Mary said that she was tired. But this isn't a particularly helpful way of thinking about their relation to other kinds of noun. For one thing, some of the most important pronouns are ME (i.e. either me or I - see pronoun form), YOU and US, but there is no ordinary noun that could have been used instead of these. For another, the traditional view implies that in some sense (1) had a previous existence in which Mary occurred before was, only to be `replaced' by she. This doesn't make sense. Most modern linguists have abandoned this way of thinking about pronouns in favour of one in which pronouns are nouns, alongside common nouns and proper nouns. They can refer to the same range of people, places, things, etc as other nouns, and can be used in all the same syntactic positions. `Pronoun' covers a wide range of sub-types, shown below. |
|
See below for the subclasses `definite' and `indefinite'. The sub-subclasses are traditional and uncontroversial (except that possessives can be used indefinitely; e.g. `my finger' refers to any old finger belonging to me and does not imply that I only have one). The definite pronouns that are not further subclassified include the traditional classes `personal pronoun' and `demonstrative pronoun/adjective'. There is no traditional name for all the unclassified indefinite pronouns, but some are called `quantifiers'. One unsolved problem is the classification of interrogative WHOSE (as in Whose book did you use?); as an interrogative it is indefinite but as a possessive it is definite, so we cannot classify it simply as `interrogative and possessive'. One of the peculiarities of pronouns is that some of them may take a common noun as complement; for example, YOU may be used either on its own or with a common noun, and similarly for SOME: (2) You (students) live in interesting times. (3) Some (pronouns) take a complement. In other modern grammatical theories, those that do allow a complement (or, alternatively, that actually have a complement - a different matter) are called `determiners'. Another peculiarity is the variation of shape found in many pronouns:
Most of the ordinary definite pronouns (alias `personal pronouns') have different forms according to the pronoun's grammatical function:
(1) She saw me/*I. (2) She came with me/*I. (3) I/*me came. The second form in each of these pairs is only used as the subject of a tensed verb, so there is no doubt (or dispute) that the `basic' form is the non-subject one, e.g. me; this is therefore used as the name of the lexeme: ME, US, HER, HIM, THEM. For some (archaic or super-formal) speakers, the interrogative or relative pronoun WHO also has two distinct forms:
In this case the special form is whom, which is used with all the functions except subject of a tensed verb; unlike the personal pronouns, the lexeme is named after the subject form, WHO. In my opinion, this contrast of forms is not an example of case. There are two main reasons for believing this:
(4) Me/*I and her/*she went together. This kind of interaction is not found in any language (such as German) that has `proper' case, and is a real problem for any analysis in terms of cases imposed `top-down' on every noun. This would require a very odd rule indeed: (5) The case of the subject of a tensed verb is nominative, unless it is coordinated; then it is accusative. It is relatively easy to accommodate (though still a bit problematic) if the alternative forms are selected by specific rules that apply only to a handful of lexical items: instead of a rule like (6), we have a more specific one like (7). (6) When ME is used as the subject of a tensed verb, its form is I. (7) When ME is used as the subject of a tensed verb and not coordinated,
its form is I. Notation: `nN' Proper nouns (a traditional term) are names, and are generally written with a capital letter - Pat, Wednesday, London, Thames, England. The main syntactic difference between them and common nouns is that, although they are usually singular and always countable, they do not need a determiner - or else they have to have one specific determiner, the. (In English this is true for rivers, mountains and some countries; in many languages it also applies to personal names.) On the other hand, unlike pronouns they can be modified by adjectives and can combine with some determiners (e.g. poor John, last Wednesday, our Pat, this Wednesday, central London, the mighty Thames, southern England). `Proper noun' can be divided into several distinct subclasses according to the kind of referent they allow (people, days, towns, etc) and how they can combine with other words. For example, given names and family names can combine with each other, and also with a preceding title, giving combinations like Mrs Mary Brown, or Mary Brown, or Mrs Brown, but not *Mrs Mary. The syntax of proper nouns remains to be written. See also: definiteness.
One reciprocal pronoun is each other. Although written as two
words, this is probably just one. The other reciprocal pronoun is one
another. Reciprocal pronouns are covered by the same principles of
anaphora as reflexive
pronouns. Most tensed auxiliary verbs have two forms, a full one (e.g. is, will) and a reduced one ('s, 'll). The distinction is important for sociolinguistics because of the formality difference, the reduced forms being much more casual than the full ones. From this point of view, negative auxiliaries (formed with -n't) are also reduced forms, so both positive and negative verbs may be reduced. See also: auxiliary inflection. The reflexive pronouns all end in -self (or selves):myself, yourselves, etc. They are used to refer anaphorically to another noun, its antecedent, so the relationship between the two is an example of co-reference; e.g. in Mary saw herself, herself is co-referential with Mary, meaning that both words refer to the same person. In simple cases the reflexive pronoun is the object of a verb and its antecedent is the subject of the same verb; but other more complex relationships are also possible:
The relations between reflexive pronouns and their antecedents are handled
by Binding theory. The kind of wh-pronoun which is used in relative
clauses (e.g. which in the present sentence). The relation
between a relative pronoun and the noun that it depends on, its antecedent,
may be co-reference; e.g. in Mary,
who is a linguist, the pronoun who refers to Mary, so the words
Mary and who have the same referent. However this relation
varies according to whether the relative clause is non-defining
(as in this example) or defining, as in a student who is a linguist.
In the latter case, the relative pronoun's referent is the same as the
sense of the antecedent - i.e. the concept
'student who is a linguist', rather than the specific person referred
to. One controversial question is whether the relative pronouns include that. The standard analysis takes that as a mere 'complementiser' - i.e. like THATsub. However I think on balance that at least some times, and for some speakers, it is a relative pronoun when it introduces a relative clause, as in the book that I bought. See the entry for that for discussion. Sentence adverbs are adverbs such as FORTUNATELY, THEREFORE or ACTUALLY whose meaning is a comment on the whole sentence rather than contributing to the sense of some word. For example, in Fortunately he passed, the word fortunately is a comment on the fact that he passed, in contrast with (say) Recently he passed, where recently helps to define the event described. In WG, sentence adverbs have just the same syntactic analysis as other kinds of adjunct of the root verb, which explains why they can occur in the same positions as other adjuncts, including the pre-adjunct position between subject and verb (or just after a tensed auxiliary).
The peculiarities of sentence adverbs emerge in the semantic analysis, not in the syntax. Fortunately he passed has much the same meaning, and semantic structure, as (2).
See number. The term `subjunctive' is inherited from traditional grammar, where it plays an important part in the analysis of highly inflected Indo-European languages like Latin, German and French, all of which do have verb inflections which are typically used in subordinate clauses - 'subjoined' to the main clause. For example, in French some subordinating conjunctions such as afin que, 'in order that', take a subjunctive verb (rather than the usual ordinary tensed verb). Some grammarians think that English has 'subjunctive' forms as well, but I think it's misleading to apply the term to English for just the same reasons that it is misleading to say that English has case: the relics of the old Indo-European subjunctive are so partial and marginal that they are easier to handle as minor lexical exceptions. In any case, different grammarians apply the term to different patterns, so the term is very confusing. Both the following are sometimes said to contain a subjunctive (underlined): (1) If the President were here, he would take the chair. (2) I propose that Pat be secretary. These patterns are both quite marginal - the first is restricted to the one verb, BE, and the second is limited to a few parent verbs (such as PROPOSE), and seems to be outside the competence of many speakers, especially in the UK. These forms do need an analysis, but they are rare and need not take high priority. (The fixed expression If I were you is more common but many people say If I was you, and the meaning is the same as the one available to any past tense verb, so it seems best to analyse it as a past tense.) See super-lexeme. See preposition. 'Subordinator' is an informal word for any word which 'introduces' a subordinate clause - i.e. a word which:
The subordinators in English are:
These words can all be analysed as the root of the subordinate clause. It is unclear whether 'subordinator' should be recognised as an 'official' part of a WG grammar. If it was, then most subordinators would be mixed categories, as they would belong to some other word class as well. Some adjectives and adverbs have inflections which are traditionally called 'superlative' and which are typically signalled by the suffix -est: biggest, soonest, most. Superlative adjectives are syntactically distinct from other adjective inflections in that:
These noun-like characteristics suggest that superlative adjectives may in fact be a mixed category which is both an adjective and a noun. See also comparative. A lexeme which subsumes other lexemes, which are its `sub-lexemes'. E.g.
HAVE subsumes HAVEperfect, HAVEexperience
and HAVEto. Tense is the contrast between `past' and `present'. Every tensed verb has a tense, and its tense is generally decided by its form rather than by its meaning; for example, was and took are always past tense, even when used with a non-typical time-reference. (See past.) Tense is related to the time of the event, which is shown in the semantic structure, but the relationship is quite loose: These classes apply to single words, so there's no `future' tense in English. An example such as will go consists of two words, a present-tense WILL followed by an infinitive GO. It's true that present-tense will usually refers to a time in the future, but this is because of the meaning of WILL rather than tense inflection. Any present-tense event verb (not a state verb) can refer to an event in the future: (1) The train leaves soon. A present-tense verb in a dependent clause may take its time reference from a superordinate verb. (2) When the train leaves I shall be on it. The sequence-of-tenses rule allows a past-tense verb to refer to a situation in the present, by a kind of superficial tense-agreement with a superordinate verb. For example, in (3) the only reason for was rather than is is the past tense of forgot: (3) I forgot that today was my birthday. Most verbs have a distinct inflectional form for their past tense (walk-ed, took), but there are a handful of irregular verbs such as HIT whose present and past tenses are the same; so for example (4) could be either present or past tense: (4) They hit us. It's easy to decide the tense of such cases, however: just make the subject singular. If it is present tense, the rule of subject-verb agreement will produce an -s on the verb; and if no -s appears, it must be past. (5) He hits us. (present) (6) He hit us. (past) See also: modal verb, past, present. A tensed verb is a verb that has a tense - past or present. Tensed verbs are finite, and contrast with the other type of finite verb, imperative verbs. Traditionally a distinction is made between `transitive verbs', which do have an object, and `intransitive verbs', which don't. (A third category of `ditransitive' verbs which allow two objects is also recognised.) This information was previously given in WG entirely by valency statements about the individual verbs, so there was no need to duplicate it by recognising `transitive' and `intransitive' as distinct word-classes as well. See word-class for the logic behind this conclusion. However more recently some of us (led by And Rosta) have tended to be more sympathetic to the traditional approach because this allows much more general statements about valency - what other theories call linking rules. Maybe a verb-class called 'transitive' would help us to generalise about the meanings that objects tend typically to express. For example, if objects normally express the 'er' of the verb's sense, then any grammar should say so; but the only link between the syntactic dependency 'object' and the semantic relation 'er' is via the verb-type 'transitive'. `Valency' is a term which is part of the European tradition of dependency
grammar. It deals with the dependency relations that are expected
by individual words, so `valency dictionaries' have been produced which
list all the possible valency patterns for thousands of words (especially
verbs). A word is said to have such-and-such a valency, which shows how
it `bonds' with other words. (The term `valency' is borrowed from chemistry,
where it refers to the bonding properties of atoms.) The term certainly
includes all complements and also
all subjects (our `valents'), and certainly
does not include adjuncts; but valency
linguists disagree about the kind of information that should be included
in a word's valency. For example, should it show the word-classes and
inflectional features of the various valents? Can it include details of
word order? and so on. It remains to be seen whether these arguments matter.
The most important word-class of all. (In Latin, verb-um means simply `word'.) Verbs are important because:
See also: full verb, auxiliary verb. In English, verbs have more inflections than any other word classes, but even so the number is small compared with many other languages. The main contrast distinguishes between finite and non-finite verbs, but each of these general inflections is further subdivided, and further subdivisions apply to finite auxiliary verbs. The inflections are shown in the following table, together with the letters that are used to abbreviate the inflection. (These letters make up the almost-memorable words spam or maps for finites, and ring for non-finites!) This table does not include gerunds, which have their own special complications; the notation for a gerund is :N (e.g. if walking is a gerund it is V:N). |
|
|
morphology | letter | example | ||
| regular | irregular | ||||
| finite | tensed
(present) |
stem | stem | p | (We) applaud. |
| imperative | stem | stem | m | Applaud! | |
| tensed, | stem + -ed | stem (+ ed) | a | (We) applauded/chose. | |
| tensed, | stem + -s | stem + -s | s | (She) applauds. | |
| non-finite | stem | stem | i | (We may) applaud. | |
| perfect | stem + -ed | stem + -en | r | (We have) applauded/ chosen | |
| participle
(active) |
stem + -ing | stem + -ing | g | (We are) applauding. | |
| participle, | stem + -ed | stem + -en | n | (We were)
applauded/chosen. |
|
The diagram below shows the hierarchical relationships among these inflections.
|
Notes
This is the (very) traditional (and not very helpful) name for the contrast between active and passive verbs. In WG it is applied just to participles: `passive' therefore contrast with the other kind of participle, which are traditionally called 'present participles' but could be called 'active participles', but which in WG are treated simply as default ('ordinary') participles. For example, for the verb SHOW, voice contrasts the passive shown with the default participle showing. Both can be used in much the same range of dependency relationships, which is why they are called participles. For example: (1) Pat was showing visitors round the college. (active) (2) Pat was shown round the college. (passive) In both these examples Pat is the subject of the participle as well as of was (see raising). Similarly, in the next examples people is the subject of both the participles (see relative clause). (3) The people showing visitors round were all lecturers. (active) (4) The people shown round were all visitors. (passive) The difference between the two participles lies in the semantic role of the subject. For an active participle, the subject has the normal semantic role (e.g. the `show-er' in the above examples) but for the passive participle the normal subject role is replaced by another semantic role - typically, but not necessarily, that of the normal object (the `show-ee'). This difference is handled in WG by giving the grammatical function `object' to the subject; so in (2), the word Pat is both subject and object of taken. As for the `normal subject' (the `show-er'), with a passive verb this is either not expressed at all, or is expressed by a noun linked to the verb by the word BY. (5) Pat was shown round by a lecturer. See also: passivization, prepositional passive. A pronoun which is spelt with an initial wh. (How is also included, though the h and w are in an unusual order!):
Wh-pronouns are either interrogative pronouns or relative pronouns. Wh-pronouns allow extraction, meaning that they have a double dependency: they depend on the first verb as its `extractee', but they also have some kind of dependency which would normally put them later in the clause, e.g. as object of this verb or of some later verb.
`Word' is the basic unit of language, not only in Word Grammar but also in virtually every other theory. Moreover, this agreement is not just a matter of terminology; we almost always agree about how to distinguish words from non-words at least when discussing English. This is partly because we have been using word-spaces since we first learned to read and write, but the conventions for using word-spaces generally fit the realities of the language well. Dividing a text into words is one of the most important parts of grammatical analysis, and it is also one of the most difficult (ask anyone who has tried to get a computer to segment speech into words!); but it is a task which we tend to take for granted. What makes words stand out in grammatical analysis in contrast with units that are either smaller than words (morphemes, phonemes) or larger than words (phrases, sentences) is that they typically combine a very large number of different characteristics. Unfortunately for grammarians, there are exceptions, but the regular tendencies are so overwhelming that these exceptions don't diminish the importance of the word.
One of the most controversial characteristics of WG is its view of words
as actions - a word is something you do, like tying your
shoe-lace or waving good-bye, and not something separate from the doing.
A word in the grammar is an abstract internal representation, but it is
a representation of a particular kind of action; for example, DOG is a
complex action in which the production of /dog/ is accompanied by certain
ideas and intentions (including the meaning `dog'). One of the attractions
of this approach is that words fit into the general conceptual
hierarchy under `action' and `communication', whereby they inherit
various characteristics which are important in semantics (e.g. deixis)
and sociolinguistics - speakers,
addressees, times and places.
Notation: a word's word-class is shown by a label (e.g. `V' for `full verb') immediately beneath the word. This label may be followed by `:' then an inflection label. A word-class is (not surprisingly) a class of words, a collection of words that allows generalisations that are not otherwise possible. Standard examples are `noun', `verb', `adjective', `pronoun' and `relative pronoun'. These terms are used by virtually all linguists as names for virtually the same lists of words, and recent work in linguistics has only added a few terms (`determiner' and `complementiser' are the obvious examples) to the list which we have inherited from the Greek and Roman grammarians. This agreed list is something to be appreciated (especially by novices!), but not to be taken for granted. We stand on the shoulders of the intellectual giants who recognised the classes centuries ago, but we must remember that they were first recognised in languages that were structurally quite different from modern English. Some confusion has crept in over the centuries, especially in the jump from Latin to English, and we need to be clear about the nature and purpose of the word-classes. Here are some general principles which arise directly out of the rather harmless-looking definition in the first sentence of this paragraph, but which are definitely not obvious to all novices, or even to all professional linguists! The members of a word-class are individual words, not pairs of words. Therefore it is nonsense to talk about `the verb TO LOVE'. TO is one word, and LOVE is another; LOVE certainly is a verb, but the classification of TO is less obvious. Similarly for will love as a form of `the verb TO LOVE' and `the preposition BECAUSE OF'. (There may be sequences that we write as two words but which are really just one, such as out of parallelling into, or each other; but that is another question.) The justification for a word-class is to allow us to express `generalisations that would not otherwise be possible'. For example, `noun' allows us to say that the object of a verb must be a noun, and also that the complement of a preposition must be a noun; without `noun' we should have to list all the words which can have each of these grammatical functions, which would be inconvenient; but the really serious problem is that the two lists of words (for objects of verbs and for complements of prepositions) would be the same. An analysis without a word-class containing all these words would miss the generalisation that words which can occur as objects of verbs can also occur as complements of prepositions. This would not just be awkward, it would be wrong. The problem would deepen as the number of rules that apply to the same list of words increased; for a `good' word-class there may be dozens of such rules. Some word-classes are very `good' in this respect - examples are `noun', `verb' and `auxiliary-verb'. On the other hand, if we take this principle seriously (as we should) then we must accept a number of consequences.
The WG word-classes for English are organised in an inheritance hierarchy below `word'. It contains six main classes, two of which are then further subdivided, but alongside these word-classes the hierarchy also includes a large number of class-less lexemes. The overall system is shown in the figure, where the letters under the lowest names are the ones for use in sentence analysis. |
|
Further subdivisions are given in the entries on `pronoun',
`proper noun', 'numeral'
and `auxiliary verb'. Word-formation is the (study of the) relationships among lexemes that are partially similar to each other. For example, FARM and FARMER must belong to different lexemes because they have different stems and senses and these connections could not be expressed if they were treated as a single lexeme; but they are closely related in both form and meaning. Many other pairs of lexemes are related in the same way, so we need to be able to express generalisations about relations between lexemes by means of `word-formation rules'. Some such rules involve regular morphological variation, in which case they belong to derivational morphology. Others involve no morphological difference at all (e.g. between the WALKn and WALKv, the noun and verb), or an irregular one (e.g. MALE: FEMALE). The name `word-formation' is used because these patterns often form the basis for creating new lexemes (eg. XEROXER), but there are many other ways of creating new vocabulary, not least of which is borrowing material from another language. One difficulty with the study of word-formation is that it lies uncomfortably between 'rule-governed behaviour' (e.g. XEROXER) and 'creativity' (e.g. borrowing).
|
