Moar Verbs

« previous post | next post »

A couple of days ago, Geoff Pullum noted that William Zinsser's On Writing Well echoes the Strunkish advice that "Most adverbs are unnecessary" and "Most adjectives are also unnecessary" ("Awful book, so I bought it", 3/21/2015). I share Geoff's skepticism about this anti-modifier animus, and indeed about all writing advice based on parts of speech.

But it occurred to me to wonder whether (various types of) good and bad writing actually do tend to differ in how much they use various parts of speech — and in particular, whether there's any evidence that bad (or at least less accessible) writing tends to use more adjectives and adverbs. Given how pervasive part-of-speech writing advice is, I decided to waste an hour exploring the question empirically.

The results are a bit surprising.  At least in this small and conceptually-limited pilot exploration, I found that writing regarded as bad (and perhaps also certain kinds of technical writing) tends to have more adjectives but fewer adverbs, and more nouns but fewer verbs.

The "more nouns and fewer verbs" effect seems to be especially strong — but I've never seen any writing guide that tells us to "write with adverbs and verbs, not adjectives and nouns".

In order to count things, I used the Stanford POS Tagger, which outputs tags from the Penn Treebank tagset, using the "wsj-0-18-bidirectional-distsim.tagger" models. I counted JJ, JJR, & JJS as "adjectives"; RB, RBR, & RBS as "adverbs", NN, NNS, NP, & NPS as "nouns"; and VB, VBD, VBG, VBN, VBP, & VBZ as "verbs".

As a baseline, I started with 8 novels by admired authors:

Author Title %Adj %Adv %N %V
Jane Austen Pride and Prejudice
6.6%
8.3%
20.5%
19.99%
Jane Austen Mansfield Park
6.9%
8.3%
20.4%
19.58%
Charles Dickens Bleak House
7.1%
7.3%
21.4%
19.22%
Charles Dickens David Copperfield
6.6%
7.0%
20.2%
19.97%
Mark Twain Tom Sawyer
6.4%
8.0%
18.0%
20.05%
Mark Twain Huckleberry Finn
5.2%
8.5%
18.0%
20.05%
Ernest Hemingway The Sun Also Rises
5.6%
6.9%
21.3%
21.30%
Ernest Hemingway The Old Man and the Sea
6.6%
7.6%
18.2%
19.95%
Margaret Atwood Oryx and Crake
7.2%
6.8%
22.5%
20.30%
Margaret Atwood Lady Oracle
6.5%
6.7%
20.1%
21.19%
Mean   6.46% 7.54% 20.34% 20.19%

The results are pretty consistent, especially the verb percentages, which are amazingly close, between 19.22% and 21.30% for this sample.

Edward Bulwer-Lytton's Paul Clifford, the source of the famous (and allegedly bad) "It was a dark and stormy night" opening, is a bit higher in adjectives and nouns, and a bit lower in adverbs and verbs:

Author Title %Adj %Adv %N %V
Bulwer-Lytton Paul Clifford
8.2%
6.6%
23.7%
17.28%

I next looked at some good non-fiction:

Author Title %Adj %Adv %N %V
Harry Frankfurt On Bullshit
8.0%
8.0%
21.3%
18.60%
Steven Pinker The Source of Bad Writing
8.7%
6.2%
22.6%
19.66%
Geoffrey Pullum [last two LLOG posts]
6.8%
6.1%
23.7%
19.97%
Alan Turing Computing Machinery and Intelligence
8.1%
6.5%
21.8%
18.52%
 Mean    7.92%  6.69%  22.35%  19.19%

Pretty much the same as the good fiction — a tad more adjectives and nouns, and a smidgen fewer adverbs and verbs.

William Zinsser's book goes a bit further away from adverbs and verbs:

Author Title %Adj %Adv %N %V
William Zinsser On Writing Well
7.8%
5.8%
24.9%
18.73%

I then took the 1997 and 1998 winners of  Dennis Dutton's "Bad Writing" contest, the Google top hit for "leveraging our assets" (a phrase that made it to the finals of the 2012 Forbes Jargon Madness tournament), and Janet Yellen's recent report to Congress (which is quite clear and well crafted, in my opinion, but is also highly technical and formal):

Author Title %Adj %Adv %N %V
Judith Butler Further Reflections on Conversations of Our Time
(Winner, 1998 Bad Writing Contest)
8.1%
4.9%
27.0%
14.85%
Frederic Jameson Signatures of the Visible*
(Winner, 1997 Bad Writing Contest)
12.4%
6.7%
25.8%
12.41%
Robert Houbeck Jr. Leveraging our assets
9.7%
5.0%
29.0%
15.90%
Janet Yellen Semiannual Monetary Policy Report to the Congress 11.7% 4.9% 31.8% 14.43%
Mean    10.47%  5.39%  28.40%  14.40%

[*Introduction & Chapter 1]

That sample is definitely higher in adjectives and nouns, and lower in adverbs and verbs.

I suspect that the differences in POS distributions are a symptom, not a cause, and that  attempts to improve writing by using more verbs and adverbs would generally make things worse. But still.

 

 



57 Comments

  1. FM said,

    March 24, 2015 @ 1:14 pm

    A lot of Strunk-and-White-ish advice, though, boils down to "avoid being in a situation where the nouns are doing all the work" (like that phrase in the quotes that I just wrote! :P) That's presumably what advice like "avoid the passive" boils down to, once you translate it from bad-grammar-ese.

  2. Bob Ladd said,

    March 24, 2015 @ 1:14 pm

    An early suggestion that bad writing is characterised by lots of nouns is here in Verbatim. Google "nounspeak" for some more suggestions.

    [(myl) And Gertrude Stein did write that "any one can see that verbs and adverbs are more interesting than nouns and adjectives". Of course she also wrote that "Nouns are the names of things and so nouns are the basis of poetry."

    However, the specific complaint in the Verbatim piece is about long complex nominals like "patient starter package" and "Increased labor market participation rates" — in fact these are relatively rare in the particular types of highly noun-y writing that I tested (like this example).]

  3. Y said,

    March 24, 2015 @ 1:25 pm

    Is it easy to extract statistics for multiple modifiers to the same head? For example, do nouns in purple prose pull longer trains of adjectives behind them?

    [(myl) You could look at POS ngrams; or run a parser over the original text. But I don't really have a good sample of "purple prose", other than Bulwer-Lytton, and my curiosity about this stuff has (at least for the moment) dipped below threshold.]

  4. Y said,

    March 24, 2015 @ 1:27 pm

    BTW, Pullum did not note that the adverb and adjective laden sentence he quoted from Zinser is also in the dread passive voice.

  5. leoboiko said,

    March 24, 2015 @ 1:48 pm

    Poor Kēlen.

  6. jasper said,

    March 24, 2015 @ 2:22 pm

    The real clincher is the proportion of verbs that are finite (lower=better unless there are technical factors pulling in other direction).
    Of course technical factors are key!

  7. David L said,

    March 24, 2015 @ 2:58 pm

    I have some experience of bureaucratic writing, which (for reasons I can't explain) tends to shun phrases like "respond quickly" in favor of wordier equivalents like "respond in a timely manner." So instead of one adverb you get one adjective and a noun. I wonder whether this sort of circumlocution contributes to the effect you find.

    [(myl) We may find some clues in this random sentence from Judith Butler's article:

    The move from a structuralist account in which capital is understood to structure social relations in relatively homologous ways to a view of hegemony in which power relations are subject to repetition, convergence, and rearticulation brought the question of temporality into the thinking of structure, and marked a shift from a form of Althusserian theory that takes structural totalities as theoretical objects to one in which the insights into the contingent possibility of structure inaugurate a renewed conception of hegemony as bound up with the contingent sites and strategies of the rearticulation of power.

    Note that this sentence contains none of the greater-than-two-nouns complex nominals that Price's "Noun Overuse Phenomenon Article" complains about.]

  8. Sean Bentley said,

    March 24, 2015 @ 3:14 pm

    I read aloud the entire Harry Potter series to my kids, and around the second book I began omitting the adverbs, because virtually every verb had one. So, not only horrible writing but horrible (or no) editing. Rowling didn't trust the reader to intuit the mood or tone of the speaker (since so much of the books consisted of dialogue) or her own descriptions of the scene, perhaps recognizing her inability to simply use evocative nouns and verbs.

  9. Gosse Bouma said,

    March 24, 2015 @ 3:16 pm

    Here is a paper showing that dutch literary novels contain more nouns than chick lit.
    http://www.aclweb.org/anthology/W13-1410
    the authors did not look at verbs, adjectives, and adverbs, but did investigate clause types.

  10. J. W. Brewer said,

    March 24, 2015 @ 3:52 pm

    It seems a pity that the Bad Writing competition used as a resource here lapsed after 1998 – it can't be that they had run out of worthy nominees.

  11. Dominik Lukes (@techczech) said,

    March 24, 2015 @ 3:53 pm

    I'm a bit surprised that any of this is surprising. Nominalisation is a well known feature of certain written genres – bureaucratic and academic prose are particularly good examples of it (I like it vs. I admit to my liking for it. Peter ate it. vs. Peter engaged in the act of eating it.) . It doesn't inevitably make it bad writing – just a particular kind of writing. I suspect you'd find PG Wodehouse's prose chock full of hilarious nouns. A random sample (I'm not sure how it works out quantitatively but his use of nominal phrases like 'with great promptitude' or 'in the immediate neighbourhood of point' gives his prose a particular punch:

    "Mike, who had gone in first as the star bat of
    the side, had been run out with great promptitude off the first ball of
    the innings, which his partner had hit in the immediate neighbourhood
    of point."

    [(myl) Maybe so — but in terms of overall POS statistics, My Man Jeeves is pretty much right in there with the rest of the fiction writers:

    Adj    Adv    Noun    Verb
    5.3% 7.0% 20.4% 21.4%

    ]

  12. Charles said,

    March 24, 2015 @ 3:56 pm

    Imagine telling Beethoven. "Use more eighth notes and fewer open fifths;" or, "invert themes one fourth above the original tonic of the chord." Maybe his publisher should've told Woody Guthrie to expand his vocabulary.

    Having poked fun at the idea of exhorting specific POS usage, I do believe that good writing differs from bad writing and that good guidance on the subject is available. I offer Helen Sword, "Stylish Academic Prose," Steven Pinker, "The Sense of Style," and Thomas and Turner, "Pure and Simple as the Truth".

    Sword, in particular, remarks on the incidence of abstract nouns, lack of examples and positions of nouns and their associated verbs as indicators of the return on reader investment that authors should expect. Her conclusions are based on systematic, careful research that included readers and writers.

  13. Dan said,

    March 24, 2015 @ 4:40 pm

    Presumably neither the authors nor the judges in the Bad Writing Contest are actually native speakers writers of Bad English though…

    It would be interesting to compare against something like bad fan fiction.

  14. James said,

    March 24, 2015 @ 4:59 pm

    Dan, the authors are indeed native speakers.
    This wasn't the Bulwer-Lytton Fiction Contest run by the San Jose State English Department. The Dutton Bad Writing entries were caught in the wild.

  15. D.O. said,

    March 24, 2015 @ 4:59 pm

    Presumably, the direct audience of this writing advice consists of undergrads taking English 00000001 and not intended for turning Frederick Jameson into Jane Austin. The best writing advice, of course, is to read a lot of exemplary prose and try to emulate, but it is hard and time consuming so simple recipes are called to rescue. All this means that the best comparison samples would be good and bad examples of students' writing.

    [(myl) I don't know of any non-anecdotal evidence of good outcomes from telling beginners to eliminate modifiers, and I'm skeptical that this advice would work any better than telling them think carefully about their choice of words in general — or nearly as well as Pinker's advice to think about what their readers do and don't know.

    As for the samples used, I wish there were an available collection of "good" and "bad" student writing…]

  16. Jason said,

    March 24, 2015 @ 7:05 pm

    So late-period Hemingway is indeed a "just the verbs and nouns" man, just like the stereotypes say.

  17. Jason said,

    March 24, 2015 @ 7:07 pm

    Well, actually, it looks the other way round from this sample, doesn't it? Hmmm.

  18. christoll said,

    March 24, 2015 @ 8:19 pm

    I find it fascinating (although completely unsurprising) that both you and the people behind the Dutton Bad Writing contest basically equate "bad writing" with "writing that is difficult to read". You yourself note that Janet Yellen's report is "quite clear and well crafted, in my opinion, but is also highly technical and formal" – a statement that implies that any text that is "highly technical and formal" is ipso facto "bad writing".

    This strikes me as a very English-language conception of bad writing. In German and in French, "bad writing" is most definitely not equated with "writing that is hard to read". If it were, then Thomas Mann and Marcel Proust would be regarded as two of the worst writers of all time. Being difficult to read is absolutely no obstacle to being an admired writer in German and French; indeed, it is more likely to assist your elevation to greatness.

    More seriuosly, surely it would be more interesting to compare "bad" and "good" writing within the same genre of text? For example, compare Margaret Atwood and Hilary Mantel with Dan Brown and E. L. James? (If that isn't deemed unacceptable on grounds of literary snobbery, of course.)

    [(myl) But Proust's style is grammatically quite lucid — if it's hard to read, it's only because long stretches go by where not a lot happens in the external world.

    I mean, "Longtemps, je me suis couché de bonne heure" is much more like "It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife" than it is like "The move from a structuralist account in which capital is understood to structure social relations in relatively homologous ways to a view of hegemony in which power relations are subject to repetition, convergence, and rearticulation brought the question of temporality into the thinking of structure, and marked a shift from a form of Althusserian theory that takes structural totalities as theoretical objects to one in which the insights into the contingent possibility of structure inaugurate a renewed conception of hegemony as bound up with the contingent sites and strategies of the rearticulation of power."

    And the usual English translation of Du Coté de Chez Swann has roughly the same POS distribution as the other novels considered:

    ADJ   ADV   NOUN  VERB
    7.0% 7.0% 20.0% 18.1%

    ]

  19. Chris C. said,

    March 24, 2015 @ 9:28 pm

    I wonder if we're seeing here advice intended for novices to correct common beginner errors, elevated to the level of dicta for all? If you read a lot of amateur or beginner fiction, there's a distinct tendency toward the purple, as if a noun without an adjective or a verb without an adverb is too naked for decency. "Fewer adverbs and adjectives" is one way to suggest to them how to tone it down.

    [(myl) My experience with beginners' writing suggests that excessive modifier usage is low on the list of problems, when it's on the list at all.]

    It's when you extend this to all writing regardless of its overall merits that it becomes a senseless rule.

    Perhaps "Eye of Argon" could serve as a counterexample? http://www-users.cs.york.ac.uk/susan/sf/eyeargon/eyeargon.htm

  20. William Berry said,

    March 24, 2015 @ 10:56 pm

    @christoll:
    "You yourself note that Janet Yellen's report is "quite clear and well crafted, in my opinion, but is also highly technical and formal" – a statement that implies that any text that is "highly technical and formal" is [ipso facto] "bad writing." [emphases added]

    ???

    I'll just put this down to your being a whole hell-of-a-lot better at detecting implication than I am.

  21. Eli Nelson said,

    March 24, 2015 @ 11:21 pm

    @Chris C:
    I know what you mean about misused adjectives/adverbs or other modifiers in badly written fiction. But I think there's a possible issue with simply advising less use of adjectives or adverbs. There might be an observation bias such that modifiers that are badly used are simply more noticeable, while the unobjectionable ones used by better writers don't draw as much attention. So we would have to verify if bad writers really use more modifiers on average; maybe the real issue is bad choice of which ones to use or where to use them.

  22. Eli Nelson said,

    March 24, 2015 @ 11:26 pm

    @William Berry:
    Well, the implication is probably not only drawn from that sentence, but from the context of this article, which groups together "writing regarded as bad (and perhaps also certain kinds of technical writing)". While it doesn't exactly state that highly technical or less accessible writing is "bad writing", this implies that it is a bad or at least undesirable property of writing.

  23. William Berry said,

    March 25, 2015 @ 12:18 am

    @Eli:

    Really?

    Well, I was taking the text (not the con-text) to be the thing, but whatever.

  24. Michael Johnson said,

    March 25, 2015 @ 12:48 am

    @christoll

    That's a really dull straw man. I didn't see Liberman putting Proust on his "bad writing" list.

    He said *certain kinds* of technical writing. Indeed, it's a genuine hypothesis that difficult to read, dry, formal reports like Yellen's are bad writing, and it's some confirmation of that hypothesis that they pattern with the other more stereotypical bad writing.

    It's that straightforward. I'm pretty sure neither the author nor the readers of LL need a lecture on how some difficult to read things are nevertheless well written.

  25. Bob Ladd said,

    March 25, 2015 @ 2:26 am

    @Michael Johnson

    I don't think christoll's argument is just a straw man. I think there there really is a difference between English and a number of other European languages in the extent to which simple, direct prose is highly valued. The difference can be seen within the same genre (technical academic writing) – I know a number of academics who are native speakers of other European languages but do most of their professional writing in English and report difficulties getting back into the style required for academic prose in their native language. Whether this is a matter of POS frequency is another question – as Mark said in his original post, POS frequency may just be a symptom of bad writing rather than being the thing that makes it bad – but I think the differences are real.

  26. Milan said,

    March 25, 2015 @ 5:04 am

    @christoll: Highly admired English language writers, in contrast, always produce clear, lucid and easily accessible prose, as can be seen from the likes of James Joyce and Thomas Pynchon.

    I wonder if the advice about adverbs is actually only concerned with those adverbs derived from adjectives by adding -ly. After all, many a reader (and writer) of those books will have been told in school that "an adjective is a word that ends in -ly." Intuitively, I would say that a high number off those really makes a text appear rather clumsy, though that is probably not the main problem with most of the bad writing out there.

    I also wonder, if there has been any experimental research on the matter. It seems feasible enough: Let the participants perform a task they have never done before and give each group different but equivalent instructions. Then compare which group spend how much time understanding the task, and how well they did in the end. Or give them some informative texts, and compare their performance in test about the text's topic.

  27. David Morris said,

    March 25, 2015 @ 5:32 am

    One of my jobs involves editing to a strict word count, and I have found many times that adjectives and adverbs can be cut from a sentence without injury, in a way that nouns, verbs, determiners and prepositions can't. If something simply has to go – for the word count or for physical space – it's probably an adjective or adverb. (Some prepositional phrases are also expendable.)
    That said, I've got no grouch against adjectives or adverbs in real life. They are a linguistic resource to be used – well, badly or over- – like any other linguistic resource. If I was writing a usage guide, I would warn about the *overuse* of adjectives and adverbs, because, clearly, adjectives and adverbs can be multiplied in a way that nouns and verbs can't.

  28. NW said,

    March 25, 2015 @ 7:09 am

    I think there's something in the idea that bad writing advice is adequate writing advice given to beginners. A young child who writes 'said' all the time is advised to vary their speech tags. By the time they graduate to fan fiction you've got to threaten to break their fingers if they ever use 'queried' or 'questioned' as a speech verb.

    Incidentally, that structuralist rearticulation passage is impressive on the Flesch-Kincaid scale: reading ease of minus 47 (out of a possible 0 to 100) and requiring 43 years of education. Doesn't beat the sentence I once proof-read, from a law firm, that was minus 91 and grade level 55.

  29. Mark Dowson said,

    March 25, 2015 @ 7:44 am

    Dan Brown's Da Vinci Code – on all grounds probably the worst written best-seller of the century – can't be beaten for overuse of adjectives and adverbs. I started by wondering why he didn't have an editor, but ended up fantasizing frantic emails from his editor along the lines of "More adjectives, more adverbs, or you'll never make it to 450 pages".

    [(myl) It's simply not true that Dan Brown uses adjectives and adverbs at an abnormally high rate. The statistics for The Da Vinci Code:

    ADJ   ADV   NOUN   VERB
    6.9% 5.7% 28.3% 19.8%

    This seems to be a classic example of stereotype formation and confirmation bias: "Bad writers use too many modifiers; Dan Brown is a bad writer; Hey look, there's a modifer! Just as I thought!" ]

    Like Sean Bentley I read the Harry Potter series to our kid, but didn't find the adjectives and adverbs too obtrusive, at least in the early books. But I couldn't make it past the first chapter of Rowling's The Casual Vacancy which adds (presumably for grown-ups) incredibly convoluted sentence structures to the flood of adjectives/adverbs.

  30. D-AW said,

    March 25, 2015 @ 9:01 am

    Apropos of Stein's comment about poetry, running the nltk tagger on 100,000 words of modern English (UK) poetry, I get the following breakdown:

    ADJ ADV NOUN VERB
    6.1% 7.4% 29.1% 14.4%

  31. rvman said,

    March 25, 2015 @ 9:17 am

    People like Joyce and Pynchon are the literary equivalent of high-wire acrobats – they do all sorts of things which would be clearly stupid and out-of-bounds for normal people, even normal publishable writers, but get away with it through massive amounts of skill, practice, and artistry. Raising them as counterexamples for any sort of strunk-and-whitening kind of misses the point of their prose, which is, in essence, "the rules don't apply when you rock this much".

    [(myl) I'm puzzled as to why people seem to think that Pynchon's style is complex and difficult. On the contrary, Pynchon's style is generally simple and even colloquial. Thus the opening of Inherent Vice:

    SHE CAME ALONG THE ALLEY AND UP THE BACK STEPS THE WAY she always used to. Doc hadn't seen her for over a year. Nobody had. Back then it was always sandals, bottom half of a flower-print bikini, faded Country Joe & the Fish T-shirt. Tonight she was all in flatland gear, hair a lot shorter than he remembered, looking just like she swore she'd never look.

    Or the opening of V:

    Christmas Eve, 1955, Benny Profane, wearing black Levi's, suede jacket, sneakers and big cowboy hat, happened to pass through Norfolk, Virginia. Given to sentimental impulses, he thought he'd look in on the Sailor's Grave, his old tin can's tavern on East Main Street. He got there by way of the Arcade, at the East Main end of which sat an old street singer with a guitar and an empty Sterno can for donations. Out in the street a chief yeoman was trying to urinate in the gas tank of a '54 Packard Patrician and five or six seamen apprentice were standing around giving encouragement.

    Strunk might disapprove, but it would be on grounds of excessive vulgarity. On a sentence-by-sentence or paragraph-by-paragraph basis, Pynchon is more like a genre writer than like a practitioner of High Style.

    His subject material is sometimes complex; and some of his novels are long, though others are short. But the poster child for acrobatic writing? I don't think so.]

  32. Sam said,

    March 25, 2015 @ 11:11 am

    @Charles

    Actually, in certain styles of composition the sorts of rules you mention are very common. For example, "avoid parallel fifths" is a rule of thumb Beethoven would have followed most of the time. Writing advice like "use fewer nouns" sounds much sillier to my ears..

  33. robert said,

    March 25, 2015 @ 11:30 am

    First off, in many cases of bad writing the individual sentences are perfectly fine. The problem is the way the sentences are put together. In factual writing, they may form to fail a coherent account of the subject. In fiction, they may be assembled into a cliché-strewn story, with cardboard characters.

    Those varieties of bad writing simply won't show up in a sentence-level analysis like this. At most, it can only hope to detect poorly constructed sentences.

    Note, poorly constructed isn't the same as ungrammatical. Garden path sentences are grammatical, but they're poorly constructed because they force the reader to backtrack,.

    More generally, poorly constructed sentences are those that fail in the basic duty of the writer: to convey the intended meaning clearly. This failure can be because of overly convoluted sentence structure, an high percentage of obscure words, or ambiguities which a fair-minded reader can't resolve. (Ambiguities which would only be noticed by someone deliberately hunting for them aren't a problem.)

    An abundance of obscure words is easy enough to spot, just by comparing the text with a vocabulary list, but the other two features are harder for a simple measure to identify. It's certainly conceivable that they might lead to notably different frequencies for the various word classes, but it's not something I'd expect.

  34. GH said,

    March 25, 2015 @ 12:22 pm

    @myl

    "On a sentence-by-sentence or paragraph-by-paragraph basis, Pynchon is more like a genre writer than like a practitioner of High Style."

    Is he?

    They have begun to move. They pass in line, out of the main station, out of downtown, and begin pushing into older and more desolate parts of the city. Is this the way out? Faces turn to the windows, but no one dares ask, not out loud. Rain comes down. No, this is not a disentanglement from, but a progressive knotting into–they go in under archways, secret entrances of rotted concrete that only looked like loops of an underpass . . . certain trestles of blackened wood have moved slowly by overhead, and the smells begun of coal from days far to the past, smells of naphtha winters, of Sundays when no traffic came through, of the coral-like and mysteriously vital growth, around the blind curves and out the lonely spurs, a sour smell of rolling-stock absence, of maturing rust, developing through those emptying days brilliant and deep, especially at dawn, with blue shadows to seal its passage, to try to bring events to Absolute Zero . . . and it is poorer the deeper they go . . . ruinous secret cities of poor, places whose names he has never heard . . . the walls break down, the roofs get fewer and so do the chances for light. The road, which ought to be opening out into a broader highway, instead has been getting narrower, more broken, cornering tighter and tighter until all at once, much too soon, they are under the final arch brakes grab and spring terribly. It is a judgment from which there is no appeal.

    I find passages such as these difficult as a matter of literary style, if not perhaps in terms of grammatical structure. Though he also has a thing for long, rambling sentences, which can be exhausting:

    Bloat is one of the co-tenants of the place, a maisonette erected last century, not far from the Chelsea Embankment, by Corydon Throsp, an acquaintance of the Rossettis' who wore hair smocks and liked to cultivate pharmaceutical plants up on the roof (a tradition young Osbie Feel has lately revived), a few of them hardy enough to survive fogs and frosts, but most returning, as fragments of peculiar alkaloids, to rooftop earth, along with manure from a trio of prize Wessex Saddleback sows quartered there by Throsp's successor, and dead leaves off many decorative trees transplanted to the roof by later tenants, and the odd unstomachable meal thrown or vomited there by this or that sensitive epicurean-all got scumbled together, eventually, by the knives of the seasons, to an impasto, feet thick, of unbelievable black topsoil in which anything could grow, not the least being bananas.

  35. Jerry Friedman said,

    March 25, 2015 @ 12:41 pm

    In keeping with what Dominck Lukes said, Wilson Follett (or Jacques Barzun) discussed the "noun plague", referring specifically to the overuse of abstract nouns that could be replaced by verbs (or saying specifically that many writers overuse abstract nouns that could be replaced by verbs). If I knew who repeats, converges, and rearticulates power relationships, I might have a go at Judith Butler's sentence.

    When I was looking up condemnations of adverbs after reading GKP's post (carefully), I found that most of them referred specifically to -ly adverbs. That might be an interesting test.

    As an sf reader, I must point out that Oryx and Crake is genre writing and might appropriately be compared to, I hate to mention names, Stephen Baxter or Gentry Lee, I'm not sure whether Dan Brown is in the same genre.

  36. Jerry Friedman said,

    March 25, 2015 @ 12:50 pm

    This random page of Gravity's Rainbow might count as acrobatics from Pynchon: sentences connected by commas and by ellipses, dialogue without any previous hint of the speaker, "bastard enumeration" as Fowler called it, onomatopoeia is thrown in without grammatical relation to the rest of the sentence, copulas deleted…

  37. Jerry Friedman said,

    March 25, 2015 @ 1:32 pm

    Dominik Lukes: I'm sorry about all the misspellings.

    By the way, "Most adverbs are unnecessary" doesn't necessarily mean good writers use fewer adverbs than bad writers. It just means good writers use necessary adverbs and bad writers use unnecessary ones.

    Also, I hadn't seen GH's comment when I linked to a very unstrunkian page of Pynchon. (And I meant "onomatopoeia thrown in ptoov! without grammatical relation…")

  38. Dick Hudson said,

    March 25, 2015 @ 2:06 pm

    I did some research on these questions (downloadable from http://dickhudson.com/papers/) reported in "About 37% of word-tokens are nouns" (Language 70, 1994), where I used other people's data to show that the percentage of nouns is remarkably consistent, especially if you collapse nouns proper with pronouns; it rises gently with style level, in a way that I now think can be explained in terms of the increasingly broad vocabulary.

    Re adjectives and adverbs, one corpus gave a constant 13% if you collapse them into 'adwords', but the balance favoured adjectives in 'informational' prose and adverbs in 'imaginative' prose. In children's speech, there was a clear progression, with more adwords as they got older from 6-12 (going up from 14% to 17%). So maybe what children have to learn in writing is actually to use fewer adwords than when speaking? But this research says nothing about quality.

    But these differences are surely tiny, and wouldn't make a significant difference to the perceived quality of a particular passage.

  39. Mark S said,

    March 26, 2015 @ 8:52 pm

    A friend of mine wrote a non-fiction book and tried to get it published. One publisher told him to reduce the number of adverbs (among other requirements). Even though he went to some trouble to comply, they still refused it.

    Coincidentally, about the same time, I read about a computer program called "Hemingway" that makes style recommendations on written text. It was also very down on adverbs. Ironically, it had many complaints about a passage from one of Hemingway's novels.

  40. Jason Eisner said,

    March 26, 2015 @ 10:45 pm

    @robert identifies "overly convoluted sentence structure" as one of several writing sins.

    I'd love to see measurements of syntactic processing complexity on each corpus. As a start, it wouldn't be too hard to estimate the distribution of syntactic dependency lengths.

    A long dependency means that two syntactically related words are separated by a lot of intervening material. The intervening material can be distracting and can make it hard to see that the two words are related. I'd expect to find some long dependencies in the good 19th-century texts (Austin, Dickens, Twain) — more than in the good 20th-century texts. But perhaps the bad texts have even more.

    Judith Butler's sentence seems to have many long dependencies. In this case, that's because it embeds lengthy descriptions of several concepts:

    "a structuralist account in which capital is understood to structure social relations in relatively homologous ways"
    "a view of hegemony in which power relations are subject to repetition, convergence, and rearticulation"
    "a form of Althusserian theory that takes structural totalities as theoretical objects"
    "one in which the insights into the contingent possibility of structure inaugurate a renewed conception of hegemony as bound up with the contingent sites and strategies of the rearticulation of power"

    Butler probably found it easy to reel out these long descriptions. They're just pointers back to concepts she's already introduced. She knows which concept she wants to reference. She may even think it's helpful to spell it out again in case we've forgotten the details.

    Unfortunately, Butler's sentence is trying to state relationships among these concepts, and we're distracted by trying to identify the concepts at the same time. Giving short names to the concepts would have made it much easier to identify the relationships, as they now correspond to shorter dependencies:

    The move from the Static View to the Dynamic View brought the question of temporality into the thinking of structure, and marked a shift from Social Structure Theory to Social Process Theory.

  41. Peter Taylor said,

    March 27, 2015 @ 3:15 am

    The results are pretty consistent, especially the verb percentages, which are amazingly close, between 19.22% and 20.30% for this sample.

    should say "between 19.22% and 21.30%".

  42. christoll said,

    March 27, 2015 @ 4:21 am

    @ Michael Johnson:
    I'm by no means suggesting that all technical writing is difficult to read. My point is that Mark's post gives the impression that "bad" and "good" in writing come down to a question of how difficult the writing is to read, which is an entirely arbitrary way of defining it. Why not castigate unimaginative writing? Or writing that is sloppy, awkward, or clumsy?

    @Milan:
    James Joyce is an interesting exception in the English literary canon and one which did occur to me. I will content myself with noting that his work is inordinately popular among Germans, who like to read it in the original English, which if you ask me takes a love of difficult reading to the level of certifiable masochism.
    In his comment on my comment above, Mark seems to have clarified his assertion of "bad writing = hard-to-read writing" to mean "bad writing = writing which is not grammatically lucid." But that definition is no less arbitrary, and I would argue that Proust's French frequently leaves a great deal to be desired in the lucidity department. Admittedly the last three volumes of A la recherche du temps perdu were never edited as much as Proust would have liked, but even in the earlier volumes there are things like this:

    Or, précisément, que dans une phase aussi peu rapide, les visages non plus emportés dans un tourbillon, mais calmes et distincts, me parussent encore beaux, cela m'empêchait de croire, comme je l'avais fait si souvent quand m'emportait la voiture de Mme de Villeparisis, que, de plus près, si je me fusse arrêté un instant, tels détails, une peau grêlée, un défaut dans les ailes du nez, un regard benêt, la grimace du sourire, une vilaine taille, eussent remplacé dans le visage et dans le corps de la femme ceux que j'avais sans doute imaginés ; car il avait suffi d'une jolie ligne de corps, d'un teint frais entrevu, pour que de très bonne foi j'y eusse ajouté quelque ravissante épaule, quelque regard délicieux dont je portais toujours en moi le souvenir ou l'idée préconçue, ces déchiffrages rapides d'un être qu'on voit à la volée nous exposant ainsi aux mêmes erreurs que ces lectures trop rapides où, sur une seule syllabe et sans prendre le temps d'identifier les autres, on met à la place du mot qui est écrit un tout différent que nous fournit notre mémoire.
    (from vol II, A l'ombre des jeunes filles en fleurs)

    Okay, it's not quite Judith Butler, but it's far from easy to read. Proust's writing, like Thomas Mann's, is nearly always beautiful and the sentence above is no exception, in my opinion. But it is also hard to plough through. The two qualities are not necessarily mutually exclusive, yet most native English speakers seem naturally inclined to think that they are. Perhaps it would help if they read more Joyce. (Or perhaps not.)

  43. Marek said,

    March 27, 2015 @ 4:38 am

    Bear in mind that POS taggers usually assign most unknown words to the "noun" category. So the high proportion of nouns for 'bad writing' may be in some part due to a large number of unrecognized words, e.g. due to typos.

    And since no-one else pointed it out:
    "On Writing Well" contains an adverb in the very title.

  44. David Eddyshaw said,

    March 27, 2015 @ 7:28 am

    The noun/verb ratio certainly varies markedly between languages.
    Iroquois languages for example characteristically use verbs where more familiar languages have nouns, and furthermore a lot of what are syntactically nouns are morphologically verbs too.

    Languages vary too in how much of the excitement (as it were) is expressed through verb choices rather than nouns, quite apart from the overall frequencies. In the far-off days when people actually learnt to compose Latin prose, a frequent mantra from the teachers was "strengthen your verbs."

    Quite a few languages have only a limited closed class of verbs, with broad generic meanings which can be narrowed down by association with more noun-like elements. (Warlpiri is like this.)

  45. Lisa Cox said,

    March 27, 2015 @ 7:49 am

    If you're looking for a writing guide, Richard Lanham's Revising Prose teaches students to focus on verbs. But for sheer pleasure, go straight to Lanham's Style: An Anti-Textbook.

  46. Bloix said,

    March 27, 2015 @ 1:50 pm

    So what is going here? FM argues that the advice-givers are on to something but don't know enough to express it. Sean Bentley and Mark Dowson give a concrete example of a popular writer who over-uses adverbs in a specific context. So maybe there is something that can be said to distinguish good and bad uses of adverbs and adjectives, but the supposed experts don't know what it is.

    The way that this issue is approached on this blog is that Strunk/White/Zinsser etc. are pompous fools who make shit up and that there's no way to make an evidence-based distinction between good and bad writing.

    But White, at least, was one of the great essayists of the 20th century, and one of its most beloved children's authors. He knew a lot more about how to write well than JK Rowling does, although he couldn't explain what he knew. Most of what he thought about how he did what he did was wrong, yet still, he did it. Is there nothing that science can tell us about what White could do, and Rowling can't?

    When I was a young lad we used to tell Tom Swifties – remember those? "The fire is out of control!" Tom said hotly. "I replaced the bulb," Tom said brightly. The joke was that the style of the Tom Swift series over-used adverbs to characterize dialogue – the same fault that Sean Bentley attributes to JK Rowling. This is a simple and common example of bad writing. Is there nothing that science can tell us about the use of adverbs to modify "said" in written dialogue?

    Often, when I pick up a paperback thriller or genre novel, I find myself mentally editing to cut through the thickets of needless verbiage. After a while, I just stop reading. Any competent editor could have made the prose serviceable. Is it possible that science cannot explain what thousands of editors know how to do?

  47. Andrew (not the same one) said,

    March 27, 2015 @ 3:07 pm

    The truth-value of statements like 'most adjectives are unnecessary' clearly depends on context. Nobody can have intended to say 'no matter how few adjectives you use, most of them are unnecessary'. Rather, presumably, they meant that most of the adjectives they found, in the kind of writing they typically read, were unnecessary. I can totally believe that in some contexts this is true. There is a style that makes excessive use of adjectives, and it is sensible advice to people tempted to write in it that they should use fewer. Likewise with adverbs, and indeed with the passive voice; I have seen prose which cast everything in the passive voice to make sentences longer and more complex, and this is surely a bad thing. Of course, the advice to use only necessary adjectives and adverbs, and to use the active voice where this can be done without damage to sense, is imprecise; it doesn't tell you when these things are necessary. Finding a more precise way to explain it is, as Bloix says, a problem we don't seem to have solved yet.

    I think part of the problem may arise from the fact that a lot of the advice in the advice books, understood in the way that it was first intended, has been taken, so that people are no longer so widely writing the sort of prose it was intended to apply to: but the rules go on being repeated without their original context, and are applied to adjectives, adverbs and uses of the passive voice which are in fact necessary.

  48. Jason Eisner said,

    March 27, 2015 @ 3:54 pm

    On Mark's post versus Zinsser's book: Natural language processing (NLP) made a lot of progress when we started extracting grammatical knowledge from data, rather than asking experts to write it down for us. We should do the same for stylistic knowledge. The experts do provide great insights into what kinds of patterns might be relevant to grammar or style … but their informed hypotheses about what to look for are not a substitute for actual looking.

    @Bloix, you ask, "Is it possible that science cannot explain what thousands of editors know how to do?" Certainly what makes "good" writing (or good music or good art) is a legitimate and exciting scientific question.

    Improving writing is also a killer app for NLP. It seems within our reach to diagnose or fix some kinds of problems with sentences, though fixing others would require dauntingly deep understanding of what the bad writer was trying to say.

    There's been certainly some work in this direction, although I don't know how mature it is. Try searching the ACL Anthology for terms such as "readability," "simplification," "writing assistant." There have even been some workshops on topics like Computational Linguistics and Writing and Predicting and Improving Text Readability for Target Reader Populations.

    We may be limited by not having the right datasets yet. It's tricky to collect sufficiently large datasets simply by asking readers to rate sentences. The quality of a sentence is a function of the text and the reader. And both sentences and readers vary a great deal. So you need quite a lot of (sentence, reader) pairs to cover the space. Furthermore, what you really want is (sentence 1, sentence 2, reader) triples — i.e., rather than asking readers to rate sentences of varying quality on some absolute scale (which is an unnatural task), you'll get more robust judgments if you ask readers to judge whether a certain edit improves a sentence. But obtaining these triples requires either collecting people's edits or having an existing mechanism to propose plausible edits.

  49. Chris C. said,

    March 27, 2015 @ 4:55 pm

    My experience with beginners' writing suggests that excessive modifier usage is low on the list of problems, when it's on the list at all.

    When nearly every noun or verb seems unable to stand without a modifier, telling them to cut down on modifier usage is at least someplace to start. Once the prose is a lighter shade of purple you can start to talk about what else needs attention.

  50. chris said,

    March 27, 2015 @ 7:33 pm

    The way that this issue is approached on this blog is that Strunk/White/Zinsser etc. are pompous fools who make shit up and that there's no way to make an evidence-based distinction between good and bad writing.

    No way simple enough to be condensed into a tweet, certainly. Or coded into a computer program in the present state of the art.

    Distinguishing between good and bad writing is complicated and difficult — and that's assuming that there's a "real" definition of good and bad writing and not just someone's (or some committee's) subjective opinions, in the first place. Joyce is a good example here, actually — there are a number of people who can't stand him, and even some who think the art of writing was set back by his example.

    I mean, aside from "writing that leaves the reader unable to determine what was meant", or something like that, because that clearly doesn't apply to Brown or Rowling. Although it sometimes does to Joyce.

  51. Shoe said,

    March 28, 2015 @ 2:17 am

    Zinsser should probably have hedged his bold claim that most adverbs and adjectives are unnecessary. But what he actually has to say on the subject seems like good advice to me:

    Most adverbs are unnecessary. You will clutter your sentence and annoy your reader if you choose a verb that has a specific meaning and then add an adverb that carries the same meaning. Don't tell us that the radio blared loudly; "blare" connotes loudness. Don't write that someone clenched his teeth tightly; there's no other way to clench teeth. Again and again in careless writing, strong verbs are weakened by redundant adverbs.

    Most adverbs are unnecessary. Like adverbs, they are sprinkled into sentences by writers who don't stop to think that the concept is already in the noun. This kind of prose is littered with precipitous cliffs and with lacy spiderwebs, or with adjectives denoting the color of an object whose color is well known: yellow daffodils and brownish dirt.

    Sean Bentley upthread remarks on skipping the adverbs when reading aloud from Harry Potter. Following is an extract from a Guardian article about Rowling's bad prose that addresses this very issue.

    Here, from page 324 of The Order of the Phoenix, to give you a typical example, are six consecutive descriptions of the way people speak. "…said Snape maliciously," "… said Harry furiously", " … he said glumly", "… said Hermione severely", "… said Ron indignantly", " … said Hermione loftily". Do I need to explain why that is such second-rate writing?

    http://www.theguardian.com/books/booksblog/2007/jul/17/harrypottersbigconisthep

    [(myl) In fact, it seems to me, it's possible to clench one's teeth with quite a wide range of degrees of force. As evidence, this passage from Irwin Becker, "Comprehensive Occlusal Concepts in Clinical Practice":

    In figure M, the patient was asked to clench her teeth as hard as she could. Note the involvement of the neck muscles in the clenching process. Be aware that especially women tend to clench their teeth several times harder while sleeping than while awake.

    So Zinsser is wrong physiologically as well as stylistically.]

  52. J:imbino said,

    March 28, 2015 @ 11:18 am

    While Hemingway is cited as "admired," I've found his fiction to consist mainly of strings of nouns and adjectives joined by "and," and to disfavor dependent clauses and direct speech..

    An example of his awful writing along with my corrections:

    If a writer of prose knows enough of what he is writing about he may omit things that he knows and the reader, if the writer is writing truly enough, will have a feeling of those things as strongly as though the writer had stated them. The dignity of movement of an ice-berg is due to only one-eighth of it being above water. A writer who omits things because he does not know them only makes hollow places in his writing.
    —Ernest Hemingway in Death in the Afternoon [4]

    While a writer of prose who knows enough of what he is writing about may omit things he knows, if he is writing truly enough, the reader will have a feeling of those things as strongly as if the writer had stated them. The dignity of movement of an ice-berg is due to the fact that only one-eighth of it is above water. A writer who omits things because he does not know them only makes hollow places in his writing.

  53. Emily said,

    March 28, 2015 @ 1:15 pm

    Regarding Harry Potter and dialogue adverbs, the fan fiction "Hogwarts School of Prayer and Miracles" seems to be parodying Rowling's style. Hardly anyone ever "says" anything– instead they "query innocently" and "utter empathetically" and "lie dishonestly."

  54. Andrew (not the same one) said,

    March 28, 2015 @ 4:36 pm

    Emily: It could be a parody of Rowling, but 'never say said' is in any case a quite widespread maxim in fanfic circles (though not all agree).

  55. haamu said,

    March 29, 2015 @ 4:25 pm

    @myl: Re "[I]t's possible to clench one's teeth with quite a wide range of degrees of force…. So Zinsser is wrong physiologically as well as stylistically."

    Zinsser's book may be full of nonsense, but let's at least have the nuance to give him this one point. His assertion is not physiological. It's about the meaning and usage of words.

    I doubt he'd deny that there's a clenching gradient. I'm sure he'd have no problem, for instance, with Tennyson's recognition of it: "[She] clench'd her fingers till they bit the palm."

    Zinsser would simply say that all points on the clenching spectrum are tight. The endpoints are tight and excruciatingly tight. If you're below the former, you're simply not clenching. You might be closing or holding or flapping your jaw, or doing some other sort of orthodontic operation, but you aren't clenching.

    I doubt, too, that he'd object to tightly if it helped to add more information: "She clenched her jaw so tightly the cords in her neck stood out." But merely adding tightly to clench, without more, is pointless. Every sense in OED's clench, v.1 entry has some element of tight, firm, final, or secure in it. If all you're doing is redundantly restating the default sense of the verb, you're asking your reader to parse more, for zero payoff. (In my view, too, you're helping to bleach the verb, by implying it doesn't convey this sense all by itself.)

    If we could boil away all the twaddle you folks have exposed in Strunk & White, and now Zinsser, over the years, I think we'd still find something interesting at the bottom of the pot: Add words if they add value, and don't if they don't. To me, that seems like pretty solid stylistic advice. But value is tricky to define. It's easy to rack up examples and counterexamples, in both good and bad writing, if all we're doing is counting instances, but mere counting doesn't clench the argument, because arguments about style are axiological arguments. That's a problem for science. There's ample evidence that many scientists, even great ones, don't really get axiology or understand why science doesn't invalidate it as a pursuit.

    As a layman, I've learned a lot by lurking on this blog for the last 10 years or so. I totally get that when Prescriptivist P grounds his argument in (1) it isn't or hasn't been done, or (2) "good writers" don't do it, then he can and should be thoroughly debunked with solid data.

    The problem comes when P is really saying (3) it simply shouldn't be done. That might be an aggravating statement, but it's tough to attack with data. It isn't even enough to show with data that P doesn't follow his own rule. That's a well-worn tactic here, but all it establishes is that P is sloppy or maybe a hypocrite, not that his advice is actually bad.

    Any data-driven refutations, if they're possible at all, need to be directed at the reasons for the shouldn't. If P is saying that readers will be misled or won't comprehend, then we can't just count words; we need to ask readers if they've really been misled or haven't comprehended. If P is saying that a given phrasing is simply better or more beautiful than another, then maybe we can try to measure people's aesthetic reactions, but, honestly, P hasn't given us a lot of measurable ways to prove him "wrong" – or at least he's given us a really interesting challenge to find those ways by figuring out what he really means (whether he knows it or not).

    Again, this isn't about whether Zinsser's book, as a whole, is bad. It's about whether we dismiss everything he says. If there are nuggets of truth, they might eventually lead us to some quantitative methods for measuring "good" style that are worthy of investigation. As some of the commenters to this thread suggest, that could be a fascinating pursuit.

    [(myl) You've clearly given clenching 100 times more thought than Zinsser did. But I think your conclusion about all clenching being "tight" is wrong, all the same: there's nothing incoherent about massage instructions advising us to "Apply 5 to 7 deep circular friction strokes to the extensor muscles of the forearm from wrist to elbow using the heel of your hand or a loosely clenched fist." Nor should we censure Virginia Woolf for writing "She noticed the lips just parted, the fingers loosely clenched, the whole attitude of rapt contemplation, which fell like a veil between them."

    But at least you took the time to develop a theory, even if it's a wrong one. Zinnser was just spinning out a cliché about redundant modifiers, without any evidence that he gave more than a few seconds of thought to his examples.]

  56. Milan said,

    March 31, 2015 @ 7:47 am

    @christoll: I am sorry for my my somewhat caustic retort to your comment. I (wrongly, I can say in hindsight) sensed a kind of German arrogance towards anglophone culture in it. German myself, and a professed Anglophile, I am probably a bit too touchy in that area.

  57. Pacing your Prose Part 1: Using Description  | bardicblogger said,

    March 31, 2015 @ 4:01 pm

    […] Moar Verbs […]

RSS feed for comments on this post