Advertisement
Advertisement

How should a cell with three sentences in Excel be counted and delivered?

2026-02-10 02:00:29

Count the three sentences as source text, then return them in the same cell, in the same order, with the row and column untouched. A cell is a container. The billing unit is still the language inside it: English by the word, Chinese by the character. Counting rows will understate a cell that holds a short procedure, and counting the cell as a single word will throw the procedure away.

A buyer sent a spare-parts workbook of about six hundred rows and asked for a price per row. Column G looked narrow on screen. Opened up, dozens of those cells held three sentences: a fit note, a torque value, and a storage warning. Row 214 alone told the fitter which way the seal faced, the newton-metres, and a temperature limit. A per-row quote would have treated that cell as one unit and left no time to keep the three instructions from trading places.

Count the sentences that need translating

Export or copy the column into a count of source words or source characters, and ignore cells the client marks as codes, part numbers, or already-final text. Three sentences in one cell are three sentences. They often share a subject that is written only once, so splitting them onto three rows for the translator’s convenience can break the grammar. Keep them together for drafting, and show the count as language, not as a row tally.

There is no public unit price to multiply by the row count. Send the workbook, say which columns are in scope, and the quote uses the source count. Exact warning sentences that repeat down the column and are truly reused can sit near a tenth of the rate. A sentence that changes only the torque number is a new sentence for the purpose of checking that number.

Deliver the cell so it still opens as a cell

Put the translation back into the cell it came from. Keep the line breaks the client uses to show the three sentences, or agree on one convention and use it down the column. Do not deliver a Word document of extracted sentences unless the client asked for a rewrite of the sheet. The workbook is the delivery. Widened columns, wrapped text, and a taller row are layout, and they are a separate line if the client needs the sheet to print in a fixed template.

Part numbers, units, and decimals stay as written unless the language convention for separators is in scope. A translator who retypes a part number into the sentence can corrupt the next import into the client’s system. The three sentences are translated. The code beside them is checked, then left alone.

Watch the row that looks empty and the cell that holds a comment

Hidden columns, filtered rows, and notes stored as comments are easy to miss when someone counted “visible rows.” Ask whether comments, a second language already sitting in column H, and the header row are in scope. A header that changes is a global name, and it belongs in the term list before anyone fills six hundred cells with a personal choice.

Return the workbook with a short note of columns touched, rows skipped, and any cell whose three sentences would not fit the print template. That note is what the next import relies on.

Should the three sentences be split into three columns?

Split them only if the client wants a new structure. The translation job returns three sentences in the cell that held three sentences. A redesigned sheet is a different order, quoted as layout or as a data change, and it waits until the wording is accepted.

How do repeating warnings down the column get billed?

An exact repeat that is actually reused is treated at about a tenth of the translation rate, and it is still glanced at for the row it sits on. A warning that looks familiar and carries a different number is billed and read as source text. The workbook should make those two cases visible in the quote.

The related service page is desktop publishing.

Quick FAQ: AI Translation Accuracy

  • How accurate are AI translators? Accuracy is often high for repetitive or general content, while domain-sensitive content still needs expert review.
  • How to improve AI translation quality? Use glossary control, domain prompts, QA checks, and human post-editing in one workflow.
  • Where does human translation still win? Legal, medical, and high-stakes brand content usually requires human nuance and accountability.