How many words make a sentence?

February 2016

Earnsy Liu, TechCommNZ member and GDID student, looks for evidence (not just opinions) to help you manage the daily conundrums we face in our profession. If you have a question for Earnsy to tackle, please email our Comms Coordinator.

How many words should we have in a sentence? A common plain English guideline says an average of 15–20 words (Cutts, 2009; Plain English Campaign, 2015; Plain Language Association InterNational, 2015). Such sentences are clearer, less intimidating, and easier to scan (Cutts, 2009; Vincent, 2014).

Is there any current evidence that the 15–20 word guideline applies? Why is sentence length important anyway, and just how important is it? Should we count words or chunks of information? And how can we manage sentence length?

Does the 15–20 word guideline apply?

There seems to be only one study within the last ten years on sentence length. Despite finding several references to the article, I couldn’t locate it, so here’s the news second-hand. According to Wylie (2009), an American Press Institute (API) study found readers understood more when sentences were shorter:

  • 100 per cent of information when sentences averaged 8 words or fewer
  • over 90 per cent of information when sentences averaged 14 words
  • less than 10 per cent of information when sentences averaged 43 words.

This suggests the 15–20 word guideline should either stand, or be lowered to 14.

Having recent research is important because things change:

Source Period or person Words per sentence
Carroll (2013) Elizabethan times (excl. Shakespeare) 60
150 years ago 30
2013 20
Moore (2011) Chaucer 49
Dickens 20
JK Rowling 12
Some contemporary science writers 15-20
Sherman, in DuBay (2004) Pre-Elizabethan times 50
Elizabethan times 45
Victorian times 29
1890 23

When I contacted API to try and access their research, the correspondent mentioned the importance of recent evidence (personal communication, December 12, 2015):

… an old study of how people read sentences in print newspapers may no longer hold much real meaning about how people read today on different devices and screens, and with the overall evolution of texting and communication styles.

Crikey. I was asking about a study between 2006 and 2009. If that’s old – and possibly no longer valid – then we’re in trouble, because there’s no obvious evidence. Or perhaps the guideline is accepted enough that researchers don’t feel the need to re-visit the topic (C. Staudt, personal communication, January 19, 2016).

There is current research on readability (for example, on patient consent forms and online privacy forms) but readability is more than sentence length. Readability measures examine other variables too, usually the number of syllables (for example, Flesch Reading Ease, Gunning’s FOG Index, SMOG). There’s also information on literacy, but that’s different again, often covering reading and writing in the workplace, or health or financial literacy.

Why is sentence length important?

Examining the lengths of existing sentences merely tells us what’s been written. Does sentence length really matter for readers? Rudolf Flesch (n.d.), the creator of the Flesch Reading Ease measure, explained why long sentences are harder:

  • There’s more to remember. When reading, readers digest the sentence as they go, up to what they’ve read. Only when they finish the sentence do they digest it as a whole. So longer sentences mean having to remember more along the way.
  • Longer sentences are more complex grammatically.

Technical communication expert Cindy Staudt echoed his second point: ‘… you can’t write a long, winding, prepositional-phrase-heavy sentence if you’re minding how many words you’re using’ (personal communication, January 19, 2016).

Plain English champion Martin Cutts recommends writing for 13-year-olds (2009), which equates to a Flesch Reading Ease score of 60–70. The table below shows that when Flesch related the editorial content of popular magazines to his Reading Ease Score, articles scoring 60–70 had about 17 words per sentence. This fits with the 15-20 word guideline.

Flesch’s magazine review (in DuBay, 2004, p. 22)

Just to be clear: the table shows his findings. He is not recommending that academic journals strive for scores of 30-50, or that scientific articles have 30 words per sentence.

How strong is the link?

To find out what makes books readable, William Gray and Bernice Leary tested comprehension in 800 adults, using 48 passages of 100 words each. One of their findings was a -0.52 correlation between sentence length and reading difficulty (in DuBay, 2004), meaning longer sentences are harder to read. No surprises there.

But the size of the correlation bothers me. After all, a correlation of 1 shows that two variables are perfectly and positively related to each other: one goes up, the other goes up. On the other hand, a negative correlation of -1 shows they are perfectly related too, but in opposite ways: one goes up, the other goes down. And a 0 correlation shows no relationship whatsoever.

So doesn’t a correlation of -0.52 – halfway between a perfect negative correlation (-1) and no correlation (0) – mean the relationship is moderate at best? If so, are we over emphasizing sentence length? Note though, that the study was published in 1935 (a lifetime ago!), so who knows how what the correlation is now?

Should we count words or chunks?

Geoffrey Marnell (2014) has proposed measuring sentence length in chunks of information (units of meaning) rather than in words, arguing that we can only remember so much information. He describes a brief experiment he conducts in his classes. He also refers to research demonstrating that most people remember four to seven chunks of information in the short term.

His proposal appeals to me, but I have reservations. First, other than his experiments, there does not seem to be any research relating information chunks to sentence length. Second, how easy and practical would it be to count chunks instead of words? Will we all understand and count chunks in the same way? We don’t even count words the same way: different word counting tools give different results.

Judy Knighton (2015) responded to Marnell’s proposal by vigorously defending sentence-length research. But she refers to readability, which, as she points out, looks at more variables than just sentence length. Do they muddle the picture? But she’s right that there is plenty of research. What’s missing is recent research.

How can we manage sentence length?

Regardless of the precise number of words to aim for – 14, or 15-20 – and whether we count words or chunks, these techniques may help manage average sentence lengths:

  • Write in plain English. Minimise the use of the passive voice and nominalizations. For example, go for ‘we decided’ instead of ‘a decision was made’.
  • Be concise. For example, choose ‘during’ over ‘for the duration of’, and ‘because’ over ‘in view of the fact that’. Ditch filler words like ‘We are pleased to inform you’ that don’t add value.
  • Shorten long sentences. Split sentences and use connecting words like ‘however’, ‘but’,’ so’, ‘also’, ‘yet’, ‘further’, or ‘and’ (Cutts, 2009; Moore, 2012). Use colons or similar (Moore, 2012).
  • Avoid starting sentences with qualifiers like ‘although’, ‘because’ or ‘since’ (Moore, 2012).
  • Use a list (Cutts, 2009).
  • Keep an eye on the maximum sentence length. Use one of these recommendations as a guide:
    • Cutts (2009) - no maximum, but don’t exceed 40 words regularly
    • GOV.UK (Vincent, 2014) - 25 words
    • Plain Language Network (2015) - 30-35 words
  • Vary sentence lengths (Carroll, 2013). Newell (2014) elaborates:

Long sentences work best when your reader’s interest is piqued, as these sentences have the advantage of flow but require more focus on the part of the reader. Meanwhile, short sentences grab your reader’s attention … but too many consecutive short sentences are jarring. Mixing these two sentence types keeps your audience engaged throughout the paragraph . . . . ensure that each sentence is an appropriate length for the idea it expresses.

  • Use a word counting tool, such as:
    • Word’s readability statistics feature. Enable the readability stats (Options ? Proofing ? Show readability statistics) to see the average number of words per sentence, although that’s only shown after a spell check.
    • Count Wordsworth. It offers dozens of metrics over several screens, although finding the one you’re after is a challenge.
    • Word Counter. This shows the basics like the number of words and sentences, as well as words per sentence (you need to specify this under options).

    The tools give different results, possibly because of different ways of counting, so stick to one.

Finally, remember the guideline is about average length; it is not a target for every sentence.

Acknowledgements

Many thanks to Emma Harding for her invaluable comments.

References

Carroll, K. (May 1, 2013). Vary your sentence length to make your writing more interesting. Retrieved November 28, 2015, from Content Advantage website: http://ken-carroll.com/2013/05/01/vary-your-sentence-length-to-make-your-writing-more-interesting/

Cutts, M. (2009). Oxford guide to plain English (3rd ed.). New York, NY: Oxford University Press.

DuBay, W. H. (2004). The principles of readability. Retrieved November 23, 2015, from http://www.impact-information.com/impactinfo/readability02.pdf

Flesch, R. (n.d.). How to write plain English. Retrieved November 19, 2015, from University of Canterbury website: http://www.mang.canterbury.ac.nz/writing_guide/writing/flesch.shtml

Knighton, J. (2015, February). In praise of scientific rigour in plain English guidelines. Southern Communicator (34), 11-12. Retrieved from https://www.tcanz.org.nz/Category?Action=View&Category_id=452

Marnell, G. (2014, October). Does length matter? Southern Communicator (33), 9-10. Retrieved from https://www.tcanz.org.nz/Category?Action=View&Category_id=431

Moore, A. (December 22, 2011). The long sentence: A disservice to science in the Internet age. Retrieved November 23, 2015, from Wiley website: http://exchanges.wiley.com/blog/2011/12/22/the-long-sentence-a-disservice-to-science-in-the-internet-age/

Newell, C. (2014). Editing tip: Sentence length. Retrieved November 19, 2015 from American Journal Experts website: https://www.aje.com/en/author-resources/articles/editing-tip-sentence-length

Plain English Campaign (2015). How to write in plain English. Retrieved November 28, 2015, from http://www.plainenglish.co.uk/how-to-write-in-plain-english.html

Plain Language Association InterNational (2015). What is plain language? Retrieved November 28, 2015, from http://plainlanguagenetwork.org/plain-language/what-is-plain-language/

Vincent, S. (2014). Sentence length: Why 25 words is our limit. Retrieved November 19, 2015. from GOV.UK website: https://insidegovuk.blog.gov.uk/2014/08/04/sentence-length-why-25-words-is-our-limit/

Wylie, A. (January 14, 2009). How to make your copy more readable: Make sentences shorter. Retrieved November 19, 2015, from http://comprehension.prsa.org/?p=217