Case Contamination

Case contamination is the blending of two constructions that would separately assign different cases. A speaker begins from one syntactic plan but a competing word, idiom, or continuation pulls the phrase toward another. Ancient authors sometimes exploit or tolerate such blends, but contamination is primarily an analytical diagnosis—not a productive rule that lets a writer choose cases freely.

English marks many relations by word order and prepositions, so English-speaking learners are tempted to decide what a noun “means” before finding what governs it. Latin requires the opposite procedure: identify each possible governor, write out the two complete constructions, and only then decide whether the transmitted case is ordinary, attracted, contaminated, or corrupt.

The two-frame test

Suppose a writer has both ūtor illā viā “I use that route” and sequor illam viam “I follow that route” available. A mixed phrase that joins ūtor directly to illam viam would combine the verb from one frame with the accusative from the other. That is the logic of contamination; it is not recommended Classical composition.

The minimal contrast shows why translation is insufficient: both English clauses have a direct-looking “that route,” but the Latin predicates govern different cases. A form that crosses the frames cannot be justified from English.

Ordinary variation is not contamination

Some Latin words genuinely allow more than one construction, often with a difference of meaning or period. Dōnō may be organized around the gift or the recipient: dōnō librum puerō “I give the boy a book” and dōnō puerum librō “I present the boy with a book.” Each frame has its own structure. Calling either one “contaminated” would hide a lexical fact.

Likewise, a place expression may use a bare locative or a prepositional phrase according to the noun and meaning. Variation supported by the lexicon is not a mistake and does not need a rhetorical label. Check the governing word before invoking a rare exception.

Relative attraction: a visible neighboring pressure

Relative clauses provide the clearest Classical cases in which one case frame influences another. Normally a relative pronoun agrees with its antecedent in gender and number, but takes case from its own clause:

Litterās and quās are both accusative here, but for independent reasons: the first is the object of accēpī, the second of scrīpsistī. To test the rule, change the inner government: eās litterās, quibus respondistī has accusative antecedent litterās but dative quibus with respondeō.

Occasionally the relative is attracted into the case of its antecedent. A letter from Lucceius preserved in Cicero's correspondence reads:

The fully explicit analysis would lead us to expect eōrum quae cōnsuēstī agere: partitive eōrum belongs to aliquid, while quae would be the accusative object of the understood agere. Instead quōrum shares the genitive of its antecedent. Traditional grammar calls this case attraction; it can also be described as pressure between two case frames. “Contamination” is the broader mechanism, while “attraction” names its direction. The regular agreement rule and its principal exceptions are developed in Case Attraction and Case Agreement.

Inverse attraction of the antecedent

Pressure can run the other way: the antecedent appears in the case required inside the relative clause rather than in the case expected in the main clause. Vergil's famous (literary) line is traditionally read this way, although Servius also records an analysis that supplies an omitted vultis “do you want?”:

On the inverse-attraction analysis, the main predicate vestra est calls for nominative urbs: “the city is yours.” But urbem is accusative, the case of quam as object of statuō. The antecedent has been drawn into the relative clause's case. On the elliptical analysis, urbem quam statuō is instead the object of an understood vultis, followed by the complete clause vestra est. English word order makes “the city” look straightforwardly like the main subject, so a translation-first parser can miss why the Latin has generated competing analyses.

Plautus supplies another traditional example, Naucratem, quem convenire volui, in navi non erat “Naucrates, whom I wanted to meet, was not on the ship” (Amphitruo 1009; source spelling reproduced without added macrons), naturally at home in (informal/comic) dialogue. The main clause would ordinarily have nominative Naucratēs ... nōn erat. Accusative Naucratem anticipates quem, object of convenīre. The construction is marked, but its two frames are recoverable. This is not evidence that sum generally takes an accusative subject.

Contamination, anacoluthon, and ellipsis

These categories overlap at the edges. Anacoluthon describes a sentence whose initial structure is abandoned for another; contamination emphasizes the hybrid form produced by rival constructions. Ellipsis supplies an unexpressed constituent while preserving a coherent frame. Attraction copies a feature from a salient neighbor. The same difficult passage may receive different analyses from different grammarians because an ancient speaker’s production process is not directly observable.

Here fronted quās rēs belongs inside the relative clause, while resumptive eās fills the main clause. There is no case problem: both are accusative, and each has its own governor. Latin information structure permits the relative material to come first. Do not diagnose contamination merely because English would order the clauses differently.

Cui is dative under dedistī; nominative is is the main subject. The two cases are expected. A learner who sees only that both refer to one person may try to make them match, but coreference does not entail case agreement.

Discourse and register constraints

Clear case attraction is marginal in Classical Latin prose compared with ordinary independent government. Inverse attraction is conspicuous in (literary) verse and (informal/comic) dialogue, where fronting, interruption, and dramatic immediacy are more welcome. Later inscriptions and non-literary texts can show wider case fluctuation, sometimes because phonological change weakened distinctions. Those later developments should not be projected backward as a free Classical rule.

For production, choose the regular frame unless imitating a securely attested passage. For interpretation, preserve the manuscript reading long enough to test it; immediately “correcting” the case can erase meaningful information structure. At the same time, not every anomaly is intentional artistry: manuscript transmission, spelling, and editorial punctuation can matter.

A parsing routine

First, mark every finite and non-finite predicate. Second, assign its arguments from Latin lexical government. Third, identify antecedent–relative pairs by gender and number without yet forcing their cases to match. Fourth, compare the expected cases with the forms transmitted. Finally, ask whether attraction, a blended frame, ellipsis, or a textual problem best accounts for the mismatch.

The tempting English sequence “the man whom you entrusted the money” may suggest accusative quem. Latin crēdō takes a dative recipient, so cui belongs to its own clause, while eum is accusative under videō. Heads and governors, not linear proximity, determine the cases.

Common Mistakes

1. Copying English object syntax

2. Making a relative copy its antecedent automatically

3. Treating inverse attraction as normal agreement

4. Calling every alternation contamination

Key Takeaways

Case contamination becomes plausible only when two identifiable Latin frames compete. Ordinary lexical variation, fronting, and coreference do not by themselves create it. Relative attraction offers the clearest diagnostic cases: compare the relative’s internal role with its antecedent’s case, and determine the direction of pressure. In production, prefer independently licensed Classical government; in reading, let morphology and constituent structure lead before English translation.