How can syntactic ambiguity with pronouns be avoided?

Question

Due to the way pronouns work in place of any particular noun (subject or object) in the sentence, this often leads to ambiguous grammatical constructions.

Take this phrase in English for example:

Adam sent the essay to James to help him with his writing

Do him and his refer to Adam (where James will help edit Adam's essay or give him suggestions) or James (where Adam is sending James an exemplar to help James see examples of good writing)?

How would ambiguities like this be avoided in constructed languages? I know a simple solution is to just eliminate pronouns from the language but is there an alternative solution to allow pronouns to be used unambiguously?

Answer 1

Here are three options you might want to consider:

Noun classes assign each noun a class (well known examples include the gender systems of Indo-European languages, and also the more elaborate systems of Bantu languages). Having pronouns agree with the noun class of the nouns talked about greatly helps reduce ambiguity, though it would likely not help in your specific example with two male human participants.

Obviation is the system of having two distinct third person pronouns, one (the proximal) referring to the most salient (important, topical) referent and one (the obviative) to less salient ones. In your example, likely Adam would be the proximal, and James the obviative. There are nice examples on the relevant Wikipedia article.

A third option, which is to my knowledge only attested in signed languages is Indexing. This is basically the human language equivalent of variables in programming languages: a new variable, usually a location in the sign space, is assigned to every important referent, and referred to by pointing at the location or otherwise involving the location during signing of relevant words. I believe Lojban does something similar to this as well.

Answer 2

In addition to the options mentioned by Adarain and Jan, various reflexives and reflexive-like operations can often be of use in dealing with such situations. English already has some reflexives, providing some amount of disambiguation, Danish goes a little further has a compulsory reflexive/non-reflexive distinction in 3rd person possessives, and while the specific example you mentioned doesn't work well in translation, a similarly problematic sentence "Adam sent James his essay" can be easily somewhat disambiguated with this system:

Adam_i sendte James_j sin_i stil

Adam_i sendte James_j hans_j(/k) stil

It is possible for reflexive systems to be significantly more broad than this, rather than just referring to antecedents in the same clause, even if it's omitted via some other rule, e.g. in "Jonh saw Adam and shot himself in the foot", where there is no overt reference to John in the second clause, because it's been omitted due to subject coreference (such deletion rules, when constraints are put on them is actually another way of reducing ambiguity, I have relatively recently written a rather long forum post on this which goes into much more detail about them), it's possible to have "long-distance reflexives" which take antecedents outside of the clause.

In Mandarin, the reflexive ziji can refer to both local and non-local antecedents in some cases:

Zhangsan renwei Lisi kan-bu-qi ziji
Zhangsan think Lisi look-not-up REFL
"Zhangsan_i thinks Lisi_j looks down on self_i/j"

Certain things block this though, for example shifts in perspective:

Zhangsan renwei wo zhidao Wangwu xihuan ziji
Zhangsan think I know Wangwu like REFL
"Zhangsan_i thinks I_j know Wangwu_k likes self_*i/*j/k"

In Igbo, in complements of communication, entities that are coreferential with the source of information are marked by what is known as "logophoric pronouns":

ọ́ sị̀rị̀ nà ọ́ byàrà
he said that he came
"he_i said that he_j came"

ọ́ sị̀rị̀ nà yá byàrà
he said that LOG came
"he_i said that he_i came"

Gokana also has such a system, though the marking is on the verb, though the constraints on what can be coreferential are quite free:

aè kɔ aè dɔ
he said he fell
"he_i said that he_j fell"

aè kɔ aè dɔ-ɛ
he said he fell-LOG
"he_i said that he_i fell"

aè kɔ oò div-èè e
he said you hit-LOG him
"he_i said that you hit him_i

aè kɔ oò ziv-èè a gĩ́ã́
he said you stole-LOG his yams
"he_i said that you stole his_i yams"

The wikipedia article on logophoricity goes into more depth.

It is also possible to have two different reflexives, requiring respectively local and non-local antecedents. Danish has a rather limited case of this in bare complements to perception verbs, where the two reflexives sig and sig selv which are usually either only different in level of emphasis or in complementary distribution, respectively require non-local and local antecedents:

Holger hørte Peter tale om sig.
"Holger_i heard Peter_j talking about him_i."

Holger hørte Peter tale om sig selv.
"Holger_i heard Peter_j talking about himself_j."

Holger hørte Peter tale om ham.
"Holger_i heard Peter_j talking about him_k."

"4th Person systems" such as seen Eskimo languages are rather similar, though more general as they are used in all subordinate clauses are also another type of reflexive-like construction (the "fourth" person is often labeled 3R) and they are also used intraclausally in possessive marking. In all of these cases they mark that a participant or possessor is coreferential with the subject of the main clause (which is defined in terms of S/A despite the case marking being ergative). An example from Siberian Yupik:

esghaghyagu quyaaq
see:CNSQ:3s>3s happy:IND.3s
"when he_i saw him_j, he_k was happy"

esghaghyamigu quyaaq
see:CNSQ:3Rs>3s happy:IND.3s
"when he_i saw him_j, he_i was happy"

esghaghyatni quyaaq
see:CNSQ:3s>3Rs happy:IND.3s
"when he_i saw him_j, he_j was happy"

In addition to all these various reflexive constructions, switch reference, the overt marking of whether the "subject" is either the same or different, either between two coordinate clauses, or a subordinate and main clause can also resolve some ambiguity. Take for example this pair of sentences from Hua:

ebgi-Ø-na korihie
hit-SAME-3sg.ANTICSU ran.away.3sg
"he_i hit him_j and he_i ran away"

ebgi-ga-na korihie
hit-3sg.DIFF-3sg.ANTICSU ran.away.3sg
"he_i hit him_j and he_j ran away"

Switch reference can only deal with one set of coreference though what exactly is tracked varies. In some languages, it is specifically the actor that is tracked, while others may track the topic or some other pragmatically prominent NP, or may even differ in what is tracked in the controlling clause as opposed to the clause recieving the marking. Additionally the marking may occasionally also be sensitive to changes in things like time, place, discourse coherency and/or reality status, and as a result, switch reference may fulfill may other roles than simply dealing with ambiguity. This paper, particularly chapters 5 and 6, goes into quite a bit of detail about the highly varied and interesting usage of switch-ref in Papua New Guinean languages.

Answer 3

In addition to Adarain’s answer, the ambiguity can also be resolved in an out-of-the-box way. For example, English can use the former or the latter in place of an inflected form of he to distinguish between the two cases. Languages like German go a step further and would just use this instead of the latter in most cases. These words do not replace pronouns, they just complement them. This gives the nice feature of preserving a possible ambiguity where it is intended for poetic or dramatic purposes while removing it entirely where clarity is desired.

Another out-of-the-box way is to restrict the way in which such constructions can be constructed. Trying to think of how German handles the situation I realised that the sentence translated directly is unambiguous in German:

Adam hat James seinen Aufsatz geschickt, um ihm beim Schreiben zu helfen.

The only way this can be interpreted is Adam helping James just by the way the grammar works. Adam is the sentence’s subject and the infinitive construction can only be tied to the subject. If you wanted James helping Adam, you need to work around by using a subordinate clause:

Adam hat James seinen Aufsatz geschickt, damit er ihm beim schreiben hilft.

Answer 4

For a conlang, it seems like one other way to avoid ambiguity that I do not see noted in previous answers is to have the language enforce grammatical rules such that a pronoun must always refer to a specific type of referent. That is, consider these four possible types of grammar rules (they may be other rules a language may follow):

1. A pronoun always refers to the last referent, any mention of a prior referent must be renamed. So in your example, the grammar would remove the ambiguity as such a rule would refer to James in your original example:

Adam sent the essay to James to help him with his writing

And if Adam was the intent, then it would need to be one of these:

Adam sent the essay to James to help Adam with his writing

or

To James, Adam sent the essay to help him with his writing

2. A pronoun always refers to the far referent when two referents are in view, and is left off when the near referent is intended. So of James getting help, no pronoun is used, and the sentence would roughly translate simply as:

Adam sent the essay to James to help with writing

But of Adam, the pronoun would be used:

Adam sent the essay to James to help him with his writing

3. A pronoun always refers back to a subject, never an object (which must be restated if intended). So of James:

Adam sent the essay to James to help James with James's writing

Though I suspect something like #2 for James might be stated as well; or the language would form such a statement to place the main player in the subject position in a passive construction:

James was sent an essay by Adam to help him with his writing

For Adam, it would be as you originally stated:

Adam sent the essay to James to help him with his writing

4. A pronoun always refers back to an object, never the subject (which is restated if intended). So of James it is as you already had:

Adam sent the essay to James to help him with his writing

But for Adam, it might be this:

Adam sent the essay to James to help Adam with his writing

Notice in that previous example by restating "Adam" as the object of "to help," it then put Adam in the most recent object position and allows for the "his" to then refer to Adam (who was also the subject).

So for some of these grammatical ways of reducing ambiguity, the sentence structure in the language becomes more important (which may or may not be desirable in the language one is constructing).

Answer 5

Here are some solutions, roughly in descending order of naturalness.

1. distance-specific determiners or inflections
2. topic-prominence and a dedicated pronoun for referring back to the topic
3. switch-reference on nonfinite verbs
4. pro-forms incorporating some feature such as the first sound of a word.

---

1 distance-specific determiners or inflections

You can solve this problem in a reasonably natural way using demonstratives or distance-specific determiners. This is similar to proximate and obviative pronouns, but not identical.

For instance, suppose all non-vocative uses of names require some kind of definite determiner. And suppose there are proximal and distal determiners ("this" and "that"). I'm also assuming that these determiners can form NPs without an accompanying noun, like in English.

then you could say something to the effect of

this Adam sent essay to that James to help that with that's writing.

There's a lot of variation in existing natural languages when it comes to contrasts in demonstratives. Some languages surface distinctions such as visibility (e.g. Malagasy) or audibility (e.g. Khaling, a Sino-Tibetan language from Nepal) or altitude/elevation.

---

2 topic-prominence and a dedicated pronoun for referring back to the topic

The reflexive pronoun solution described above normally tracks co-reference with the syntactic subject of the matrix clause.

If that solution is not flexible enough, you can track co-reference with the topic instead and explicitly mark the topic in some way.

For example, suppose James is the topic and TOP is the pronoun that refers to the topic. Let the postposition/clitic は mark the topic of a clause.

to James は Adam sent essay to help TOP with TOP's writing.

Or you could do something like the following if you want は to obscure the role of the topicalized noun.

James は Adam sent essay to help TOP with TOP's writing.

As far as I know, no topic prominent natural languages actually do this.

---

3 switch reference on non-finite verbs

Switch reference marks whether the subject of a clause is the same as the previous clause or not. Suppose each verb is marked SS for same subject or DS for different subject. I'll assume utterance-initial words are marked with DS.

This requires some care in defining what exactly "previous clause" means.

Adam sent.DS essay to James to help.SS him with his writing.DS .

so in this sentence, sent.DS marks that sent is in an utterance-initial clause. help receives a same subject marker because help and sent have the same subject, sort of, and writing has a DS marker because James is writing, not Adam.

Getting the details right on how exactly you want pervasive switch-reference to work might be a bit tricky.

---

4 pro-forms incorporating some feature such as the first sound of a word.

Suppose you have some forms consisting some constant + ona, which can refer back to a noun beginning with that particular consonant sound (or ona itself if the word begins with a vowel). Then you can do something like:

Adam sent essay to James to help JONA with JONA's writing.

the choice of JONA over ONA means that James is the one whose writing we're attempting to improve.

Answer 6

I was thinking about the ease of Blissymbols in this context ... although Charles K. Bliss never mentions the case, one could imagine that the person symbol can be followed by an index 4 (or even 5) ... allowing in all cases just one reading ... but with some problems for doing so out loud ...

There is a nice Facebook post from Blissymbols about pronouns at https://www.facebook.com/298159500318475/photos/a.298161243651634/1360473830753698/?type=3

... I made an archive for acccessibility : https://archive.is/PI4PY