Versus vs. Verses: Results

« previous post | next post »

Following up on "Versing" (6/19/2012)  and "Vers(e|u)s" (6/20/2012), here are the perception-test results from the 56 people who sent me their answers before I posted the answer key.

Some overall statistics follow.

524/1120 answers were "versus" (47%), while 596/1120 answers were "verses".

Overall, 616/1120 answers were correct (55%). This happens to correspond with the percentage of "versus" stimuli (11/20).

The percentage of "versus" answers by stimulus (numbers 1-10 in the first row, 11-20 in the second) was

42.9 32.1 71.4 35.7 35.7 64.3 33.9 35.7 53.6 50.0
57.1 26.8 44.6 35.7 44.6 67.9 76.8 44.6 62.5 19.6

The percentage of correct answers by stimulus was

42.9 67.9 71.4 35.7 35.7 35.7 66.1 64.3 46.4 50.0
42.9 73.2 44.6 35.7 55.4 67.9 76.8 44.6 62.5 80.4

The percent of correct answers by subject was

60 30 85 50 50 55 55 40 95 50
35 60 85 50 50 30 55 50 55 65
60 30 65 50 55 50 65 60 60 45
55 45 65 65 65 50 40 55 55 70
45 55 30 70 35 60 45 55 55 80
45 55 55 60 65 60

The route to successful guessing in this experiment was to ignore the consonants and vowels, and concentrate on the prosody. The two words have very different functions — in these examples, versus is a funny kind of coordinator, while verses is a plural noun — and this translates into different expected patterns of pitch and (especially) duration.

Subject #9, who scored 95%, was DDB, a "33-year-old Canadian", who wrote:

I tried not to use contextual cues, but can’t be sure I didn’t.
I have a feeling I correlated longer-duration speech with ‘verses’.

Subject #9, who scored 85%, was RDK,  from Florida, Georgia, Tennesse, Montana, and New Jersey, who wrote:

I feel like I did use "distributional" cues (to the extent that that's possible with words in isolation), e.g., duration, but I suppose that's a moot point if I scored at or below chance.  I do feel like I distinguish these words in speech, but I'd have to actually look at my own production to determine if that impression is based in fact.

Several other subjects who scored well reported using similar strategies.

Overall, it's clear that there's some signal here. The binomial estimate of the 95% confidence interval for the true probability of success in this experiment is 55% ± 3%, so that respondents are certainly doing better than 50% overall. (On the other hand, you could score 55% by guessing "versus" all the time — though that is clearly not the strategy people followed — so depending on what you take the null hypothesis to be, subjects either did or didn't do better than chance…)

Several people complained about the stimuli being "truncated" or "cut off" — that comes with the territory here, since words as pronounced in context are very different from words pronounced in isolation. And since the usage of these two words in these examples is so different, including the context would generally give the answer away.

The original question was whether it's plausible that language learners are justified in mis-analyzing things like

Tonight's game: Philadelphia versus Toronto


Tonight's game: Philadelphia verses Toronto

where "verses" is construed as the third-person singular present form of a verb "to verse", meaning "to oppose" or "to play against". Tony from Toronto observed that

I've never heard of this usage before and am baffled by it since versus and verses are quite distinct when written and when pronounced.

I felt that this was probably an example of the difficulty that most people have in distinguishing between how a word feels to them when they think about it, and how it actually sounds when they say it. My belief, which I think is largely supported by these results, is that in American English the two word-forms "versus" and "verses" are nearly or completely homophonous in practical terms.

Indeed, given that the subjects who scored best generally explained that they focused on the prosody, the relevant cases (like "Philadelphia vers(u|e)s Toronto") would be less distinct than the stimuli used in the experiment.

As reader LB wrote when sending in her answers,

Interesting experiment! I thought I'd have some sort of grip on it, since the two words taste different to me, but after listening to the clips I doubt I'll beat random.

In fact, LB scored 65%.


  1. lukys said,

    June 23, 2012 @ 1:33 pm

    I've just remembered something which is relevant.
    I'm from North-West England, and one used to say "this team vee that team". Seemingly stemming from the abbreviation "X v. Y" on football scores and such.

  2. Suzanne Kemmer said,

    June 24, 2012 @ 12:50 am

    I thought the 2 words were going to sound distinct but discovered I could not distinguish them—I heard virtually all of them as pronounced with a final /z/, although I guessed 'versus' sometimes, based on prosodic cues.

    But I'm left wondering, do the final consonants sound like /z/ (to me) because they actually are phonetically [z] (meaning that 'versus' has simply acquired a voiced final consonant, matching 'verses'?) Or is there something more complicated going on?

    [(myl) Hi Suzanne — one of the cues to s/z voicing is the length and strength of the frication; and in both cases the frication length would be longer in utterance-final position; so to some extent your hearing all the words as ending in [z] may be an artefact of hearing them cut out of their context. But I think that there would still be a fair amount of overlap — and maybe a complete merger for some speakers — if we could create a pair of contexts for "versus" and "verses" with the same flanking words and with similar prosody. One example might be something like "…the Cummings verses Auden admired" versus "the Cummings versus Auden controversy". Alas, the collections of conversational speech available to me didn't happen to include any such examples…]

    The 'answers' to the quiz show which word the speakers were intending to say, but we don't know what final sound they were targeting for each of the two words – /s/ or /z/; and we don't know (without phonetic analysis) what were the actual segments produced – [s], [z], or something in between (e.g. a [z] in which the voicing shuts off earlier than normal).

    I think it is fairly clear what has happened morphologically: reanalysis of a noun to a new, semantically similar verb based on 1) phonetic resemblance of the final sounds in the noun to a verbal inflectional ending; 2) overlap in semantics (both are about oppositions of two competitors); and 3) contexts of syntactic/semantic similarity that would support the reanalysis.

    But I am not at all sure what is going on phonologically and phonetically in this merger, or whether the same thing is going on for everybody speaking American English. (Do others who can't distinguish them hear all [z], all [s], or maybe something in between?)

    All in all, a great example of why we need data when answering a question like 'are these two words homophones?' Our intuitions won't always give us the right answer; and also, to understand what's going on, we have to distinguish production from perception, as well as individual from group.

  3. D.O. said,

    June 24, 2012 @ 1:28 am

    Your analysis is focused more on individual people's strategies. What about stimuli? It seems that items 20, 17, 12, and 3 (all of them guessed more correctly than not) were especially easy. Anything special about them?

  4. Rachael said,

    June 24, 2012 @ 3:04 am

    I didn't bother to listen to the samples because they were presumably in American English. I do think the two words sound the same in US English, just not in UK English. I would expect to be unable to distinguish them in US English.

  5. danny o'bloom said,

    June 24, 2012 @ 9:13 am

    Brings back memories…as a Jewish kid in the 1950s and 60s, and not being real familiar with the New Testament way of reading out loud various passages, since I never read that book in synagogue, I always was confused with the pastor or father at some church I was visiting start talking about chapter and verse. I thought it was something like chapter One vs Verse 4. I think I was born to be an A-Theist.

  6. Saskia said,

    June 25, 2012 @ 1:13 am

    Interesting! – I was too late to put my results into the mix but I doubt I would have distinguished them – they all sound the same to me.
    In Australia I think the two forms are pretty much identical – I certainly didn't realise that it was "versus" until I was well into my teens. I had always parsed it as "verses". As I think my brothers and school friends all had as well.

  7. TomParmenter said,

    June 28, 2012 @ 3:01 pm

    The issue is at least tangentially related to

    'He's adverse to showering too much' vs* 'He's averse . . . '

    *Abbv. for?

RSS feed for comments on this post