book excerptise:   a book unexamined is wasting trees

A student's introduction to English grammar

Rodney D. Huddleston and Geoffrey K. Pullum

Huddleston, Rodney D.; Geoffrey K. Pullum;

A student's introduction to English grammar

Cambridge University Press, 2005, 312 pages

ISBN 0521848377, 9780521848374

topics: |  linguistics | english | grammar

Texts on linguistics differ wildly in the assumptions they bring to the task. This text is notable for the lack of overt theoretical dependencies. It is "minimalist" in the sense of attempting to make the minimal assumptions needed to show show you some of the best practices and general principles for analyzing not only English, but any language.

It tries not to get embroiled in armchair theories that haven't been able to get very far. It is pragmatic and the goal is to come up with a compact analysis of English, and present arguments for the decisions made at every step. You may not agree with their arguments, but the non-dogmatic presentation leaves you room for you to argue your own case as well.

Reason in Linguistics

In the typical grammar book in the mould of Nesfield, say, you are thrown into the turbulent river of a language, and very soon find that the familiar has started to become quite unfamiliar. This bewildering process results in your memorizing a large set of terms and arcane mechanisms, which are given to you as the only way things should be done.

Huddleston and Pullum on the other hand, try to reason with you, giving you arguemnts for every decision they are taking. This means that there is no certainty, no solid ground as you progress. You find yourself floundering in endless irregularities. But the reasons are given for every decision, and in the end, if you manage to keep going, you emerge much more knowledgeable about linguistics than you were when you jumped in...

Indeed this may be one of the finest texts for learning linguistics.

Difficulties in trying to discover a logic for language


The five-decade long attempt to find a formal logic that underlies
language has in these days, floundered on the rock of ambiguities, and
eventually had to take recourse to limiting its subject matter to that which
fit the theory (what Lakoff has derisively called "the
performance-competence game"). 

The siren call of logic has attracted us to language, but after the logicism
of every Panini, there comes the mysticism of a Bhartrihari.  The logic of
languagee seems so abundantly clear, yet it is an illusion.

    Life is... a trap for logicians. It looks just a little more mathematical
    and regular than it is; its exactitude is obvious, but its inexactitude
    is hidden; its wildness lies in wait.  - GK Chesterton, The Paradoxes of
    Christianity

Huddleston and Pullum reveal the wildness in language from their very
opening pages.

The inexactitudes of language analysis


The difficulties of trying to discover a logic for language are driven home
from the start.  The book opens with an analysis of the "past tense":

  The term past tense refers to a grammatical category for verbs:
    likes : present tense form
    liked : past tense form
  Usual definition:  past tense expresses or indicates a time that is in the
 	past.

  But it's hardly as straightforward as that. The relation between the
  GRAMMATICAL category of past tense and the SEMANTIC property of making
  reference to past time is much more subtle.  p.6

The difficulty is illustrated with examples like:
	a. If he said that, he was wrong.
	b. If he said that, she wouldn't believe him.
where said in (a) refers to a time in the past, but in (b)
we are talking about something hypothetical, more aligned with the future
than the past.

The term 'past tense' (and most grammatical terms) - are not unique to the
grammar of English.  Therefore,

    At one level we need to identify what is common to the forms that qualify
    as past tense in different languages. We call this the GENERAL level.

    At a second level we need to show, for any particular language, how we
    decide whether a given form belongs to the past tense category. This is
    the LANGUAGE-PARTICULAR level.

Thus it is made very clear that this is not a syntax-centered guide;
semantics informs its decisions at every step.

Though it is focused on English, the presentation is lucid and balanced,
and highlights the many dichotomies and ambiguities facing any attempt to
come up with a systematic procedure for specifying which linguistic
structures constitute the "standard" variety for any language.

Although no ideological stance is taken, there are no references to Chomsky
whatsoever.  Jespersen is referred to in areverential terms - he "made many
important innovative proposals, too many of which were overlooked."  Henry
Street is applauded, and of course, Randolph Quirk.




Excerpts


Many varieties of English, but little variation in grammar

The many varieties of English spoken around the world differ mainly in
PRONUNCIATION (or 'accent'), and to a lesser extent in VOCABULARY, and those
aspects of language (which are mentioned but not covered in detail in this
book) do tend to give indications of the speaker's geographical and social
links. But things are very different with GRAMMAR, which deals with the form
of sentences and smaller units: clauses, phrases and words. The grammar of
Standard English is much more stable and uniform than its pronunciation or
word stock: there is remarkably little dispute about what is GRAMMATICAL
(in compliance with the rules of grammar) and what isn't.

very few differences between Standard forms of
American English (AmE) and British English (BrE), e.g.
	BrE:   She may have done
	AmE:   She may have
but for the most part using Standard English doesn't even mark this
difference

Non-standard Englishes

Alongside Standard English there are many robust local, regional, and social
dialects of English that are clearly and uncontroversially NON-STANDARD. They
are in many cases familiar to Standard English speakers from plays and films
and songs and daily conversations in a diverse community. In [1] we contrast
two non-standard expressions with Standard English equivalents, using an
exclamation mark (!) to indicate that a sentence belongs to a non-standard
dialect, not the standard one.

[1] 	STANDARD 				NON-STANDARD
   i a. I did it myself.                   b. !I done it myself.
  ii a. I haven't told anybody anything.   b. !I ain't told nobody nothing.

[All non-standard forms: marked with asterisk "*":]
	*Ran the away dog

Done in [ib] is a widespread non-standard 'past tense' form of the verb do,
corresponding to Standard English did - in the standard dialect done is what
is called a 'past participle', used after have (I have done it) or be (It was
done yesterday).

In [ii] there are two differences between the standard and non-standard
versions.  First, ain't is a well-known non-standard form (here meaning
"haven't"); and second, [iib] exhibits multiple marking of negation: the
clause is marked three times as negative (in ain't, nobody, and nothing),
whereas in [iia] it is marked just once (in haven't).

... many languages (e.g. French, Italian, Polish, and Russian) show
multiple marking of negation similar to that in [ 1 ii]. It's a special
grammatical fact about Standard English that it happens to lack multiple
negation marking of this kind.

[in-between : a [Standard English] form used by some speakers but not all:
	%It mayn't happen
such forms indicated by "%".

Formal and informal style


The distinction between standard and non-standard dialects of English is
quite different from the distinction between formal and informal style, which
we illustrate in [2] :

[2]
      i a. He was the one with whom she worked.  [FORMAL]
        b. He was the one she worked with.       [INFORMAL]

     ii a. She must be taller than I.            [FORMAL]
        b. She must be taller than me.           [INFORMAL]

In these pairs, BOTH versions belong to the standard dialect, so there is no
call for the exclamation mark notation. Standard English allows for plenty of
variation in style depending on the context in which the language is being
used.

Perhaps the key difference between style and dialect is that switching
between styles within your native dialect is a normal ability that everyone
has, while switching between dialects is a special ability that only some
people have. Every speaker of a language with style levels knows how to use
their native language more formally (and maybe sound more pompous) or talk
informally (and sound more friendly and casual). But to snap into a different
dialect is not something that everyone can do.

Descriptive vs Prescriptive : Grammaticality as a degree

Perhaps the most important failing of the bad usage books is that they
frequently do not make the distinction we just made between STANDARD vs
NONSTANDARD dialects on the one hand and FORMAL vs INFORMAL style on the
other. They apply the term 'incorrect' not only to non-standard usage like
the [b] forms in [1] but also to informal constructions like the [b] forms
in [2] .  But it isn't sensible to call a construction grammatically
incorrect when people whose status as fully competent speakers of the
standard language is unassailable use it nearly all the time. Yet that's what
(in effect) many prescriptive manuals do.

[3] Such common expressions as it's me and was it them? are incorrect,
    because the verb to be cannot take the accusative: the correct
    expressions are it's I and was it they?  But general usage has led to
    their acceptance, and even to gentle ridicule of the correct version.4
    	  - B.A. Phythian, A Concise Dictionary of Correct English (1979).

This book [suggests] that there is a rule of English grammar requiring a
nominative form where a pronoun is 'complement' of the verb be. But there
isn't any such rule ... just about everyone says It's me.

Grammar : Terms


SYNTAX is the study of the principles governing how words can be assembled
     into sentences
       I found an unopened bottle of wine is admissible but
	*I found a bottle unopened of wine is not);

MORPHOLOGY deals with the internal form of words (unopened has the parts un',
     open, and ·ed, and those parts cannot be combined in any other order).5

Defining grammatical terms: e.g. Past tense


Though terms such as noun, verb, pronoun, subject, object, tense, may be
known, their understanding is often unsatisfactory.

The term past tense refers to a grammatical category for verbs:
    likes : present tense form
    liked : past tense form
Usual definition:  past tense expresses or indicates a time that is in the
	past.

But it's hardly as straightforward as that. The relation between the
GRAMMATICAL category of past tense and the SEMANTIC property of making
reference to past time is much more subtle.

[4]  DEFINITION WORKS			DEFINITION FAILS
  i a. The course started last week.   b. I thought the course started next week.
 ii a. If he said that, he was wrong.  b. If he said that, she wouldn't believe him.
iii a. I offended the Smiths.		 b. I regret offending the Smiths.

The usual definition works for the [a] examples, but it completely fails
for the [b] ones.

In [i] the past tense started in the [a] case does locate the starting in
past time, but in [b] the same past tense form indicates a (possible)
starting time in the future.

So not every past tense involves a past time reference.

In [ii] we again have a contrast between past time in [a] and future time in
[b].  In [a] it's a matter of whether or not he said something in the
past. In [b] it's a matter of his possibly saying it in the future: we're
supposing or imagining that he says it at some future time; again, past
tense, but no past time.

In [iii] we see a different kind of contrast between the [a] and [b]
examples. The event of my offending the Smiths is located in past time in
both cases, but whereas in [a] offended is a past tense form, in [b]
offending is not. This shows that not every past time reference involves a
past tense.

It is important to note that we aren't dredging up strange or anomalous
examples here. The examples in the [b] column are perfectly ordinary. You
don't have to search for hours to find counterexamples to the traditional
definition: they come up all the time. They are so common that you might well
wonder how it is that the definition of a past tense as one expressing past
time has been passed down from one generation to the next for over a hundred
years and repeated in countless books.

What we've shown in [4] is that the traditional definition fails badly at the
language-particular level: we'll be constantly getting wrong results if we
try to use it as a way of identifying past tense forms in English. But it is
on the right lines as far as the general level is concerned.

Correlation between grammatical form and meaning : Primary meanings

What we need to do is to introduce a qualification to allow for the fact
that there is no one-to-one correlation between grammatical form and
meaning. At the general level we will define a past tense as one whose
PRIMARY or CHARACTERISTIC use is to indicate past time. The examples in the
right-hand column of [4] belong to quite normal and everyday constructions,
but it is nevertheless possible to say that the ones in the left-hand
column represent the primary or characteristic use of this form.  That's
why it is legitimate to call it a past tense.

But by putting in a qualification like 'primary' or 'characteristic' we're
acknowledging that we can't determine whether some arbitrary verb in
English is a past tense form simply by asking whether it indicates past
time. At the language-particular level we need to investigate the range of
constructions, such as [4ib/iib] , where the forms used are the same as
those indicating past time in the [a] construction - and the conditions
under which a different form, such as offending in [iiib] , can be
associated with past time.

Imperatives


Typically defined as a "command" - but may cover a much broader range of
situations:
   - commands (Get out!), offers (Have a pear),
   - requests (Please pass pass the salt),
   - invitations (Come to dinner),
   - advice (Get your doctor to look at it),
   - instructions (To see the picture click here), etc.

The broader term DIRECTIVE is more suitable.

But even with this change from 'command' to 'directive', the definition still
runs into problems:

[5] DEFINITION WORKS		   DEFINITION FAILS
  i. a. Go to bed. 		   b. Sleep well.
 ii. a. Please pass me the salt.   b. Could you pass me the salt?

In [i] [b] is not a directive - I'm not directing you to sleep well, I'm just
wishing you a peaceful night.

In [ii] we have the opposite kind of failure. Both examples are directives,
but in terms of grammatical structure, [b] is an interrogative (as seen in
questions like Are you hungry? or Could you find any tea?). But it is
being used as a directive - I'm asking for the salt.

Again the textbook definition is along the right lines for a general definition but,
as before, we need to add an essential qualification. An imperative can be defined at
the general level as a construction whose PRIMARY or CHARACTERISTIC use is to
issue directives.

Relation of meaning to grammar


A definition or explanation for English must specify the grammatical
properties that enable us to determine whether or not some expression is
imperative.  And the same applies to all the other grammatical terms we will
be making use of in this book.

In dismissing the two meaning-based definitions we just discussed, we don't
mean to imply that meaning will be ignored in what follows. We' ll be very
much concerned with the relation between grammatical form and meaning. But we
can only describe that relation if the categories of grammatical form are
clearly defined in the first place, and defined separately from the kinds of
meaning that they may or may not sometimes express.


2 a rapid overview


Two kinds of sentence


The syntactically most straightforward sentences have the form of a single
clause or two or more coordinated clauses, joined by a coordinator (e.g., and,
or, but). We illustrate in [ 1 ] :

[i] CLAUSAL SENTENCES (having the form of a clause) I
    a. Kim is an actor.
    b. Pat is a teacher.
    c. Sam is an architect.

ii COMPOUND SENTENCES (having the form of a coordination of clauses)
   a. Kim is an actor; but Pat is a teacher.
   b. Kim is an actor; Pat is a teacher; and Sam is an architect.


Clause, word and phrase


clause = subject followed by a predicate.  Subj/Pred may be single words, or
larger.

[2]   Things change.         Kim  left.           People complained.
      Subj   Pred            Subj Pred            Subj   Pred

[3]   [All things] change.   Kim  [left early].   [Some People] [complained about it].
      Subj         Pred      Subj Pred            Subj          Pred
      NP                          VP              NP            VP

NP = NOUN PHRASES, with a noun as their HEAD
     elements other than HEAD are DEPENDENTs.
     can also be just a noun like Kim or things.
     [older grammars used to distinguish between single words from phrases]

VP = VERB PHRASES: with verb as head.   "early" and "about it" are
     dependents.

SUBJECT: default position: before the verb
	 in INTERROGATIVES : just after the verb
	    [The clock has stopped  --> Has the clock stopped?]

One useful TEST for finding the subject of a clause, therefore, is to turn
the clause into an interrogative and see which expression ends up after the
(first or only) verb.

Functions and Categories


Subject is a FUNCTION = relation with other parts of the clause
NP is a CATEGORY: grammatical category (independent of function, not an NP of
	anything).

the correspondence between function and category is often subtle and
complex. There may be clear tendencies (like that the subject of a
clause is very often an NP), but functions may be filled by many categories.

[5] ONE FUNCTION, DIFFERENT CATEGORIES       ONE CATEGORY, DIFFERENT FUNCTIONS
    i a. [His guilt] was obvious.            b. [Some customers] complained.
   ii a. [That he was guilty] was obvious.   b. Kim insulted [some customers]

Words and lexemes


The term 'word' is commonly used in two slightly different senses.
The difference can be seen if we ask how many DIFFERENT words there are in a
sentence such as:

   [6] They had two cats and a dog; one cat kept attacking the dog.

In this book:
    cat and cats are different WORDS but the same LEXEME.

INFLECTION:
- cat and cats are different INFLECTIONAL FORMS of the same lexeme, CAT.
- take, takes, took, taking, taken: inflectional forms of verb lexeme TAKE.

Word categories (parts of speech)


The traditional term 'parts of speech' applies to what we call categories
of words and lexemes. Leaving aside the minor category of interjections (covering
words like oh, hello, wow, ouch, etc., about which there really isn't
anything interesting for a grammar to say), we recognise eight such categories:

[7]i NOUN          The DOG barked.       That is SUE.           WE saw YOU.
  ii VERB          The dog BARKED.       It IS impossible.      I HAVE a headache.
 iii ADJECTIVE     He's very OLD.       It looks EMPTY         I've got a NEW car.
  iv DETERMINATIVE THE dog barked.       I need SOME nails.     ALL things change.
   v ADVERB        She spoke CLEARLY.    He's VERY old.         I ALMOST died.
  vi PREPOSITION   It's IN the car.      I gave it TO Sam.      Here's a list OF them.
 vii COORDINATOR   I got up AND left.    Ed OR Jo took it.      It's cheap BUT strong.
viii SUBORDINATOR  It's odd THAT they   I wonder WHETHER       They don't know IF
                      were late.           it's still available.   you're serious.


[Note: the PoS need not be same in all languages.
 HINDI: adverbs can becoe PP  actually -> vAstav mein  ]

categories i-vi can be the head of a corresponding phrase (NP, adjP etc).
Coordinators and subordinators can't.

Nouns

In any language, the nouns make up by far the largest category in
   terms of number of dictionary entries... (about 37 per cent of the words
   in almost any text).

 - meaning: physical objects or abstract (absence, fact, idea, computation)
 - inflection: singular / plural : cat / cats; woman / women;  but not all
 - function: mostly as head of NPs, and NPs can be subject.

unlike traditional grammars, proper nouns, and even pronouns are also treated
as Nouns here.

Verbs


SITUATION = whatever is expressed in a clause

 - meaning: the verb is the chief determinant of what kind of situation it is
	- action (I opened the door),
	- other event (The building collapsed),
	- state (They know the rules), etc.
 - inflection: most distinctive grammatical property of verbs.
   	inflectional contrast of TENSE between PAST and PRESENT.
	A past tense marked by inflection is called a PRETERITE.
		Preterite: She worked in Paris.  He knew the answer.
        present tense has two forms, if subj is singular or plural.
		She works in Paris. They work in Paris.
		He knows the answer. They know the answer.
	other inflections also.  e.g. They are working in Paris.
 - function: head of VP which acts as predicate in a clause.
   	As head, determines other arguments.  e.g.
		She left the airport.  But *She arrived the airport.

AUXILIARY VERBS: subclass of verbs. in interrogatives, can occur before the
	subject.  Can, do, are, has.  It WILL rain.
	CAN you speak French?  ; * Speak you French?  ; DO you speak French?
other verbs are LEXICAL VERBS

Adjective

 - meaning: properties of concrete or abstract things.
	if combined with verb be can describe a state (Max was jealous).

 - two functions:
   - ATTRIBUTIVE (HOT soup)  [modifier to a following noun / NP]
   - PREDICATIVE (the soup is HOT)  [after be or a small class of similar
		verbs: become, feel, seem, etc. )

 - GRADABLE: most typical (central) adjectives are GRADABLE - e.g. big,
	good, hot, jealous, old - all denote properties that can be possessed
	in varying degrees...

    The degree can be indicated by a modifier, as in fairly big,
    surprisingly good, very hot, extremely jealous, three years old - and
    can be questioned by how: How big is it?, etc .

    One special case of marking degree is by comparison, and with short
    adjectives this can be expressed by inflection of the adjective :

			Kim is old.
	COMPARATIVE :	Kim is older than Pat.
    	SUPERLATIVE :	Kim is the oldest of them all.

    This inflectional system is called GRADE.  But it is not only
    adjectives that can be gradable, on the other hand the attributive and
    predicative functions are more specific to adjectives.

Determinatives


(a) Definiteness

	The two most common determinatives are the words the and _a_.

	These function as determiner in NP structure. The marks the NP as
	definite and _a_ as indefinite.  I use a definite NP when I assume
	you will be able to identify the referent. I say Where's the dog?,
	for example, only if I'm assuming you know which dog I'm referring
	to. There's no such assumption made with an indefinite NP, as in
	I could hear a dog barking.

(b) Determinative vs determiner
	Notice that determinative is the name of a category (a class of
	words), while determiner is the name of a function. There are other
	determinatives besides the and a: e.g. this, that, some,
	any, many, few, one, two, three, etc . They can likewise function
	as determiner, but that isn't their only function. In
		    It wasn't that bad,
	for example, the determinative that is modifier of the adjective bad.

(c) Differences from traditional grammar
	Traditional grammars generally don't use the term 'determinative'.
	The words in that class are treated as a subclass of the
	adjectives.  But in fact words such as the and a are very different in
	grammar and meaning from adjectives as above,

Adverbs


Related to adjectives:
   The most obvious adverbs are those derived from adjectives by adding ·ly:

		careful    -> carefully
		certain    -> certainly
		fortunate  -> fortunately
		obvious    -> obviously
		rapid 	   -> rapidly
		usual	   -> usually

   But there are also a fair number of adverbs that do not have this form:
   almost, always, not, often, quite, rather, soon, too, and very.

(b) Function

	It is mainly function that distinguishes adverbs from adjectives. The
	two main functions of adjectives are
	attributive and predicative, but adverbs do not occur in similar
	structures: compare *a jealously husband and *He became
	jealously. Instead adverbs mostly function as modifiers of verbs (or
	VPs), adjectives, or other adverbs.

  i Modifying Verb or VP    She [spoke CLEARLY].         I [OFTEN see them].
 ii Modifying adjective     a [REMARKABLY good] idea     It's [VERY expensive].
iii Modifying an adverb     She spoke [QUITE clearly].   It'll end [QUITE soon].

Prepositions


Meaning
	The most central members of the preposition category have primary meanings
	expressing various relations of space or time:
	[14] across the road    after lunch	  at the corner   before Easter
	     in the box		off the platform  on the roof	under the bridge

Function
	Prepositions occur as head of preposition phrases (PPs), and these in
	turn function as dependents of a range of elements, especially verbs
	(or VPs), nouns and adjectives.
	Examples: (uppercase: preposition; brackets: PP; italics: element on
		which PP is dependent: )

  i DEPENDENT ON VERB OR VP   I sat   [BY the door].    I saw her [AFTER lunch].
 ii DEPENDENT ON NOUN         the man [IN the moon]     the day [BEFORE that]
iii DEPENDENT ON ADJECTIVE    keen    [ON golf]         superior [TO the others]

(c) Differences from traditional grammar

In traditional grammar the class of prepositions only contains words that
combine with nouns (actually, in our terms, NPs).  Later, in Ch. 7, §2, we
drop this restriction and extend the membership of the preposition
category. We'll show that there are very good reasons for doing this.

Coordinators


The central members of the coordinator category are and, or, and but - in
traditional grammar they are called 'coordinating conjunctions' . Their
function is to mark the coordination of two or more expressions, where
coordination is a relation between elements of equal syntactic status. This
syntactic equality is typically reflected in the ability of any one element
to stand in place of the whole coordination, as in:

[16]  i  We need a long table and at least eight chairs.
     ii  a. We need a long table. b. We need at least eight chairs.

Thus both a long table and at least eight chairs can occur in place of
the whole.  Because the elements are of equal status, neither is head:
coordination is not a head + dependent construction.

Subordinators

(a) Function

The most central members of the subordinator category are that, whether, and
one use of if - the one that is generally interchangeable with whether (as in
I don't know whether/if it's possible).  These words serve to mark a clause
as subordinate.  Compare, for example:

[17] 	MAIN CLAUSE 		SUBORDINATE CLAUSE
	a. He did his best. 	b. I realise [that he did his best].

He did his best in [a] is a main clause - here forming a sentence by itself.

Addition of the subordinator that changes it into a subordinate clause.
Subordinate clauses characteristically function as a dependent element within
the structure of a larger clause. In [b] that he did his best is a
dependent of the verb realise.

That is often optional: in I realise he did his best the clause he did his
best is still subordinate, but it is not overtly marked as such in its own
structure.

(b) Differences from traditional grammar

One minor difference is that we follow most work in modern linguistics in
taking subordinators and coordinators as distinct primary categories, rather
than subclasses of a larger class of 'conjunctions' . More importantly, we
will argue in Ch. 7, for a redrawing of the boundaries between subordinators
and prepositions - but again we will in the meantime confine our examples to
those where our analysis matches the traditional one in respect of the
division between the two categories.



Grammatical classes as prototypes


each class is distinguished not by one property but by a cluster of
properties.   e.g. equipment is surely a noun, but it does not have a plural
form like most nouns.

PROTOTYPICAL: central or core members of a category that do have the full set
     of distinctive properties.

Applies not only to parts of speech but also to function, such as Subject.

	PROTOTYPICAL			NONPROTOTYPICAL
nouns: Cat and dog.     		equipment
verbs: Go, know, tell 			must
	[because (e.g.) must has no preterite form
	(*1 musted work late yesterday); and it can't occur after to

adjectives: Big, old, happy , 		asleep
	[because it can't be used attributively (*an asleep child).]


Subject: HIS GUILT was obvious		THAT HE WAS GUILTY was obvious
	[cannot be inverted with was as in Was his guilt obvious?]



Phrase structure


A phrase normally consists of a head, alone or accompanied by one or more
dependents. The category of the phrase depends on that of the head: a phrase
with a noun as head is a noun phrase, and so on.

We distinguish several different kinds of dependent

6.1 Complement and modifier

The most general distinction is between complements and modifiers,
as illustrated for VPs and NPs in [18], Complements are capitalized,
modifiers italicized:

[18] i VP  He [kept HER LETTERS for years].
    ii NP  She regularly gives us [very useful advice ON FINANCIAL MATTERS],

Complements are related more closely to the head than modifiers. In the
clearest cases, complements are obligatory : we cannot, for example, omit
her letters from [i].

In [ii] the complement is optional, but its close relation to the head is
seen in the fact that the particular preposition on which introduces it
is selected by advice: advice takes on, fear takes of, interest takes in,
and so on. A more general account of the distinction between complements
and modifiers will be introduced when we come to look at clause structure
in Ch. 4.

6.2 Object and predicative complement

two TYPES of COMPLEMENT (usu. in a VP):
	the OBJECT and the PREDICATIVE COMPLEMENT

Objects are found with a great number of verbs, while predicative complements
occur with a limited number of verbs, with be by far the most frequent. The
constructions differ in both meaning and syntax.

[19] :  OBJECT 				PREDlCATIVE COMPLEMENT
   i a. I met A FRIEND OF YOURS. 	b. She was A FRIEND OF YOURS.
  11 a. Sam appointed A REAL IDIOT. 	b. I felt A REAL IDIOT.
 iii a.     		     		b. They seemed VERY FRIENDLY.
		[note: very friendly can't be an object]

A prototypical object refers to a person or other entity involved in the
situation.  In [ia] there was a meeting between two people, referred to by
the subject and object, while in [iia] we have a situation involving Sam and
a person described as a real idiot. A predicative complement, by contrast,
typically expresses a property ascribed to the person or other entity
referred to by the subject.  In [ib] a friend of yours is a property of
"she", while in [iib] a real idiot doesn't refer to a separate person but
describes how I felt.

The most important syntactic difference is that a predicative complement can
have the form of an adjective (or AdjP), as in [iii b], whereas an object
cannot.
Thus we cannot have, say, *I met very friendly or *Sam appointed very
friendly.

6.3 Determiner (in NP)

marks the NP as definite or indefinite.  Certain kinds of singular noun usually
require the presence of a determiner. In The dog barked or I need a key,
the determiners the and a are obligatory.

The determiner function is usually filled by DETERMINATIVES but it can also
have the form of a GENITIVE NP, as in Fido's bone or the dog's owner, where
's is the marker of the genitive.

Non-canonical clauses


Canonical clauses:  Subj + Pred. e.g. Kim [left early]

non-canonical: may differ along one of the following:

Polarity


	Positive		NEGATIVE
	He is very careful	He isn't very careful

Polarity may be marked on the verb as in n't, or by not, or a negative
word (e.g. Nobody liked it).

Non-declarative clause


     DECLARATIVE:
     she can mend it		INTERROGATIVE: can she mend it?
		[subject follows verb]
     You are patient		IMPERATIVE:    be patient.
		[subject is missing (covert); different inflection of verb]


Subordinate (relative) clauses


All canonical clauses are main clauses.

Subordinate clauses characteristically function as a dependent within a
larger clause, and very often they differ in their internal structure from
main clauses, as in the following examples:

[23] MAIN SUBORDINATE (non-canonical)
   i a. She's ill. 		   b. I know THAT SHE'S ILL.
  ii a. We invited the Smiths. 	   b. INVITING THE SMITHS was a mistake.
 iii a. Some guy wrote the editorial. b. He's [the guy WHO WROTE THE EDITORIAL].

In [ib] the subordinate clause is complement of the verb know. It is marked
by the subordinator that, though in this context this is optional: in I
know she's ill the subordinate clause does not differ in form from a main
clause.  In [iib] the subordinate clause is subject. Its structure differs
more radically from the main clause: the subject is missing and the verb
has a different inflection.

The subordinate clause in [iiib] is a RELATIVE CLAUSE. The most
straightforward type of relative clause functions as MODIFIER within the
structure of an NP and begins with a distinctive word such as who, which,
when, where, etc., that 'relates' to the head of the NP - who in our
example relates to guy.

Coordination


One clause may be coordinated with another, the relation usually being marked
by means of a coordinator such as and or or. Canonical clauses are
non-coordinate.  Compare:

[24] NON-COORDINATE	COORDINATE (non-canonical)
      That's Bill.
      I'm blind. 	That's Bill or I'm blind.

coordinator = or.

Here there is no difference between the coordinate clause and the main clause
form. just as subordinate clauses do not necessarily differ from main ones.

7.5 Information packaging [Perspective]

In many cases, one can say essentially the same thing through syntactically
different constructions. This allows us to present or package the information
in a variety of ways [e.g. present a particular perspective].  Canonical
clauses always present the information in the syntactically most elementary
way.  There are many such constructions (see ch. 15); here
we illustrate: passive, preposing, and extraposition.

(a) Passive clauses

[25]    ACTIVE 			PASSIVE (non-canonical)
	a. The dog bit me. 	b. I was bitten by the dog.

These have practically the same meaning; if used in the same context it would
be impossible for one to be true while the other was false.

ACTIVE: the subject of the active version (in [a] the dog) denotes the active
	participant, the performer of the action,
PASSIVE: the subject  (in [b] I) denotes the passive participant [patient], the
	undergoer of the action.

Syntactically the passive version is clearly more complex than the active by
virtue of containing extra elements : the auxiliary verb was and the
preposition by.  Hence we take the passive as a non-canonical construction.

(b) Preposing
   [26] BASIC ORDER :               a. I gave the others to Kim.
        PREPOSING (non-canonical):  b. The others I gave to Kim.

Here the two versions differ simply in the order of elements.

In [a] the object occupies its default position after the verb.
In [b] it is preposed, placed at the beginning of the clause, before the subject.

Canonical clauses have their elements in the basic order, with departures from this
order being handled in our account of various types of non-canonical clause, such
as the preposed complement construction in [b] .

(c) Extraposition

 [27]   BASIC           a. THAT I OVERSLEPT was unfortunate.
        EXTRAPOSITION   b. It was unfortunate THAT I OVERSLEPT.

In [a] the subject is a subordinate clause - occupying the usual subject
position.  In [b] the subject position is occupied by the pronoun it and the
subordinate clause appears at the end: it is called an EXTRAPOSED SUBJECT.

In pairs like this, the version with extraposition is much more frequent than
the basic one, but we still regard version [a] as syntactically more
basic.

The extraposition construction is virtually restricted to cases where
the basic subject is a subordinate clause. It's the [a] version that matches
the canonical structure of clauses with NPs as subject, e.g., THE DELAY was
unfortunate. And [b] is (slightly) more complex in structure: it contains the
extra word it.

7.6 Combinations of non-canonical features

Non-canonical clause categories can combine, so that a clause may differ
from a canonical one in a number of different ways at once:

    i a. Sue can swim.     b. He says that SUE CAN'T SWIM.
					[both subordinate and negative.]
   ii a. Kim took the car. b. I wonder whether the car was taken by Kim.

(ii b) is interrogative and passive as well as subordinate, marked by the
	subordinator whether, though subject is not after the verb.

Word structure


Words are made up of elements of two kinds: BASES and
AFFIXES.  For the most part, bases can stand alone as whole words whereas affixes
can't:

[29] en·DANGER 		     BLACK.BIRD·s.
     SLOW·ly. 		     un·GENTLE·MAN·ly.
     un·JUST
     WORK·ing

Affixes may be PREFIXES or SUFFIXES.
When citing them individually, we indicate their status
by putting · after prefixes (en· , un· ) and before suffixes (·ly, ·ing).


--Exercises

3. Assign each word in the following examples to one of the part-of-speech
   categories: noun (N), verb (V), adjective (Adj), determinative (D), adverb
   (Adv), preposition (Prep), subordinator (Sub), coordinator (Co).

	  i She lives in Moscow.
	 ii The dog was barking.
	iii Sue and Ed walked to the park.
	 iv I met some friends of the new boss.
	  v We know that these things are extremely expensive.

4. Construct a plausible-sounding, grammatical sentence that uses at least
   one word  from each of the eight categories listed in the previous
   exercise (and in [7] in the text of this chapter).

5. Is it possible to make up an eight-word sentence that contains exactly
  ONE word of each category? If it is, do it; if not, explain why.




3 Verbs: tense, aspect, and mood


Inflection:

paradigm: set of inflectional forms of a lexeme that can be
	inflected, together with their grammatical labels.
Unlike some languages, English verb paradigms they are fairly simple.  he
great majority of verbs in English have paradigms consisting of six
inflectional forms.

	 a. Kim has flown home. 	b. Kim flew home.

in (a) only "flown" is possible, so it is required.  In (b) other forms are
possible (e.g. "flies").

[2] 		  PARADIGM		EXAMPLE   SENTENCE
-----------------------------------------------------------------------
		  PRETERITE		walked 	  She walked home.
PRIMARY FORMS 	  3RD SINGULAR PRESENT 	walks 	  She walks home.
		  PLAIN PRESENT 	walk   	  They walk home.
-----------------------------------------------------------------------
		  PLAIN FORM		walk 	  She should walk home.
SECONDARY FORMS   GERUND-PARTICIPLE 	walking   She is walking home.
		  PAST PARTICIPLE 	walked 	  She has walked home.
-----------------------------------------------------------------------

we note that walked and walk each appear twice in the paradigm. To cater for
this we need to draw a distinction between an inflectional form and its
shape.

By shape we mean spelling or pronunciation: spelling if we're talking about
written English, pronunciation if we're talking about spoken English.

The preterite and the past participle are different inflectional forms but
they have the same shape walked. Similarly for the plain present and the
plain form, which share the shape walk.

In the case of the preterite and the past participle there is a very obvious
reason for recognising distinct inflectional forms even though the shape is
the same: many common verbs have DIFFERENT shapes for these inflectional
forms. One is fly, (example [1]) : its preterite form has the shape flew,
while its past participle has the shape flown.

The reason for distinguishing the plain present from the plain form is less
obvious (discussed later).

Primary vs secondary forms


With one isolated exception that we take up in §8.4, PRIMARY forms show
inflectional distinctions of tense (preterite vs present) and can occur as
the sole verb in a canonical clause.

SECONDARY forms have no tense inflection and cannot occur as the head of a
canonical clause

PRETERITE = INFIECTIONALLY MARKED PAST TENSE
	marked by a inflectional form and not by a separate auxiliary verb.

By a past tense we mean one whose most central use is to indicate past
time. The preterite of take is took, and when I say I took them to school I
am referring to some time in the past.  (but in English, this relation is far
from straightforward).


6 Adjectives and Adverbs


1 Prototypical adjectives

Semantics:
Adjectives typically denote properties of objects, persons, places, etc . :
properties relating to age (old, young), size (big, small), shape (round,
flat), weight (heavy, light), colour (black, blue), merit or quality (good,
bad), and so on.

Syntactically, prototypical adjectives in English can have Function, Grade
and Modification.

(a) Function

Attributive adjectives function as internal pre-head modifier to a following
noun; predicative adjectives function mainly as predicative complement in
clause structure:

 ATTRIBUTIVE   an old car         black hair           good news
 PREDICATIVE   The car is old.    Her hair is black.   The news is good.

(b) Grade
They either inflect for grade, showing a contrast between plain, comparative
and superlative forms, or else form comparative and superlative adjective
phrases (AdjPs) marked by more and most:

PLAIN           COMPARATIVE                     SUPERLATIVE
She is tall.    She is taller than you.         She is the tallest of them all.
This is useful  This is more useful than that.  This is the most useful one.

(c) Modification

They can be modified, usually by adverbs, as in [3]
	 TOO old    REMARKABLY tall     EXTREMELY useful to us


adjective -> adverb via -ly


enormous set: e.g. remarkable vs remarkably.
where the word without the ·ly modifies a following noun,
the one with ·ly that modifies a following adjective:

  N   a remarkable judge   its incredible size    this wonderful silk
ADJ   remarkably wise      incredibly big         wonderfully smooth

Sometimes nouns can also function as attributive modifiers or predicative
complements.  However, ADJs cannot head phrases in SUBJ or OBJ positions

	SUBJECT			OBJECT
    N	Its size amazed me.	I like silk.
  ADJ	* size amazed me.	*I like smooth.


Overlap between the categories: many lexemes can belong to both categories,
e.g. cold:
	This soup is cold / I caught a bad cold).

But only Nouns take determiners or can function as a head :

[8]               ADJECTIVE         NOUN
i  INFLECTION     colder, coldest   colds
ii DETERMINERS                     my cold, which cold?
iii MODIFIERS	  terribly cold     a terrible cold
iv FUNCTION	                    The cold was nasty. Don't catch a cold.


The fused modifier-head construction


One complication in distinguishing between adjectives and nouns is that a
limited range of adjectives can appear as fused modifier-head in an NP:
[9] i SIMPLE  The first version wasn't very good but [the SECOND] was fine.
11 PARTITIVE  I couldn't afford [even the CHEAPEST of them].
 iii SPECIAL  This tax cut will benefit [only the RICH).

Precisely because they are in head position in NP structure, the emphasized
words might at first glance be thought to be nouns. But they're not nouns :
they're adjectives.

In the simple and partitive constructions this is fairly easy to see:
Note the possibility of adding a repetition of the noun version in [i] .
In [ii] we have a superlative form, cheapest, which certainly can't be a
noun.

In most SPECIAL cases as in [iii], the form can be clearly identified as an
adjective.

* modifier test: the EXTREMELY rich: takes a modifying adverb
• The only determiner permitted is the - a person who is rich can't be
	referred to as *a rich or *some rich.

And although the NP the rich is plural (hence the verb agreement in The rich
are the beneficiaries), it doesn't have plural inflection on rich - two rich
people can't be referred to as *two riches. Rich thus behaves very
differently from a noun.

Overall, there is strong evidence that the rich in [iii] contains an
adjective but no noun.



16 Morphology: words and lexemes


INFLECTIONAL MORPHOLOGY deals with the differences between the shapes of
the inflectional forms of variable lexemes; for example, the formation of
the verb-forms endangers, endangered and endangering from the lexical base
endanger.

LEXICAL MORPHOLOGY deals with the formation of lexical bases - with the
formation of endanger, for example, from en· and danger. This includes the
formation of the lexical bases of invariable lexemes, such as cleverly. This
doesn't inflect (there are no forms *cleverlier or *cleverliest), but the
fact that it is made up out of clever and ·ly is a fact of lexical
morphology.

   Lexeme			Inflectional Forms
 i friend (N) :	     friend    friends     friend's  friends'
ii friendly (Adj) :  friendly  friendlier  friendliest

horizontal: inflectional (more tightly determined by rules)
vertical (friend -> friendly) : derivational

[2] i a. She's been a good ___.      b. Their _____ own car was a vw.
   ii a. He's ____ than his brother.  b. He's the _____ of them all.

To use friend in the gaps:
ia:  plain singular form friend,
ib: a genitive form (sg friend's or pl friends')

iia: comparative friendliest
iib: superlative friendliest
	(no other choices are allowed)

But instead of friend in (i) or friendly in (ii) other nouns / adjectives
can be used.

But is there is no rule saying that the adjective appearing in [ii]
must be formed from a noun (e.g. [iia-b] could just as well be older and
oldest) - whose lexical base is morphologically simple.

Similarly, the lexical base of nouns for [i] can morphologically compound,
e.g. teacher. So the internal structure of the lexical base is not very
significant. All that matters is whether you have picked a syntactically
admissible inflectional form of the lexeme you decide on.

To summarise, INFLECTIONAL morphology ties in mainly with syntax, while
LEXICAL morphology is mainly relevant to the content of the dictionary.

lexical base used in inflection is usually a stand-alone form (like friend).
exceptions: binoculars, trousers, auspices, scissors, credentials etc.

--Verbs:

FINITE:
She brings her own food.   			[primary form]
Bring your own food. 				[plain (imperative)]
We insist [that she bring her own food].	[plain (subjunctive)]

NON-FINITE:
It 's rare [for her to bring her own food]. 	[plain (infinitival)]
She regrets bringing her own food.   		[gerund-participle]
This is the food [brought by my sister].	[past participle]



from the Glossary


ACCUSATIVE. The inflectional case of the pronouns me, him, her, us, them,
  whom.  Contrasts with NOMINATIVE.
ADJECTIVE. A category of lexemes characteristically denoting properties of
  persons or objects (old, big, round, blue, good). The prototypical
  adjective can be used both ATTRIBUTIVELY and PREDICATIVELY (hot soup, The
  soup is hot), participates in the system of GRADE (occurs in the
  COMPARATIVE and SUPERLATIVE), and takes adverbs as modifier (extremely hot,
  very useful).
ANTECEDENT. Constituent whose meaning dictates the meaning of a pronoun or
  other such expression in cases of anaphora.
CASE. An inflectional system of the noun with the primary use of marking
  various syntactic functions, such as subject and object in clause structure
  or determiner in NP structure. In English the main distinction is between
  genitive and plain case (e.g. Kim's vs Kim). A few pronouns have distinct
  nominative and accusative cases (e.g. I vs me) instead of a plain case.
CLOSED INTERROGATIVE. A subtype of interrogative clause characteristically
  used, in main clauses, to ask a closed question . Marked in main clauses by
  subject-auxiliary inversion (Is it raining? ; Is he alive or dead?), and in
  subordinate clauses by whether or if (I wonder whether it's raining; I
  don't know if he's alive or dead).
CLOSED QUESTION. Question with a closed set of answers : Is it raining?
  (answer s: Yes, No); Is he alive or dead? (answers : He's alive, He's
  dead).
COMPLEMENT. A kind of dependent that must be licensed by the head. In It
  shakes the building, the building is a complement because it's allowed only
  with a certain kind of head verb: shake licenses dependents of this kind,
  but quake doesn't (*It quakes the building).
CONTINUATIVE PERFECT. A use of the perfect indicating a situation lasting
  over a period starting before a certain time and continuing up to it: She
  had been in bed for two hours when we arrived means she was in bed two
  hours before we arrived and continued to be until we arrived (and possibly
  after that).
DEFINITE ARTICLE. The determinative the. Prototypically functions as
  determiner in NP structure with the sole meaning of indicating that the
  head is sufficient in the context to identify the referent: when I ask,
  Where's the car?, I assume you know which car I' m referring to.
DEFINITE NP. NP marked by the definite article the or by certain other
  determine rs (e.g., this, that, my), or with no determiner but having a
  proper noun as head.  Characteristically used when the content of the NP is
  sufficient in the context to identify the referent.
DEICTIC. Used in a way that allows the interpretation to be determined by
  features of the act of utterance like when and where it takes place, and
  who the speaker and addressee are; e.g. l (refers to the speaker), now
  (refers to a time that includes the time of utterance).
FINITE CLAUSE. Clause that is either headed by a primary verb-form
  (Ed is careful) or is imperative (Be careful) or is subjunctive
  (I insist that he be careful).   Main clauses are always finite,
  subordinate clauses may be finite or non-finite.
FRONTED PREPOSITION. Preposition placed along with its complement at the
  front of the clause: [To whom] are you referring?
GENITIVE. An inflectional case of the noun whose primary use is to mark an NP
  as determiner within the structure of a larger NP: Kim's book. Some
  pronouns have two genitive forms : dependent genitive (my) and independent
  genitive (mine).
HEAD. The function of the most important element in a phrase. Often stands
  alone without any dependents, as in Dogs were barking: the subject NP
  contains just the head noun dogs.
INDEFINITE ARTICLE. The determinative a (or an), prototypically used as
  determiner in count singular NPs indicating that the content is not
  sufficient to identify a specific referent: q bus.
INFINITIVAL CLAUSE. Subordinate clause containing a plain form of the verb
  (subject marked by for if there is one).  Covers to-infinitivals
  (To err is human) and bare infinitivals (I will go).
NON-FINITE CLAUSE. SUBORDINATE CLAUSE headed by GERUND-PARTICIPLE (his
  writing it), past participle (having written it), or plain form in the
  infinitival construction (to write it).
PASSIVE CLAUSE. Prototypically, a clause with auxiliary be followed by a past
  participle followed optionally by by + NP, and having an active
  counterpart: The record was broken by Lance (compare active Lance broke the
  record).
PREDICATE. The head of a clause, a function filled by a verb phrase: We
  washed the car.
RELATIVE CLAUSE. Subordinate clause of which the most central type functions
  as modifier to a noun: I've met the woman who wrote it. The noun serves as
  antecedent for an element within the relative clause which may be overt
  (like who in the above example) or merely understood (as in I've met the
  woman )!Qll.  are referring to).
SPEECH ACT. An act like making a statement, asking a question, or issuing a
  directive.  Stranded preposition. Preposition which is not followed by the
  NP that is understood as its complement: Who did you give it to?; This is
  the book [/ was talking about] .
SUBJECT. The function in clause structure (usually filled by an NP; before
  the predicate in canonical clauses) that in active clauses describing
  action normally denotes the actor: Ed ran away.
TENSE. A system marked by verb inflection or auxiliaries whose basic use is
  to locate the situation in time: I liked it (past tense, past time), I like
  it (present tense, present time).
TO-INFINITIVAL CLAUSE. Infinitival clause containing the marker to: I want to
  see them; We arranged for them to meet.
VOICE. The grammatical system contrasting active and passive clauses: Ed
  broke it is in the active voice, It was broken by Ed is in the passive
  voice.
VOWEL. Speech sound produced with unimpeded smooth airflow through the mouth.
  Vowel symbol. Letter or sequence of letters representing a vowel: u is a
  vowel symbol in hut, but not in quick.





Review of CGEL by Peter W. Culicover, Language 2004


The Cambridge Grammar of the English Language (CGEL) is a monumentally
impressive piece of work.  [Rodney Huddleston was the original conceiver and
planner of this work, and Geoff Pullum joined him in the project in 1995.]

Already published reviews of this work do not overstate its virtues: ...
	one of the most superb works of academic scholarship ever to appear
	on the English linguistics scene... a monumental work that offers
	easily the most comprehensive and thought-provoking treatment of
	English grammar to date. Nothing rivals this work, with respect to
	breadth, depth and consistency of coverage. 

I fully agree with these sentiments. Huddleston, Pullum and their
collaborators definitely deserve a prize for this achievement.
... And they have in fact won a prize: the LSA's Leonard Bloomfield book
award for 2001-2003! (See http://www.lsadc.org/index2.php?aaa=lsanews.htm for details.)]

I try to convey here a sense of what it feels like to work with and through
CGEL, and what one might plausibly conclude from this exercise about how
language works. I also outline the theory of grammar that is explicit and
implicit in CGEL and speculate a bit on what we might conclude from this
theory about what it means to know a language. In particular, I explore the
possibility that CGEL is actually the basis for a complete description of
the knowledge that a native speaker of English has of English, and the
consequences of that possibility.

CGEL is organized into twenty chapters:

	Ch. 1 ‘Preliminaries’,
	Ch. 2 ‘Syntactic overview’,
	Ch. 3 ‘The verb’,
	Ch. 4 ‘The clause: complements’,
	Ch. 5 ‘Nouns and noun phrases’,
	Ch. 6 ‘Adjectives and adverbs’,
	Ch. 7 ‘Prepositions and preposition phrases’,
	Ch. 8 ‘The clause: adjuncts’,
	Ch. 9 ‘Negation’,
	Ch. 10 ‘Clause type and illocutionary force’,
	Ch. 11 ‘Content clauses and reported speech’,
	Ch. 12 ‘Relative constructions and unbounded dependencies’,
	Ch. 13 ‘Comparative constructions’,
	Ch. 14 ‘Non-finite and verbless clauses’,
	Ch. 15 ‘Coordination and supplementation’,
	Ch. 16 ‘Information packaging’,
	Ch. 17 ‘Deixis and anaphora’,
	Ch. 18 ‘Inflectional morphology and related matters’,
	Ch. 19 ‘Lexical word-formation’,
	Ch. 20 ‘Punctuation’.

CGEL is similar in its organization to the work that it aims to supplant,
Quirk, et al. 1985 (Q85). Although the two are of roughly the same length
(Q85 has 1789 pages and CGEL 1859), Q85 feels almost superficial compared
with CGEL. The level of detail of CGEL is such that the reader may begin to
feel that s/he is being told everything that one could possibly know about
the topics that it covers. And while common sense tells us that this cannot
be, in many cases it is difficult to think of what else one would want to
say in factual terms about a particular construction, or form. (Theoretical
excursions are something else entirely, of course.) There were only a few
points here and there that I felt could have been mentioned but were not;
further research almost invariably showed that they were in fact covered
somewhere in the text.

Ch.1 introduces a number of critical points that guide the approach throughout.
There is a basic introduction to the concepts of constituent structure and
syntactic (lexical 3 and phrasal) categories. The typical clause is composed of
a noun phrase followed by a verb phrase. Crucially, each constituent of a
phrase has not only a category, but a grammatical function. These are
represented simultaneously on a phrase marker, as in (1) (26:[13]).

(1)           NP
            /     \
          Det:    Head:
           D        N
           |        |
         some     children

Here, some is of the syntactic category D(eterminative), and bears the
grammatical function Determiner in the phrase, while children is of the
category N(oun), and bears the grammatical function Head of the phrase. Other
functions include Subject, Predicate, and Object.

In the view of CGEL, syntactic categories are determined strictly by formal
and distributional criteria; function is completely orthogonal.5 The
distinction is of course not novel, and is even found in some theoretical
work (cf. Specifier and Complement in Chomsky 1972), but CGEL observes it
rigorously, in the easy cases and in the hard cases.  Sometimes the results
are familiar, sometimes they are quite novel. For example, prehead
adjectives are modifiers (happy dog), pre-head nouns are modifiers (biology
syllabus), and they are of different syntactic categories (537). The, a,
this, that, some, etc. are determinatives, not adjectives, on the basis of
their distributional characteristics; so are many, few, much and little
(539).

In a sleeping child, sleeping is a verb since it cannot function as a
predicative adjective, in contrast to disturbing in some disturbing
news. Similarly for pre-head heard and worried (541). Clausal complements
are not NP objects. The reasons: (i) V PP S is canonical, V PP NP is not,
(ii) Some verbs take only S, not NP: marvel, vouch, wonder, charge, (iii) V
P NP is grammatical, while V P S is not.

Pursuing the logic of categorization, CGEL argues that the prepositions
that head phrases like before I got home are just that, prepositions, and
not traditional ‘subordinating conjunctions’. They argue that just as
remember is a verb regardless of whether it takes an NP complement or a
clause complement, so after is a preposition (600). Then, given that
prepositions take such complements, and are not nouns or verbs, and are the
heads of phrases that function as adjuncts, CGEL arrives at the conclusion
that there are many prepositions besides the familiar before, after, in,
to, at, on and so on.  There are intransitive prepositions, such as
downstairs, prepositions derived from adjectives, such as opposite, ahead,
contrary, and prepositions derived from verbs, such as owing (to), barring,
counting, including, excluding, given, concerning, provided, etc.
(606-610).

And while one might not initially be inclined to say that barring, as in
barring accidents for example, is a preposition, it is hard to argue with
the distributional facts. It does not have the full range of forms of a
verb and lacks control (*Having barred accidents, we would have succeeded),
it certainly heads an adjunct that alternates with PPs, it is in the same
head position as a prototypical preposition. Granted, the prototypical
preposition typically has a thematic function, but that could be taken to
show that is that there are at least several semantically differentiated
subclasses of prepositions, some thematic and some not.

A consequence of strict application of the form-function distinction is
that a single form may have more than one grammatical function at the same
time. Such a situation is what CGEL refers to as ‘fusion’. For instance, in
few of her friends, few bears the Head function as well as the Determiner
function. (412:[7a]).



[thus the structure is no longer a "tree"]

Similar analyses are given for such expressions as someone (Det-Head, fusing
some one), (the) second (Mod-Head, fusing second one), what (I said)
(Head-Prenucleus, fusing thing which), and the rich (Adjective-Head, fusing
rich folk).

Fusion is a clean but not very deep solution to the problem of how to analyze
these constructions. It neatly sidesteps the question of whether there is
deletion (the rich folk -> the rich folk) , empty proforms (the second pro),
movement and substitution (for free relatives – I won’t try to give a blurb for
a derivation here). Moreover, it takes the forms to be sui generis, which
avoids the problem of explaining why they don’t always mean what they would
mean if the derivation did not occur.

In addition to syntax, there is a lot of informal semantics in CGEL. Sometimes
the presentation is simple and elegant; for example, the rule for the
interpretation of the (368) is that the speaker expects the hearer to be able
to identify the referent. Similarly, the rule for indefinite a (371) is that
‘[t]he addressee is not expected to be able to identify anything.’ (The
complexity is then presumably in defining under what circumstances one can
reasonably hold these expectations.) At times the presentation is complex and
detailed. Although space is limited, I must cite one representative passage
because without an example it is difficult to appreciate just how much detail
there is.

    The perfective/imperfective contrast is particularly important in the
    present tense because of the constraint that a present time perfective
    interpretation is normally possible only when the situation is of short
    enough duration to be co-extensive with the utterance:

    [4] i His daughter mows the lawn. [salient reading: serial state]
     	 ii His daughter is mowing the lawn. [salient reading: single occurrence]

    Mowing the lawn does not satisfy that condition, so that a single
    occurrence reading is not normally available for [i], which we interpret as
    a serial state, with habitual lawn- mowing. The imperfective meaning in
    [ii], by contrast, allows for Td to be included within Tsit, giving the
    interpretation where a single occurrence of mowing is now in progress. In
    the present tense, therefore, the progressive is much the more frequent
    aspect for dynamic situations. It would, however, be a mistake to see
    ‘habitual’ vs ‘non-habitual’ as the difference in meaning between [i] and
    [ii] (or, worse, between the present non-progressive and the present
    progressive generally). A single occurrence interpretation of [i] is not
    semantically excluded, but merely pragmatically unlikely: it could occur as
    a timeless or historic present or as a futurate - and if embedded, for
    example in a conditional construction, it could easily take a single future
    occurrence interpretation. Nor does [ii] exclude a serial state
    interpretation: compare His daughter is mowing the lawn until he is well
    again. The ‘habitual’ vs ‘non-habitual’ contrast is thus a difference in
    salient interpretations arising from the interaction between the meaning of
    the aspects and the pragmatic constraints on present perfectivity. Note
    that in the preterite the nonprogressive His daughter mowed the lawn allows
    a single occurrence reading as readily as the progressive His daughter was
    mowing the lawn. (164)

Much of the discussion of the semantics of time, aspect, negation, modality,
and quantity is on this order of detail, as is that of verbal
semantics. Remarkably, I find myself constantly in agreement with the
distinctions drawn, down to the very finest points. I say ‘remarkably’, because
the question of how CGEL and I (and, I would presume, virtually all other
native speakers of English) managed to come to the same judgments about what
things mean down to the finest details is a non-trivial one.

[...]

[In] many of the factual observations in CGEL; the expert reader will no doubt
often experience with a shock of recognition some data in CGEL that was first
pointed out by, say, Chomsky7 in 1977, or Klima in 1964, or whatever.

[FN: The possibility of such a reaction was not unanticipated by the
author/editors. On page xvi they refer explicitly to their policy of not citing
sources, even in footnotes. And on the Cambridge University Press website
(http://uk.cambridge.org/linguistics/cgel/faqs.htm) one finds the following
exchange: ‘I don't see any references to the literature in the grammar's
pages. How come?’ Answer: ‘This is a reference grammar, not a monograph about
linguistics. No references to the literature are given in the body of the
work. A modest attempt at attribution of key ideas is made in the Further
Reading section at the end, together with a list of references, but the idea of
including a complete bibliography of the gigantic field of English grammar
could not even be considered.’ My own view is that this policy is a mistake.]

[Curiously, CGEL does not even cite Chomsky in the references given in Further
Reading (1765-78).]

While the particular citations are no doubt of limited interest to the reader
who is consulting CGEL as a reference grammar (for that is what it is), and not
as a treatise on syntactic theory, it would be unfortunate if readers formed
the view that CGEL is the first place where such data is cited, the disclaimers
in the preface and the section on Further Reading notwithstanding.


There are remarkably few errors in CGEL; see
http://people.ucsc.edu/~pullum/errata.html for current errata.

In an eerily prescient passage, Pullum 1984 offers the following description of
what linguistics is: ‘Suppose you wanted to program a computer to understand
plain English, like the HAL 9000 computer in Stanley Kubrick's film 2001: A
Space Odyssey.  Linguistics is the subject that figures out what you’d need to
know about a language in order to do that, for English or any other language,
in a general and theoretically principled sort of way.’
 

bookexcerptise is maintained by a small group of editors. get in touch with us!
bookexcerptise [at] gmail [] com.

This review by Amit Mukerjee was last updated on : 2015 Dec 15