# What is the word “li” in Toki pona, grammatically?

In Toki pona, words are never inflected, and many words can act as a noun, verb or adjective in different contexts. The word "li" usually separates the subject from a predicate, and “e” goes before the object (in SVO order). With soweli (animal), moku (eat, food) and kili (fruit), one can say:

soweli li moku e kili

The animal eats the fruit.

This is the usage listed in the dictionary section of the official book, Toki Pona: the Language of Good:

li. PARTICLE (between any subject except mi alone or sina alone and its verb; also to introduce a new verb for the same subject)

But there is another usage, as seen in phrase:

soweli li ijo

Animals are things (ijo means thing).

It seems that this usage increases the ambiguity, because

soweli li moku

can mean both ”The animal is eating” (with moku as an intransitive verb), or ”animals are food”.

What role does the li actually have in this phrase?

Li in these examples could be described as a copula - this is the name for a word whose function is to link a subject and predicate.

• I guess the lack of an explicit verb for “to be” is what makes this confusing. Are you saying li is a copula both when it means ”animals are eating” and ”animals are food”? In the first case, li goes between the subject and verb, in the second case there is no visible verb, just “animal li food” - it's more like in this case, li means “to be”? Or is it maybe “moku” that goes from meaning “eating” to mean ”be food”, rather than just “food”? – peterorme Feb 22 '18 at 8:29
• "To be" does function as a copula in English (among other things), but I would not say that this means li means "to be", as it is not a verb and is not used in the same ways. Many languages lack a verb for "to be", and Toki Pona is one of them. However, it serves a similar function when in a NOUN = NOUN context. – Sparksbet Feb 22 '18 at 14:07

Mapping word class jargon from one language to another is a dicy project. Lojbanists decided that they needed to invent entirely new words for word class jargon. That said...

All the toki pona particles (o, li, e, pi, la) act most similar to Japanese particles or English clitics, eg. "the". Clitics bind to a phrase. There are other tests for what counts as a clitic.

They are semantically bleached, do not act like content words. "li" in particular introduces a noun phrase or a verb phrase.

Calling li a particle or predicate marker is correct and not misleading. A Toki Pona clause can have more than one predicate. li never marks a non-predicate. It also appears on every predicate except when that predicate is the first one in the clause and the subject is exactly mi or exactly sina (101) or when preceded by the vocative/imperative particle o (102).

I think a good way to analyze this would be to say that every indicative predicate obligatorily has a predicate marker, but sometimes it's null.

mi  moku  li   pakala. (101)
I   food  LI  mistake.
I eat and destroy.

o   lukin     e    ni. (102)
IMP   see  DObj  that.
watch this.


Here's a fragment of (my best guess of) the Toki Pona grammar. I haven't seen any examples of stacking imperative phrases, so I don't know whether look and say would be o lukin li toki or o lukin o toki.

Also, in Toki Pona verb phrases and noun phrases are different syntactically. For instance, *sina moku e soweli ala soweli. cannot be used to mean are you a vegetarian.

LI        := 0 | "li"
PredInd   := LI VP {PP} | LI NP
ClauseInd := NP [PredInd]
PredImp   := "o" VP {PP} | "o" NP
ClauseImp := NP PredImp | PredImp


I'm not thrilled with this analysis because of the asymmetry between indicative and imperative clauses. Unifying the clause types, however, would predict that *li moku. is valid. As far as I can tell, Toki Pona does not permit null subjects in indicative clauses.

I can think of two potential alternative analyses for li: A) it's a pronoun (like in Example 4 Longgu on this WALS article or B) it's a preposition. I'll try to show that both of those analyses lead to bad predictions.

A) li as pronoun.

This falls apart for a couple of reasons. It predicts that mi moku li pakala would be invalid and that the correct form would be mi moku mi pakala. In reality, both forms are fine.

li as a pronoun also predicts that *li moku would be valid since it would contain an explicit subject under that analysis.

Also, *mi moku li. cannot be used to mean I eat it.

B) li as preposition.

If li were a preposition, (103) and (104) would both be valid.

Mi pakala e moku. (103)
I ruin the food.

*Mi e moku li pakala. (104)
*I ruin the food.


If li were a preposition, the following would be invalid.

Ona mute li toki ala toki? (105)
Are they speaking?


The syntactical li structure was inspired by Toki Pisin ...

and in an online Tok Pisin grammar I found :

[...] Tok Pisin has its own grammatical rules.

First of all, look at the following sentences:

Mi wok. 'I worked.' Yu wok. 'You worked.' Em i wok. 'He/she worked.' Tom i wok. 'Tom worked.'

Note that the last two sentences have the little word i before the verb. (Remember that in Tok Pisin, i is pronounced something like "ee".) This little word is called a "predicate marker", and it must occur in a sentence when subject is em or a noun (like "Tom" or "the bicycle"). This rule is certainly different from anything found in English.

http://www.hawaii.edu/satocenter/langnet/definitions/tokpisin.html

So I presume that 'li' is also a "predicate marker" ...

• I hope participants here need not be told to pronounce the letter ‹i› as IPA /i/ by default! – Anton Sherwood Oct 14 '18 at 4:38
• I'd guess that i marker evolved from English 3rd person pronouns? Of course, that doesn't mean it's still a pronoun in Tok Pisin, since reanalysis is very much a thing in language evolution. – Ilmari Karonen Mar 28 '20 at 21:38

In some sentences it's a lot like 'は' in Japanese 'タナカさん先生です' Mr Tanaka is a teacher. The toki pona equivalent would be 'jan Tanaka li jan pi pana sona'.