One term, agreed once, rendered إنهاء eighteen times and فسخ four.
Those are not synonyms. One ends the contract going forward; the other unwinds it from the beginning and brings a different remedy with it. Four clauses of this agreement say something nobody negotiated, and reading either version on its own will never show you which.
8 defined terms, 128 uses, both directions
Flip it. The list is not symmetrical, and the asymmetry is the point.
3 of 8 terms split reading this way
128 uses · 187 clauses- Deliverables
27 uses · 2 renderings
- المخرجات22×
- التسليمات5×
Close synonyms with no obvious consequence here. Flagged because consistency is cheap and an inconsistent defined term invites an argument that costs more than the fix.
- Termination
22 uses · 2 renderings
- إنهاء18×
- فسخ4×
فسخ is rescission, not termination. It unwinds the contract from the beginning rather than ending it going forward, and the restitution that follows is a different remedy entirely.
- Affiliate
14 uses · 2 renderings
- الشركة التابعة11×
- الطرف ذو العلاقة3×
Related party is a wider class than subsidiary. Three clauses therefore bind a larger group in Arabic than in English, including the confidentiality undertaking.
- Confidential Information
31 uses
- المعلومات السرية31×
- Business Day
14 uses
- يوم عمل14×
- Obligation
9 uses
- التزام9×
- Material Breach
7 uses
- إخلال جوهري7×
- Undertaking
4 uses
- التزام4×
Two English terms collapse into one Arabic word. Nothing is wrong on the Arabic side; the drift is on the English, and it is invisible until the concordance is read the other way round.
Flip the direction. A term that looks consistent one way round can be two terms collapsing into one the other way — and that drift is on the English side, where nobody thinks to look for it.
3 terms split reading English to Arabic and 1 split reading back. The one that only appears in reverse is التزام, where two English terms have collapsed into one Arabic word — a drift on the English side that an English reader has no way to notice.
Not every divergence is worth your afternoon
8 terms produced 3 splits, and only some carry a consequence. Deliverables rendered two ways is untidy. Termination rendered two ways is a different contract in four places.
So findings arrive ordered by what the difference does, not by how many times it occurs — 16 of 128 uses are flagged, and the largest count on the page is not the most serious one.
Intended · 18 uses
إنهاء
Ends the agreement prospectively. Accrued rights survive; what has been performed stays performed.
Actually used · 4 uses
فسخ
Rescission. Unwinds the contract from the start and opens a restitution argument nobody priced.
Both translate to "termination" in every general-purpose dictionary, and a competent non-lawyer translator will use them interchangeably. This is the class of error that survives review by everyone except a bilingual lawyer reading both columns at once.
"English shall prevail" does less than you think
Three questions that come up on every bilingual deal, answered without the comfortable version.
- 01 The contract says English governs. Does it?
- Between the parties, generally yes — a governing-language clause is enforceable. In front of a Saudi court it is less decisive than it reads, because proceedings are conducted in Arabic and the court works from an Arabic text. The clause tells the judge which version wins an argument; it does not spare anyone the translation.
- 02 So which version should be right?
- Both, which is the only answer that survives contact with a dispute. Nominating a governing language is a tie-breaker for the cases you failed to prevent, and firms that treat it as a substitute for aligning the two versions are relying on a clause to fix a drafting problem.
- 03 What if the Arabic was produced by a translator after signature?
- Then it was never negotiated, and it is the version a court is most likely to read. This is the common and expensive case: an English agreement argued line by line, and an Arabic version nobody on either side has checked.
Aligning them is the hard part
Everything above assumes you can put clause 14.2 next to its Arabic counterpart. On a negotiated agreement that assumption fails early.
- Clause numbering does not survive
- Versions renumber independently once either side inserts a clause. Alignment is done on structure and content rather than on numbers, because the numbers stop agreeing early and quietly.
- One clause becomes two
- Arabic drafting frequently splits a long English clause, or merges two short ones. Alignment has to be many-to-many or it reports a false divergence on every split.
- Schedules drift furthest
- The body gets negotiated and the schedules get pasted. Most of the divergences on a real agreement are behind the signature page.
- Unaligned sections are reported, not skipped
- Where a section in one version has no counterpart in the other, that is the finding. Silently dropping it is how a missing indemnity stops being missing.
What a flag is not
It finds where two documents stop agreeing. It has no view on which of them is right.
- A flag is not an opinion
- It reports that two words differ and what the difference usually means. Whether it matters in your clause, in your deal, is judgement — and the فسخ example is only serious because of where it sits.
- It compares versions, not intentions
- Where the two sides deliberately agreed different wording for a reason, it will flag that too. Suppressing the flag is your call and it is recorded against the clause.
- Dialect and register vary legitimately
- Saudi drafting conventions differ from Egyptian or Levantine ones, and a term that reads unusual is not necessarily wrong. Consistency within one agreement is the standard applied, not conformity to a house dictionary.
- It will not fix the Arabic for you
- The output is a list of places the two versions stop agreeing, ordered by consequence. Deciding which version is right, and redrafting it, is the work.
The rest of the platform
Assistant
Ask the file a question. Every answer cites its source.
Research
Nizam, regulations and ministerial decisions.
Drafting
From your own precedents, not a generic bank.
Contract review
Marked against your playbook, not somebody else's.
Document sets
Thousands of pages, ranked by what matters.
Matters
Every document, deadline and hour against the file.
Time and billing
Captured as you work, not reconstructed on Thursday.
Deadlines
Counted from the trigger, not from memory.