The table below lists all known typographical errors in The Cambridge Grammar of the English Language. Most are very small: one-character spelling mistakes, spacing errors, character transpositions, and the like. On page 1036, in a fit of typographical nerdiness, we even correct an en dash (–) to a hyphen (-).

Only a handful of the errors involve facts (and one of those, a glaring slip on page 518, involves not grammar but the geography of the Big Apple).

Cases of the book giving or implying a wrong syntactic analysis are very rare indeed, but here are four examples:

1. On page 332 the tree in [11] shows the PP from Lloyds as a complement ("Comp:") licensed by the noun salary, but it should surely be analysed as a modifier, and the tree structure is compatible with that. The PP should have the function label "Mod:" above the category label "PP".

2. On page 560, in [21], the use of laureate as a postpositive adjective is illustrated with Nobel laureate, but in that phrase the noun laureate is the head and Nobel is an attributive modifier. We should have cited poet laureate, where the head noun is poet and laureate does function as a postpositive adjective. (Notice that the plural of Nobel laureate is Nobel laureates. The plural of poet laureate — rare, because any given country selects a unique poet for that title — would be poets laureate under the analysis we are talking about, though many people would treat the phrase as if it were a word, and write poet laureates. It is clear that laureate as an adjective is being lost from the language, even in postpositive function with the head noun poet.)

3. On page 912, the term "Determinative whose" is incorrectly used, twice: whose is the genitive inflected form of the pronoun who, so although it functions as determiner, it belongs to the pronoun category, not the determinative category.

4. On page 1365, the example set [3] is introduced by noting that "canonical and non-canonical versions do not invariably have the same truth conditions". The intent was to point out that a passive (non-canonical) will not always have the same meaning as its active (canonical) counterpart. But our illustrative sentences involve negation: we note that the active clause Kim didn't sign many of them could be false (if Kim signed thousands of them) while the passive clause Many of them weren't signed by Kim was false (if only three remained unsigned), so the two don't have the same truth-conditional meaning. But since negative clauses are defined as non-canonical (p. 46), "canonical" was the wrong word to use. The sentence preceding [3] should have said that non-canonical clauses do not invariably have the same truth conditions as their syntactically more basic counterparts.

Even in these cases, however, the correction involves changing just a single word.

A second table is given following the main one which shows a few entries that we now think should have been in the lexical index but were missed in the first printing (they were added in the second), plus a few further entries that we now think should have been added.

Some of the errors have been spotted by the authors, and others by alert readers around the world – and we mean literally around the world: we have heard from readers in Australia, Canada, China, Germany, Japan, Korea, Nigeria, Papua New Guinea, the UK, and the USA. To be specific, we thank Adam Albright, Randy Alexander, Michael J. Corrigan, Peter Culicover, Jim Donaldson, Stephen Drage, Vanja Dunjko, Dennis During, Simon Goodwin, Lloyd Humberstone, Benson Ibe, Dae-Ho Kim, Hideki Kishimoto, Audrey Laroche, Chungmin Lee, Geoffrey Leech^†, Stan Legacy, Maurice McCallum, Joybrato Mukherjee, Will Oxford, Paul Postal, Brett Reynolds, Adrian Stenton, Roland Sussex, and Jonathan Swinton for their eagle eyes and their kindness in taking the trouble to write.

We list in total about 100 errors here. The book has 1,860 pages, so that's about one error every 19 pages.

We'd be grateful for messages about any further errors that might be discovered. They should be emailed to pullum@gmail.com.

Table of Errata and Corrigenda

We write "\\" to signify a line break. A prefixed + means we definitely know this was corrected in the second reprinting or a subsequent one. Each line has four main fields: page number, location in the page, erroneous text, and corrected text.

	Page	Where	What it says	What it should say
+	viii	line 7	Üniversität	Universität
+	xii	line 14	(p. 48)	(p. 49)
+	xiii	line 11 up	[24] what you insisted that we need 1098	[24] what you insisted that we need 1089
+	11	fn. 3	the London/Oslo/Bergen (LOB) corpus	the Lancaster/Oslo/Bergen (LOB) corpus
+	24	5 lines up	the  photographs	the  photographs
+	47	3 lines below [3]	the passive [iiib]	the passive [iiia]
	50	tree [8b]	PredComp	PredComp:
	62	line 6	a request to close the door.	a request to close the window.
+	79	line 8	in Ch. 16, §10.3	in Ch. 16, §10.1.3
+	140	6 lines below [3]	T12	Tr2
	150	line 11	the accepting her offer	the accepting of her offer
	167	5th line below [13]	if Kim's situation had been conceptualised	if Jill's situation had been conceptualised
+	176	2 lines above [5]	it important	it is important
+	217	11 lines up (text)	think in a passive clause	consider in a passive clause
+	218	4 lines below [6]	think	consider
+	219	line 2	die	leave
	235	3rd line after header (c)	[iv] with precede/follow	[v] with precede/follow
+	241	example [11], line (f)	/   X   /   X	N/A   N/A   N/A   X
+	247	4 lines below [4]	her	Jo
	252	4th line of last full paragraph	the be of [3i]	the be of [3a]
+	261	example [26]	almost raw	almost raw
	269	end of shaded box	(cf. Ch. 12, §5.4)	(cf. Ch. 12, §6.4)
	299	4th line after [10]	the Dean met (with) Pat	the Dean met (with) Kim
	332	tree in example [11]	Comp:	Mod:
+	349	example display at line 12	[51]	[53]
+	412	example [7b]	Nom:	Nom
	440	lines 6-7	The one exception is the noun denizen...	There are no known exceptions.
+	446	example [13ii]	fifty miles an hours	fifty miles an hour
+	446	line 10	indefinities	indefinites
	474	4th line up	Mary's letter in [iii]	Mary's letter in [xiii]
+	479	below example [63]	the suffix's attaches	the suffix  's  attaches
+	518	line 9	the Bronx naming a district in Manhattan	the Bronx naming a borough of New York City
	529	line below example [3]	preceding the def-	preceding the indef-
+	529	6 lines below example [3]	An adjectives that	All adjectives that
	530	3 lines below example [7]	predicative PPs with idiomatic meanings, such as in a bad	predicative PPs with idiomatic meanings, such as in a bad
+	530	12 lines up	Predicative adjuncts in front position . . .	█ Predicative adjuncts in front position . . .
	540	last line	Ch. 16, §10.3	Ch. 16, §10.1.3
+	547	example [33ii]	It surely isn't [That important]	It surely isn't [that important]
+	560	in [20], second line	drunk	drunk BrE
+	560	line 3 up (not counting footnote)	a Nobel laureate	the poet laureate
+	560	line below [20]	Those in [20] have to do with medical health or condition.	Those in [i] have to do with medical health or condition.
+	561	line 9	We take these to involved	We take these to involve
	588	line before [45]	the relevant concept is not "except", but "more than"	the relevant concept is not "except", but "no more than"
+	599	line 8	Adj Ps	AdjPs
	620	tree (c) in example [9]	NP (at the right child of the root PP node, under Comp:)	PP
+	626	bottom line of text	In [i] the	In [a] the
+	627	line 2	in [ii]	in [b]
+	629	example [7] ii a.	you can certainly rely __ oni	you can certainly rely on __i
	640	line 11	require a complement – a for phrase	require a complement – a from phrase
+	692	line 1	[iii] even has a path	[ii] even has a path
	704	3rd line after [6]	And in [iib]	And in [iiib]
	733	2nd line in [35v]	see Ch. 15, §§10–11	see Ch. 15, §§2.10–2.11
+	912	line 10 up, header (a)	Determinative whose	Determiner whose
+	912	line 5 up	With determinative whose,	With determiner whose,
+	1036	2 lines below [8]	gerund–participials	gerund-participials
+	1037	line 20	The meaning in both [i] and [ii]	The meaning in both [ii] and [iii]
	1042	5 lines after [17]	into an NP	in an NP
	1043	example [20] ii	I felt the need \\ need of a better knowledge of Hebrew and archaeology.]	I felt the need \\ of a better knowledge of Hebrew and archaeology.]
	1045	example [32] iii	[where they went last year ___i].	[where they went ___i last year].
	1050	example [51] iii	the knoll behind the missioni,	the knoll behind the missioni,
	1055	line 18	in [ib]	in [iib]
	1082	1st line after [10]	While [i] is fully acceptable, [ii] is ungrammatical.	While [a] is fully acceptable, [b] is ungrammatical.
	1105	line 5	more satisfactorily than we did last year	better than we did last year
+	1173	example [2] iii	subordinate	embedded in a larger clause
	1218	line 2 of text	whereas in [ii]	whereas in [b]
	1223	18 lines up	very unlikely in [iii]	very unlikely in [iic]
+	1229	example [14]	help (B) NS	help (B) NS
	1232	example [28]	ackowledge	acknowledge
	1247	2nd line of blue-shaded section	but only of convince	but only of accept
	1253	3 lines up	likely, probably, certain,	likely, probable, certain,
	1263	12 lines up	Wantstraightforwardly	Want straightforwardly
+	1276	example [5] i	[on Monday,	on [Monday,
+	1276	example [5] ii	[on Monday	on [Monday
+	1280	example [14] i b.	[on Tuesday or Wednesday].	on [Tuesday or Wednesday].
+	1284	example [28] i	[the premiers of Queensland and Tasmania]	[the premiers of Queensland and Tasmania]
	1305	2nd line below [38]	[iia/iib]	[ib/iib]
	1321	note 41	similar to that of yet	similar to that of yet
	1325	2nd line below (d)	[85] of §2.2	[85] of §2.11
	1349	line 4	‘__ unlocked the	__ unlocked the
	1349	7th line below [39]	The nonce-constituent analysis	Nonce-constituent coordination
	1365	sentence just above [3]	Secondly, the canonical and non-canonical versions do not invariably have the same truth conditions:	Secondly, non-canonical clauses do not invariably have the same truth conditions as their syntactically more basic counterparts:
	1389	3 lines above section header 5.3	(cf. the three days later of [11vi]),	(cf. the three days later of [1vi]),
	1458	4th line below [18]	in [i] refers	in [18] refers
+	1587	line 4 up	length, moth, strength	length, strength
+	1587	line 5 up	lath, oath, sheath,	lath, moth, oath, sheath,
+	1605	example [44]	thrive	thrive R
+	1623	line 23	consraints	constraints
	1652	7th line after [15]	comparable to that seen in [14] above	comparable to that seen in [11] above
+	1725	line 23	(' or ' ')	(' or ")
	1736	example [4] ii b	he did not want have to	he did not want to have to
	1741	foonote, bottom line	to whit	to wit
+	1755	example [11] ii	'She	She
	1771	line 8	Sumney	Summey
	1777	Sumney entry	Sumney	Summey
	1778	last entry	Anna	Anne
+	1796	column 3	let 208, 271n,	let 208, 270n,
+	1809	line 3, col. 2	tortelleni 1594	tortellini 1594
+	1817	line 3	co-indexing 49, 68, 1037, 1039, 1085, 1088	co-indexing 49, 68, 1037, 1039, 1085, 1088, 1454
+	1823	‘fused-head’ entry	384,	384-5,
	1828	line 9 of column 2	metalinguistic negation 724, 1101n	metalinguistic negation 724, 790, 1101n
+	1837	line 21, col. 1	1582n	1581n

Additions to lexical index (these references were missing in 2002 but have been added in recent reprintings)

+	Page	Where	What it says	What it should have said
+	1784	line 6 of column 3	buy 230, 232, 235, 248, 260, 285,	buy 220, 230, 232, 235, 248, 260, 285,
+	1786	nearly halfway down column 2	contain 167-8, 1432	contain 167-8, 220, 1432
+	1788	line 10, col. 2	dive 296	dive 296, 1604
+	1795	line 11 of column 2	inquire 975-6, 978, 1027, 1529	inquire 220, 975-6, 978, 1027, 1529
+	1806	col. 3, below line 14		spit 1604
+	1811	one third down column 3	wonder 170, 600, 871, 882, 958	wonder 170, 220, 600, 871, 882, 958

Finally, this is as good a place as any to state a general warning that a few lists of lexical items that are claimed not to have some property are longer than they should have been — they get shorter each time we look at a larger corpus. One example is the list of strictly transitive verbs, those that take a truly obligatory object (see section (b), Selective obligatoriness, on page 246). We include the verb use as strictly transitive; but in connection with illegal drugs an objectless use has developed (Amy is using again). Verbs taking strictly obligatory objects are extremely rare, especially if one considers secondary forms (occurrences in non-finite clauses like infinitivals or participials are more likely to be able to dispense with the object NP).

Another example is the list of monosyllabic adjectives that do not inflect for comparison (see page 1583, [9]). Inflected forms of the adjectives we list there (cross, fake, ill, like, loath, prime, real, right, and worth) are certainly very rare; but crosser definitely occurs (and was more frequent in British writing about a century ago); faker and iller and realer can occasionally be found; and so on. Although *worther and *worthest do not appear to exist at all, in general, monosyllabic adjectives that absolutely never inflect for grade are extraordinarily scarce; we only list ten, and even that is too many. We should probably have listed just worth, loath, and perhaps prime.