Linguistic Problems in Esperanto

The following article by Ken Miner appears here in a slightly modified version for a general readership.  Original reference is given below. 

The Esperanto world is one of those many small worlds that have more substance to them than outsiders think but less than most insiders think.  The two rarely meet.  Most linguistic attention to Esperanto is written in Esperanto itself and therefore inaccessible to outsiders.  For this reason I summarise here some of the work which I and others have done.

My affiliation with Esperanto has been somewhat unusual.  I learned the language from the age of about fourteen, but regarding the movement – the attempt to advance Esperanto as a serious solution to the world’s “language problem” – from the sixties I favoured instead an inward-turning approach.  So I acknowledged the futility of getting it “recognised” and simply wanted to nurture the language and its community of speakers.

Not being an “Esperanto salesman” (see Language Log for November 21, 2011) I was not disturbed when my later linguistic work, in part following on that of others, revealed aspects of the language well outside its usual portrayal.  Throughout most of history of the movement Esperanto has been promoted as regular and easy to learn, with intuitive word-formation, reducible to miniature booklets (called  “keys”).  These have been available in a wide range of languages and usually contain the basic grammar and lists of morphemes.  But the regularity of Esperanto is only in its inflectional morphology.  Its derivational morphology can be quite capricious.  Certain aspectual and pragmatic matters are actually un-determined.  As for ease of learning, the excellent practical grammar, Plena Manlibro de Esperanta Gramatiko [Complete Handbook of Esperanto Grammar]  by the tireless Bertil Wennergren, runs in print to nearly 700 pages. 


Esperanto is supposed to have only minor allophonic variation.  The basic principle is “one letter, one sound”.  This quite overlooks what happens in poetic elision, where final /-o/ (marking nouns) may optionally drop, being replaced in writing by an apostrophe and leaving final clusters, including voiceless obstruent + /r, m/: /kapr/, /ŝultr/, /supr/, /pastr/, /patr/, /astr/, /majstr/, /ventr/; /abism/, /skism/, /komunism/, etc.[1]  These are no problem if a vowel follows in the next syllable, but we find lines of verse like

  • El supr’ ĉiela mi ne ĵetos fulmojn
    from heavenly height I will not cast bolts
  • Kun ĉefpastr’ kapitol’-supron irados plu
    with Monsignor will continue going to the top of Campidoglio
  • Nigra aperas abism’ senradia
    black appears an abyss with no ray [of light]

The pronunciation of such lines without introducing an extra syllable is not provided for.  Presumably the sonorants will be devoiced, resulting in voiceless allophones.  Zamenhof apparently gave little thought to the results of elision (although his own usage has been examined).  Since any root or radical can be left standing, even things like line-final /latv/, /lingv/, /pugn/ and /himn/ (with non-syllabic /n/), /binokl/ (with non-syllabic /l/) etc. are theoretically possible. Admittedly most of the objectionable actual examples are from the hand of a single poet-translator, the prolific Antoni Grabowsky (1857-1921).  But since for many Esperantists the language has been chiefly or only a written one, there has been carelessness about elision.  One translator of Catullus’ line “vivamus mea Lesbia atque amemus” even came up with Ni vivu kaj ni amu, mia Lesb’  which is going to give you either /lezb/ or /lesp/, since obviously once voicing ceases within a syllable, it cannot begin again in the same syllable; this will result in morphophonemic variation, no less, since /s/, /z/, /p/ and /b/ are all phonemes.


Here is where the most interesting issues are.  They involve lexical categories and the word as a morphological unit – conflicts between what appear to have been the original ideas of Esperanto and subsequent development of the language, some of it quite early.  I will begin with lexical categories, discussed by Grimley Evans (1997) and others.  Nouns are marked by a suffix /-o/, adjectives by a suffix /-a/, verbs in the infinitive by /-i/.  Thus we have /detruo/ ‘destruction’, /detrua/ ‘destructive’, /detrui/ ‘to destroy’.  This leaves the root, /detru/, in theory a “naked root” similar to those of Semitic, supposedly unmarked for category.  However, the actual development of the language has decided against “naked roots”.  Roots turn out to belong in themselves to nominal, adjectival, or verbal categories.  This shows up when suffixes other than /-o/, /-a/, /-i/ are attached to them.  For example there is a suffix /-iĝ/ meaning ‘become whatever the root signifies’.  Take again /detru/.  So /detruiĝi/ means ‘to be destroyed’.  If /detru/ were nominal, /detruiĝi/ would mean something like ‘to become destruction’.  If /detru/ were adjectival, /detruiĝi/ would mean ‘to become destructive’.  Therefore /detru/ is inherently verbal.  Take /virg/ ‘virginal’.  /Virgo/ does not mean ‘virgin’ but if anything ‘virginity’ (though the word actually used with that meaning is /virgeco/, with /-ec/ ‘characteristic’).  A virgin is a /virgulino/ (/-ul/ ‘person’, /-in/ ‘feminine’).  Thus /virg/ is inherently adjectival.

Two realities ensue.  Contrary to what it seems Zamenhof intended, when one learns a new root one must also learn what its inherent category is, and secondly, all lexical words other than adverbs are doubly marked for category.  Thus /detrui/ is marked twice as a verb, once in the inherently verbal root and once in the suffix /-i/.  /Virga/ ‘virginal’ is marked twice as an adjective, once in the inherently adjectival root and once in the suffix /-a/.  And so on.

In addition, the categories turn out to be arbitrary.  This can be shown by selecting pairs of roots belonging to the same semantic field, such as /bros/ ‘brush’ and /komb/ ‘comb’.  Here the categories of the roots can be determined with the help of the suffix /-il/ ‘tool’.   One brushes with a /broso/ but one combs with a /kombilo/.  This shows that /bros/ is inherently nominal, while /komb/ is inherently verbal.  Yet the meanings of the roots, which are so similar, could scarcely be the reason for the difference in category.

A similar pair, again from the same semantic field, is /elegant/ and /dand/.  Here we can use the suffix /-ul/ ‘person’, mentioned above, to reveal the categories.  A /dando/ is a dandy.  A form /dandulo/ would therefore be redundant.  An elegant person however is an /elegantulo/.  This shows that /elegant/ is inherently adjectival while /dand/ is inherently nominal.

Though there is a long story behind all this (as Wennergren has extensively documented), we can explain this arbitrariness simply by saying that when roots were taken from the national languages, they were taken with their categories, not as “naked roots”.  Transitivity is also arbitrary for the same reason.  Here a useful pair from the same semantic field is /dron/ and /sufok/.  The useful suffix in this case is /-ig/, ‘cause’.  /Droni/ means ‘to drown’ and is intransitive. /Sufoki/ means ‘to suffocate’ and is transitive.  Therefore /dronigi/ meaning ‘cause to drown’ is fine while /sufokigi/ could only mean ‘to cause someone to suffocate someone else’ and I haven’t seen it used.

Most Esperantists today, especially those who are linguistically sophisticated (and many are) [2], are willing to live with arbitrary root categories and root transitivity.  However, due to compounding, lexical category and transitivity must also be properties of words.  In the case of lexical category, there is no issue. The final element of the compound determines the category of the whole – as in many languages.  Not so with transitivity.

Many simple verbs are transitive: /filtri/ ‘to filter’, /ŝovi/ ‘to shove, shift’, /ĝeni/ ‘to disturb’, /generi/ ‘to generate’.  Many simple verbs are intransitive: /babili/ ‘to chat’, /ĝemi/ ‘to groan’, /stari/ ‘to stand’, /esti/ ‘to be’, /halti/ ‘to stop’.  Some are both: /zumi/ ‘to hum’, /eskapi/ ‘to escape’, /fiŝi/ ‘to fish’, /afekti/ ‘to pretend’, /paroli/ ‘to speak’, /nebuli/ ‘to (be)cloud’, /ludi/ ‘to play’.  Since all these simple verbs consist only of root and verbal suffix, the transitivity (submorphemic) must be marked on the root: /filtr/ +tr ; /babil/ -tr ; /zum/ ±tr, etc.  

However many compound verbs show that transitivity must be marked also on the entire word.  Take for example /postkuri/ ‘to chase’ from /post/ ‘after’ and /kuri/ ‘to run’.  /Kuri/ is intransitive, but /postkuri/ is transitive.  Clearly /postkuri/ is transitive as a word.  Among similar examples are /preterkuri/, ‘to outrun’, /antaŭiri/ ‘to precede’, /ĉirkaŭiri/ ‘to go around’, /eniri/ ‘to enter’, all based on the intransitive verbs /iri/ ‘to go’ and /kuri/.  But other verbs made with /iri/ and /kuri/ are intransitive: /deiri/, ‘to start from’, /antaŭeniri/ ‘to go forward’, /foriri/ ‘to go away’, /forkuri/ ‘to run away’.  Similarly /perlabori/ ‘to earn’, /ellabori/ ‘to develop’ based on the intransitive verb /labori/ ‘to work’.

But the word is not a morphological unit in what we might call Esperantic theory.  The unit is the morpheme.  Words are in theory linearly built up.[3]  Yet from the beginning, dictionaries were compiled, and still are, based on the word.  This is necessary, because the meanings of many words (as in most languages) cannot be determined only from their component parts.  A /vortaro/, from /vort/ ‘word’ and /-ar/ ‘collection’, is not just any collection of words, but a dictionary.  Thus it appears that there has long been a tension in the language between word-based grammar and morpheme-based grammar.

Grimley Evans (1997) has shown that prefixes must sometimes modify not merely the following element, but the entire word.  One example is /alglui/ ‘to glue something to something’, from /al-/ ‘to’, /glu/ ‘glue’.  /Glu/ is an inherently nominal root.  The structure has to be [al[glui]] and not [[alglu]i], since [alglu] by itself is not meaningful.  This fortifies the notion of the word as a unit, but it leaves us with the problem of the scope of prefixes.  There is no way to predict the structure of a given word.  Some go one way, some the other.  In the following, the prefix modifies its following element (see the Appendix for a detailed explanation of all the examples):

  • ekskoloniano [[ekskoloni][ano]], not *[[eks][koloniano]] ‘excolonial’ (person)
  • ĉefministrejo [[ĉefministr][ejo]], not *[[ĉef][ministrejo]] ‘prime minister’s office’
  • plibonigi [[plibon][igi]], not *[[pli][bonigi]] ‘improve’
  • malavarega [[malavar][ega]], not *[[mal][avarega]] ‘extremely generous’


But in the following, the prefix modifies the entire word:

  • ĉefoficejo [[ĉef][oficejo]], not *[[ĉefofic][ejo]] ‘main office’
  • pludaŭrigi [[plu][daŭrigi]], not *[[pludaŭr][igi]] ‘continue further’
  • mallaborema [[mal][laborema]], not *[[mallabor][ema]] ‘lazy’
  • eksklubano [[eks][klubano]], not *[[eksklub][ano]] ‘ex-club-member’

The above examples are more or less standard words.  Their meanings are fixed by usage.  In the case of more spontaneous formations, wherein lies the oft-noted creativity of the language, one often cannot be sure of the structure and therefore of the precise meaning:

  • fibestejo: [[fibest][ejo]] or [[fi][bestejo]]? A place for contemptible animals or a contemptible place for animals?
  • fireĝido: [[fireĝ][ido]] or [[fi][reĝido]]? The son of a contemptible king or the contemptible son of a king?
  • praspecano: [[praspec][ano]] or [[pra][specano]]? A member of a primitive species or a primitive member of a species?
  • eklegigi: [[ekleg][igi]] or [[ek][legigi]]? To make someone suddenly read or to suddenly make someone read?
  • mismemorigi: [[mismemor][igi]] or [[mis][memorigi]]? To make someone remember wrongly or to wrongly make someone remember?
  • ĉeffakulo: [[ĉeffak][ulo]] or [[ĉef][fakulo]]? A specialist in the main field or the main specialist?

There are other morphological problems here and there.  A standard word is /ekfloro/ ‘a sudden flowering’ from /ek-/ ‘suddenly; incipiently’ and /flor/ ‘flower’. /Ek-/ is normally attached to verbs.  The root /flor/ is nominal, so /floro/ is a flower.  However, there is a verb /flori/ ‘to flower’.  If we insist that the root is the same in the verb and the noun, then in /ekfloro/ we have /ek-/ attached to a nominal root.  I will admit that ‘a sudden flower’ is very poetic, but /ekfloro/ does not mean that.

Much of what I have written above would seem to support the notion, now popular among thoughtful Esperantists, that Esperanto, though created as an artificial, regular and morphologically ideal language, has through its century and a quarter of use become more like a natural language.  Except in the areas of tense & aspect, telicity and pragmatics (on these see below), I would argue there is something to be said for that idea.

With respect to tense and aspect there was long a problem about compound tenses.  It survives today in places.  The passive system is not symmetrical with the active system.  This asymmetry had two solutions.  Some speakers went one way, some the other, probably influenced by their native languages.  The problem was officially resolved by the Esperanto language academy.  Probably in favour of the majority of users.  But to this day speakers differ.  Therefore one is never certain whether /mia aŭto estis riparata/ means ‘my car was repaired’ or ‘my car was being repaired’.  The issue of aspect involves a suffix /-ad/ ‘duration’ and is too complex to go into here.

A further problem that has never been resolved, however, is that of telicity or ultimate purpose.  A simple descriptive assertion like /li iris en la domon/ ‘he went / was going into the house’ for some speakers is telic, for some atelic, as can be seen from very lengthy disputes in the newsgroup soc.culture.esperanto in past years.  But telicity was only really understood after Zeno Vendler’s work beginning in the fifties (e.g., Vendler 1957).  Zamenhof and early Esperantists could not have been expected to be aware of Aktionsarten at all, even if Zamenhof had paid any attention to the linguistics of his time, which he may not have.

Another area developed only in recent times is pragmatics, including factivity, whether truth is assigned or not.  It is not clear in Esperanto which verbs and head nouns are factive in cases where their meanings do not make it obvious.  Factivity is not clear in, for example, /indigni ke/ ‘resent (it) that’, /lamenti ke/ ‘lament that’, /plendi ke/ ‘complain that’, /priplori/ ‘bemoan’, /grumbli ke/ ‘grumble that’, /koleri ke/ ‘be angry that’, /montri ke/ ‘show that’.  The entire area of pragmatics in relation to Esperanto is in need of attention.  Presumably speakers simply apply the pragmatics of their native languages to Esperanto. 

Here then are the problems in any notion of Esperanto’s acquired naturalism.  Speakers of natural languages can be presumed to know (tacitly of course) their compound tenses, which verb phrases are telic and which are atelic, and which verbs are factive.  And they have apparently mastered rich systems of pragmatics.  It is difficult to see how such problems could ever be resolved for Esperanto, even by a language academy.  A planned language has to stop somewhere, while linguistic insights into natural languages continue to develop.

Native speakers

There are people whose first language was Esperanto, and these are often referred to by movement Esperantists as “native speakers” (in part because in Esperanto the standard term for “native speaker” is /denaska parolanto/ and /denaska/ means literally ‘from birth’).  Obviously they are not native speakers.  They did not grow up in a language community speaking Esperanto, and they are not a source of grammaticality and meanings as are native speakers of natural languages.  Their situation is often no different from that of an acquaintance of my youth, whose best language was American English, but whose first language was Croatian, of which he remembered only a few phrases.  The fallacies are, apparently, (a) the notion that one’s first language is necessarily one’s native language and/or (b) the belief that one learns one’s native language primarily from one’s parents.  I have encountered two denaskaj speakers of Esperanto.  One, whose father is influential in the movement, kept up his interest in the language and blogged in it for a time, but of course his Esperanto is no better than mine.  The other also still used the language but seemed rather embarrassed at being constantly touted as a “native speaker”, and sometimes needed a dictionary.  I am told that many of these people never use the language in adulthood at all.  George and Paul Soros are said to be examples.

Esperanto was devised as an auxiliary language, therefore its ideal use would supposedly be that of a monolingual.  Yet the best Esperantists are, as one might expect, polyglot and/or linguistically sophisticated.  They quickly pick up on turns of phrase used in the national languages.  These enter Esperanto and enrich it.  Perhaps as a result of this enrichment, the language works quite well in spite of the points of vagueness I have mentioned, even among non-Europeans, especially if they are polyglot.  But even after a century and a quarter, it is still likely that the comprehension of Esperanto is best between speakers in the same branch of Indo-European: Germanic to Germanic, Romance to Romance, Slavic to Slavic, etc.  I say “likely” because, oddly enough, the comprehension of Esperanto has never been systematically examined to my knowledge at all.

Would it work in some kind of official capacity?  Probably the main problem is vocabulary.  Many thousands of words are added every year to the important national languages, mainly due to the rapid development of science and technology.  Esperanto vocabulary development is geared to a somewhat smaller world, although in theory any “international word” is automatically an Esperanto word (properly Esperantized of course).  But it seems that if there were any interest at all on the part of the EU countries in an artificial auxiliary language, there would be talk of Interlingua, which is quite widely readable for the educated even without study and which has no potentially embarrassing ideological and emotional baggage.  But to my knowledge there isn’t.


One thing that seems to offend amateur linguists about Esperanto is that it uses the pronominal form /mi/ even in the nominative, as if it were a pidgin or a creole.  In most of Indo-European, m-forms are found only in the oblique cases.  But Celtic (at least Goidelic), where the VSO order obscures it, and, for that matter, Hindi-Urdu, where the SOV order doesn’t obscure it, also have m-forms in the nominative.  In fact the pronoun “I” in Scots Gaelic happens to be identical with the Esperanto.  Some are bothered by constantly seeing /la/ ‘the’ used with nouns ending in /-o/ (since all nouns end in /-o/).  Esperanto has no grammatical gender, but in Occitan final /a/ becomes [ɔ]. So Esperanto is not as odd as it at first seems.


[1] I append here a glossary, arranged alphabetically, of the morphemes used in the examples some of which are not glossed in the text:

/abism/ ‘abyss’; /-an/ ‘member’; /astr/ ‘heavenly body’; /avar/ ‘greedy’; /best/ ‘animal’; /binokl/ ‘binoculars’; /bon/ ‘good’; /ĉef-/ ‘main’; /daŭr/ ‘duration’; /-eg/ augmentative; /-ej/ ‘place’; /eks-/ ‘ex-, former’; /-em/ ‘tending toward’; /fak/ ‘specialty’; /fi-/ ‘contemptible’; /himn/ ‘hymn’; /-id/ ‘offspring’; /-ig/ ‘cause to become’; /-iĝ/ ‘become’; /kapr/ ‘goat’; /klub/ ‘club’; /koloni/ ‘colony’; /komunism/ ‘communism’; /labor/ ‘work’; /latv/ ‘Latvian’; /leg/ ‘read’; /lingv/ ‘language’; /majstr/ ‘master’; /mal-/ ‘opposite’; /memor/ ‘remember’; /ministr/ ‘minister’; /mis-/ ‘wrongly’; /ofic/ ‘office’ (position); /pastr/ ‘pastor’; /patr/ ‘father’; /pli-/ ‘more’; /plu/ ‘further’; /pra-/ ‘early, primeval’; /pugn/ ‘fist’; /reĝ/ ‘king’; /skism/ ‘schism’; /spec/ ‘kind, species’; /ŝultr/ ‘shoulder’; /supr/ ‘above’; /-ul/ ‘person’; /ventr/ ‘stomach’

[2] Indeed there are professional linguists other than myself who are Esperantists: Jouko Lindstedt, Cyril Brosch, Marc van Oostendorp, Probal Dasgupta, John C. Wells, Ilona Koutny, Liu Haitao, Detlev Blanke, to name a few.  Amateur linguists have contributed a great deal to most of the topics I discuss here.  This article is bibliographically weak in that I no longer remember who pointed out what.  As far as I know, mine are: (1) the phonological observations, (2) the claim that wide scope of prefixes is quite common, (3) noting uncertainty of structure in nonce forms, (4) the points about tense and aspect, telicity in general, factivity and general pragmatics, and (5) the point about “native speakers”.  For the rest I cannot claim originality.

[3] There seems to have been much dependence on “logic” in the early days.  In 1913 the Academy accepted a proposal of René de Saussure (the brother of Ferdinand de Saussure, often called the father of structural linguistics), to the effect that “word-formation in Esperanto is founded on the logical and direct construction of each separate word and not on so-called rules of derivation.”  Significantly, in early writings on Esperanto grammar, morphemes were called “words”.

Appendix: details of examples

ekskoloniano [[ekskoloni][ano]], not *[[eks][koloniano]] ‘excolonial’ (person)

The first bracketing denotes a member of an entity which used to be a colony.  The second bracketing denotes someone who used to be a member of a colony.  A native of Congo would be designated by the first; a former native of Anguilla would be designated by the second.

ĉefministrejo [[ĉefministr][ejo]], not *[[ĉef][ministrejo]] ‘prime minister’s office’

The first bracketing denotes a place where a prime minister lives or works.  The second bracketing denotes the main abode or workplace of a minister or ministers.  10 Downing Street would be designated by the first; 70 Whitehall would be designated by the second.

plibonigi [[plibon][igi]], not *[[pli][bonigi]] ‘improve’

The first bracketing denotes the activity of making better; the second bracketing denotes the activity of augmenting an improvement.  If I add a wing to my house, I am doing the first.  If I add floor heating to that wing, I am doing the second.

malavarega [[malavar][ega]], not *[[mal][avarega]] ‘extremely generous’

The first bracketing describes someone who is very generous.  The second describes someone who is abstemious to the point of sainthood (the opposite of rapacious).  Warren Buffet might be designated by the first; St. Francis of Assisi might be designated by the second.

ĉefoficejo [[ĉef][oficejo]], not *[[ĉefofic][ejo]] ‘main office’

The first bracketing denotes a main office.  The second denotes the abode or workplace of the main office-holder. The main office of a corporate headquarters would be designated by the first; the Oval Office in the White House would be designated by the second.

pludaŭrigi [[plu][daŭrigi]], not *[[pludaŭr][igi]] ‘continue further’

The first bracketing denotes the activity of continuing further to make something last.  The second denotes the activity of making something last longer. Continuing a filibuster would be an example of the first; painting a house would be an example of the second.

mallaborema [[mal][laborema]], not *[[mallabor][ema]] ‘lazy’

The first describes someone who is lazy; the second is really not a possibility since [mallabor] suggests that there is an activity that is the opposite of work – play perhaps; but /mallabori/ is not used with that meaning (/ludi/) other than in jest.

eksklubano [[eks][klubano]], not *[[eksklub][ano]] ‘ex-club-member’

The first denotes a person who was formerly a member of a club. The second denotes a member of an entity which was formerly a club.  The British Prime Minister, David Cameron, formerly a member of the Bullington Club, is an example of the first.  If the Sierra Club becomes incorporated, its members will be examples of the second.

If some of the distinctions seem hazy, there clearly are differences in sense, even if their referents were to be the same.


Grimley Evans, Edmund, “Lingvaj notoj” [language notes], La Brita Esperantisto (Mar-Apr 1997, pp 57-59).

Van Dijk, Ziko, Esperanto Sen Mitoj [Esperanto Without Myths], Flandra Esperanto-Ligo, 1999.  Unfortunaely not yet available in English.

Vendler, Zeno. “Verbs and Times”, The Philosophical Review 66(2):143-160 (1957).

How to cite the original post from which the above has been extracted.

Miner, Ken. 2015. Esperanto: some observations of a speaker-linguist. History and Philosophy of the Language Sciences

Ken Miner (b. 1936) formerly of University of Kansas