Dative Substitution and Argument-Structure Change

There is a feature of modern Icelandic that prescriptive grammar calls a disease — þágufallssýki, literally "dative sickness" — and that historical linguistics calls something quite different: one of the cleanest examples anywhere of a case system regularising itself in real time. The phenomenon is simple to state. A small set of verbs assigns its experiencer "subject" the accusative in the standard (mig langar "I want," mig vantar "I lack," mig dreymdi "I dreamt"), and a very large share of speakers replace that accusative with the dative (mér langar, mér vantar, mér dreymdi). The register page on usage debates covers the social verdict — hear mér langar, write mig langar. This page does something that page deliberately does not: it treats the substitution as a piece of grammar, asks why the dative and not some other case, and shows that the direction of the change is not random at all. It is the single most regular thing the language could do.

The two experiencer classes, and the mismatch between them

Start with the fact that creates the pressure. Icelandic has a whole family of verbs whose logical subject — the experiencer, the person who feels, perceives, or undergoes something — is not nominative. That family splits into two case-classes, and they are wildly unequal in size.

The dative-experiencer class is large and productive. finnast "find/think," líða "feel (be)," leiðast "be bored," þykja "deem," líka "like," ofbjóða "be appalled," blöskra "be shocked," sárna "be hurt/offended," hlýna "warm up (to)" — verb after verb, all taking a dative experiencer. This is the default shape of an Icelandic experiencer-subject verb; if you coin a new one or borrow one, the dative is where the experiencer lands.

The accusative-experiencer class is tiny and closed. In practice it is langa "want," vanta "lack/need," dreyma "dream," and the now-archaic hungra/þyrsta "be hungry/thirsty." That is essentially the whole list (see verbs/accusative-subject-verbs). A handful of high-frequency verbs sit in an otherwise empty box.

So the experiencer-subject system contains one big, regular, dative-marking pattern and one small, irregular, accusative-marking pattern. That is exactly the configuration in which analogical levelling happens: a minority class gets pulled into the majority class. The "dative sickness" is the accusative experiencers being absorbed into the dative experiencers — the small box emptying into the big one.

The change, framed as regularisation rather than error

Watch the substitution happen and label it for what it is. The standard form preserves an inherited accusative; the innovative form replaces it with the dative that the rest of the experiencer system uses. Nothing else about the sentence changes — the verb is the same, the meaning is the same, the object (where there is one) keeps its case. Only the experiencer's case is updated, and it is updated towards the regular pattern.

The directionality is the tell. If þágufallssýki were mere sloppiness, you would expect errors in every direction — speakers occasionally putting mér finnst into the accusative (*mig finnst), or randomly swapping nominatives for datives elsewhere. That is not what happens. The flow is overwhelmingly one-way: accusative experiencers drift to dative, essentially never the reverse. mig finnst for mér finnst is not a thing; mér langar for mig langar is everywhere. A change with a consistent direction is not noise — it is the system reorganising, and the direction points straight at the larger class.

The set of affected verbs — and the ones moving the other way

Within this picture you can predict which verbs are vulnerable: precisely the accusative-experiencer minority, because they are the irregular ones. langa, vanta, and dreyma are the core targets, and they are exactly the verbs the standard insists keep the accusative.

There is a second front in the same war, and it confirms the analysis by showing the dative attracting even nominative experiencers. A few verbs take a nominative experiencer subject in the standard — hlakka til "look forward to," kvíða (fyrir) "dread" — and these too are being pulled towards the dative (mér hlakkar til) for the same reason: the dative is the experiencer magnet. The starting case differs (accusative for langa, nominative for hlakka), but the destination is identical, which is precisely what the "dative is the prototypical experiencer case" account predicts and a "random error" account does not.

That two unrelated starting points (accusative langa, nominative hlakka) converge on the same dative outcome is the strongest internal evidence that this is a coherent change driven by one factor — the experiencer-case prototype — rather than a scatter of independent slips.

Why this is theoretically interesting: case can be lexical and mobile

The deepest point connects this page to case preservation. There, the lesson was that lexical (quirky) case is fixed — it clings to a noun through every syntactic operation, which is why the passive of hjálpa yields a dative subject. So how can a lexically-assigned case change at all? The answer is that lexical case is fixed within a grammar but can be re-specified across generations. Each verb stores, in its lexical entry, which case it gives its experiencer. langa historically stored "accusative." A learner acquiring the language, swamped by dative experiencers everywhere else, can store "dative" instead — re-writing the lexical entry to match the productive pattern. Once re-specified, that dative is just as fixed and quirky as any other: a speaker with mér langar will preserve that dative under raising and the passive exactly as the standard preserves its accusative. Lexical case is a per-verb, per-grammar fact that is stable for a speaker but renegotiable at the point of acquisition. þágufallssýki is what renegotiation looks like when it is caught mid-stream.

This reframes the whole prescriptive debate. The standard is not defending "correct grammar" against "broken grammar"; it is defending an older, more irregular state of the lexicon against a newer, more regular one. The innovators are not failing to learn a rule — they have inferred a cleaner rule (experiencers take dative) and applied it across the board. Whether that is "decay" or "improvement" depends entirely on whether you value historical continuity or systemic regularity, which is a social choice, not a grammatical fact. No learner resource frames it this way, and it is the single most clarifying thing you can know about the controversy.

Where the change stands

The sociolinguistic profile fits the regularisation story. þágufallssýki was noticed and named in the 1920s and has advanced steadily since; in present-day Icelandic it is reported as essentially universal as a tendency — almost every speaker produces it at least sometimes, many have it as their default for langa and vanta, and acceptance is highest among the young. That is the textbook age-grading of a change spreading from below. The accusative survives where it survives largely because the school system actively maintains it: it is taught, tested, and corrected, which is exactly the support an irregular form needs to resist levelling. Remove that institutional pressure and the prediction is plain — the dative wins, because the system wants it to.

The practical upshot for an advanced learner is unchanged from the register page (write the accusative), but the understanding is richer: you are not memorising an arbitrary exception, you are choosing to reproduce a historically older state of the experiencer system that the living language is steadily abandoning.

Common Mistakes

These are conceptual errors specific to understanding the change, since the production rule (write mig langar) is covered on register/usage-debates.

The substitution is systematic. Calling it "just an error" misses that it flows one way, targets exactly the irregular verbs, and lands on exactly the case the rest of the experiencer system uses.

Key Takeaways

• þágufallssýki ("dative sickness") is not random error but a directional, systematic change: the experiencer subjects of langa, vanta, dreyma (standard accusative mig) are being regularised to the dative (mér).
• The driver is class size: Icelandic has a large, productive dative-experiencer class (mér finnst, mér líður, mér leiðist, mér líkar) and a tiny accusative-experiencer class (langa, vanta, dreyma). The dative is the prototypical experiencer-subject case, so the minority class is being levelled onto it.
• The change is one-way: accusative (and even some nominative, ég hlakka til → mér hlakkar til) experiencers move to the dative; dative experiencers never move to the accusative. Convergent direction from different starting cases is the signature of regularisation.
• Theoretically, this shows lexical (quirky) case is fixed within a grammar but renegotiable across acquisition: a learner can re-specify langa's stored case as dative, after which it behaves like any other preserved quirky case (complex/case-preservation).
• The reframing: standard mig langar is the irregular form, innovative mér langar the regular one — prescription defends an inherited irregularity, which is why the change is so robust and why it is, by most assessments, slowly winning.
• Practically (per register/usage-debates): you will hear mér langar, but should write mig langar. Now you know it is a choice to reproduce an older state of the system, not an arbitrary rule.