Can legal translators be replaced by machines?

Posted on by
11

In recent months at least, it seems that machine translation, based on huge databases of sample translations (neural networks), has massively improved. DeepL is one example. Professional translators would avoid using this as their translations might be integrated into the system, which would be a breach of their client confidentiality. But I do suspect that any law firm processing a huge pile of exhibits in a foreign language and wondering which pages would be worth translating can have the whole lot rapidly machine-translated, then zoom in on the most relevant bits and have them machine-translated.

Peter Winslow, a legal translator with a penchant for Karl Kraus, has posted in a Beck Verlag forum three translations of a sentence, two of which are machine translators and the third by a human translator with to me dubious qualifications:

Nur eine der nachstehenden Übersetzungen ins Deutsche wurde von einem menschlichen Übersetzer angefertigt, die anderen zwei stammen von maschinellen Übersetzungssystemen (vor mindestens sechs Monaten). … Erkennen Sie, welche Übersetzung der menschliche Übersetzer angefertigt hat? Der Mensch ist Deutscher und deutscher Muttersprachler. Er ist Diplom-Übersetzer – sogar für die englische Sprache allgemein beeidigt und öffentlich bestellt – und gibt an, mehr als fünf Jahre Berufserfahrung als freiberuflicher Übersetzer zu haben.

Presumably most readers of this quiz will be German lawyers, and of course they will ask themselves how to know whether a translator can be relied on. It isn’t easy. Someone who has studied translation at a German university will probably have learnt little about legal translation, although you may need to show legal knowledge to be qualified to translate for the courts. It would be better to find a translator with specific legal experience or qualifications, and experience in doing legal translations. But I think one problem is that lawyers specialize, whereas legal translators tend to specialize only in law, not in a narrow area of law. They may have years of experience in a particular area of legal translation, or they may not. I hope most big law firms that do a lot of international work will have inhouse translation teams including trained translators, who will know how to evaluate any software systems used for translation. With smaller firms it is less likely.

The sentence taken as an example is “This policy defines the specific server roles required to implement the server program.” This sentence is hardly typical of legal translations.

(I am guessing, like Prof. Dr. Müller, that the second version is the native speaker of German – the answer has not yet been revealed).

One problem at the moment seems to be that agencies are using MT and the occasional sentence is quite wrong. They then require “proof-reading” from a freelance, but if only the final product is reviewed, in English for example, the error may not be evident, although the review will be cheaper than if it were compared with the original.

(With thanks to Igor Plotkin for posting this on a mailing list)

Aldi’s ‘London’ cheese cake

Posted on by
20

Aldi really has a lot of very British things in its selection, but I had not seen this cheesecake. This is the ultimate British product!

I remember this cheese cake as a regular item at the bakery in the 50s and 60s. It is sometimes called ‘London cheese cake’ to distinguish it from cheesecake that contains cheese. This one, as I remember it, was round and consisted of layers of flaky pastry topped with (usually dry) icing and coconut ‘shreds’. In between pastry and icing there was a blob of sponge cake. The strings of coconut must have been from a paste that was extruded in some way.

The Aldi version is rectangular and has luscious fondant icing and jam under it.

Following this, I decided to try the Greggs’ one, and on the way to Greggs I passed Kingcotts bakery, where they had their own.

The Kingcotts one was as I remember them, round, dryish but relieved by a plug of sponge cake i the middle.

The Greggs one was squarish and had jam but no cake. It was rather thin and meagre.

T

The coconut shred/strips on top vary. The Greggs ones do look like desiccated coconut, very dry.

Kingcotts are the makers of the famous ‘real’ bread:

It is actually real bread. I recommend the cloudy white sandwich tin (sourdough).

Recommendations: the Kingcotts is the genuine cheesecake (£1), but only if you are in Upminster. You may have to try a local baker. The Aldi is delicious (I have forgotten the price) but not authentic: the puff pastry is slightly moist (in the direction of baklava) and the fondant icing with coconut dominates. I am sorry the Greggs is not quite right (80p). I must next find out why Godfreys in Hornchurch say theiir Tottenham cake is not the real thing, and whether Greggs is: I believe the pink colour has to come from a particular mulberry tree in Tottenham.

Legal Integration and Language Diversity: book on translation in EU lawmaking

Posted on by
2

Legal Integration and Language Diversity: Rethinking Translation in EU Lawmaking, by C.J.W. Baaij – Oxford University Press, coming out in February

This book should be interesting. It comes to the conclusion that particularly after Brexit, it would be a good idea for English to be the original language of all legislation.

  • Introduces the first comprehensive quantitative analysis of the case law of the Court of Justice of the European Union, spanning 50 years, focusing on interpreting and solving discrepancies between language versions of EU legislation

  • Integrates a variety of analytic methods and gathers data from both policy document analyses, interviews, and quantitative and qualitative examination of the EU’s Institutional Multilingualism

  • Builds a normative theoretical framework from legal translation studies and comparative law, general translation theory and language philosophy, and European studies

  • Proposes three EU policy changes that question mainstream thinking, from both political and theoretical vantage points

  • Argues that Brexit provides an additional reason in favor of rather than against recognizing English as the primary official language of the EU

(Via Wildy & Sons newsletter)

Simplifying contract language

Posted on by
Reply

The Case for Plain-Language Contracts, by Shawn Burton, Harvard Business Review Jan/Feb 2018

I’ve read a lot of arguments about the use of plain English, and I haven’t often been convinced by them. Now this article by Shawn Burton is at first glance an interesting one (thanks to Inge for recommending it on Twitter), but contains some problems.

Are pages of definitions; words like “heretofore,” “indemnification,” “warrant,” and “force majeure”; and phrases like “notwithstanding anything to the contrary herein,” “subject to the foregoing,” and “including but in no way limited to” necessary for an agreement to be enforceable? Is there some counterintuitive value in useless boilerplate language? Does a contract really need 15-word strings of synonyms; all-cap, italicized, bolded sentences that span multiple pages; awkward sentences containing numerous semicolons; and outdated grammar to be worthy of signature? In my opinion, the answer is a resounding no.

Of course, it would be a good idea to remove archaic words like ‘heretofore’, but what about words with a legal meaning like ‘force majeure’. (The ’15-word strings of synonyms’ are one of the reasons some legal translators prefer to translate from German to English even if German is their native language, because German contracts are simpler, partly because terms are backed up by the Civil Code and other legislation.)

Burton writes: ‘Business leaders should not have to call an attorney to interpret an agreement that they are expected to administer.’ I have my doubts about that. And the ‘litmus test’ was whether a ‘high-schooler’ could understand the contract. Maybe this worked with simplified contracts for customers, but surely not for every type of contract.

Here’s an example from the end of the article:

Before and After

Language in the indemnification clause of a services contract was revised to be clearer and much more concise:

BEFORE

Customer shall indemnify, defend, and hold Company harmless from any and all claims, suits, actions, liabilities, damages and costs, including reasonable attorneys’ fees and court costs, incurred by Company arising from or based upon (a) any actual or alleged infringement of any United States patents, copyright, or other intellectual property right of a third party, attributable to Customer’s use of the licensed System with other software, hardware or configuration not either provided by Company or specified in Exhibit D.3, (b) any data, information, technology, system or other Confidential Information disclosed or made available by Customer to Company under this Agreement, (c) the use, operation, maintenance, repair, safety, regulatory compliance or performance of any aircraft owned, leased, operated, or maintained by Customer of (d) any use, by Customer or by a third party to whom Customer has provided the information, of Customer’s Flight Data, the System, or information generated by the System.

AFTER

If an arbitrator finds that this contract was breached and losses were suffered because of that breach, the breaching party will compensate the non-breaching party for such losses or provide the remedies specified in Section 8 if Section 8 is breached.

In context, it appears that ‘AFTER’ does not indicate the final version. Shawn Burton was made general counsel of GE Aviation’s digital-services unit in 2013 – not very long ago. Three digital-services businesses had to be consolidated, and all three had complex contracts containing text like ‘BEFORE’ above. Three long contracts were replaced by one five-page one. But the content was different too! It was vetted by Weil, Gotshal & Manges and ‘refinements’ were made. Burton lays weight on the amount of time that was saved in negotiation.

It was clear that this would not be the first time Burton had written about the use of plain language in GE Aviation contracts, and sure enough a web search revealed more. It also revealed a blog post by Ken Adams critiquing this very article. dated yesterday:

Optimal Contract Language Requires More Than Enthusiasm: My Critique of Shawn Burton’s Article in the Harvard Business Review

That article contains other useful links.

Merry Christmas

Posted on by
Reply

Here’s wishing a partially completed Merry Christmas to all readers.  (Ebor Street)

Here is some information on VAT on Christmas trees in Germany:

Plastic tree: 19%

Real tree sold on the market 7%

Real tree not deliberately grown, sold by farmer 5.5%

Real tree deliberately grown, sold by farmer 10.7%

Private sale or sale by small non-VAT-registered business: 0%

(Corrections to this translation may be added in the comments)

Die Höhe der Umsatzsteuer, die Sie für Ihren Tannenbaum bezahlt haben, hängt ab von der Herkunft und dem Verkäufer des Baums:

  • Der ganz normale Umsatzsteuersatz in Höhe von 19% wird fällig bei künstlichen Bäumen, das dürften meist Weihnachtsbäume aus Plastik sein.

  • Der ermäßigte Umsatzsteuersatz von 7% wird angewendet, wenn es sich um einen echten Baum handelt, der zwar artgerecht aufgewachsen ist, aber durch einen Gewerbetreibenden (zum Beispiel einen Baumarkt) oder einen nicht-pauschalierenden Landwirt verkauft wird.

  • 5,5% Umsatzsteuer will der Fiskus sehen, wenn der Baum zufällig irgendwo im Wald aufgewachsen ist und von einem Landwirt verkauft wird, der sich für die Pauschalierung der Vorsteuer entschieden hat.

  • Kaufen Sie bei einem pauschalierenden Landwirt, der den Weihnachtsbaum in einer Sonderkultur großgezogen hat, fallen 10,7% Umsatzsteuer an.

  • Falls der Weihnachtsbaum-Verkäufer Ihres Vertrauens Privatverkäufer oder Kleinunternehmer ist, zahlen Sie gar keine Umsatzsteuer.