ELONIO
ELONIO
Mission—Team—Privacy—Terms—Sponsors—Grammar—Typing Course—Donate
© 2026 Elon Automation B.V.
  1. Grammar
  2. /Icelandic Grammar
  3. /Information Structure
  4. /Object Shift: Holmberg's Generalisation in Detail

Object Shift: Holmberg's Generalisation in Detail

You already know the basic shape of object shift from syntax/object-shift: an unstressed object pronoun jumps leftward past ekki (ég sá *hann ekki), a full noun phrase does not (ég sá ekki manninn*), the shift is obligatory for an unstressed pronoun, and stress can cancel it. This page does not repeat that. It takes the one fact that turns object shift from a curiosity into one of the most-studied phenomena in Scandinavian syntax: object shift is not an independent rule about pronouns — it is parasitic on verb movement. A pronoun may shift past the negation only when the main verb has itself left the verb phrase. That single conditional — Holmberg's Generalisation — predicts when the shift happens, when it is blocked, and why the same word behaves in opposite ways in two sentences that look almost identical. We will state it precisely, derive the compound-tense blocking from it, separate obligatory pronoun shift from optional full-NP scrambling, and show how focus stress fits in.

Holmberg's Generalisation, stated precisely

The loose version — "object shift needs the verb to have moved" — is enough to get the right answers, but the precise version is sharper and worth internalising, because it explains cases the loose version misses. Holmberg's Generalisation (HG) says:

An object cannot shift across any phonologically overt material that the verb itself has not crossed.

In other words, the shifted object is not allowed to "overtake" anything that the verb stayed behind. If the verb has moved up and left the negation behind it, the pronoun may follow it up over the negation — they crossed the same thing. But if a phonologically overt element (a supine main verb, a particle, an indirect object that itself didn't shift) sits between the landing site and the pronoun, and the verb did not cross that element, the pronoun cannot cross it either. Object shift is licensed only in the wake of the verb's own movement.

💡
The precise rule, in one line: a shifted object may not cross anything overt that the verb did not also cross. Object shift rides in the slipstream of verb movement — it can go exactly as far as the verb has cleared the way, and no further. This is why a single principle predicts both the shift in sá hana ekki and its absence in hef ekki séð hana.

Simple tense: the main verb moves, so the pronoun shifts

In a simple tense the finite main verb itself raises to the second (V2) position: sá, las, þekki, hitti all sit up high, having left ekki and the adverbs below them. The verb has crossed the negation. By HG, the object pronoun is therefore free to cross the negation too — and because pronoun shift is obligatory for an unstressed pronoun, it must. The result is verb → pronoun → ekki.

Ég sá hana ekki í allan dag.

I didn't see her all day. — simple past: the finite main verb 'sá' has raised past 'ekki', so the pronoun 'hana' may (and must) follow it over 'ekki'.

Hann þekkir mig ekki lengur.

He doesn't know me anymore. — finite main verb 'þekkir' high; unstressed 'mig' shifts past 'ekki' in its wake.

Við skiluðum henni aldrei peningunum.

We never gave her back the money. — the dative pronoun 'henni' shifts past 'aldrei'; the finite main verb 'skiluðum' has cleared the way.

The key is which verb is finite. In the simple tense it is the lexical verb that does the work, and that verb is the one that has moved. The pronoun shifts because its own selecting verb went up first.

Compound tense: the main verb stays low, so the shift is blocked

Now the heart of the matter. In a compound tense the finite slot is filled by an auxiliary — hef, hafði, mun, get, var (að) — and the lexical main verb appears as a supine or infinitive that stays down low inside the verb phrase: séð, lesið, talað, gert. The element that has raised is the auxiliary, which is not the verb that governs the object. The governing verb (the supine) has not crossed the negation. By Holmberg's Generalisation, the object pronoun therefore cannot cross the negation either: it is stranded with its supine, to the right of ekki.

TenseFinite slotObject-selecting verbCrossed ekki?Pronoun shifts?
Simple (sá)main verb itselfsá (high)YesYes — sá hana ekki
Compound (hef … séð)auxiliary hefsupine séð (low)NoNo — hef ekki séð hana

Ég hef ekki séð hana í allan dag.

I haven't seen her all day. — perfect: the supine 'séð' stays low and never crossed 'ekki', so the pronoun 'hana' cannot cross it either — it sits after the supine. No shift.

Hann hefur aldrei þekkt mig almennilega.

He has never really known me. — the supine 'þekkt' is low; 'mig' is stranded after it, right of 'aldrei'.

Við munum ekki skila henni peningunum.

We won't give her back the money. — future with 'munu': the infinitive 'skila' is low, so 'henni' stays right of 'ekki', after 'skila'.

Look at the minimal pair the two tenses make. Sá *hana ekki — pronoun before *ekki. Hef ekki séð hana — the same pronoun hana, the same negation, but now after ekki and after the supine. Nothing about the pronoun changed; what changed is whether its governing verb moved. In the simple past the lexical verb raised and dragged open a path; in the perfect the lexical verb is the low supine and no path exists. The same pronoun, the same verb-root (sjá/séð), opposite behaviour — and one principle covers both. This is exactly why object shift is a teaching staple: it is a clean, falsifiable dependency between two movements, visible on the surface of ordinary sentences.

💡
Diagnostic you can apply instantly: look at the finite verb. If it is the lexical verb (simple tense), the pronoun shifts left of ekki. If it is an auxiliary followed by a supine/infinitive (compound tense), the pronoun stays right of ekki, after the supine. Auxiliary + supine = no shift.

Pronoun shift is obligatory; full-NP scrambling is optional

So far we have spoken of pronouns, because pronoun object shift is the strict, grammaticalised case: with an unstressed pronoun in a simple tense, shifting is obligatory (\ég sá ekki hann is ungrammatical for neutral "I didn't see him"). But Icelandic also allows a *full noun phrase to appear to the left of ekki under the right conditions — this is object scrambling, and it is a different animal. It is optional, it is not forced, and it is driven by information structure: a full NP shifts only when it is given, topical, backgrounded — already established in the discourse — and it stays in its base position (right of ekki) when it is new or focal.

Ég las bókina ekki.

I didn't read the book (the one we were just talking about). — full NP 'bókina' scrambled left of 'ekki': possible, and signals that the book is given/topical. Optional, not required.

Ég las ekki bókina.

I didn't read the book. — neutral base order: 'bókina' stays right of 'ekki'. This is the default; it does not flag the object as old information.

Þennan disk hef ég ekki sett í þvottavélina.

This plate I haven't put in the dishwasher. (topicalised, separate from scrambling)

The contrast with pronouns is the whole point. A pronoun must shift (it is inherently given — that is what a pronoun is); a full NP may shift, but only to mark itself as given, and may freely stay put. So the two phenomena differ on two axes at once: obligatory vs optional, and purely grammatical vs information-structural. Both, however, obey Holmberg's Generalisation: a scrambled full NP, like a shifted pronoun, cannot cross ekki in a compound tense, because the supine still hasn't.

Ég hef ekki lesið bókina.

I haven't read the book. — compound tense: even the full NP cannot scramble past 'ekki', because the supine 'lesið' is low. HG blocks NP-scrambling just as it blocks pronoun shift.

💡
Two distinct operations, one shared condition. Pronoun shift = obligatory, automatic, because pronouns are inherently given. Full-NP scrambling = optional, and it means something — it flags the NP as old, topical information. Both are licensed only when the verb has moved (HG), so both are blocked in compound tenses.

Stress and focus: the override that proves the rule

Why is a pronoun "inherently given"? Because in normal use it points back to something already in the discourse — that is exactly what makes it shiftable. But if you stress a pronoun, you turn it into focus (new, contrastive information), and a focal element behaves like a new full NP: it does not shift. It stays right of ekki, in the base position, even in a simple tense where shift is otherwise obligatory.

Ég sá hana ekki.

I didn't see her. — neutral, unstressed 'hana': obligatory shift, left of 'ekki'.

Ég sá ekki HANA, heldur systur hennar.

I didn't see HER, but rather her sister. — focal, stressed 'HANA': the shift is cancelled; it stays right of 'ekki', exactly like a new full NP.

This is the elegant confirmation of the whole picture. The deciding factor in object shift was never "is it a pronoun?" but "is it given?" An unstressed pronoun is given (shifts); a full NP is given only when context makes it so (scrambles optionally); a stressed pronoun is focal, hence new, hence behaves like new material and does not shift. Information structure governs the whether; Holmberg's Generalisation governs the whether-it's-even-possible. The two interact: you can only shift a given element, and you can only shift at all if the verb has moved.

The case never changes, wherever the pronoun lands

A practical reminder that survives all this movement: the shifted (or unshifted) pronoun keeps the case its verb assigns. Moving leftward in the wake of the verb does not promote the pronoun or alter its grammatical relation — it is still the object, still accusative or dative as the verb demands. sjá → accusative hana; hjálpa → dative henni; whether before or after ekki, the case is identical.

Ég hjálpaði henni ekki.

I didn't help her. — 'hjálpa' is a dative verb, so the shifted pronoun is dative 'henni', not accusative 'hana'. Shifting does not change the case.

Ég hef ekki hjálpað henni.

I haven't helped her. — compound tense: 'henni' stays right of 'ekki' (HG), still dative. Position changed, case did not.

Common Mistakes

❌ Ég hef hana ekki séð.

Holmberg violation — in a compound tense the supine 'séð' is low and never crossed 'ekki', so the pronoun cannot either. Correct: 'Ég hef ekki séð hana.'

✅ Ég hef ekki séð hana.

I haven't seen her.

The flagship error for English speakers who have learned the simple-tense rule and then over-generalise it. Object shift is not "pronouns go before ekki"; it is "pronouns go before ekki when the verb has moved." In the perfect the lexical verb is the low supine, so the pronoun is stranded after it.

❌ Ég sé ekki hann (neutral 'I don't see him').

Missing obligatory shift — in a simple tense an unstressed pronoun MUST shift past 'ekki'. Correct: 'Ég sé hann ekki.' Leaving it right of 'ekki' is only allowed under contrastive stress.

✅ Ég sé hann ekki.

I don't see him.

The mirror error: failing to shift in a simple tense. Because the finite verb sé has moved, the unstressed pronoun is obligatorily dragged along over ekki.

❌ Ég las hana ekki bókina.

Double object error — you can shift the pronoun 'hana' OR keep the full NP 'bókina', never stack both. Correct: 'Ég las hana ekki' (it) or 'Ég las (bókina) ekki' / 'Ég las ekki bókina' (the book).

✅ Ég las hana ekki. / Ég las ekki bókina.

I didn't read it. / I didn't read the book.

A shifted pronoun and a full-NP object of the same verb cannot both appear — they are the same argument. Choose the pronoun (and shift it) or the full NP (and scramble it or leave it in place).

❌ Ég hjálpaði hana ekki.

Case error — 'hjálpa' assigns dative, so the shifted pronoun must be dative 'henni', not accusative 'hana'. Correct: 'Ég hjálpaði henni ekki.'

✅ Ég hjálpaði henni ekki.

I didn't help her.

Shift never re-cases the pronoun; it keeps the dative hjálpa gives it.

❌ Ég sá hann ekki, HANN, heldur Pál.

Focus/position clash — a contrastively stressed pronoun must NOT shift; it stays right of 'ekki'. Correct: 'Ég sá ekki HANN, heldur Pál.'

✅ Ég sá ekki HANN, heldur Pál.

I didn't see HIM, but rather Páll.

A focal pronoun is new information and behaves like a new full NP: no shift. Shifting it contradicts the focus you are trying to mark.

Key Takeaways

  • Object shift is parasitic on verb movement. Holmberg's Generalisation: a shifted object cannot cross any overt material the verb has not also crossed. Shift is licensed only in the slipstream of the verb's own movement.
  • Simple tense → shift. The finite main verb has raised past ekki, so the unstressed pronoun obligatorily follows it: sá hana ekki.
  • Compound tense → no shift. The finite slot holds an auxiliary; the lexical verb is a low supine that never crossed ekki, so the pronoun is stranded after it: hef ekki séð hana. The same pronoun, opposite behaviour, one principle.
  • Pronoun shift is obligatory and grammatical; full-NP scrambling is optional and information-structural — a full NP shifts only when given/topical. Both obey HG and are blocked in compound tenses.
  • Stress overrides. A focal, stressed pronoun is new information and does not shift, behaving like a new full NP — confirming that the real trigger is givenness, gated by HG.
  • The pronoun keeps its object case wherever it lands — accusative hana with sjá, dative henni with hjálpa.

Related Topics

  • Object Shift and Pronoun PlacementB2 — Object shift in Icelandic — an unstressed pronoun object moves leftward past ekki and the sentence adverbs (ég sá hann ekki) while a full noun-phrase object stays put (ég sá ekki manninn); Holmberg's Generalisation explains why the shift is blocked in compound tenses (hún hefur ekki lesið hana); and stressing the pronoun cancels the shift, tying word order to focus.
  • Where Negation Goes: Main vs SubordinateB1 — A placement drill for ekki and sentence adverbs across clause types — after the finite verb in main clauses (hann kemur ekki), before it in careful subordinate clauses (... að hann ekki komi), and between auxiliary and main verb in compound tenses (hann hefur ekki komið).
  • Case Preservation and Quirky Case in DepthC2 — The single most-cited fact in Icelandic syntax: a lexically case-marked argument KEEPS that case across every syntactic operation — passive, raising, control, and ECM. The passive of a dative-object verb produces a DATIVE SUBJECT (Honum var hjálpað 'he was helped'; Bílnum var stolið 'the car was stolen'), and raising carries a quirky dative up unchanged (Honum virðist leiðast). This preservation is the clinching proof that some case is lexical, not structural — a property found in almost no other well-studied language, and the crown jewel of the field.
← PreviousDefiniteness Effects and SpecificityNext →The Subjunctive in Depth: Mood Selection