Line-final hyphens have different roles
A hyphen may mark layout division or belong to a compound, name or code. Its line position alone cannot decide. Read context and compare the source.
Deleting every hyphen-plus-newline embeds an unverified linguistic choice. Use only on checked equivalent cases. Collect candidate joins first and separate headings, tables and abbreviations. Restore actual words rather than pursuing hyphen-free text or lower totals.
Visible hyphenation differs from copied characters
CSS hyphens controls rendering, including automatic and suggested break opportunities. Display alone cannot establish copied characters.
Paste a sample into plain text: it may contain the whole word, hyphen plus break or a discretionary character. Programs may export visually identical breaks differently. OCR may also misrecognize marks or adjacent letters; joining fragments cannot repair an earlier substituted letter.
Versions of the Italian word comunicazione
Literal co-, LF, municazione has fourteen site characters and two words. Deleting LF gives co-municazione, fourteen characters and one word because internal hyphens are allowed within tokens. Spelling is still not the intended continuous word.
comunicazione has thirteen characters and one word. One word does not certify spelling. Check the exact sequence and consult a suitable language source or intact original for unfamiliar words rather than treating tokenization as a dictionary judgment.
Soft hyphens can be invisible
U+00AD SOFT HYPHEN suggests a break and can remain invisible until needed. It differs from U+002D. The table’s [SHY] label is explanatory rather than part of measured text.
Current-engine comu + U+00AD + nicazione yields fourteen characters and two words: it is not automatically removed. This is a concrete copied-text limitation. For continuous-word delivery, intentionally remove it in a copy and verify. Invisible does not mean absent.
Similar marks need different treatment
Marks between numbers can indicate ranges, subtraction or identifiers; name hyphens may be required. Similar glyphs can have different Unicode codes. Appearance-only rules can delete the wrong mark or miss the intended one.
Record code and neighboring text for doubts. See invisible characters for precise reporting. Do not expand deletion to every connecting symbol. Codes require authoritative data references, not linguistic guesses.
Reconstruct in controlled steps
Work on a copy and collect line-final hyphens. Check expected words, layout origin and exceptions. Apply shared rules only to genuinely equivalent groups; preserve or individually repair exceptions.
Find soft hyphens separately using their actual character or code. Agree whether new layout should retain discretionary points or continuous words. Keep space and newline cleanup separate to identify which transformation repaired words or changed paragraphs.
Broken words may evade search
Continuous-word searches may miss hyphen-plus-break or discretionary sequences, depending on search rules. Search a recognizable fragment before concluding content is absent.
Check repaired words against source and genuine hyphens that should remain. Repairing split words while corrupting codes is not adequate. For columns, restore fragment order first; one column’s ending cannot be safely joined to another’s beginning.
Count the correct version
Run the counter on intended delivery, inspect known samples and real joins, and identify which fragments caused reductions. Smaller totals do not automatically demonstrate successful repair.
For untouched received-file measurement, retain that count with a note about divisions and discretionary marks. Original and reconstructed copies are different objects. Do not present corrected totals as intact-file counts. See joining lines for paragraphs, lists and quotations.
Word totals alone cannot establish repair
| Sample notation | Characters | Words | Check |
|---|---|---|---|
| co-\nmunicazione | 14 | 2 | Hyphen and LF division |
| co-municazione | 14 | 1 | LF removed; hyphen remains |
| comu[SHY]nicazione | 14 | 2 | U+00AD remains |
| comunicazione | 13 | 1 | Continuous sample word |
Sources and checks
- MDN: hyphens · CSS hyphenation and discretionary U+00AD soft hyphens