Why is service pronounced the way it is and not like device even though the last 4 letters of the words are identical?

I would think that if they end the same way, the same pronunciation rules should apply as well.

I also checked Wiktionary.org and it says both words evolved from Old French where 'service' originated from 'servis' and 'device' originated from 'devis'. Again identical end of the word spelling.

  • 3
    In English orthography there are no rules, only tendencies! Commented Aug 16, 2012 at 12:53
  • 2
    Why is "service" pronounced so much like "crevice" – is that what you're asking?
    – J.R.
    Commented Aug 16, 2012 at 13:15
  • 10
    "I would think that if they end the same way, the same pronunciation rules should apply as well." Just like head and bead? Or bough, cough, rough, though, and through? It's hard to make that kind of assertion about a language that gives us the likes of live and live, read and read, and wind and wind.
    – J.R.
    Commented Aug 16, 2012 at 13:30
  • 4
    @J.R. Do you mean that English pronunciation is kind of chaotic? Let's see...
    – Julien Ch.
    Commented Aug 16, 2012 at 14:45
  • 3
    "the same pronunciation rules" I don't think we generally have pronunciation rules in English. Spellings were normally made to represent pronunciations, rather than the other way around.
    – bdsl
    Commented Dec 28, 2014 at 20:04

5 Answers 5


I believe this is due to the Great Vowel Shift.

In Middle English the "i" in device used to be prounounced as in service but later shifted to a /ai/ diphtong because it was bearing the stress.

  • 2
    I can't speak to the history, and I'm not an OF scholar, so I can't say what stress, if any, the words had before they migrated across the Channel. But certainly anyone who used the words would be aware that the orthographic identity of the endings is accidental: service is serv- with a suffix, device is vis- with a prefix. That sets the stress, after which the GVS takes over. Commented Aug 16, 2012 at 12:50
  • 1
    This has less to do with GVS, more to do with stress, as John Lawler described.
    – RainDoctor
    Commented Aug 16, 2012 at 17:59
  • 1
    Well I think it has to do with both, the sounds used to be the same (length excepted) and diverged during the GVS because one was stressed and the other was not.
    – Julien Ch.
    Commented Aug 17, 2012 at 7:05

It's entirely a matter of syllable stress. English is a stress-timed language. Only a stressed syllable can have a full vowel in English; unstressed syllables are centralized and reduced as much as possible, especially in rapid speech.

SERvice is stressed on the first syllable, so the vowel in the first syllable gets fully pronounced ['sɝ], while the second unstressed syllable is reduced to shwa [vəs].

On the hand, deVICE is stressed on the second syllable, so the vowel in that syllable gets fully pronounced ['vəys], while the first unstressed syllable is reduced to shwa [də].

Spelling has very little to do with pronunciation in English; its spelling was invented for a different language and doesn't work at all well for modern English. This fact explains a lot of other things, including why you shouldn't be worried if it's not logical. In fact, you're right. -- it's not logical.

  • How do unstressed full vowels work?
    – tchrist
    Commented Aug 16, 2012 at 17:43
  • I suspect they're less common in American than in British speech. I rarely come across them. Commented Aug 16, 2012 at 18:28
  • Are you sure? You reduce chromosome, ambition, humanity, tofu, Pleistocene, pharoah, kryptonite, plaintain? I don’t believe I’ve heard such an accent. I bet you don’t reduce those.
    – tchrist
    Commented Aug 16, 2012 at 18:43
  • ['kroməˌsom, ˌæm'bɪʃn, ˌyʊ'mænəɾi, 'toˌfu, 'playstəˌsin, ... ] Apparently secondary stress is against someone's religion; but not mine. Commented Aug 16, 2012 at 18:50
  • 1
    The most interesting part of what you said is the bit about how seriously they expect to be taken. I don’t know that I’ve ever realized that was a factor.
    – tchrist
    Commented Aug 16, 2012 at 23:12

The pronunciation of device might actually be considered less predictable than the pronunciation of service (if we compare them to words like novice and crevice). But then again, in the word advice we have the same pronunciation as in device. The stress pattern of device and advice may be related to their morphology: in both words, the first syllable was originally a prefix. It may also be related to the existence of the verbs devise and advise; if so, another similar example may be abuse n. and abuse v.

I'm going to ignore the presence of the letter "v", since I don't think it adds much information. So, here are some overall tendencies I see for the pronunciation of words that end with -ice.

Monosyllabic words ending in -ice all have /aɪs/ (unless the i is part of a vowel digraph, as in voice, juice, sluice). Some examples: ice, price, spice, slice.

Polysyllabic words derived from these also have /aɪs/, such as de-ice, re-price, overspice.

Most other polysyllabic words ending in -ice have fully unstressed /ɪs/ or /əs/. Of the words pronounced like this, most are nouns. There are some verbs, such as notice and apprentice, but they all have corresponding nouns spelled and pronounced the same way. In U.S. English, the verb practice also falls in this category. In British English the verb is spelled practise, although standardly pronounced the same as the noun. The OED has an interesting note about practise:

The word was originally stressed on the second syllable [...] and this is still the case in some regional varieties, especially in Scots (hence such spellings as practize, practeeze, practeese). The stress was subsequently shifted to the first syllable, with devoicing of the final consonant, probably by association with practice n.

Some examples, with the stressed vowel in bold: crevice, justice, lattice, notice, novice, artifice, avarice, edifice, orifice, accomplice, apprentice, interstice.

A small number of polysyllabic words have /aɪs/; as mentioned in other answers, these words always have primary or tertiary stress on the last syllable (and none of them have primary stress on the second-to-last syllable). This category has a greater proportion of verbs. (The reason why I've discussed part of speech is that in English verbs are generally more likely than nouns to have some degree of stress on the final syllable; we can see this in contrasts like associate n. versus associate v., or the noun ending -ity versus the verb ending -ify.) As StoneyB mentions in a comment, the stress may also be affected when the first syllable of a word comes from a Latin prefix. Here is a list of all the words like this I found, with the primary-stressed syllable in bold:

  • advice n. (pairs with advise v.; from Latin past participle vīsum with the preposition/prefix a(d-) )
  • device n. (pairs with devise v.; apparently ultimately from Latin dīvīs-/dīvidere, from videre with the prefix dis-)
  • entice v. (from Vulgar Latin intitiāre, with the prefix in-)
  • suffice v. (from Latin sufficere, with the prefix sub-)
  • sacrifice n., v. (from French, ultimately from Latin sacrificium. The OED says the verb was derived from the noun)
  • satisfice v. (note: this word was coined by combining satisfy and suffice)
  • cockatrice n. (however, this word is also pronounced with /ɪs/)

There are also a couple of words spelled with -ice that are usually pronounced with stressed /iːs/: police, caprice. They were borrowed relatively recently from French, and don't show the signs of the Great Vowel shift.


I'm going out on a limb a little but, I'm fairly sure that words that have two consonants followed by "ice" are pronounced like service. Words where "ice" is preceded by a vowel and consonant are pronounced like device.

  • 9
    I believe just one example will suffice to disprove that theory :-) Commented Aug 16, 2012 at 12:49
  • 3
    That's cheating ;) it's really one letter. One sounded letter anyway. I need to go and hide my shame in a crevice.
    – Chris
    Commented Aug 16, 2012 at 13:51
  • Please play nice.
    – bib
    Commented Aug 16, 2012 at 13:58
  • 1
    While your theory is enticing, I can't help noticing counterexamples. Commented Aug 16, 2012 at 17:47
  • 2
    Had a look. There's only about a dozen or so of words where the letter pattern is vowel, consonant and followed by "ice." The only exception for pronunciation in this group is "sacrifice." The sixy or so words that aren't in the above group don't follow much pattern in pronunciation. I think i've proven my answer thoroughly wrong.
    – Chris
    Commented Aug 17, 2012 at 0:31

The Great vowel shift split them,people used to say them the same then they took different roads

Service sɛrviːs > sɜɹvɪis > sɚvəs/sɜves(sairvees > sairvihys > servihs > srrvuhs/suhvuss)

Device dɛviːs > devɪis > dɪvəɪs>dəvæɪs(dehvees > dehvihys > duhvuhys > duhvaahys)

I am not exactly sure how and at what combination did the vowels change but the above is pretty much what happened)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.