Esperanto phonology

The original Esperanto lexicon contains 23 consonants, including 4 affricates and one, /x/, which has become rare; and 11 vowels, 5 simple and 6 diphthongs. A few additional sounds in loan words, such as /ou̯/, are not stable.

The Esperanto alphabet is nearly phonemic. The letters, along with the IPA and nearest English equivalent of their principal allophone, are,

Esperanto has many minimal pairs between the voiced and voiceless plosives, /b d g/ and /p t k/; for example, pagi "pay" vs. paki "pack", baro "bar" vs. paro "pair", teko "briefcase" vs. deko "group of ten".

On the other hand, the distinctions between several Esperanto consonants carry very light functional loads, though they are not in complementary distribution and therefore not allophones. The practical effect of this is that people who do not control these distinctions are still able to communicate without difficulty. These minor distinctions are ĵ /ʒ/ vs. ĝ /d͡ʒ/, contrasted in aĵo ('concrete thing') vs. aĝo ('age'); k /k/ vs. ĥ /x/ vs. h /h/, contrasted in koro ('heart') vs. ĥoro ('chorus') vs. horo ('hour'), and in the prefix ek- (inchoative) vs. eĥo ('echo'); dz /d͡z/ vs. z /z/, not contrasted in basic vocabulary; and c /t͡s/ vs. ĉ /t͡ʃ/, found in a few minimal pairs such as caro ('tzar'), ĉar ('because'); ci ('thou'), ĉi (proximate particle used with deictics); celo ('goal'), ĉelo ('cell'); -eco ('-ness'), eĉ ('even'); etc.

Belarusian seems to have also provided the model for Esperanto's diphthongs, as well as the complementary distribution of v (restricted to the onset of a syllable), and ŭ (occurring only as a vocalic offglide), although this was modified slightly, with Belarusian oŭ corresponding to Esperanto ov (as in bovlo), and ŭ being restricted to the sequences aŭ, eŭ in Esperanto. Although v and ŭ may both occur between vowels, as in naŭa ('ninth') and nava ('of naves'), the diphthongal distinction holds: [ˈnau̯.a] vs. [ˈna.va]. (However, Zamenhof did allow initial ŭ in onomatopoeic words such as ŭa 'wah!'.) The semivowel j likewise does not occur after the vowel i, but is also restricted from occurring before i in the same morpheme, whereas the Belarusian letter i represents /ji/. Later exceptions to these patterns, such as poŭpo ('poop deck'), ŭato ('watt'), East Asian proper names beginning with ⟨Ŭ⟩, and jida ('Yiddish'), are marginal.[note 4]

The distinction between e and ej carries a light functional load, in the core vocabulary perhaps only distinctive before alveolar sonorants, such as kejlo ('peg'), kelo ('cellar'); mejlo ('mile'), melo ('badger'); Rejno ('Rhine'), reno ('kidney'). The recent borrowing gejo ('homosexual') could contrast with the ambisexual prefix ge- if used in compounds with a following consonant, and also creating possible confusion between geja paro ('homosexual couple') and gea paro ('heterosexual couple'), which are both pronounceable as [ˈɡeja ˈparo]. Eŭ is also uncommon, and very seldom contrastive: eŭro ('a euro') vs. ero ('a bit').

Within a word, stress is on the penultimate syllable, with each vowel defining a syllabic nucleus: familio [famiˈli.o] ('family'). An exception is when the final -o of a noun is elided, usually for poetic reasons, because this does not affect the placement of the stress: famili’ [famiˈli].

On the rare occasions that stress needed to be specified, as in explanatory material or with proper names, Zamenhof used an acute accent. The most common such proper name is Zamenhof's own: Zámenhof. If the stress falls on the last syllable, it is common for an apostrophe to be used, as in poetic elision: Oĝalan’.

There is no set rule for which other syllables might receive stress in a polysyllabic word, or which monosyllabic words are stressed in a clause. Morphology, semantic load, and rhythm all play a role. By default, Esperanto is trochaic; stress tends to hit alternate syllables: Ésperánto. However, derivation tends to leave such "secondary" stress unchanged, at least for many speakers: Ésperantísto or Espérantísto (or for some just Esperantísto) Similarly, compound words generally retain their original stress. They never stress an epenthetic vowel: thus vórto-provízo, not *vortó-provízo.

Within a clause, rhythm also plays a role. However, referential words (lexical words and pronouns) attract stress, whereas "connecting" words such as prepositions tend not to: dónu al mí or dónu al mi ('give to me'), not *dónu ál mi. In Ĉu vi vídas la húndon kiu kúras preter la dómo? ('Do you see the dog that's running past the house?'), the function words do not take stress, not even two-syllable kiu ('which') or preter ('beyond'). The verb esti ('to be') behaves similarly, as can be seen by the occasional elision of the e in poetry or rapid speech: Mi ne ’stas ĉi tie! ('I'm not here!') Phonological words do not necessarily match orthographic words. Pronouns, prepositions, the article, and other monosyllabic function words are generally pronounced as a unit with the following word: mihávas ('I have'), laknábo ('the boy'), delvórto ('of the word'), ĉetáblo ('at table'). Exceptions include kaj 'and', which may be pronounced more distinctly when it has a larger scope than the following word or phrase.[6]

Within poetry, of course, the meter determines stress: Hó, mia kór’, ne bátu máltrankvíle ('Oh my heart, do not beat uneasily').

Emphasis and contrast may override normal stress. Pronouns frequently take stress because of this. In a simple question like Ĉu vi vídis? ('Did you see?'), the pronoun hardly needs to be said and is unstressed; compare Né, dónu al mí and ('No, give me'). Within a word, a prefix that wasn't heard correctly may be stressed upon repetition: Né, ne tíen! Iru máldekstren, mi diris! ('No, not over there! Go left, I said!'). Because stress doesn't distinguish words in Esperanto, shifting it to an unexpected syllable calls attention to that syllable, but doesn't cause confusion as it might in English.

As in many languages, initialisms behave unusually. When grammatical, they may be unstressed: k.t.p. [kotopo] ('et cetera'); when used as proper names, they tend to be idiosyncratic: UEA [ˈuˈeˈa], [ˈu.e.a], or [u.eˈa], but rarely *[u.ˈe.a]. This seems to be a way of indicating that the term is not a normal word. However, full acronyms tend to have regular stress: Tejo [ˈte.jo].

Lexical tone is not phonemic. Nor is clausal intonation, as question particles and changes in word order serve many of the functions that intonation performs in English.

A syllable in Esperanto is generally of the form (s/ŝ)(C)(C)V(C)(C). That is, it may have an onset, of up to three consonants; must have a nucleus of a single vowel or diphthong (except in onomatopoeic words such as zzz!), and may have a coda of zero to one (occasionally two) consonants.

Any consonant may occur initially, with the exception of j before i (though there is now one word that violates this restriction, jida ('Yiddish') which contrasts with ida "of an offspring").

Any consonant except h may close a syllable, though coda ĝ and ĵ are rare in monomorphemes (they contrast in aĝ’ 'age' vs. aĵ’ 'thing'). Within a morpheme, there may be a maximum of four sequential consonants, as for example in instruas ('teaches'), dekstren ('to the right'). Long clusters generally include a sibilant such as s or one of the liquids l or r.

Geminate consonants generally only occur in polymorphemic words, such as mal-longa ('short'), ek-kuŝi ('to flop down'), mis-skribi ('to mis-write'); in ethnonyms such as finno ('a Finn'), gallo ('a Gaul') (now more commonly gaŭlo); in proper names such as Ŝillero ('Schiller'), Buddo ('Buddha') (now more commonly Budho); and in a handful of unstable borrowings such as matĉo ('a sports match'). In compounds of lexical words, Zamenhof separated identical consonants with an epenthetic vowel, as in vivovespero ('the evening of life'), never *vivvespero.

Word-final consonants occur, though final voiced obstruents are generally rejected. For example, Latin ad ('to') became Esperanto al, and Polish od ('than') morphed into Esperanto ol ('than'). Sonorants and voiceless obstruents, on the other hand, are found in many of the numerals: cent ('hundred'), ok ('eight'), sep ('seven'), ses ('six'), kvin ('five'), kvar ('four'); also dum ('during'), eĉ ('even'). Even the poetic elision of final -o is rarely seen if it would leave a final voiced obstruent. A very few words with final voiced obstruents do occur, such as sed ('but') and apud ('next to'), but in such cases there is no minimal-pair contrast with a voiceless counterpart (that is, there is no *set or *aput to cause confusion with sed or apud). This is because many people, including the Slavs and Germans, do not contrast voicing in final obstruents. For similar reasons, sequences of obstruents with mixed voicing are not found in Zamenhofian compounds, apart from numerals and grammatical forms, thus longatempe 'for a long time', not *longtempe. (Note that /v/ is an exception to this rule, like in the Slavic languages. It is effectively ambiguous between fricative and approximant. The other exception is /kz/, which is commonly treated as /ɡz/.)

Syllabic consonants occur only as interjections and onomatopoeia: fr!, sss!, ŝŝ!, hm!.

All triconsonantal onsets begin with a sibilant, s or ŝ. Disregarding proper names, such as Vladimiro, the following initial consonant clusters occur:

• Stop + liquid – bl, br; pl, pr; dr; tr; gl, gr; kl, kr
• Voiceless fricative + liquid – fl, fr; sl; ŝl, ŝr
• Voiceless sibilant + voiceless stop (+ liquid) – sc [st͡s], sp, spl, spr; st, str; sk, skl, skr; ŝp, ŝpr; ŝt, ŝtr
• Obstruent + nasal – gn, kn, sm, sn, ŝm, ŝn
• Obstruent + /v/ – gv, kv, sv, ŝv

And more marginally,

Consonant + /j/ – (tj), ĉj, fj, vj, nj

The affectionate suffixes -ĉj- and -nj-, which retain remnants of the Slavic palatalized consonants, may very occasionally be used as words in their own right, as in mia ĉjanja popolo ('my dear nation'), in which case they may be word initial and not just syllable initial.

Although it does not occur initially, the sequence ⟨dz⟩ is pronounced as an affricate, as in edzo [ˈe.d͡zo] ('a husband') with an open first syllable [e], not as *[ed.zo].

In addition, initial ⟨pf⟩ occurs in German-derived pfenigo ('penny'), ⟨kŝ⟩ in Sanskrit kŝatrio ('kshatriya'), and several additional uncommon initial clusters occur in technical words of Greek origin, such as mn-, pn-, ks-, ps-, sf-, ft-, kt-, pt-, bd-, such as sfinktero ('a sphincter' which also has the coda ⟨nk⟩). Quite a few more clusters turn up in sufficiently obscure words, such as ⟨tl⟩ in tlaspo "Thlaspi" (a genus of herb), and Aztec deities such as Tlaloko ('Tlaloc'). (The /l/ phonemes are presumably devoiced in these words.)

As this might suggest, greater phonotactic diversity and complexity is tolerated in learnèd than in quotidian words, almost as if "difficult" phonotactics were an iconic indication of "difficult" vocabulary. Diconsonantal codas, for example, generally only occur in technical terms, proper names, and in geographical and ethnic terms: konjunkcio ('a conjunction'), arkta ('Arctic'), istmo ('isthmus').

However, there is a strong tendency for more basic terms to avoid such clusters, although cent ('hundred'), post ('after'), sankta ('holy'), and the prefix eks- ('ex-') (which can be used as an interjection: Eks la reĝo! 'Down with the king!') are exceptions. Even when coda clusters occur in the source languages, they are often eliminated in Esperanto. For instance, many European languages have words relating to "body" with a root of korps-. This root gave rise to two words in Esperanto, neither of which keep the full cluster: korpuso ('a military corps') (retaining the original Latin u), and korpo ('a biological body') (losing the s).

Many ordinary roots end in two or three consonants, such as cikl-o ('a bicycle'), ŝultr-o ('a shoulder'), pingl-o ('a needle'), tranĉ-i ('to cut'). However, these roots do not normally entail coda clusters except when followed by another consonant in compounds, or with poetic elision of the final -o. Even then, only sequences with decreasing sonority are possible, so although poetic tranĉ’ occurs, *cikl’, *ŝultr’, and *pingl’ do not. (Note that the humorous jargon Esperant’ does not follow this restriction, because it elides the grammatical suffix of all nouns no matter how awkward the result.)

Within compounds, an epenthetic vowel is added to break up what would otherwise be unacceptable clusters of consonants. This vowel is most commonly the nominal affix -o, regardless of number or case, as in kant-o-birdo ('a songbird') (the root kant-, 'to sing', is inherently a verb), but other part-of-speech endings may be used when -o- is judged to be grammatically inappropriate, as in mult-e-kosta ('expensive'). There is a great deal of personal variation as to when an epenthetic vowel is used.

With only five oral and no nasal or long vowels, Esperanto allows a fair amount of allophonic variation, though the distinction between /e/ and /ei̯/, and arguably /o/ and /ou̯/, is phonemic. The /v/ may be a labiodental fricative [v] or a labiodental approximant [ʋ], again in free variation; or [w], especially in the sequences kv and gv (/kw/ and /gw/, like "qu" and "gu"), but with [v] considered normative. Alveolar consonants t, d, n, l are acceptably either apical (as in English) or laminal (as in French, generally but incorrectly called "dental"). Postalveolars ĉ, ĝ, ŝ, ĵ may be palato-alveolar (semi-palatalized) [t̠ʃ, d̠ʒ, ʃ, ʒ] as in English and French, or retroflex (non-palatalized) [t̠ʂ d̠ʐ ʂ ʐ] as in Polish, Russian, and Mandarin Chinese. H and ĥ may be voiced [ɦ, ɣ], especially between vowels.