I'm an armchair linguist. By this I mean it's been an absorbing hobby for decades by reading books and online and playing with many languages. But I've never taken any course or other kind of formal training. So I'm lacking knowledge of things which are best learned in a formal setting or which require special equipment.

I am now interested in knowing exactly where on the cartesian vowel chart each of the vowel phones of my personal idiolect occur.

I have learned that they don't always seem to be the same as those conventionally indicated by IPA symbols in broad transcriptions. I'm a speaker of Australian English but I'm not even sure my vowels match General Australian. We don't really have dialects here but there are some regional differences, sociolects, generational differences, and a couple of mergers and splits that some have and some do not. On top of that I've spent much of the past 25 years travelling or surrounded by foreigners.

Since vowels are more or less relative rather than having a set of features as with consonants, there seems to be more room for making incorrect assumptions about where one's own phones are located in regards to height and backness. (Obviously rounding is a bit easier.)

What methods are used to self-analyse vowels? I'm interested both in traditional methods and technological ones, especially if I only need a PC and free software.

  • 1
    This is a great question by itself, but as a side note, I think that everyone can produce any vowel in a well-known trapezoid vowel chart (with maybe minimum training), since it's about human's vocal tract. The difference is how one's vowels are perceived by a native-speaking listener. Commented Jun 18, 2014 at 9:36
  • Yes I'm playing with a language where I have no native speakers handy or good quality online samples. Descriptions compare with certain English vowels but I'm not certain which I have. I've been in the same situation before so it's time to learn how to deal with it (-: Commented Jun 18, 2014 at 10:33
  • 1
    There really is no good simple way to do this; learning enough phonetics to accurately describe one's own usage inevitably stretches one's native phonemic system beyond recognition. In my case, at least, it was only after several years that I gained enough perspective to figure out what my own English repertoire was. Start collecting minimal pairs now, though.
    – jlawler
    Commented Jun 18, 2014 at 17:33
  • 1
    It's not too hard to do some instrumental analysis of your own vowels. If you grab Praat and the Akustyk plug-in, you can record your minimal pairs and pretty easily generate a plot of their F1-F2 and see what you get. Commented Jun 19, 2014 at 12:26
  • 1
    Related: linguistics.stackexchange.com/questions/2269/… Commented Jun 19, 2014 at 17:18

2 Answers 2


I don't think it would be exactly possible to pin down a certain inventory of vowels in your own idiolect because even then they can vary so much depending on the environment that they appear in.

Take, for example, the vowel contrast of /u/ and /ʊ/. I can pronounce those just fine, and I can easily note the difference in the vowel between "book" /bʊk/ and "nude" /nud/. However, if you asked me to contrast between "pool" and "pull," which in Standard American English are /puːl/ and /pʊl/, respectively, I am physically incapable of pronouncing /puːl/. I pronounce both words as /pʊl/.

And then there are all the influences on any given vowel, like nasalization before a nasal consonant, rounding because of the environment in which the vowel appears, etc. Saying three words that have the same vowel are still probably going to all be slightly different.

This little gem from "lol, my thesis" basically sums it up: "It appears, based on experimental evidence, that vowel perception is pretty much magic."

However, if you want to somehow try and compare your vowels relative to one another, the best way to do that is to download Praat, fire it up, and start recording. I have no resources on how to properly analyze frontness-backness or vowel height (sorry, I mostly use it for suprasegmental stuff), but it's a place to start. With that you could start comparing all the different vowels you can produce.

  • 1
    That's interesting. I pronounce /pʊl/ and /pʊl:/ for "pull" and "pool". Before laterals I distinguish only length but in other environments by two "u" sounds differ by both length and quality. But right now I'm not sure if my "short e" sound is more of an /e/ or more of an /ɛ/ and I'm dabbling in a language which has both but I have no native speakers to practice with right now. Commented Jun 18, 2014 at 10:29
  • Sorry that should be /pʊl/ and /pʊ:l/ )-: Commented Jun 19, 2014 at 1:34

Part of what makes your question a bit tricky to answer is that hidden beneath it is a much more general question--"How do I do phonetics?"--which is obviously an absurdly broad question. The issue is narrowed by the fact that you are focusing on a concrete problem, but addressing it still entails a lot of knowledge about how speech and language work.

That said, several valid points have been brought up in comments and in @Nick Anderegg's answer, and I thought I'd take a crack at formulating a cohesive answer that incorporates the various points and hopefully clarifies some issues for you. But after reading my answer you might find that you have a bunch of follow-up questions!

Remember that the phonemes /e/ and /ɛ/ are phonological constructs. As @TKR notes, they are only meaningful inasmuch as they contrast in a particular phonemic inventory. The symbols a phonologist chooses to use for the phonemes are really just mnemonic labels--the symbol e indicates that the vowel in question tends to be realized as higher (and maybe fronter) than the one labeled as ɛ given the same environment. Just to be clear: in a given language, /ɛ/ may have an allophone that is just as high/front as or higher/fronter than one of the allophones of /e/. In such a case, the phonologist might characterize the allophone as [e], i.e., she might say that /ɛ/ is realized as [e] in certain environments.

So hopefully at this point you understand how the names of phonemes are purely mnemonic. But even the labels assigned to phones, such as [e] and [ɛ], do not have an absolute meaning in a vacuum. They are still labels of relativity imposed on the realizations of vowels by the phonetician, and they are language-specific. The phonetician measures the formants (and other parameters, such as duration) of many instances of the vowels in question, and she sees that they appear to cluster into two groups when everything else about the environment is the same (including, for example, the surrounding phones). The ones in the higher/fronter group get labeled as [e], while those in the not-as-high/not-as-front group (that are still higher and fronter than other vowels she's measured) get labeled as [ɛ].

Remember that IPA is language-dependent, a point I harped on in my answer to 'How can I differentiate between syllable-initial ɣ and ə using Praat or other software?'. Even if Language A has vowels that are conventionally labeled as [e] and [ɛ], respectively, and Language B also has vowels labeled as such, it doesn't mean that [e] in Language A will be the same as [e] in Language B, even in the analogous environment. Icelandic is usually described as having [e] and [ɛ], as is my dialect of English. But when my Icelandic friends produce their version of the English word bed, I hear it as bad, because the formant values and relative durations of their version of [e] in that word are more appropriate for the [æ] (the allphone I have for /æ/ between /b/ and /d/) in my dialect. Note that Icelandic doesn't have a phoneme that corresponds to my /æ/, i.e. they have more height leeway when it comes to their front-ish vowels.

Are you starting to see why it doesn't make so much sense to talk in absolute terms of whether you "have [e] or [ɛ]" in your dialect? There aren't "cardinal" measures of height/backness/duration for [e] and [ɛ]--only language-specific relative tendencies. What you can do is try to see how your pronunciations of certain vowels differ from those of native speakers in the target language.

On the perceptual front, you could do the following:

  1. Find a bunch of minimal pairs that alternate [e] and [ɛ] in the target language.
  2. Pronounce them with your one front mid vowel and ask native speakers which word of the relevant minimal pair they perceive you as saying. You'll want to ask a number of native speakers, as their intuitions might vary.
  3. Analyze their responses. If most of the responses tend in one direction over the other, it will be a hint that your one mid front vowel tends to map to that vowel in the target language. If the responses are more or less 50-50, it indicates that your vowel tends to hover somewhere in the middle of where they would expect their native [e] and [ɛ] to fall.

It's not as straightforward to try to approach the issue from an acoustical perspective; @jlawler is right when he points out that there are a vast number of factors that influence phonetic measurements like formant values and relative duration. Don't fall into the trap of thinking that you can measure a bunch of native speakers' formant values and compare theirs to yours! If Speaker A and Speaker B are native speakers of the same dialect of the same language, but they have different head and throat shapes, the formant values for [e] produced by Speaker A will be different from the formant values for [e] produced by Speaker B. In fact, the formant values for Speaker A's [e] could map onto the formant values for Speaker B's [ɛ], even if both speakers agree that A is producing [e] and B is producing [ɛ].

Here's what you can do on the acoustic front:

  1. Record a bunch of repetitions of a bunch of [e]/[ɛ] minimal pairs for a bunch of native speakers. Record them spoken in a bunch of environments (surrounded by different phones, in isolated words, in words in frame sentences, in careful speech, in faster speech, etc.).
  2. Measure the formant values and relative durations (e.g., the ratio of the vowel duration to the duration of a control segment somewhere else in the utterance) of the target vowels.
  3. Analyze the results and look for patterns.
  4. Record yourself trying to produce the minimal pairs, and practice differentiating them such that you eventually find the same relative patterns in your own formant values and relative durations.

I hope most of that makes sense. As this answer is getting a bit unwieldy, I'll leave it to you to post follow-up questions you may have about the best way to measure formant values in vowels or how certain parameters like F1 and F2 change according to changes in tongue height/backness!

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.