Hiring Bilingual Office Staff: What to Test

Why this matters

"Do you speak Spanish?" "Yes." That exchange is how most shops hire bilingual office staff, and it measures nothing the job needs. The person who answers yes may hold a fluent conversation at a family dinner and be unable to write a two-line appointment text, or may have four years of classroom Spanish and freeze when a frightened caller talks fast about water coming through the ceiling. You find out which three weeks in, from a customer. A 35-minute test built from your own calls finds out before the offer.

What the job actually uses

Break "bilingual" into the four things a service-shop front desk does in the second language, because candidates are routinely strong in one and weak in another:

• Listening under pressure. Understanding a caller who is upset, talking fast, using regional words, on a bad cell connection. This is the skill that matters most and the one a conversation at the interview table never tests.
• Speaking to get information. Asking for an address, a callback number, the age of a water heater, and confirming them back without the caller having to repeat themselves three times.
• Writing short, correct customer messages. Appointment reminders, "your tech is on the way" texts, a two-sentence follow-up. Short, but customers read them, and a misspelled reminder looks like a scam text.
• Trade vocabulary in both directions. Knowing what the customer means by their word for "breaker" or "drain" and writing it into the work order in the English the techs use. See related: Regional Variation in Trade Vocabulary.

Plus one trait that is not language: calm on the phone. The test can surface it, and it is worth weighting.

What the front desk should not be doing, and so you should not test for: translating contracts, estimates, handbooks or legal documents. That is professional translation work, with its own rules. See related: Getting a Contract or Estimate Professionally Translated.

Keep the test fair, and keep it legal

A language requirement for a job that genuinely uses the language is lawful. Federal law (Title VII, enforced by the EEOC) prohibits national origin discrimination, and testing goes wrong in three predictable ways:

• Testing something the job does not need. Scoring accent, demanding "native" fluency, or requiring perfect written English for a job that writes two-line notes. Score what the job uses, to the level it uses it.
• Testing candidates differently. Every candidate for the role takes the same test, scored on the same sheet, including the internal candidate everyone likes and the applicant who "obviously" speaks the language.
• Asking where someone is from. Proficiency is what you test. Where they learned it is not your question.

See related: National Origin and Citizenship Discrimination in Hiring. Many states add their own protections on top of the federal floor and reach smaller employers than Title VII's 15-employee threshold, so the fairness rules above are the safe default everywhere. If the bilingual skill earns extra pay, decide that before you post the job; see related: Setting a Language Pay Differential Fairly.

Who scores it

Someone who is proficient in the second language and is not the hiring manager scores the language parts. If nobody in the shop qualifies, pay an outside assessor (a freelance interpreter or a language-testing service) for an hour; it is cheaper than the hire going wrong. A shop owner who does not speak the language cannot score Part A by listening to how confident it sounds, and confidence is exactly what fools people.

The test, built from your own calls

Four parts, 35 minutes, 100 points. Write the scripts from real calls your shop gets, in your trade.

Part A: the live call (10 minutes, 40 points). The scorer plays a customer, in the second language, calling with an urgent problem: in a plumbing or HVAC shop, a water heater leaking into a finished basement. The scorer talks at a natural speed, interrupts, uses one regional word, and is upset. The candidate must get the customer's name, address, callback number, a description of the problem, whether anyone is at risk, and give the right first safety instruction: turn off the cold-water valve feeding the heater, and if there is any gas smell, everyone leaves the house and calls the gas utility from outside. Score against a checklist: each required item captured correctly, confirmed back, and handled without making the caller repeat it.

Part B: the write-up (8 minutes, 25 points). Two pieces. First, the work order note from the Part A call, in English, for the tech: what is wrong, where, how urgent, what the customer was told to do. Second, a two-sentence appointment confirmation text in the second language with a date, a time window and the shop name. Scored for accuracy of facts, spelling and clarity, not elegance.

Part C: numbers and words (5 minutes, 20 points). The scorer reads aloud, in the second language, an address with a unit number, a phone number, and a time window ("between eight and ten, but not before eight thirty"), and the candidate writes them down. Then ten trade terms in the second language, chosen from your shop glossary, which the candidate writes in English. Numbers are where front-desk errors cost real trips: a transposed house number sends a truck to the wrong street.

Part D: the hard call (5 minutes, 15 points). A customer disputing a charge, or a caller who wants to hand the phone to a teenage son to translate. Scored for staying calm, staying polite, and knowing when to say "let me have the office manager call you back" rather than inventing a policy.

Pass rule: 70 points or more overall, AND at least half the points in every part. The floor exists because a strong writer who cannot handle the live call is not a front-desk hire, however high the total.

The scoring sheet, filled in

Two candidates for a 9-person HVAC and plumbing shop's front desk, Spanish as the second language, scored by a freelance interpreter.

BILINGUAL FRONT DESK TEST - SCORE SHEET
Part (max, half)       Cand. A      Cand. B
A Live call (40, 20)     36           22
B Write-up  (25, 12.5)   13           23
C Numbers   (20, 10)     17           18
D Hard call (15, 7.5)    12            7
TOTAL       (100, 70)    78           70
Every part at half+?     YES          NO (D)
RESULT                   PASS         FAIL
Notes A: missed unit no. once, fixed on readback.
  Text had 3 spelling errors. Calm throughout.
Notes B: asked caller to repeat address 3 times.
  Promised a refund on the dispute call.

Reading the two results

Candidate A grew up speaking Spanish at home and never studied it in school. She got every required item on the live call, confirmed the address back and caught her own missed unit number, and stayed level when the scorer interrupted. Her written Spanish text had three spelling errors. Her total, 36 plus 13 plus 17 plus 12, is 78. Part B at 13 clears its half-point floor of 12.5 by half a point.

Candidate B studied Spanish for four years and wrote a nearly perfect confirmation text and the best English work order note of anyone tested. On the live call she asked the caller to repeat the address three times and never gave the gas instruction. On the dispute call she promised a refund the shop does not offer. Her total, 22 plus 23 plus 18 plus 7, is exactly 70, which meets the overall bar, but Part D at 7 falls below its 7.5 floor. She fails on the floor rule, not the total.

Candidate B scored higher than Candidate A on two of the four parts, and in an interview she would likely have come across as the stronger candidate. The test shows why the job would have gone badly: the live call is most of the work, and the dispute answer was a policy she invented under pressure.

Why Candidate A's weakness is the acceptable one. Her spelling errors are in short, repetitive texts. The shop can fix that in a week with template messages the office manager approves once, and every reminder after that is a template. Part A's skill, understanding a frightened caller and getting the facts right, does not come from a template.

What changes the weights

The 40/25/20/15 split fits a front desk that mostly answers the phone. Change it when the job changes:

• A role that is mostly written (emailing estimates, answering web chat, managing review replies): move points from Part A to Part B, something like 25 and 40, and keep the half-floor on every part.
• A dispatcher role that talks to techs as much as customers: add a Part C item where the candidate relays a schedule change to a tech in English, because the tech-facing English matters as much as the customer-facing Spanish.
• A language with fewer speakers in your area (Vietnamese, Korean, Haitian Creole): the pool will be small, so consider hiring at a lower Part B score and supporting writing with professionally translated templates.

How to know the test is working

After six months, pull the front desk's calls that went wrong in the second language: wrong addresses, missed safety instructions, complaints. If they cluster in a skill the test scored well, the script for that part is too easy; rewrite it from the call that went wrong. And re-test any internal candidate moving into the role on the same sheet, so the test is the standard, not the gate for outsiders only.

References

• U.S. EEOC guidance on national origin discrimination and employment testing under Title VII.
• See related: National Origin and Citizenship Discrimination in Hiring; Setting a Language Pay Differential Fairly; Handling Calls in Another Language at the Front Desk; Getting a Contract or Estimate Professionally Translated.