Estonian phonology

Estonian vowel chart, from Asu & Teras (2009:368). For some speakers, /ɤ/ can be more back (closer to /o/), or more back and higher (closer to /u/).

There are 9 vowels and 36 diphthongs, 28 of which are native to Estonian.[1] All nine vowels can appear as the first component of a diphthong, but only /ɑ, e, i, o, u/ occur as the second component. A vowel characteristic of Estonian is the unrounded back vowel /ɤ/, which may be close-mid back, close back, or close-mid central.[2]

• Vowels /ɑ, e, i, u/ can occur in both initial and non-initial syllables. Vowels /o, ɤ, æ, ø, y/ generally occur in initial syllables; they can occur in non-initial syllables only in compound words, exclamations, proper names and unnaturalized loanwords.[2]
• The front vowels /æ, ø, y/ are phonetically near-front [æ̈, ø̈, ÿ].[3]
• Before and after /j/, the back vowels /ɑ, u/ can be fronted to [ɑ̽, u̟].[2]
• The unrounded vowel transcribed /ɤ/ can be realized as close back [ɯ], close-mid central [ɘ] or close-mid back [ɤ], depending on the speaker.[2]
• The mid vowels /e, o, ø/ are phonetically close-mid [e, o, ø].[3]
• Word-final /e/ is often realized as mid [e̞].[2]
• The open vowels [ɑ, æ] are phonetically near-open [ɑ̝, æ̈].[3]

There are very few instances of vowel allophony; for instance, the long /y/ is pronounced as the diphthong [yi] before [j].

Simple vowels can be inherently short or long, written with single and double vowel letters respectively. Diphthongs are always inherently long. Furthermore, long vowels and diphthongs have two suprasegmental lengths. This is described further below.

• /m, p, pː/ are bilabial, whereas /v, f, fː/ are labiodental.[5]
• The fricatives /f, ʃ/ appear only in loanwords. Some speakers merge /ʃ/ with /s/ to [s].[3]
• /n/ is realized as velar [ŋ] before a velar consonant (e.g. panga /pɑnkɑ/ [pɑŋɡ̊ɑ] 'bank [gen.sg.]').[3]
• /k, kː/ are velar, whereas /j/ is palatal.[5]
  • When it appears between non-high vowels, the intervocalic /j/ can be realized as a non-syllabic close-mid front unrounded vowel [e̯].[3]
• The stops are voiceless unaspirated, but the short versions may be partially [p̬, t̬, t̬ʲ, k̬] or fully [b, d, dʲ, ɡ] voiced when they appear before or between vowels.[5]
• In spontaneous speech, word-initial /h/ is usually dropped. It is mostly retained in formal speech, and can be realized as voiced [ɦ] between two voiced sounds.[3]

Like the vowels, most consonants can be inherently short or long. For the plosives, this distinction is reflected as a distinction in tenseness/voicing, with short plosives being voiced and long plosives being voiceless. This distinction only applies fully for single consonants after stressed syllables. In other environments, the length or tenseness/voicing distinctions may be neutralized:

• After unstressed syllables or in consonant clusters, only obstruents can be long, other consonants are always short.
• In consonant clusters, voiced plosives are devoiced when next to another obstruent. That is, voiced plosives only occur if they are next to sounds exclusively from the classes of vowels and sonorants.
• Word-initially, obstruents are always voiceless, while the remaining consonants are always short. Recent loanwords may have voiced initial plosives, however.

In addition, long consonants and clusters also have two suprasegmental lengths, like the vowels. This is described below.

Non-phonemic palatalization generally occurs before front vowels. All word-internal and word-final alveolar consonants (except /r/) can be palatalized. In addition, about 0.15% of the vocabulary features fully phonemic palatalization, where palatalization occurs without the front vowel (e.g. /palʲk/ 'log, beam' vs /palk/ 'wage'). A front vowel did historically occur there, but was lost, leaving the palatalization as its only trace (a form of cheshirization). It occurs word-finally or word-medially. Thus, palatalization does not necessarily need a front vowel, and palatalized vs. plain continuants can be articulated. Estonian has pre-palatalization: palatalization occurs before rather than after the consonant and is characterized by a longer i-like transition from vowel to consonant.[3] Palatalization is not indicated in the standard orthography.

The stress in Estonian is usually on the first syllable, as was the case in Proto-Finnic. There are a few exceptions with the stress on the second syllable: aitäh ('thanks'), sõbranna ('female friend'). In loanwords, the original stress can be borrowed as well: ideaal ('ideal'), professor ('professor'). The stress is weak, and as length levels already control an aspect of "articulation intensity", most words appear evenly stressed.

A secondary stress, which is not always predictable, falls usually on odd-numbered syllables or some suffixes, e.g. -mine (laulmine [ˈlɑuːlˌmine] ‘singing’), -line (aluseline [ˈɑluˑseˌline] ‘basic, alkaline’), -lik (ohtlikku [ˈohːtˌlikːku] ‘dangerous (part. pl.)’). Words of more than three syllables can consist of combinations of monosyllabic, disyllabic and trisyllabic feet.[2]

Syllables can be divided into short and long. Syllables ending in a short vowel are short, while syllables ending in a long vowel, diphthong or consonant are long. The length of vowels, consonants and thus syllables is "inherent" in the sense that it is tied to a particular word and is not subject to morphological alternations.