Many languages lack phonemic glottal stops, but regularly insert them. For example:

English invariably inserts glottal stops before utterance-initial vowels, and often before word-initial vowels when enunciating:

/ˈɔw ˈnɔw/ [/ˈɔwˈnɔ́ẁ/] 'Oh no!'

and also when needed to break up adjacent identical vowels:

/ði ˈir/ [ði̠ˈʔiɰ˞] 'the ear'

Japanese is similar, except it also allows a glottal stop utterance-finally, especially in emphatic utterances:

/ee/ [ʔèéʔ] ⟨ええ?⟩ 'huh!?'

This seems to be an under-studied phenomenon despite its widespread occurrence. Do you have any references comparing how this works in different languages?

  • 1
    Interesting question. I would just note that it's a bit too strong to say 'invariably' for the English and Japanese cases. In spontaneous speech in both languages, the glottal stop may or may not be observed and its presence often depends on level of emphasis, as you mention. It would be interesting to find out if there are cases of truly obligatory glottal stop insertion that is phonologically conditioned. Commented Nov 20, 2012 at 15:03
  • 3
    I only realised the other day that intervocalic glottal stop insertion across word boundaries is a major contributing factor to Germans not being able to lose their accent when speaking English. It was one of those 'click' moments! (I'm English, by the way).
    – legatrix
    Commented Nov 28, 2012 at 23:21
  • It seems and its pair are quite uncommon, they render as boxes for me in the question. There are at least two alternatives: and (〉 and 〉 without monospace font). The last is a CJK character so could be the correct one to use with Japanese? Commented Dec 21, 2012 at 22:27
  • @hippietrail: I've looked into this a bit. U+232A is a compatibility character (canonically equivalent to the CJK one), so it shouldn't be used. The CJK one, U+3009, is apparently used as a sort of quotation mark in Chinese. There's also U+203A (single guillemet), which is also a quotation mark. The mathematical angle brackets ⟨⟩ seem most semantically correct, and are what Wikipedia uses. Commented Dec 23, 2012 at 22:21
  • @Mechanicalsnail: Hmm I can see them here in the comments, just in the actual answer they are displayed as boxes. (Windows 7 / Google Chrome) Commented Dec 23, 2012 at 22:31

2 Answers 2


I don't have references, but the glottal stop before word-initial vowels is present also in Polish and Norwegian. And also in Sidamo, it is always present before word-initial vowels, but generally it is phonemic (can occur within words).

In Polish the glottal can actually claimed to be phonemic (even though I haven't seen anybody claiming it), because it makes the sound of a non-syllabic semivowel be interpreted as a syllabic vowel:

łgana [wgana] "lie" (feminine adjectival participle form), 2 syllables vs.

Ugana [ʔwgana], [ʔugana] "Ugana, a populated place in Papua New Guinea", 3 syllables

This is the only minimal pair I can think of, because semivowels that don't belong to diphtongs are rare at word-initial position in Polish.


I'm not sure if this is as understudied as you might think. What you're describing is a laryngeal feature used to mark the edges of an utterance. That's pretty common. The laryngeal feature used, though, might vary quite a bit between languages. They can be glottal stops, pitch, aspiration, devoicing, creaky voice, etc. Searching for something like laryngeal markers at utterance boundaries might give you more interesting hits.

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.