In French, phonemes like /e/ and /ɛ/ are so similar in sound. In English, phonemes like /o/ and /ɔ/ are so similar too. Briefly, almost any language, contains phonemes which are very similar to each other. The question is: how much comprehension does a listener lose, if a speaker pronounces them the same?

PS: I'm asking this question because I'm working on something I call SPA (Simplified Phonetic Alphabet), which is based on reductionism, and aims to facilitate language learning process in first steps by removing similar sounds for foreign learners.

Update: I'm talking about non-native speaker, and a native-listener.

  • By 'loose' do you mean 'lose'? Are you asking if a listener's ability to comprehend the speech of a speaker suffers if the speaker pronounces two phonemes the same all the time in all phonological environments? Commented Oct 24, 2011 at 4:03
  • 7
    Why do you think that will facilitate learning process? One of the weaknesses often found in non-native speakers is the failure to distinguish similar phonemes. If you intentionally remove similar sounds from teaching, you basically intentionally teach them a foreign accent.
    – Louis Rhys
    Commented Oct 24, 2011 at 7:02
  • @musicallinguist, yeah, I meant "lose". My question is as you said. Commented Oct 24, 2011 at 7:09
  • 1
    It's not a true answer, but I can give my not so educated opinion on one of the examples you quote. My experience with non-native speakers of French made me think that the merging of /e/ and /ɛ/ is not so troublesome (maybe it has something to do with the fact that even important minimal pairs à la <été> /ete/ ≠ <étais> /etɛ/ aren't clearly marked by all native speakers), but that some similar mistakes really impede communication (I know Germans who can't make the difference between /ɔ̃/ and /ɑ̃/ and Italians who are shaky about /s/ and /z/: that one really makes things hard).
    – JPP
    Commented Oct 31, 2011 at 11:07
  • 3
    And a more general remark: I think that hiding a phonemic complexity is really a bad service to do to (even beginning) learners of a foreign language. I feel like I would have learned English much more efficiently if I had been told earlier about all those funny-sounding vowels...
    – JPP
    Commented Oct 31, 2011 at 11:19

7 Answers 7


A native speaker may or may not be able to understand you; it depends on the phonemes, the language, the accent, the skill of the native speaker, and more. A few things to consider:

  1. One should keep in mind that phonemes which sound very similar to a foreigner may sound very different to native speakers. For example, I once pronounced Navarra as Navara in Spanish, and my Spanish friend had no idea at all what I was talking about. After my third try (it was admittedly in a noisy bar), he said, "ohh, you mean Navarra! Then why didn't you say so?" To me, r and rr sound like the same phoneme—but certainly not to Spaniards.

  2. It depends on how many relevant minimal pairs there are in the target language. If there are many words, or a few very frequent words, in which only [e] and [ɛ] are different, it will be more difficult to understand for the native speaker if one is used for the other.

  3. It also depends on whether the native speaker is aware of the fact that the foreigner is apt to confuse these two phonemes. For example, since the Romance accents in English are fairly well known, an Englishman will not be misled as easily if a Frenchman keeps confusing this (/ðɪs/) and these (unaccented before consonant: something like /ðis/). And there are probably more factors to consider.

  • 6
    Cerberus's point #2 here is getting at the concept of functional load. If a language has frequent minimal pairs for [e] and [ɛ], you can say the distinction has a high functional load in that language. If it doesn't, the distinction has a low functional load. Note, though, that different distinctions carry different amounts of functional load in different languages. (E.g. the θ/ð distinction has a very low functional load in English but a high functional load in Greek.) Commented Oct 24, 2011 at 14:52
  • 1
    +1 @dan. It's also important to note that functional load is not always easily quantified and is influenced by a variety of factors. When I learned Mandarin we were taught the Pinyin romanization system, which marks tone on every syllable using diacritics. Three semesters into the program, I had an instructor who had the habit of writing e-mails completely in Pinyin without any tone diacritics! To my astonishment, I had no trouble comprehending the e-mails. Although lexical tone is thought to have quite a high functional load in Mandarin, in that context it didn't affect comprehension at all! Commented Oct 24, 2011 at 16:08
  • 1
    @Cerberus: these is pronounced /ðiː z /. Commented May 22, 2012 at 0:45
  • @Mechanicalsnail: It is often pronounced voiceless, or virtually indistinguishable from voiceless, when the word is unaccented and it comes before a consonant (otherwise the s/z contrast would indeed be audible). Try saying this quickly: "they killed not some, but all of these people in a single night". I can post some real examples if you need more convincing.
    – Cerberus
    Commented May 22, 2012 at 2:18

Another way to find "safe" mergers in a language is to look for mergers that some native speakers of that language make. For instance, in English, many speakers merge "cot" and "caught" (i.e. fail to distinguish /ɑ/ and /ɔ/). And in the US at least, those of us who don't have the cot/caught merger are still accustomed to hearing it from other people. So a foreign learner who merged /ɑ/ and /ɔ/ would still be intelligible here.

Keep in mind though that this suggestion is relative to a specific language. So in US English, merging /ɑ/ and /ɔ/ is "safe," but that doesn't mean it's equally safe in other languages. For instance, I don't know of any French dialects that merge /ɑ/ and /ɔ/, so I doubt that merger would be as safe in French.

I don't think that you're going to find any mergers that are "universally safe" — i.e. safe in all languages — and so if you want your SPA to be a universally applicable system, you're going to run into trouble pretty quickly. On the other hand, a simplified phonetic notation for a specific language could succeed. So you might end up with "SPA-English" (merging /ɑ/ with /ɔ/) and "SPA-French" (merging /œ̃/ with /ɛ̃/ and /ɑ/ with /a/) and so on.

  • 2
    (In fact, it's a little more complicated than that, because even speakers with the cot/caught merger distinguish /ɑ/ and /ɔ/ in a few environments. Perhaps a better example would have been the merger between "witch" and "which"; many speakers have merged /w/ and /ʍ/ in all environments, and a foreign learner of English really doesn't need to bother with the distinction at all.) Commented Oct 24, 2011 at 15:23
  • 3
    In French, merging /a/ with ɑ is pretty safe (many native speakers do it); merging /ɔ/ with /o is fairly safe (some native speakers do it) but is likely to hinder comprehension somewhat; merging /ɑ/ and /ɔ/ is definitely not safe. Or, to give another example that many non-native speakers hit due to the relative rarity of the vowel /y/: merging /u/ and /y/ is absolutely not safe, as witnessed by the minimal pair “dessus, dessous” (/dəsy/, /dəsu/, meaning “above, below”). Commented Oct 24, 2011 at 22:23
  • @Gilles: which native speakers are you alluding to about that /ɔ/-/o/ merger? It's very clear that the border between those phonemes isn't the same in Paris and Marseilles but do you consider that they can be considered as completely merged in some southern dialects, for example?
    – JPP
    Commented Oct 31, 2011 at 11:14
  • 1
    @JPP Merging the phonemes /ɔ/ and /o/ (also /ɛ/ and /e/) is fairly standard in southwestern accents. They may still exist as distinct phones (as described e.g. here), though I believe a minority of speakers don't have distinct sounds at all. Commented Oct 31, 2011 at 18:12

Standard Italian has 7 vocalic phonemes: a i u è é ò ó, some Northerners and most Southerners merge open and close e's and o's, but this is not a prestigious pronunciation (and it's strange-sounding to people from Emilia Romagna, Marche, Umbria, Rome or Tuscany). Nonetheless, you can find in many books for foreigners: ''Italian vowels are easy: ah, eh, eeh, oh, ooh.'' It's called dumbing down.


To enlarge on what Cerberus said: your premise is wrong. Different phonemes in a language are by definition different in sound (for at least some phonological environments). The fact that speakers of some other language may find them similar is a fact about that other language, not about the target language.

I completely agree with Louis Rhys' comment, that a notation which confuses two different phonemes in a language is a particularly bad notation for that language - especially for learning it. (Conventional orthographies for languages sometimes do confuse phonemes in this way, but in these cases materials for learners will often disambiguate them.)

Edit: added rider about phonological environment - thanks @Musicallinguist - and not about conventional orthographies.

  • You make a good point about the perception of similarity by speakers of different languages, but I would suggest a slight rewording of your answer. Different phonemes are not necessarily pronounced differently in all environments. In fact, two phonemes can be pronounced the same in a majority of environments and only distinguished in a small subset of environments. Commented Oct 24, 2011 at 12:12
  • @musicallinguist is spot on. It is not uncommon for a contrast between two phonemes in a given language to be completely neutralised in some environments, maybe even the majority. Word-final stops in German are a classic example here but it's very common. Commented May 23, 2012 at 1:50
  • @GastonÜmlaut: although in fact, when I started learning some German words, I was surprised by how often the voicing of word-final stops is predictable (at least in native words) due to historical reasons. Word-final [p] is generally /b/, word-final [k] is usually /k/ after a short vowel or after /n/, and /g/ after a long vowel. Word-final [t] is fairly unpredictable though. Commented Sep 28, 2015 at 5:13
  • @sumelic yes the realisation may be in some cases be conditioned by the environment and in others be free variation, but that doesn't affect the fact that there is no contrast. Commented Sep 28, 2015 at 6:00

I was reading a paper called Loanword adaptation as first-language phonological perception for a term paper I'm currently writing. I don't know enough about the literature to know if this is a particularly novel idea, but I think that their OT-based perception analysis (as seen in section 3 of that paper) is relevant here. I'll describe the theory first and then I'll talk about how I think it applies to your problem.

For people who don't know, Optimality Theory is basically an alternative to rule-based systems where instead of having specific inputs undergo specific transformations to become specific outputs, there is simply a bunch of preferences where some have a higher priority than others (and thus are less likely to be violated).

Usually when people talk about OT they're talking about the production process (from UR to SR). However, that paper mentions an OT-based approach to perception. So, for example, consider a language that does not contain /ɛ/.There must be some highly ranked constraint *[ɛ]/ɛ/ (meaning the phonetic sound [ɛ] must not be interpretted as the phonological unit /ɛ/). This prevents speakers of that langauge from perceiving /ɛ/ when they hear [ɛ]. This would be desirable because, due to the lack of /ɛ/ in that language, any SR containing /ɛ/ would be unexpected.

Let's compare this to a language that does contain /ɛ/ as well as /e/ (like French). It is important to note that in OT all constraints exist across all languages only the rankings change. Therefore, that same *[ɛ]/ɛ/ must exist in this language too except that now it is much much much lower priority. In this langauge, since /e/ and /ɛ/ are different, we know there needs to be another constraint *[ɛ]/e/ (do not interpret [ɛ] as /e/) that has higher priority than *[ɛ]/ɛ/. This means that when speakers hear [ɛ] they are more likely to perceive it as /ɛ/ because perceiving it as /e/ is a greater violation (*[ɛ]/e/ >> *[ɛ]/ɛ/).

The same paper proposes that the ranking of these constraints is somewhat dynamic and based on a "probability-matching listener". This means that, theoretically speaking, if a French speaker is systematically forced to accept [ɛ] as /e/ (for example, by speaking for an extended period of time with an L2 French speaker from a language without /e/, only /ɛ/) then the *[ɛ]/e/ constraint slowly loses priority and the French speaker slowly becomes more ready to identify [ɛ] as /e/ (and of course, the process will slowly undo itself when the French speaker returns to speakering to other native speakers). Taken to its logical extreme, this could explain why certian languages merge certian sounds while others do not (i.e. if a language does not distinguish between two sounds, violations of the relevant constraints become so lowly ranked that faithfully perceiving phonetic input becomes more costly than approximating it).

So here is how I think this applies to your problem: in English the constraints *[ɛ]/e/ and *[e]/ɛ/ are both very lowly ranked, allowing English speakers to freely interpret one as the other. Conversely, in French, where misinterpreting one as the other causes a loss of meaning, the constraints are much higher ranked and thus it is much less likely that they will confuse to two. By this approach, it isn't that speakers "lose comprehension" when faced with similar phonetic sounds, it's that the natural "error correcting" in their brain interprets the sounds as whatever sound makes the most sense in a given context.

As mentioned earlier, I'm not too familar with the literature, but I've found another author who's written a few papers around a similar idea of a productive perception as well as both sets of authors cite various other authors who came before them.

  • This is an interesting analysis, but what does it actually buy you over other probabilistic models? Commented Jun 1, 2012 at 10:45
  • My understanding from the reading the paper is that this learning algorithm approach doesn't require any implicit knowledge about which sounds are "closer" than others (as implied with P-map theory). Taken to a logical extreme, children might assign random URs to words and then as more input comes in they slowly begin to adjust and codify these URs at the same time as the constraints get ranked. Thus, English children learn to consider /e/ and /ɛ/ the same not because they're similar but because the language doesn't distinguish between them. The net result is the same, though.
    – acattle
    Commented Jun 1, 2012 at 14:20
  • In all honesty, the learning algorithms is one of the weaker parts of the paper for me as the author only seems to cite himself and I've yet to read the original paper. I just thought it was an interesting idea worth sharing.
    – acattle
    Commented Jun 1, 2012 at 14:23

A partial answer to this is that it will probably depend on how native-like the speaker's pronunciation of other parts of the utterance are. So taking a very simplistic example with your two French vowels as an example, the words "clé" and "craie", in addition to having "l" vs "r", generally differ in that the final vowel is /e/ and /ɛ/ respectively. So if a speaker does not differentiate /e/ and /ɛ/, this puts more 'pressure' on the /l/ vs /r/ distinction to differentiate these two words.

(N.B. it turns out that /e/~/ɛ/ are actually in free variation much of the time in French and pronunciation of either is highly subject to effects of assimilation or "vowel harmony", so this pair isn't necessarily the most compelling example.)

Extending this isolated example out to many different phonemes in a more complex utterance, and you'll see that the situation quickly becomes very complex. Listeners are expecting all sorts of phonetic cues to decipher what they hear, and if you eliminate one particular cue in a specific place it's difficult to predict exactly what pressure this will put on which cues elsewhere.

You would also need to think about what exactly you mean by two phonemes being "similar". If you mean specifically vowel quality, what other features of the phoneme (e.g. duration patterns, propensity to assimilation, phonotactics) are you then disregarding and does the cost-benefit ratio actually work in favour of your assumption? For example, maybe speakers have difficulty in distinguishing the quality of English /I/ vs /i/. But maybe they can easily pronounce these vowels with different durations and that it is primarily the duration that is perceptually important. (cf Tajima et al, 1997, "Effects of temporal correction on intelligibility of foreign-accented English: they found that by re-synthesising foreign accented speech to correct specifically the timings of segments, this apparently improved intelligibility even though vowel quality may been quite 'far off the mark' in some cases). In that case, you have thrown out a priori a distinction that speakers could potentially have made quite easily because your system decided that it was "too difficult to be worth making".

Or, to put it more succinctly: what you're proposing doesn't sound as though it will achieve its objectives.


Native or fluent speakers enjoy greater license in producing inaccurate pronunciations than do non-native non-fluent speakers. Native and fluent speakers and listeners have tacit command of the relative frequencies of different words and combinations thereof, so expressions which are, given the context, of relatively high frequency, will be understood even when pronounced carelessly, because they are more expected. Native and fluent speakers speak more carefully when they are delivering a less predictable message.

Non-native speakers who nevertheless speak fluently are able to make themselves understood because they know the usual things to say in a given context, and their inaccurate productions will be correctly interpreted because they only try to say things which people normally say in a given situation. I can, for example, understand Vietnamese speakers who are speaking English badly (e.g., pronouncing coda /s/, /t/, /k/, /tS/ all as unreleased [k]) as long as they are saying something that is expected in the situation.

The language-learning situation is very unforgiving for a non-native speaker who has little experience with the language and wishes to to express him/herself, only attending to what is a grammatical sentence, and not to what is customarily said in a given situation. A patient or sympathetic listener and accurate pronunciation is needed.

If you are learning a new language, and pronunciation is not your strong suit, practice saying the same things in very predictable situations so that you will have a better chance of being understood. The level of muscular-auditory coordination needed to pronounce in a standard way, whether it is a novel or mundane expression, is the same, so this will be good practice for obtaining the skill level needed to become more expressive.

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.