What language targeting actually matches
The setting sounds like it filters searches by language. It does not. Google matches on the language a person has set for their Google interface, and on the language of the page or video they are looking at when the ad is a display or video placement. The words typed into the search box are not the test.
That distinction changes everything about how the setting behaves. A shopper in Kathmandu whose Google account runs in English is treated as an English user even when the search itself is typed in Nepali. A shopper who has never changed the default may be treated as a Nepali user even though the whole session is in English. Language targeting sorts people by the settings they carry, not by the sentence they just typed.
Google also does not translate anything. Your keywords and your ad text have to already be written in the language you are aiming at. Selecting a language does not produce a version of the ad in it.
Why language targeting matters
It is one of the quietest ways to lose reach. Narrow the setting to a single language and you remove every eligible user whose interface is set to something else, including people who would happily read your ad. In markets where English interfaces are common on phones sold with English defaults, a Nepali-only setting can cut a campaign down to a fraction of its real audience with no warning in the interface.
The reverse mistake costs money instead of volume. Leaving every language selected while running ads written in one of them means paying for impressions in front of people who cannot read the offer. Neither error is visible in a normal performance report; both show up only as odd volume.
Common mistakes with language targeting
The biggest is assuming the setting controls script. It does not distinguish a search typed in Devanagari from the same words typed in the Latin alphabet, which is how a great many Nepali searches are actually written. If you need to separate those, you separate them with keywords, not with this setting.
The second is mismatching languages inside one campaign. Running English and Nepali ad copy in the same ad group makes the reporting unreadable and gives Google no clean signal about which creative belongs to which audience. Split them into separate campaigns so budget, bids and results stay separate, in the same way you would split by location targeting.
How to act on it
Match the setting to the language your keywords and ads are written in, then add English if your market commonly runs English interfaces. That is a decision to test rather than assume; widen the setting for a period, compare the volume and the quality of the enquiries, and keep whichever version performs.
If you are running the same offer across several markets, build one campaign per language and give each its own budget. That keeps the reporting honest and stops a cheaper market absorbing the spend intended for an expensive one. And check the setting on every new campaign, because a copied campaign carries the language of the campaign it was copied from, not the one you meant.