Mistranslation is both innocent and ubiquitous. Mark Twain had fun with the mistranslations of one of his short stories into French. Indeed, literary superstars frequently acknowledge that translations are invariably imperfect. Jorge Luis Borges’ line — “the original is unfaithful to the translation” — exposes that variance with irony.

Both the innocence and the inevitability are captured in the commonplace phrase “lost in translation.” But, the NLRB cannot recognize irony nor innocence. In Labriola Baking Co., 361 N.L.R.B. No. 41 (9/8/14), it concluded that a Chicago bakery interfered with a representation election based not on its message to its English-speaking workers but on the imprecisions in translating that message for its Spanish-speaking workers.

The communications occurred at a meeting shortly before voting on union representation. At that meeting, Labriola’s COO read from a scripted presentation and told its English-speaking employees:

“If you chose Union Representation, we believe the Union will push you toward a strike. Should this occurs [sic], we will exercise our legal right to hire replacement workers for the drivers who strike.”

To cover Labriola’s employees Spanish-speaking employees, Labriola tasked its payroll administrator with translating. His version recited that the Union would “push” the employees to strike and that, in the event of a strike, they would be replaced by a “legal workforce.” Here, Labriola’s perfectly proper statement on its “legal right to hire replacement workers” got muddled in translation with “legal” now modifying “workforce” rather than “right” according to the NLRB’s retranslation.

In monitoring employer campaign statements, the NLRB does not consider “good faith,” “substantial compliance,” or “innocent mistake.” Rather than looking at what the speaker meant, the NLRB applied its traditional test: whether – in the NLRB’s subjective judgment — the actual words would “tend to interfere with employee free choice.” Applying this standard, the NLRB majority noted that the “potential vulnerabilities of the Spanish speaking workers” could render this translation threatening.

There were no facts that would transform the translation as “legal workers” into a threat to the Spanish-speaking employees’ immigration status. As the dissent pointed out, the record contained no evidence to suggest that any employees lacked work authorization or feared being reported to immigration authorities; there was also no evidence that other immigration-related issues or problems existed in the workplace. Likewise, immigration-related issues were never part of the campaign.

Despite the absence of any contextual facts to support its deconstruction of the translation and despite the fact that the intended message was legally permissible, the NLRB found the translated phrase itself to be not only unambiguous but also “highly coercive.” While the NLRB acknowledged that “some imprecision inevitably arises when communicating complex issues in multiple languages,” it declined to allow any additional latitude there.

Perhaps, the NLRB’s point is best illustrated by a trailing parenthetical in a footnote quoting an earlier case involving ambiguity: “assuming arguendo that there is some ambiguity lurking in the statement, the employer is responsible for the double entendre.” This line may best explain why the employees’ 20-16 vote in favor of decertifying the Teamsters’ union was set aside and a do-over election ordered.

Employers can find multiple lessons in Labriola:

First, where translation is needed in union election campaigns, price can be no object. A bilingual supervisor lacks the precision of professional translators. Students of presidential mistranslations will remember the embarrassments of President Carter in Poland. Those mistranslations came from buying cheap: a freelancer @ $150 per day was hired as President Carter’s official translator.

Second, just as nuance is important in English (with lawyers vetting campaign fliers and speeches by unpacking each line word-by-word to insure that there are no loose ends that the NLRB might claim to be threats or promises), that is equally important in translation. Again, presidential translations provide an apt illustration. President Kennedy’s “eich bin ein Berliner” can be dissected variously as “I am a jelly doughnut” and as “I am a citizen of Berlin” based on debates over whether “ein” is improperly used for statements of citizenship. Each translated version needs equal attention.

Third, it appears from the NLRB decision that employees who heard the Spanish translation (which, unlike the English version, was never written down) testified in English to their own subjective translation of the operative words. This overlooked the layers of opportunity to play out Borges’ truism: e.g., cross-examining the fluency in Spanish of the testifying employee; confirming the proof in Spanish of the actual words said; and establishing the proper translation (via expert witness rather than rank-and-file pro-Teamster employees) of those remembered Spanish words into English.

Finally, in a world where the NLRB will decide what words “tend to interfere with employee free choice”, who will be brave enough to start creating a record with forensic linguistics? For too long, labor lawyers have been content to put words into the record and then engage in a losing debate with the NRLB over whether there is an impermissible meaning. Perhaps, the only viable option is to build a record with expert testimony ruling out such attenuated interpretations.