Esperanto phonology

The creator of Esperanto, L. L. Zamenhof, illustrated Esperanto pronunciation by comparing its letters with their equivalents in several major European languages and declaring a principle of "one letter, one sound".

With over a century of use, Esperanto has developed a phonological norm, including accepted details of phonetics,^[1] phonotactics,^[2] and intonation,^[3] so that it is now possible to speak of proper Esperanto pronunciation and properly formed words independently of the languages originally used to describe Esperanto. This norm diverges only minimally from the original ideal of "one letter, one sound"; that is, it accepts only minor allophonic variation.^[4]

Before Esperanto phonotactics became fixed, foreign words were adopted with spellings that violated the apparent intentions of Zamenhof and the norms that would develop later, such as poŭpo ('poop deck'), ŭato ('Watt'), and matĉo ('sports match').^[5] Many of these coinages have proven to be unstable, and have either fallen out of use or been replaced with pronunciations more in keeping with the developing norms, such as pobo for poŭpo, vato for ŭato, and maĉo for matĉo. On the other hand, the word jida ('Yiddish'), which was also sometimes criticized on phonotactical grounds^[6] but had been used by Zamenhof, is well established.

Inventory

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.

Consonants

		Labial	Alveolar	Post- alveolar	Palatal	Velar	Glottal
Nasal		m	n
Plosive	voiceless	p	t			k
	voiced	b	d			ɡ
Affricate	voiceless		t͡s	t͡ʃ
	voiced		(d͡z)	d͡ʒ
Fricative	voiceless	f	s	ʃ		x	h
	voiced	v	z	ʒ
Approximant			l		j
Trill			r

The uncommon affricate /d͡z/ does not have a distinct letter in the orthography, but is written with the digraph ⟨dz⟩, as in edzo ('husband'). Not everyone agrees on its status as a phoneme; Wennergren considers it as a simple sequence of d + z.^[7] Consonantal /w/ is only found in onomatopoeia and in foreign names.

Vowels

	Front	Back
Close	i	u
Mid	e	o
Open	a

There are also six historically stable diphthongs: /ai̯/, /oi̯/, /ui̯/, /ei̯/ and /au̯/, /eu̯/. However, some authors such as John C. Wells regard them as vowel + consonant combinations (/aj/, /oj/, /uj/, /ej/, /aw/, /ew/), while Wennergren regards only the latter two as diphthongs.^[8]

Slavic origins

This inventory is rather similar to that of Polish, but is especially close to Belarusian, which was historically important to Zamenhof, the creator of Esperanto. The essential difference from Belarusian (and Polish) is the absence of palatalization, although this was present in Proto-Esperanto (nacjes, now nacioj 'nations'; familje, now familio 'family') and arguably survives marginally in the affectionate suffixes -njo and -ĉjo, and in the interjection tju!.^[9] Minor differences are that g is pronounced as a stop, [ɡ], rather than as a fricative, [ɣ] (in Belarusian, the stop pronunciation is found in recent loan words); a distinction between /x/ and /h/; and the absence of a diphthong oŭ /ou̯/, though that was added to Esperanto to a minor degree after its creation. Like Belarusian, /v/ is found in syllable onsets and /u̯/ in syllable codas; however, unlike Belarusian, /v/ does not become /u̯/ if forced into coda position through compounding, though Zamenhof avoided such situations by adding an epenthetic vowel: lavobaseno ('washbasin'), not *lavbaseno or *laŭbaseno.

Orthography and pronunciation

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

Consonants				Vowels & diphthongs
Letter	English	IPA		Letter	English	IPA
b	b	[b]		a	spa	[a]
c	ts	[t͡s]		e	bet	[e]
ĉ	choose	[t͡ʃ]		i	machine	[i]
d	d	[d]		o	fork	[o]
f	f	[f]		u	rude	[u]
g	go	[ɡ]
ĝ	gem	[d͡ʒ]		aj	sky	[ai̯]
h	h	[h]		aŭ	now	[au̯]
ĥ	loch	[x]		ej	grey	[ei̯]
j	young	[j]		eŭ	*	[eu̯]
ĵ	pleasure	[ʒ]		oj	boy	[oi̯]
k	k	[k]		uj	phooey	[ui̯]
l	l	[l]			(as one syllable)
m	m	[m]		*Something similar to eŭ can be heard in exaggerated mimicry – as delivered by such American comedians as Carol Burnett – of the British pronunciation of the word oh Ŭ is a consonant in foreign names, where it tends to be Esperantized to [v], and occasionally in mimesis, as in ŭa! (waa!)
n	n	[n]
p	p	[p]
r	r (trilled)	[r]
s	s	[s]
ŝ	sh	[ʃ]
t	t	[t]
v	v	[v]
z	z	[z]

Minimal pairs

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.^[10]

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').

Stress and prosody

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.^[11]

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.

Phonotactics

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.

Allophonic variation

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. Disregarding assimilation for the moment, the more noticeable allophony among the consonants is with /r/ and /v/. The /r/ may be pronounced as either an alveolar flap [ɾ] or an alveolar trill [r], in free variation but with the flap more common. 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, 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. However, aspiration or incomplete voicing of consonants as in English or Mandarin is considered substandard, as are the English diphthongized "long" vowels [ij, ɛj, uw, ɔw] for /i, e, u, o/.

Vowel length and quality

Vowel length is not phonemic in Esperanto. Vowels tend to be long in open stressed syllables and short otherwise. Adjacent stressed syllables are not allowed in compound words, and when stress disappears in such situations, it may leave behind a residue of vowel length. Vowel length is sometimes presented as an argument for the phonemic status of the affricates, because vowels tend to be short before most consonant clusters (excepting stops plus l or r, as in many European languages), but long before ĉ, ĝ, c, and dz.

Vowel quality has never been an issue for a, i and u, but has been discussed much for e and o. Zamenhof recommended pronouncing the vowels e and o as mid [e̞, o̞] at all times. Kalocsay and Waringhien gave more complicated recommendations.^[12] For example, they recommended pronouncing stressed e, o as short open-mid [ɛ, ɔ] in closed syllables and long close-mid [eˑ, oˑ] in open syllables. However, this is widely considered unduly elaborate, and Zamenhof's recommendation of using mid vowels is considered the norm. For many speakers, however, the pronunciation of e and o reflects the details of their native language.

Epenthesis

Epenthetic glottal stops in vowel sequences such as boao ('boa') are non-phonemic, but allowed for the comfort of the speaker. They are especially common with sequences of identical vowels, such as heroo [heˈroˑʔo] ('hero'), and praavo [praˈʔaˑvo] ('great-grandfather'). Other speakers, however, mark the hiatus by a change of intonation, e.g. by raising the stressed vowel: heróò, pràávo.

It is also very common to pronounce an epenthetic [j] between an /i/ and a following vowel (mia [ˈmiˑja], mielo [miˈjɛˑlo]), but this is avoided in careful enunciation.

Poetic elision

Vowel elision is allowed with the grammatical suffix -o of singular nominative nouns, and the a of the article la, though this rarely occurs outside of poetry: de l’ kor’ ('from the heart').

Normally semivowels are restricted to offglides in diphthongs. However, poetic meter may force the reduction of unstressed /i/ and /u/ to semivowels before a stressed vowel: kormilionoj [kɔɾmiˈli̯oˑnɔi̯]; buduaro [buˈdu̯aˑɾo].

Assimilation

Zamenhof recognized two types of regressive assimilation in Esperanto:

• Place assimilation among nasals, and
• Voicing assimilation among obstruents.

In addition, he noted epenthetic glides between vowels. However, he stated that "severely regular" speech would not have assimilation, and this has led to debate over whether it "should" occur.^[13]

An example of the first type is assimilation of n before a velar, as in banko [ˈbaŋko] ('bank') or sango [ˈsaŋɡo] ('blood'). N may also palatalize before palatal /j/, as in panjo [ˈpaɲjo] ('mommy') and sinjoro [siˈɲjoˑro] ('sir'). However, although the desirability of these may be debated, the question almost never arises as to whether the m in emfazi should remain bilabial or should assimilate to labiodental f ([ɛɱˈfaˑzi]), because this assimilation is nearly universal in human language. Indeed, where the orthography allows, we see that assimilation does occur.

The debate on voicing assimilation is likewise dependent on speakers' language backgrounds. Assimilation may or may not occur in words that maintain Latinate orthography such as absolute ('absolutely') or obtuza ('obtuse'), depending on the speaker's background, despite the fact that potentially contrastive voiceless equivalents such as apsido ('apsis') and optiko ('optics') occur ([absoˈluˑte], [apsoˈluˑte] or even [abɨsoˈluˑte] for absolute; [ɔbˈtuˑza], [ɔpˈtuˑza] or even [obɨˈtuˑza] for obtuza).^[14] Instead, the debate centers around the non-Latinate orthographic sequence kz, frequently found in Latinate words like ekzemple ('for example') and ekzisti ('to exist'). It is often claimed that kz is properly pronounced as written, with mixed voicing, [kz], despite the fact that Zamenhof recognized that the k may assimilate to the z for [ɛɡˈzɛmple, ɛɡˈzisti], as in Slavic, English, French, and many other languages. The two opinions are called ekzismo and egzismo in Esperanto.^[15] In practice, most Esperanto speakers assimilate both kz to [ɡz] and nk to [ŋk] when speaking fluently.

Compound words such as okdek ('eighty'), longtempe ('for a long time'), and glavsonoro ('the ringing of a sword') are likewise more likely to retain mixed voicing, though assimilation is not uncommon in rapid speech, and may even be preferred:^[14] [ˈɔɡdɛk, lɔˑŋkˈtɛmpe, ˈɡlaˑfsoˈnoˑro]. However, in compounds of lexical words Zamenhof inserted an epenthetic vowel between obstruents with different voicing, as in rozokolora ('pink'), never *rozkolora, and longatempe, never *longtempe as with some later writers; mixed voicing only occurred with grammatical forms. V behaves as a sonorant when it occurs after another consonant, and even Slavs readily distinguish Zamenhofian kv from gv. V is also never found in coda position in Zamenhof's writing, because that would force it to contrast with ŭ.

Similarly, mixed sibilant sequences, as in the polymorphemic disĵeti ('to scatter'), tend to assimilate, sometimes completely, in rapid speech ([diʒˈʒeˑti]), though, if noticed, this would be considered incorrect.

Like the generally ignored regressive devoicing in words such as absurda, progressive devoicing tends to go unnoticed within stop–sonorant clusters, as in plua [ˈpl̥uˑa] ('additional'; contrasts with blua [ˈbluˑa] 'blue') and knabo [ˈkn̥aˑbo] ('boy'; the kn- contrasts with gn-, as in gnomo [ˈɡnoˑmo] 'gnome'). Partial to full devoicing of the sonorant is probably the norm for most speakers.

Voicing assimilation of affricates and fricatives before nasals, as in taĉmento "a detachment" and the suffix -ismo ('-ism'), is both more noticeable and easier for most speakers to avoid, so [ˈizmo] for -ismo is less tolerated than [apsoˈluˑte] for absolute.

Zamenhof also noted that glides may be inserted between dissimilar vowels, especially after high vowels as in [ˈmiˑja] for mia and [ˈpluˑwa] for plua. This is quite common, and there is no possibility of confusion, because ij and uŭ do not occur in Esperanto, but may cause confusion between gea and geja, as mentioned above.

The following allophones are thus acceptable in Esperanto:

Phoneme	a	e	o	i	u	n	m	r	v	z	ʒ	f	s	ʃ	b	d	dʒ	ɡ	p	t	tʃ	k
Allophones	ɑ	ɛ	ɔ	ij	uw	ŋ, ɲ	ɱ	ɾ	ʋ, w, f	s	ʃ	v	z	ʒ	p	t	tʃ	k	b	d	dʒ	ɡ

Loss of phonemic ĥ

The sound of ⟨ĥ⟩, [x] was always somewhat marginal in Esperanto, and there has been a strong move to merge it into [k], starting with suggestions from Zamenhof himself.^[16][17] Dictionaries generally cross-reference ⟨ĥ⟩ and ⟨k⟩, but the sequence ⟨rĥ⟩ (as in arĥitekturo 'architecture') was replaced by ⟨rk⟩ (arkitekturo) so completely by the early 20th century that few dictionaries even list ⟨rĥ⟩ as an option. Other words, such as ĥemio ('chemistry') and monaĥo ('monk'), still vary but are more commonly found with ⟨k⟩ (kemio, monako). In a few cases, such as with words of Russian origin, ⟨ĥ⟩ may instead be replaced by ⟨h⟩. This merger has had only a few complications. Zamenhof gave ĥoro ('chorus') the alternate form koruso, because both koro ('heart') and horo ('hour') were taken. The two words still almost universally seen with ⟨ĥ⟩ are eĥo ('echo') and ĉeĥo ('a Czech'). Ek- (perfective aspect) and ĉeko ('check') already exist, though ekoo for eĥo is occasionally seen.

Proper names and borrowings

A common source of allophonic variation is borrowed words, especially proper names, when non-Esperantized remnants of the source-language orthography remain, or when novel sequences are created in order to avoid duplicating existing roots. For example, it is doubtful that many people fully pronounce the g in Vaŝingtono ('Washington') as either [ɡ] or [k], or pronounce the ⟨h⟩ in Budho ('Buddha'). Such situations are unstable, and in many cases dictionaries recognize that certain spellings (and therefore pronunciations) are inadvisable. For example, the physical unit ('Watt') was first borrowed as ŭato, to distinguish it from vato ('cotton-wool'), and this is the only form found in dictionaries in 1930. However, initial ⟨ŭ⟩ violates Esperanto phonotactics, and by 1970 there was an alternate spelling, vatto. This was also unsatisfactory, however, because of the geminate ⟨t⟩, and by 2000 the effort had been given up, with ⟨vato⟩ now the advised spelling for both "Watt" and "cotton-wool". Some recent dictionaries, such as the Reta Vortaro, no longer even list initial ⟨ŭ⟩ in their index. Likewise, several dictionaries now list a newer spelling ⟨Vaŝintono⟩ for Washington.

References

1. Burkina, O. (2005): "Rimarkoj pri la prononca normo en Esperanto", Lingvaj kaj historiaj analizoj. Aktoj de la 28-a Esperantologia Konferenco en la 90-a Universala Kongreso de Esperanto
2. "PMEG". bertilow.com.
3. http://www.phon.ucl.ac.uk/home/wells/frazmelodio_plena.doc
4. "PMEG". bertilow.com.
5. These violations were the coining of a new diphthong oŭ, the use of ⟨ŭ⟩ as a "w" at the beginning of a syllable, and the use of geminate consonants outside of compound words.
6. In that j does not occur before the vowel i in other words, and this sequence is difficult for many people to pronounce.
7. "Bazaj elparolaj reguloj — PMEG". bertilow.com.
8. "Duonvokaloj kaj diftongoj". Lingva Kritiko.
9. The Belarusian letters ł, l represent /l, lʲ/ (phonetically [lˠ, lʲ]), and i, y represent /ji, i/ (phonetically [ji, ɨ]), so these are accounted for by the absence of palatalization.
10. Poŭpo, ŭato, and names such as Ŭakajama ('Wakayama') are more fully assimilated as pobo, vato, and Vakajama.
11. Edmond Privat, Esprimo de sentoj en Esperanto 1980:10
12. Plena Analiza Gramatiko de Esperanto 4th edition, 1980
13. Plena analiza gramatiko, §17
14. Indeed, modern Slavic writers may simply state that one obstruent assimilates to the next, even across lexical boundaries, as in subteni/supteni, okdek/ogdek, ekzemple/egzemple. (Miroslav MALOVEC, 1999, Gramatiko de Esperanto, §2.9.)
15. Orthographic gz does not occur in Esperanto, except in the nonce word egzismo itself.
16. Chris Gledhill. "Regularity and Representation in Spelling: the case of Esperanto". Journal of the Simplified Spelling Society 1994-1 pp 17–23.
17. R. Bartholdt and A. Christen, H. Res. 415 "A resolution providing for the study of Esperanto as an auxiliary language". Hearings before the Committee on Education, House of Representatives, 63rd Congress, 2nd Session 1914 March 17.