Note: I posted earlier versions of parts of this essay in discussions over the decade during which I collected the data I discuss.
Decades ago I went to an autograph party at an F&SF bookstore during which the two authors of a new novel were not only signing the books, they were turning to a specific page in each copy to change one letter in the text. How much can one letter matter in a text of over a million letters? A lot. That change of one letter made a difference in understanding the nature of an entire alien race. (The Mote in God's Eye by Larry Niven & Jerry Pournelle 1974 hardcover p. 465: The children should have been spared [spaced].)
The term "Grammar Police" is often used dismissively to suggest that someone is being too nit-picky, but I am a strong proponent of careful writing. The 99.44% purity of the old Ivory Soap slogan isn't good enough for text. When words are the entire medium of communication, as they are in any work of fiction without illustrations or audio, getting every word right is essential. Many errors are less critical than my opening example, but every error is a barrier to the reader getting the author's intended meaning.
I can’t help noticing typos, poor grammar, inappropriate word choices, homophone errors, continuity errors, and some classes of scientific and historical errors. Some I assume to be from the author and some from a copy-editor [yes, editing CAN add errors], but I blame the publisher for ALL errors in a published work. Some writers are better able to proofread than others, but it is harder for the original writer to spot problems in text than it is for any other reader because we tend to see what we intended rather than what we actually created. Permanent mindos (mental errors instead of typographical errors) are a whole class of errors that writers can NEVER spot because what is written is what the author believes to be correct. (Transient mindos, momentary mental errors, can be spotted by the author later. See my definition of "mindo" here.) Supplying other eyes and minds to ensure a clean text is one of the all-too-often-shirked responsibilities of publishers.
I copy-edited a small quarterly journal for over 20 years. My day job is computer programming, which requires very careful attention to exact spelling. I can't turn the mental circuits off for recreational reading. I can & do enjoy books despite errors--errors irritate me without stopping me from reading. If I stopped reading at a few errors, I would have to abandon more than half of my reading, while my actual DNF (Did Not Finish) rate is about 1 book in 1,000.
I have joined periodic discussions of errors ever since I've participated in online discussions of what I read. In 2005, after encountering a book with more errors than usual, I started keeping a record of errors I noticed during my recreational reading. (Several years before that I had started keeping an "oops" list of wrong words used that I noticed.) This is not deliberate proofreading, just what I can't help noticing. In early 2015, when I reached a full 10 years of logs, I stopped the routine logging of errors seen in all books read. (I obviously didn't record any errors that I didn't notice when I didn't know a particular usage any better than the author and publisher.) I like discussions to be based on objective data when possible, so my error tracking was an attempt to bring a little objectivity to this topic. My sample isn't randomized or controlled in a way suitable for scientific statistics, since most of my reading is in the romance & F&SF genres, but it is a reasonably large sample recorded by a single reader. Even the single reader aspect doesn't make the data completely consistent, since I started reading ebooks in 2006, changing from 100% printed books in 2005 to 26% printed books so far this year. I also encountered enough formatting problems (extra or missing spaces, bad hyphenation, broken paragraphs, etc.) in ebooks that in 2010 I stopped treating them the same as other errors and just started counting them (or estimating a count in really bad cases). The ebook formatting issues could also have been affected by a change from lrf (Sony proprietary) ebooks to epub (a more widely used standard) ebooks several years ago. Messed up formatting in ebooks has decreased over the years since I started getting ebooks.
After eliminating almost 600 duplicates from rereads, I have 3,741 books or shorter works logged with counts of copy-edits I noticed during 10 years of recreational reading (early 2005 to early 2015, my 50s, with a few added later in 2015). After cutting out anthologized or collected short stories to prevent double-counting, I have 2,672 entries.
The genre breakdown is 11 cartoon, 7 erotica, 2 historical fiction, 17 mystery, 8 non-fiction, 5 other fiction, 1,658 romance, 767 F&SF, 18 short erotica, 2 short mystery, 139 short romance, and 39 short F&SF books. There were a few other printed non-fiction books read during the same period that I didn’t log errors for, but most of my full-book non-fiction “reading” in the decade was audio that I didn’t try to track errors for (114 titles, mostly courses from The Teaching Company [aka The Great Courses]). I didn’t log errors from short or partial-book non-fiction reading.
When I sort by copyright year, there is little difference in error rates of printed books over several decades, despite complaints about declining text quality being fairly common in online discussions.
Here is a breakdown of three kinds of average of copy edits by copyright decades. Mean is what is commonly meant by average (sum divided by count), median is the middle of a sorted list, and mode is the most frequent entry.
If you add in formatting problems, ebooks look much worse. I think it is reasonable to say that the quality of EDITING of books has not changed greatly over the last several decades, though ebooks don't get quite as good editing as printed books, but that with ebooks the quality of PRODUCTION has deteriorated due to some publishers' laziness or incompetence or deliberate devaluing of ebooks.
One way of looking at text quality is that 596 of 2,672 books (22.3%) had zero copy-edits that I noticed and logged, 523 had one, 421 had two, 306 had three, 206 had four, 457 had 5-10, 161 had 11 or more, and a few had notations more complicated than a single number. Of the more complex counts, one was the same error repeated so many times I listed the error. Two others had so many problems I stopped recording specific typos or formatting problems and just estimated total errors.
Formatting errors cropped up in so many early ebooks that I started counting them separately from typos and mindos in my log, so some of the errors counted in the previous paragraph are formatting errors included as copy-edits (typos and mindos) before I separated the counts in my logs. Of the worst 328 books with over 10 errors each (adding formatting errors to typos & mindos), 43 were printed books & the rest were ebooks. None of these books were ARCs (Advance Reader Copies), so supposedly they had all been edited.
While cleaning up my error log to prepare to write this, I deleted several items I had logged as errors when reading that later dictionary or other checks showed were OK (or at least potentially OK with allowed usages). This points to a totally separate sort of discussion that sometimes comes up when readers discuss errors: false errors, where a reader says the author got something wrong and it turns out the author was right.
The errors I logged can be divided into seven major categories: 6,279 typos or word substitutions, 1,406 missing words needing to be inserted, 1,118 extra words needing to be deleted, 317 continuity or factual errors, 146 layout problems, 82 word order problems, and 46 Reader accent/character display bugs, for a total of 9,394 (after I deleted iffy items or resolved questions by looking things up).
The last category is mostly specific to an ebook device: several models of the Sony Reader could not correctly display some accented characters or other symbols outside a basic ASCII or ANSI character set. I don't know if other models or ebook readers have the same problems, though I know that the Kobo Aura I switched to after Sony stopped selling the Reader does not have the problem with the SAME characters. However, even though the Kobo Aura displays most of the characters that the Sony Reader couldn't display, I have recently encountered other character problems.
Even though I saw this problem mostly in ebooks, in a few cases I have seen equivalent errors in printed books. If you ever see the non-word faade where the word facade (with a cedilla) would make sense, that text is a victim of this problem. I think the printed books where I saw this involved electronic manuscripts or bad file conversions along the way to camera-ready copy. Some accented characters and ligatures (letter pairs like fl or nm rendered as a single character) don't translate well in some programs.
Depending on how the character translation or rendering fails, characters can simply be missing, or a blank or a question mark or a different character can take the character's position
The word order problems on my list involve having all the right words for an intended meaning, but with a pair of words reversed or a longer scramble, like "asking to him see" instead of "asking him to see" or "can't reach just down" instead of "can't just reach down".
Word order errors are definitely mindos rather than typos, though I suspect that some large fraction of them are the results of editing or rewording rather than having been typed in the bad order of the published copy.
Layout problems are mostly from the process of editing and producing a book, though some can come from the author's manuscript. This category includes missing text, duplicated text, scrambled text (which can overlap with the Word Order category), extra inserted text (which can overlap with the Extra Words category), wrong symbols, formatting glitches like italics changing in mid-word, extra or missing paragraph breaks, extra or missing spaces, inappropriate hyphenation, etc. Some layout problems look like the results of careless use of search & replace functions. Layout problems can be as short as a single character or as long as multiple chapters.
All errors in this category are mindos, not typos.
Continuity errors can involve a description of anything in a story that is inconsistent with an earlier description. This can include character eye color, clothing, body placement, durations of events, time zones, titles of nobility, names of people, places and things, etc. Quite a few of these errors are from a wrong character's name in a scene, often a character name not otherwise in the scene.
Factual errors can be anything I notice that is contrary to what I know about the story context, including anachronisms (e.g., modern psychobabble in dialog in stories set before the late 1800s or a Fabergé egg before the Fabergé who made fancy eggs was born), astronomy (e.g., the Moon in the wrong part of the sky or a wrong constellation), geography (e.g., wrong city or direction or island or ocean or princedom or valley), history (e.g., wrong ruler), numbers (e.g., twenty-seven centuries for 270,000 years), titles & forms of address, etc.
I didn't note just extra words--I also noted extra apostrophes, commas, dashes, numbers, parentheses, periods, pipe characters, quotation marks (123 of them), question marks, semicolons, slashes, spaces, and stray letters.
Many extra words are exact duplicates (of single words, pairs of words, or longer sets of words) or duplicative variations of wording, but there are also plenty of extra words that aren't duplicating any nearby text. Many duplications are directly adjacent, but many others are separated by part of a sentence--my list of errors isn't in a form where I can easily count to see whether adjacent or separated duplications are more common. As with word order errors, I suspect that some large fraction of single extra words are the results of editing or rewording.
Most extra words are minor irritants because they interrupt the flow of reading, and a few are more significant because they reverse a meaning (e.g., an extra "not").
As with extra words, missing words included plenty of punctuation marks: ampersands, apostrophes, braces, brackets, commas, dashes, parentheses, percents, periods, quotation marks (181), and spaces (158).
These are missing words noted at least 3 times in my log: 4 a or the, 107 a, 3 an, 17 and, 13 are, 39 as, 26 at, 40 be, 3 been, 7 by, 3 care, 4 do, 3 for, 4 from, 51 had, 15 have, 32 he, 12 her, 9 his, 7 I, 3 if, 23 in, 14 is, 17 it, 5 know, 3 more, 3 no, 14 not, 60 of, 10 on, 8 one, 3 other, 3 see, 10 she, 3 so, 8 than, 8 that, 61 the, 3 they, 8 to be, 134 to, 22 was, 6 were, 6 what, 10 with, 7 would, and 15 you. Everything else in this category appears only once or twice in my sorted log. None of my corrections for missing words are more than two words long, and most are just one word. Anything with more than a couple words missing would get logged as a missing text layout error because I couldn't guess exactly what words were missing.
Missing or extra quotation marks deserve a special mention. Quotation marks let the reader distinguish characters’ speech from other text, so they are one of the more critical punctuation marks. My log has 304 missing or extra quotation marks, or about 3% of the total logged errors. That’s a lot of reader confusion.
This is a summary from 6,279 errors recorded from 2,672 books read from 2005 to 2015. I usually don't record nonstandard English in dialogue as errors when I can tell it is intentional (like dialect). Many other errors involve spelling, punctuation, capitalization, formatting, word order, extra words, repeated words, or omitted words, which I summarized above. This section is mostly about word substitution errors where an incorrect real word is used. I will use slashed word pairs to indicate incorrect/correct words.
Did educators lie down on the job? Lay/lie errors are far and away the most common. My list includes 946: 285 lay/laid + 276 lay/lie + 192 laid/lay + 119 laying/lying + 28 laid/lain + 11 lie/lay + 11 lying/laying + 10 lay/lain + 8 lays/lies + 2 lies/lays + 2 lied/lay + 1 lain/lay + 1 lay/lies. This one type of error accounts for 1/10 of ALL the errors I logged in 10 years! There is much more use of past tense than present tense or other verb forms in most fiction I read, and these error counts reflect that. They also show that people mess this up in both directions, though not equally.
The intransitive verb (action by subject) is lie, lying, lay, had lain (rest, be prone).
The transitive verb (action of subject on object) is lay, laying, laid, had laid (place, put down).
The confusion of "lie" and "lay" came up years ago on a romance novel discussion list I'm on, with a list member from Germany (who learned English English) having a better understanding than most of the American members (including me). If I had ever clearly learned the proper usage in school I had long forgotten it, so I pulled a few dictionaries and style books (which I tend to collect) and made sure I understood the correct usage. I had to look in several books. Even though this is easy to keep straight when you know it, it is not always clearly taught or written about. IIRC, the best explanations were in books from Oxford University Press, not American books. I graduated from college in 1976, so I'm sure a lot has changed since I took any English classes, but I suspect that this is still a problem in many American schools.
It's a boy! It's a girl! Subject/object pronoun errors are described separately below. This paragraph is about all the other pronoun problems: typos, verb problems, gender flips (e.g., she for he), etc. I counted 212 total wrong-pronoun errors: 5 her/he, 4 his/he, 2 it/he, 10 she/he, 2 the/he, 1 they/he, 4 he/her, 6 him/her, 11 his/her, 1 them/her, 1 her/hers, 4 her’s/hers, 1 her/herself, 3 himself/herself, 6 her/him, 1 herself/him, 1 his/him, 1 me/him, 1 them/him, 4 he/his, 27 her/his, 2 hers/his, 1 him/his, 1 its/his, 1 them/his, 9 this/his, 1 It/I, 1 myself/I, 2 you/I, 1 my/me, 1 you/me, 2 her/my, 1 me/my, 1 I/myself, 2 one’s/ones, 1 ones/one’s, 2 other/others, 1 others/others’, 1 other/other’s, 2 others/other’s, 11 he/she, 4 the/she, 2 he'd/she’d, 5 he/the, 1 I/the, 2 them/the, 4 they/the, 3 her/their, 1 his/their, 4 its/their, 1 my/their, 1 theirs/their, 1 they’re/their, 3 him/them, 3 it/them, 2 it/they, 4 the/they, 1 their/they, 5 whose/who’s, 6 who’s/whose, 1 him/you, 1 I/you, 1 we/you, 5 you/your, 1 you’re/your, and 10 your/you're. I also counted 32 errors involving wrong verb conjugations in contractions with pronouns: 4 he’d/he, 2 I’ve/I, 11 she'd/she, 2 she's/she, 4 she/she’d, 2 she’s/she’d, 1 they’re/they, 1 you’ll/you, 1 you’re/you, 1 you’ll/you’d, and 3 you’re/you’ve.
Adding these 212 & 32 errors to the 87 subject/object errors I list below makes total pronoun substitution errors a strong second place after lay/lie errors. If you add the 108 pronouns in the missing word list above and 98 extra pronouns in the extra word errors (not detailed in this write-up), that brings the total of errors involving pronouns to 537--more than half of the lay/lie count.
All other errors were an order of magnitude less frequent than lay/lie errors and pronoun errors.
The next most common error makes me grit my teeth. I counted 124 instances of "grit" instead of "gritted" for the past tense, usually in "gritted his/her teeth". None of the references I checked say that "grit" is a valid past tense. None even mention "grit" or "gritted" as a problem, so I guess most writers dealing with usage consider the correct past tense too obvious to mention.
Don't subject me to this! I counted 87 subject/object pronoun problems: 33 I/me, 16 me/I, 13 he/him, 12 she/her, 4 her/she, 3 whom/who, 2 who/whom, 2 him/he, and 2 she/him and her. I could count another 9 me/I errors from one book I read, except it was clear from the text that the first-person narrator is supposed to think he knows correct usage. These problems are already written about in many places. Most instances happen with x and y constructions (e.g., Joe and I did). Just use the test of removing the "and y" to test your usage.
Into every life . . . . I counted 81 confusions of "in to" for "into" (67 into/in to, 14 in to/into). Many incorrect instances of "into" were after forms of "give": "give into" instead of "give in to" (yield to). If you include a pause when you say it, include a space when you write it. (An example: "They moved in to begin their new life together." ["moved in" + "to begin", NOT into]) I don’t know how many of these might be typos where a space bar didn’t register and how many are mindos where the proper meanings were missed. There were also 4 onto/on to, a parallel problem.
Reluctant dislike: I counted 66 instances of "loathe" instead of "loath" in descriptions of reluctance ("loath to").
Next were 62 instances of one of the old standards everyone is warned about: its vs. it's. I saw 45 it's/its, 6 its/it's, 4 it/its, 4 it’s/it, 1 it/it's, 1 it’d/it, and 1 it'd/it's--definitely unbalanced toward extra apostrophes. Always mentally expand "it's" to "it is" if you aren't sure what to write.
I saw 5 is/are, 3 are/is, 21 were/was, 25 was/were and 2 wasn't/weren't errors, for a total of 56 errors from failing to keep track of numbers.
Next is another of my pet peeves that I haven't seen mentioned in the usage books I've checked. Between requires two or more objects, yet I counted 54 instances of between with singular objects. Most of these errors are in love/sex scenes (cleavage, cleft, crevice, gap, gown, juncture, seam, slit, valley, vee and zipper are all singular).
Don't tread on me: I counted 53 problems with "tread" (tread, trod, had trodden): 34 tread/trod, 13 trod/tread, 4 tread/trodden, 1 tred/tread, and 1 trounced that I think was supposed to be trod.
Heavy metal: I saw 48 led/lead problems (45 lead/led, 2 let/led, and 1 led/lead). The present-tense verb is a homograph with the metal (lead) and the metal is a homophone with the past-tense verb (led), but the two tenses of the verb are neither homophones nor homographs (lead/led: EE sound/EH sound), yet this clearly gives people trouble.
Damp manuscripts: I logged 48 pour/pore (pore pored poring) errors (44 pore over, 4 pore through), all in the same direction. I think this is a rarity effect, since "pore" has no other modern uses as a verb. Unfortunately, every time I see these errors I think of the damage of applying liquid to books instead of studying them intently.
I object: I counted 46 adverse/averse errors, all in the same direction. Adverse usually refers to hostile or unfavorable situations or conditions. Averse (usually "averse to" in the U.S.) refers to dislike or opposition. Most appearances of averse are variations of "not averse to" (not opposed to). I think many people mess this one up just because there is almost no other use of averse in modern English.
Spreading the bomb? I saw 40 diffuse/defuse (defused defusing) errors, all in the same direction. Diffuse means spread or disperse. Defuse means remove a fuse (reduce danger). Figurative uses with situations, temper, tension, etc. should all be defused, not diffused.
Horse & buggy rulers: I counted 37 reign/rein (reined reining reins) errors, all in the same direction. Maybe our culture is simply too far from horsemanship as a common skill for writers to remember the spelling of those straps for steering horses.
Better than then. Years ago I thought this error was just an ESL (English as a Second Language) problem, but I have 36 instances (14 then/than, 11 that/than, 8 than/then, and 3 than/that). I honestly have no idea why this happens, though it might just be a common word typo problem.
It’s not all Greek to me! Sometimes one book can provide a bountiful harvest of errors. This example is a book with a Roman gladiator hero. While I applaud authors who provide varied settings, I would much prefer to see correct foreign words used. I saw several Latin problems: 3 lanistra/lanista (not a word/trainer or manager of gladiators), 2 rentarii/retiarii, 6 rentarius/retiarius (renter/gladiator using a net), 1 renarius's/retiarius's, 11 rudius/rudis (ceremonial retirement sword—both spellings are online, only rudis is in my Latin dictionary), 5 trignon/trigon (not a word/three-player ball game in baths), and 1 stolla/stola (not a word/dress, clothing), plus several English problems, including 2 grieve/greave (for armor).
Being careful about to/too/two is another standard warning, but I still counted 26: 20 to/too, 3 so/too, 1 more/too, 1 too/to, and 1 two/too.
Past time passed: I counted 24 errors with passed (12 passed/past, 11 past/passed, and 1 paned/passed). Passed and past are closely linked words, but past is an adjective and passed is the past tense of a verb. They really can't validly trade places.
Not to be borne: I saw 23 borne errors (15 born/borne, 5 bore/borne, and 3 borne/born). These are all forms of bear, so some confusion is understandable. Bore is the past tense and shouldn't be used after had. Born and borne are both past participles, but born is restricted to references to birth and borne covers all other meanings.
Too much to drink? I counted 22 bad conjugations of drink, all of the form "had drank". Drank is the past tense of drink. Drunk is the past participle of drink. Don't use the simple past tense form after had.
Ring the wringer? I saw 14 forms of ring/wring, 6 ringer/wringer, and 1 wrung/rung. To ring is to surround or to make a sound. To wring is to squeeze and twist (usually to force out liquid). They really are different.
Censor that! I saw 20 cens. . . errors: 13 censor/censure, 2 censure/censor, 2 censorship/censure, 4 censor/censer, and 1 stricture/censure. A censer is a container in which incense is burned. To censor is to cut or suppress. To censure is to criticize. People often censure the creators of material they censor, but there is a difference.
Sight the site or site the sight or cite the site? I counted 15 site/sight, 4 sight/site, and 1 cited/sighted errors. This is another standard warning.
Not one whit: I counted 19 wit/whit errors, all in one direction. Wit is intelligence. A whit is a very small amount. I suspect this is another rarity problem.
A noisy climb: I saw 17 clamor/clamber errors (13 clamor/clamber and 4 clamber/clamor). To clamber is to climb, usually awkwardly or laboriously. Clamor is loud noise. I assume there are so many of these errors because neither word is in everyday use.
Can you stand any more of this anymore? I recorded 16 anymore/any more errors. Without the space, anymore means to any further extent or any longer, with the space any more is an emphasized form of more or a variant spelling of anymore. Because of the spelling option, it is only possible to spot errors in the direction I recorded.
Lose that extra “o”: I saw 13 loose/lose, 2 lose/loose, and 1 lost/lose error. Lose is always a verb (be deprived of, become unable to find, fail to win), loose can be a verb (set free, release) or an adjective (not firmly or tightly fixed or fitted, relaxed).
Clamp down: I saw 15 vice/vise errors, all in the same direction. Vice is immoral behavior or characteristics. In American English, a vise is a tool with jaws to hold objects. (In English English both are spelled vice, so I don't list this as an error when I know a book is using English English.)
Annoying tease: I counted 15 chaff/chafe errors, all in one direction. To chafe is to rub or annoy. To chaff is to tease.
What are you implying? I don't know why imply would be difficult, but I saw 11 infer/imply errors and 4 belie/imply errors. To belie is to disguise or contradict, to infer is to deduce or conclude, and to imply is to suggest, so mixing these up reverses meanings. These mistakes are apparently common enough to get a usage note in the OAD (Oxford American Dictionary).
I logged 1 needs/need to, 1 needed/need, 2 needs/need, 1 need/needs and 11 need/needed errors, all simple failures to conjugate correctly.
Smile when you say that! I encountered 11 errors with smiled (verb past tense) used where smile (noun) should have been used, and 4 smile/smiled (wrong verb tense) errors.
I saw 10 clinch/clench errors (all verbs) and 4 clench/clinch errors (all nouns). To clench is to press tightly together. A clench is a tightening of part of the body. To clinch is to grapple or embrace. A clinch is a close scuffle or an embrace. Jaws clench. Couples on covers are often in clinches.
I saw 13 forbid/forbade errors and 1 forbid/forbidden, which are simply failures to conjugate.
I logged 3 off/of and 9 of/off errors, which are probably all typos, not mindos.
Is the fair fare fair? I counted 11 fair/fare errors, all with verb usage, so several other meanings of both words aren’t involved. To fair is to streamline a vehicle with fairings, while to fare is to perform in a specific way in a specific situation, so I suspect this is another error due to the relative rarity of the correct word.
I saw 6 looked/look and 5 look/looked errors. This is just failure to keep tenses straight.
Wash that cut! I counted 11 lathe/lave errors, all in one direction. A lathe is a machine for turning an object for shaped cutting. To lathe is to use a lathe. To lave is to wash. This error produces a disturbing mental image, since every instance is in a love scene where an author intended to use lave as an approximate synonym for lick and instead produced an image of cutting.
Had they run a grammar check? I saw 11 ran/run errors, all using the simple past form (ran) where the past perfect form (had run) was needed because of some form of the word had.
I saw 8 surprize/surprise and 3 surprized/surprised errors. According to all the printed dictionaries and most of the online sites I checked, this is a misspelling.
Comparative superlative. I logged 11 worse/worst errors. Worse is the comparative of bad and worst is the superlative of bad, so the normal usage is “think the worst” or “do her worst”. Only use worse if you are dealing with only two items.
Break the brakes? I saw 9 break/brake errors and 1 brake/break error. A break is a pause or interruption. The meaning of brake intended in these cases is a device for stopping a moving vehicle.
Hard to find deception? I saw 10 illusive/elusive errors, all in the same direction. Illusive is deceptive or illusory. Elusive is hard to find.
I logged 9 phase/faze and 1 faze/phase error. To faze is to disturb or disconcert. Phase is used more as a noun, but the verb means to carry out in stages or to resynchronize. I suspect this homophone error is another due to the relative rarity of one of the pair.
I counted 9 hoard/horde and 1 horde/hoard errors. The noun hoard is a store of valuables and the verb hoard is to amass & store valuables. The noun horde is a large group of people, such as an army of nomadic warriors. The 9 errors were all noun usages, wrongly using the treasure word for the army. The 1 error was a verb usage, wrongly using the word that is only a noun.
Ouch! I saw 10 ravage/ravish errors. To ravage is to severely and extensively damage. To ravish is to seize and carry off, to force sex, or to enrapture (fill with delight). I suspect these errors come from authors wanting to intensify a scene without realizing that these terms aren’t that synonymous.
I counted 7 roll/role and 3 role/roll errors. A role is a part played or a function assumed. To roll is to move by turning. A roll can be a cylinder, a turning movement, a prolonged sound, a kind of bread, a list of names, etc.
It’s a law! I recorded 10 statute/statue errors, all substituting a law for a sculpture.
I saw 10 suppose/supposed errors. This is just failure to keep tenses straight.
I saw 5 without/with, 4 with out/without, and 1 with/without errors. I think the 4 are simple typos, and the 6 are mindos reversing meaning.
The above notes are pretty long, so I will just give brief counts and spotty definitions for everything with under 10 instances.
3 west/east + 6 east/west
4 here/her + 5 her/here
4 liked/like + 5 like/liked
6 not/no + 3 no/not
4 opened/open + 5 open/opened
6 sunk/sank + 3 sank/sunk
8 all together/altogether (all in one place; all at once/in total)
7 bare/bear (uncover/carry; support; endure) + 1 bear/bare
8 waiver/waver (instance of surrendering a claim/shake or quiver)
3 wretch/retch (n. unfortunate person/v. vomit) + 5 wretched/retched (adj. very unhappy/v. vomited)
4 where/were + 4 were/where
5 baited / bated (taunted or prepared lures/in great suspense) + 1 bated/baited + 1 bating/baiting
7 cache/cachet (collection or storage place/prestige)
7 Here, here!/Hear, hear!
7 peddle/pedal (sell/foot-operated lever)
3 peak/pique (reach highest point/stimulate interest) + 4 peaked/piqued
7 secret/secrete (noun/verb)
7 tact/tack (sensitivity dealing with others/a boat’s course; change course)
3 used/use + 4 use/used
1 closed/close + 5 close/closed
6 dissect/bisect (cut up to study/divide into two parts)
2 beared/bore + 1 bored/bore + 3 bore/bored
4 never/ever + 2 ever/never
6 lull/loll (calm/sit-stand-lie in relaxed way)
3 make/made + 3 made/make
4 peak/peek (reach highest point/look quickly) + 2 peek/peak
6 slather/slaver (spread liberally/let saliva run)
4 straightened/straitened + 1 straights/straits + 1 strait/straight (narrow or cramped/without curves or bends)
5 thing/think + 1 think/thing
3 with in/within + 2 with/within + 1 within/with
4 wreck/wreak (cause destruction (of a ship)/cause harm; inflict vengeance) + 1 reek/wreak + 1 wreaking/wrecking
5 bided/bidden (remained or stayed somewhere/offered a price for something; commanded)
5 bows/boughs (front of ship & other meanings/tree branches)
2 imminently/eminently + 3 eminent/imminent (famous and respected/about to happen)
4 for/from + 1 from/for
4 went/gone + 1 gone/went
5 ladened/laden (I don’t think ladened is an English word)
2 or/nor + 3 nor/or
3 one/once + 2 once/one
3 peel/peal (remove skin/loud sound) + 2 pealed/peeled
1 peck/pec (stroke or bite by bird beak/pectoral muscle) + 4 pecks/pecs
4 proceed/precede (begin or continue action; move forward/come before) + 1 preclude/precede (prevent; make impossible/come before)
5 rationale/rationality (set of reasons for actions or beliefs/quality of being agreeable to reason)
1 realising/realised + 3 realized/realize + 1 realize/realized
1 spoken/spoke + 4 spoke/spoken
4 sprung/sprang + 1 sprang/sprung
3 tale/tail (story; lie/hind end of an animal) + 2 tales/tails
2 their/there + 3 there/their
3 threw/through + 1 though/through + 1 through/though
3 wanted/want + 2 want/wanted
3 ancestors/descendents (people someone descended from/people descended from someone) + 1 descendents/ancestors
4 awhile/a while
4 bath/bathe (noun/verb)
2 bit/big + 2 big/bit
4 breath/breathe (noun/verb)
4 canon/cannon (law; rule; church law; accepted collection of books/large piece of artillery)
4 canvass/canvas (get opinions or solicit votes/course cloth)
2 compliment/complement (expression of praise/thing or number that completes something) + 2 complimented/complemented
1 dessert/desert + 1 desserts/deserts + 2 desert/dessert
2 dosed/dozed (administered medicine/slept lightly)) + 1 dozed/dozing + 1 dousing/dozing (drenching/sleeping lightly)
3 ate/eaten + 1 eat/eaten
4 flaunt/flout (display ostentatiously/openly disregard)
2 floundered/foundered (struggled clumsily/sank) + 2 floundering/foundering
2 heard/hear + 2 hear/heard
2 heal/heel (make healthy/n. back of foot; v. dog follow owner; ship tilt) + 2 heals/heels
3 help/helped + 1 helped/help
1 met/meet + 3 meet/met
4 naval/navel (relating to a navy/belly button)
2 now/not + 2 not/now
1 pulled/pull + 3 pull/pulled
2 restrain/refrain (prevent someone from doing something/stop oneself from doing something) + 2 restrained/refrained
3 wrack/rack + 1 rack/wrack [Based on a usage note, either spelling is OK for the verb meaning subject to extreme stress, so I was wrong about this being an error.]
3 see/seen + 1 seen/see
4 hence/since (as time references: in the future/ago; in the time between a mentioned time and now)
4 stripped/striped (removed coverings/marked with stripes)
4 throws/throes (v. propels through air by arm-hand motions/n. intense or violent pain & struggle)
4 fob/watch (a chain attached to a watch; a small ornament attached to a watch chain/a small timepiece)
3 would/wouldn't & 1 wouldn't/would
3 ajar/a jar
3 away/a way
2 broached/brooked + 1 broaching/brooking (raising a difficult subject for discussion/tolerating or allowing something)
1 corroborate/collaborate (confirm/work jointly) + 2 corroboration/collaboration
1 counsel/council + 2 council/counsel (advisory or legislative body/advice; lawyer)
2 deduct/deduce (subtract from a total/draw a logical conclusion) + 1 deducting/deducing
3 upwind/downwind (against wind direction/in the direction wind is blowing)
3 drought/draught (prolonged shortage of water/British spelling of draft: cool air, act of pulling, act of drinking, depth of water, preliminary version)
2 easily/easy + 1 easy/easily
3 eek/eke (exclamation of surprise or horror/v. manage with difficulty)
3 epitaph/epithet (tombstone inscription/descriptive phrase or term of abuse)
2 every/ever + 1 ever/every
2 very/every + 1 every/very
2 faced/face + 1 face/faced
2 flare/flair + 1 flair/flare (special aptitude; stylishness/burst of light or flame; gradual widening)
2 gives/give + 1 give/gives
2 have/has + 1 has/have
2 incredulous/incredible (skeptical or unwilling to believe/unbelievable) + 1 incredibly/incredulously
2 jettisoned/jetted (threw or dropped from a ship, abandoned, discarded/traveled by using jets) + 1 jettisoning/jetting
1 now/know + 2 know/now
2 learned/learn + 1 learn/learned
1 lent/leant + 2 leant/lent (past of lean/past of lend)
3 lea/lee (open grassy area/shelter from wind)
2 lobbed/lopped + 1 lobbing/lopping (throwing or hitting something in a high arc/cutting off)
1 lose/loss (v. cease to have; become unable to find; fail to win/n. fact of losing something) + 2 loses/losses
2 men/man + 1 man/men
1 millennium/millennia + 2 millennia/millennium
3 knick/nick (not sure this is an English word/small cut or notch)
3 proscribed/prescribed (forbade; denounced or condemned/authorized a treatment; recommended)
2 cue/queue (signal for action/line; sequence; list) + 1 quay/queue (platform for loading ships/line; sequence; list)
3 renown/renowned (noun fame/adj. famous)
1 rouge/rogue (red cosmetic/unprincipled man) + 2 rouges/rogues
3 scene/scent (place where something occurs; sequence of action in a story/distinct smell)
3 sexton/sextant (person who looks after a church/navigation instrument)
3 shown/shone (past part. of show/past part. of shine)
1 shudder/shutter (v. tremble from fear or repugnance; n. act of shuddering/n. hinged panel outside window; device controlling camera exposure) + 2 shudders/shutters
1 snicker/snick (give a smothered laugh; snigger/cause to make a sharp clicking sound) + 1 snickered/snicked + 1 snickering/snicking
3 stealing/steeling (taking without right or permission/mentally preparing to do something difficult)
3 tortuous/torturous (twisty/causing extreme suffering)
1 undue/undo + 2 undo/undue (unfasten; reverse effects/unwarranted or inappropriate and excessive)
3 weigh/way (measure using scales; have a specific weight/method; style; road; path)
2 wherein/whereas + 1 whereas/wherein
3 Wig/Whig (hair covering for the head/historical British political party)
3 winched/winced (hoisted or hauled with a winch/shrank, flinched or grimaced from distress)
2 want/wont (desire or lack/accustomed) + 1 wont/want
2 along/a long
1 adjured/abjured + 1 abjured/adjured (renounced/urged someone to do something)
2 Incubae/an Incubus
2 protagonist/antagonist (leading or major character in fiction/active opponent of someone|something, adversary)
2 any thing/anything
2 be come/become
2 belied/betrayed (disguise, contradict, fail to justify/be disloyal, reveal)
2 bellowing/billowing (roaring loudly/(fabric) filling with air & swelling outward, (smoke) flowing outward with undulating motion)
2 blanche/blanch (a female name/make white by extracting color)
2 slur/blur (n. damaging allegation; act of speaking with sounds run together/n. something that can’t be seen clearly)
2 cashes/caches (gives or gets money for a check or money order/stores away)
2 crackled/cackled (made rapid cracking noises/gave a raucous clucking cry [or a similar sound during laughter])
2 challis/chalice (a fabric/a goblet)
2 compromised/comprised (settled a dispute by mutual concessions, weakened a reputation or principle/consisted of, made up)
2 content/contentment (adj. peacefully happy, n. state of satisfaction/n. state of happiness and satisfaction) [I was wrong about this being an error, but I still think it is a less than optimal word choice because I think of the adjective when I see “content”.]
2 contingency/contingent (n. uncertain future event/adj. subject to chance, dependent on certain events; n. group of people)
2 dent/dint (n. a slight hollow made by a blow or pressure; v. make a dent/“by dint of” = by means of)
2 duel/dual (deadly contest/two-part)
2 solstice/equinox (days sun changes direction in summer & winter/day-night length equal in spring & fall)
2 gambit/gamut (action or remark to gain advantage/complete range or scope of something)
2 grieve/greave (v. feel grief/n. armor to protect the shin)
2 green house/greenhouse
2 grizzly/grisly (n. large brown bear; adj. gray/adj. causing horror or disgust)
2 grit his jaw/gritted his teeth or clenched his jaw
2 honor/horror (high respect; esteem; privilege/intense fear, shock or disgust)
2 inclination/indication (natural tendency; slope/information that shows something; symptom)
2 importune/importunate (v. ask persistently/adj. annoyingly persistent)
2 jam/jamb (instance of becoming stuck, gathering of musicians, sweet fruit spread or preserve/side post or surface of a doorway)
2 least/lest (smallest/with intent to prevent; to avoid risk of; in case)
2 brevity/levity (concise & exact use of words; shortness of time/humor or frivolity)
2 lightening/lightning (v. making or becoming lighter or brighter/n. natural electrical discharge; adj. very fast)
2 moral/mortal (adj. concerning right & wrong; n. a lesson/adj. subject to death)
2 muzzle/nuzzle (put a guard over the face/rub gently with nose & mouth)
2 palatable/palpable (pleasant to taste; satisfactory/able to be touched or felt)
2 peripheral/periphery (adj. on the edge of something/n. edge of an area or object)
2 parish/perish (church administrative district/suffer death)
2 plain/plane (adj. simple; clear; n. large flat area of land/n. flat surface; level of existence; airplane; adj. completely flat; v. glide) [Both logged instance wanted the level of existence meaning.]
2 plumb/plump (v. measure depth; measure vertical; n. plumb bob; adv. exactly; adj. vertical; v. install pipes/adj. having a rounded shape; v. adjust a pillow to be rounded)
2 pummel/pommel (strike repeatedly with fists/knob at the end of a sword handle; upward projecting part of a saddle)
2 proliferate/profligate (increase rapidly; multiply/recklessly extravagant or wasteful)
2 rites/rights (solemn ceremonies/moral or legal entitlements)
1 skewered/skewed + 1 skewed/skewered (adj. askew; v. suddenly changed direction/fastened with a pin or a skewer [long thin wood or metal to hold food for cooking])
2 galaxy/solar system (a system of millions or billions of stars/the planets and smaller bodies orbiting one star)
2 sewn/sown (joined by stitches/planted)
1 starring/staring + 1 staring/starring (looking fixedly at something/having someone as a principal performer; having a principal role)
2 subtly/subtlety (adv. of subtle/n. quality of being subtle; a subtle feature)
2 watched/switched (observed attentively/changed position, direction, or focus of; changed; substituted)
2 swore/sworn (past of swear/past participle of swear)
2 taught/taut (past of teach/adj. stretched or pulled tight; tense)
2 thrice/trice (three times/moment)
2 truck/trunk (heavy wheeled vehicle/main woody stem of a tree; luggage space at the back of a car)
2 tube/tub (long, hollow cylinder/wide, open, deep flat-bottomed container)
2 underweigh/underway (usage error/in progress; moving through water)
2 urban/urbane (relating to cities/suave, refined)
2 venal/venial (susceptible to bribery/slight and pardonable)
2 wined/whined (entertained by offering drinks/made long, high-pitched complaints)
I won't list any proper name substitutions or any of the word substitutions that I have only once in my log.
Even though this essay is mostly about repeated errors, I also notice new-to-me errors. The supply of unique word errors never runs out.
I started keeping a list of "oops" words even before I started the book-specific log of errors seen. That list now has over 900 entries, making it a bit long to post.
The New Oxford American Dictionary from Oxford University Press (my copy is a 2001 edition 8.5"x11" hardback with 2062 pages). Usually the first one I check for current language, this has very good usage notes and shows many British / American differences.
The Compact Edition of the Oxford English Dictionary from Oxford University Press (my copy is a 1971 edition with 4116 pages each showing 4 page images, plus a 1987 supplement with 1412 four-page pages). This is the one I check for obscure or obsolete words or historical usage.
My routine alternate to the OAD is The Random House Dictionary of the English Language (The Unabridged Edition) from Random House (my copy is a 1967 edition 9.5"x12" hardback with 2092 pages).
For pronunciations: The Oxford Dictionary of Pronunciation for Current English from Oxford University Press has no definitions, but gives British and American pronunciations for many words. My 2001 edition is missing a complete legend for the phonetic symbols used, so I worked up my own legend page I could use with it.
For name questions, I usually start with A Dictionary of First Names and A Dictionary of Surnames, both by Hanks & Hodges and from Oxford University Press, after which I have several books of baby names to check.
I also routinely use dictionaries in my ebook devices. Tapping or pressing words on the screen is enough of a habit that I sometimes find myself reaching toward pages in printed books as if to tap on unfamiliar words.
For years, the family had a large Unabridged Webster's, but it was lost in a fire several years ago. I recall it as about twice the size of the OAD or RH.
I have had CDs of several large dictionaries over the years, but due to copy protection schemes or other issues, none have kept working through years of computer and OS upgrades. The search features in some were quite nice, but I didn't like the upgrade expenses (or the complete unavailability of some upgrades).
I did a quick comparison of printed books to ebooks. 1,586 ebooks have about 1.7 times the copy-edit error rate of 1,086 printed books, but when you add in formatting problems the ratio jumps to about 10.1 to 1.
Mean of copy-edit errors for 1,086 printed books is 2.70.
Mean of copy-edit errors for 1,586 ebooks is 4.60.
Mean of combined copy-edit & formatting errors for 1,086 printed books is 2.93.
Mean of combined copy-edit & formatting errors for 1,586 ebooks is 29.67.
Note: 197 of the ebooks in these counts are shorter works: novelettes, novellas, even short stories that are impractical for stand-alone printed publications but quite easy to create as stand-alone ebooks.
I worked up some counts and averages by publisher.
|Publisher||Books||with 0 Err||with >=10 Err||Mean||Median||Mode|
|G. P. Putnam's Sons||18||12||66.7%||1||5.6%||1.11||0||0|
"with 0 Err" here means with zero copy-edits and zero format problems spotted & logged.
"with >=10 Err" means with 10 or more problems logged.
1632, Inc. is a special case: all the books are issues of a bimonthly ebook called the Grantville Gazette, which has stories and articles in the 1632 universe. Since someone gave me an address years ago, I've been sending the editor my copy-edits after I read each release. I may be reading these ebooks more carefully than others just because of that. Of course, it is also possible that a bimonthly publication isn't edited the same as other books.
You can see in several means how one or two books with really bad counts can greatly raise an average. All the highest means are from really bad numbers on a few books.
Errors in texts can be viewed as increased entropy (disorder). Sometimes disorder increases from one edition of a book to another.
Some ebooks (and even some printed books) produced by OCR (Optical Character Recognition) of a scan of an earlier edition introduce a lot of errors. OCR has improved over the years, but thorough copy-editing is still desirable.
Even with digital files, incompatible rendering logic or incomplete format translation support can produce problems. E.g., all the apostrophes were lost in a format conversion of an ebook I read last month. There are a LOT of apostrophes in most text.
In print, I saw increased errors with the Harlequin Heyer reprints of the early 2000s. The books were very nice in many ways, replacing our falling-apart older copies with more readably sized text, but they introduced dozens of NEW errors. (I checked against our older copies from other publishers.) This is an example of source-specific degradation of text.
An earlier (1970s) example of quality going down was Time Enough For Love by Robert A. Heinlein. 26 of the 34 errors that I noticed in the paperback copy I reread during this project are not in my earlier hardback copy.
Production (manufacturing) errors include binding errors in printed books where a signature (block of pages) is missing or misplaced. In ebooks, I have seen missing and misplaced chapters and other large blocks of misplaced text. I have also seen smaller blocks of text missing. Missing text can only be fixed if a copy of the work containing the missing part exists somewhere or the author is available to ask. A few years ago I got a couple Harlequin ebooks from Kobo that had major blocks of text missing. When I contacted Harlequin, they said they had provided corrected ebooks to all sellers, but Kobo didn't correct their available downloads when I contacted them, so this is a different example of source-specific degradation of text.
596 books in my sample (22.3%) could be perfect texts, since they had no errors that I noticed and logged. This means one reader knocked the remaining 3/4 of the books in my sample from the possibly perfect category. I don’t know how many books might still have no noticed errors after the attention of all their readers.
For me, a text would be perfect if it conveyed exactly what the author wanted to convey, with no distracting OCR, spelling, grammar, punctuation, layout, wrong word, historical, scientific, cultural, or continuity problems in any omniscient or third-person narrative text. I wrote something about Accuracy that includes more types of errors and appeared as part of an AAR column years ago.
Error-free also means no errors of any kind in dialogue or first-person narrative text OTHER THAN those deliberately used to convey aspects of characters. Actual speech is full of uh, um, er and other filler that is rarely included in written dialogue. It also includes many errors that speakers would edit out if given the chance, which written dialogue usually does edit out.
Use of dialects and regional languages are stylistic choices rather than errors unless they descend into godwottery, but that shades into a question: is it proper to use modern American English in dialogue involving non-American characters and settings? I know there is a convention of using the main language of a story for foreign dialogue (a sort of implicit translation), but should that convention be used for speech in other flavors of English? If the speech of all characters isn't presented in American English in books for American markets, should that of non-American characters use British (and other regional/national) English spelling and punctuation?
I (and other people) have suggested before that publishing could benefit from a beta-testing system. If books had version numbers like a lot of software, I would say the finished manuscript (or file) from most authors is a 0.9 version and the manuscript after editing is a 1.0 version. (Some authors with good proofreading skills produce 1.0 level manuscripts, and some who need a lot of editing produce versions lower than 0.9.) It is almost impossible for a few people to spot all problems in a few readings of any large document. This is especially true for factual/historical errors, but can even apply to grammar. A small group of people (an author and one or a few editors) simply won’t have as much knowledge as the eventual pool of readers of a published book. Readers with specialized knowledge will spot science errors, title errors, geography errors, climate errors, botany errors, horsemanship errors, etc. If publishers provided an easy feedback mechanism their books could benefit from the knowledge of readers. Then the publishers could release version 1.1 (or 2.0 if there are a lot of changes) of the book. This should be relatively easy with ebooks. In fact, I have actually seen version numbers in some ebooks, and I sometimes get automatic updates of previously downloaded Kindle ebooks. It even happens occasionally with print publishing. Samuel R. Delany’s Dhalgren and an early Eloisa James book come to mind (if I’m recalling correctly). I seem to recall reading years ago that multiple rounds of corrections to Dhalgren wiped out profits from the paperback printings, but that could be an urban legend.
The open source model for software provides a system for having many people examine and improve programs, and the wiki system provides something similar for online non-fiction text, but it would be hard to adapt those models for fiction, which is mostly written by individual authors.
I have been sending the errata that I notice to the editors of the Grantville Gazette ever since I was given contact info. I also got thanked for errata I sent to the authors of a free story on an author web site, so I know that there are actually authors and publishers who will correct texts after a release.
I know of at least one source of ebooks that already has a system in place for readers to submit errata: Project Gutenberg. They welcome corrections, and their FAQ page describes how to submit them.
I have also seen a request for info about typos in some books from Wildside Press, with a web site and email address.
Written works vary widely in error levels, from totally clean to almost unreadable. This has been true for many decades and will probably remain true for decades to come. We can deplore or ignore or discuss or study or correct errors, but I don’t think we can expect to stop seeing them.