Long-Distance Dependencies and Clause Boundaries | Indonesian Grammar

A long-distance dependency links an expression at the edge of one clause to a role interpreted inside another clause. Indonesian allows some dependencies across selected complements, especially when the embedded role is a subject or patient made accessible by voice. It does not allow the local head + yang + gap pattern described in extraction and gaps to extend through unlimited clause depth without cost.

Categorical grammar is only part of the picture. Complement type, voice, the number of intervening subjects, discourse givenness, and processing load all shape judgments. A sentence one speaker accepts in formal prose may lead another to prefer a passive, a resumptive topic, or two shorter sentences.

Local, one-boundary, and deeper dependencies

Begin with a controlled set whose head is the understood subject of akan memimpin tim.

Itulah calon yang akan memimpin tim penelitian.

That is the candidate who will lead the research team. (local dependency)

Itulah calon yang kami kira akan memimpin tim penelitian.

That is the candidate who we think will lead the research team. (one complement boundary)

Itulah calon yang kami dengar diperkirakan akan memimpin tim penelitian.

That is the candidate who we hear is expected to lead the research team. (deeper dependency)

All three keep the head calon as the understood subject of the lowest predicate. The final example is grammatical in careful formal interpretation, but heavier: calon must remain active through kami dengar and passive diperkirakan before reaching akan memimpin.

Compare a version that repeats the participant explicitly:

Tentang calon itu, kami mendengar bahwa panitia memperkirakan bahwa dia akan memimpin tim penelitian.

As for that candidate, we hear that the committee expects that he/she will lead the research team. (formal/careful)

This is a topic chain with resumptive dia, not a single relative gap. It is longer but easier to process and makes the lower subject explicit.

💡

Increasing depth does not merely add words. Each clause boundary creates another place where the listener must preserve the head's role and resist assigning it to an intervening predicate.

Relativization from a complement

Bridge-like predicates of thought, knowledge, expectation, and report can host a complement containing the role associated with a relative head.

Orang yang saya kira tinggal di rumah sudut itu ternyata seorang dokter.

The person I thought lived in the corner house turned out to be a doctor.

Kebijakan yang kami yakini akan mengurangi kemacetan mulai diuji bulan depan.

The policy we believe will reduce congestion begins testing next month. (formal)

Calon yang diperkirakan akan memperoleh suara terbanyak berasal dari wilayah timur.

The candidate expected to receive the most votes comes from the eastern region. (formal/journalistic)

The relative heads are embedded subjects: the person lives, the policy reduces, and the candidate receives votes.

Patient dependencies need compatible embedded voice:

Laporan yang kami yakin telah diperiksa auditor itu masih belum diterbitkan.

The report we are sure the auditor has examined has still not been published.

Usulan yang saya dengar akan dibahas dewan minggu depan mendapat banyak dukungan.

The proposal I hear the council will discuss next week has received broad support.

The first uses embedded telah diperiksa auditor; the second uses passive akan dibahas dewan. An actor-voice embedded object gap would be substantially worse in neutral Standard Indonesian.

Dokumen yang kami tahu sudah ditandatangani direktur disimpan di lemari besi.

The document we know the director has signed is kept in the safe.

This (formal/careful) pattern is natural because passive voice keeps the patient head structurally accessible across the complement.

Minimal voice sets

The following set holds meaning approximately constant while changing the embedded voice.

Direktur mengatakan bahwa tim hukum menyusun peraturan itu.

The director says that the legal team drafted that regulation.

Peraturan yang menurut direktur disusun tim hukum itu mulai berlaku besok.

The regulation that, according to the director, was drafted by the legal team takes effect tomorrow.

Peraturan itu, menurut direktur, disusun tim hukum setelah konsultasi publik.

That regulation, according to the director, was drafted by the legal team after public consultation.

The relative and topic versions do not preserve the actor-voice complement unchanged. Passive disusun promotes the regulation. The parenthetical menurut direktur avoids an unnecessary deep gap and is often the most natural formal repair.

With a first-person agent, object voice can occur at the lowest clause:

Artikel yang Rina tahu sudah saya revisi itu akan terbit bulan depan.

The article Rina knows I have revised will be published next month.

Artikel itu Rina tahu sudah saya revisi sebelum tenggat.

Rina knows that I revised that article before the deadline. (topic chain)

The second is more marked in careful writing and benefits from an intonational boundary after artikel itu. A clearer formal form is:

Rina tahu bahwa artikel itu sudah saya revisi sebelum tenggat.

Rina knows that I revised the article before the deadline.

Questions across a complement boundary

Long-distance questions are clearest when the questioned phrase corresponds to an embedded subject.

Siapa yang kamu kira akan menggantikan kepala cabang?

Who do you think will replace the branch head?

Calon mana yang Anda yakin mampu menjalankan program tersebut?

Which candidate are you sure can carry out that program?

For an embedded patient, Indonesian often combines a fronted question phrase with passive or object voice:

Proposal mana yang Anda dengar akan dibahas panitia sore ini?

Which proposal did you hear the committee would discuss this afternoon?

Berkas mana yang Rina bilang sudah kamu kirim ke kantor pusat?

Which file did Rina say you had sent to head office? (neutral spoken)

The latter uses spoken bilang (informal/neutral spoken) and first/second-person object voice kamu kirim. Formal prose would prefer a menurut Rina frame or a full indirect report with bahwa.

Question acceptability declines when the dependency crosses several complements or a nonselected adjunct. Repeat the relation rather than assuming the first yang or wh phrase can control every later gap.

Topic chains are not necessarily extraction

Indonesian discourse can establish a topic and maintain it through pronouns, clitics, or omission across several clauses.

Laporan audit itu, direktur mengatakan bahwa tim hukum masih memeriksanya.

That audit report—the director says that the legal team is still examining it.

Masalah perizinan itu, kami sudah meminta kementerian menjelaskannya secara tertulis.

That licensing issue—we have asked the ministry to explain it in writing.

Proposal baru itu, menurut ketua, belum sempat dibahas dalam rapat kemarin.

That new proposal, according to the chair, did not get discussed at yesterday's meeting.

The first two use resumptive -nya. The third uses a passive gap whose topic is easily maintained. These sentences create aboutness across clauses; they need not involve one syntactic movement operation.

💡

Distinguish a syntactic gap from discourse resumption. If an overt pronoun or -nya supplies the lower role, the initial phrase is a topic linked by reference, not the filler of an empty object position.

Reported speech boundaries

Indirect reported content can participate in a long-distance dependency when its syntax remains integrated.

Kebijakan yang menurut menteri akan diterapkan tahun depan masih dikaji.

The policy that, according to the minister, will be implemented next year is still under review. (formal)

A clearer formal rewrite separates the report from the policy's present status:

Menteri menjelaskan bahwa kebijakan itu akan diterapkan tahun depan. Saat ini, kebijakan tersebut masih dikaji.

The minister explained that the policy would be implemented next year. At present, the policy is still under review. (formal)

Direct quotation is a stronger boundary. Words inside quotation marks are presented as an independent utterance; ordinary extraction does not reach into them.

Tentang kebijakan itu, menteri berkata, “Kami akan menerapkannya tahun depan.”

Regarding that policy, the minister said, ‘We will implement it next year.’

The external topic is resumed by -nya inside the quote. This is a discourse arrangement, not a relative gap across quotation.

Menteri berkata bahwa kebijakan itu akan diterapkan tahun depan.

The minister said that the policy would be implemented next year.

Choose indirect speech when the content must integrate syntactically with the surrounding sentence.

Complex noun phrases and adjuncts resist dependency formation

A gap buried inside a relative clause, a noun complement, or an adverbial clause is difficult to process and often unacceptable. Rather than extract from such an island-like domain, name the relationship.

Tentang buku itu, saya bertemu penerjemah yang mengerjakan edisi Indonesianya.

As for that book, I met the translator who worked on its Indonesian edition.

Karena laporan itu belum selesai, rapat evaluasi kami tunda.

Because that report was not finished, we postponed the evaluation meeting.

Laporan itu belum selesai. Karena itu, rapat evaluasi kami tunda.

The report was not finished. Therefore, we postponed the evaluation meeting.

The first uses a topic plus possessive resumption inside a relative. The latter two keep a whole reason clause intact rather than extracting its internal noun.

Why judgments vary

Several factors improve a long dependency:

the embedding predicate readily selects a proposition (kira, tahu, yakin);
the head is definite and already discourse-active;
the embedded role is subject or passive patient;
few overt subjects intervene;
the sentence is read in a supportive formal context.

Several factors make it worse:

repeated embeddings with similar verbs;
an object gap under actor voice;
extraction from a relative, adjunct, or quotation;
competing referents and unclear voice;
an unsupported sentence presented in isolation.

There is no honest one-line rule that predicts every speaker's judgment. At C2, the practical skill is recognizing the structural pressure and selecting a repair that preserves meaning.

Common Mistakes

Extending a local yang pattern through unlimited depth

❌ Inilah laporan yang saya dengar Rina bilang direktur menyatakan tim hukum memeriksa.

Severely degraded—the patient gap crosses several clauses under an actor-voice predicate.

✅ Inilah laporan yang menurut direktur sedang diperiksa tim hukum.

This is the report that, according to the director, is being examined by the legal team.

The parenthetical source and passive predicate remove unnecessary dependency depth.

Keeping actor voice at the embedded patient gap

❌ Dokumen yang kami yakin auditor memeriksa itu belum diterbitkan.

Incorrect neutral structure—the embedded actor voice does not license the patient gap.

✅ Dokumen yang kami yakin telah diperiksa auditor itu belum diterbitkan.

The document we are sure the auditor examined has not been published.

Extracting from direct quotation

❌ Kebijakan yang menteri berkata, “Kami akan menerapkan,” mulai berlaku besok.

Incorrect—the relative dependency cannot leave a gap inside an independent direct quotation.

✅ Tentang kebijakan itu, menteri berkata, “Kami akan menerapkannya mulai besok.”

Regarding that policy, the minister said, ‘We will implement it starting tomorrow.’

Calling a resumptive topic an empty gap

❌ Dalam ‘Laporan itu, tim masih memeriksanya’, objek memeriksa kosong.

Incorrect analysis—the object position is filled by attached -nya.

✅ Laporan itu, tim masih memeriksanya bersama konsultan luar.

That report—the team is still examining it with an outside consultant.

Ignoring processing ambiguity

❌ Semakin banyak klausa selalu membuat hubungan yang semakin jelas.

Incorrect—additional embeddings increase memory load and competing analyses.

✅ Pernyataan itu kami pecah menjadi dua kalimat agar sumber dan isinya jelas.

We split that statement into two sentences so its source and content would be clear.

Key Takeaways

Indonesian permits some dependencies across selected complements, especially for embedded subjects and passive patients.
Voice repair is crucial; actor-voice object gaps degrade rapidly across a boundary.
Topic chains with pronouns or -nya are reference dependencies, not empty-gap extraction.
Direct quotation, relatives, complex noun phrases, and adjuncts create strong boundaries.
Acceptability depends on categorical syntax, processing load, discourse support, and speaker preference; splitting is often the best formal repair.

Now practice Indonesian

Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.

Start learning Indonesian →