Wikipedia:Geselshoekie/MT

in Wikipedia, die vrye ensiklopedie
Spring na: navigasie, soek

Proposal[wysig bron]

Hi, sorry that I write in English. I ask if anyone minds if I produce articles for the Afrikaans Wikipedia using machine translation from English. If I make them, would you like me to tag them as "Unedited Machine Translation", so they can be in a category and someone can look to fix them? If people object I will stop my experiments. Thanks :) - Francis Tyers 10:45, 17 Mei 2007 (UTC)

It will depend on how accurate the translation is, I suppose. Maybe you would like to do one or two test articles? If the amount of fixing it needs is equal to the amount of effort it would take to write the article from scratch, then it obviously would not be useful. See The Star-Spangled Banner for an example of such an article (I don't know if it was automatically translated, but it sure is a big mess).
I would like to see the results of one or two articles before I made up my mind. Anrie 11:00, 17 Mei 2007 (UTC)
I wouldn't be intending to translate long articles (not just yet anyway -- the system is not up to it). Some of the articles I've translated can be found with: Spesiaal:Contributions/Francis_Tyers, I would welcome your comments. Btw, if you do try out the system, note that I edit the output (to fix minor issues) — but am in no-way a native or even proficient speaker. - Francis Tyers 11:12, 17 Mei 2007 (UTC)
They look okay, but I would personally prefer to have more than one sentence in any article. Anrie 11:15, 17 Mei 2007 (UTC)
That can be arranged. Is there a minimum amount of sentences you would like to see? Also would you like to see them tagged? (either on the page or the talk page) - Francis Tyers 11:17, 17 Mei 2007 (UTC)
Well, I love tags, so you won't hear me complaining. The main page should be appropriate. I prefer at least a short paragraph - maybe four or five sentences? Of course, there is no definite minimum. Let's give this a test run with a few articles then we can see what it looks like. Anrie 11:28, 17 Mei 2007 (UTC)
Ok. - Francis Tyers 12:17, 17 Mei 2007 (UTC)
Ok, have a look at af:Khujand. I have to simplify some of the language (in English) as my lexical coverage is currently quite poor. Do you think you could manage with post-editing an article like this per day? - Francis Tyers 12:24, 17 Mei 2007 (UTC)
The article looks okay - there are a few oddities which I already corrected, but for the most part it seems okay.
Certainly an article (or even two or three) per day would be acceptable. I'm sure you can appreciate that a flood of such articles would most certainly create a bit of problem for the few active editors here. But your suggestion sounds more than reasonable. Thank you for your effort to help improve our Wikipedia. Anrie 12:29, 17 Mei 2007 (UTC)
Great, thanks :) I certainly wouldn't want to flood you, which is why I bought it up here first. I'll see if I can take care of the oddities in the system so you will need to do less editing next time. - Francis Tyers 12:39, 17 Mei 2007 (UTC)

One question: could you, if possible, on the talk page or in the summary mention from where you are translating. Sometimes it might not be clear exactly what is meant by a certain phrase or sentence and then it is helpful to be able to access the original text. Anrie 12:31, 17 Mei 2007 (UTC)

Yep, no problem. I mention it in the edit summary, but I can leave a note on the talk page. Could you help me create a template with something like "This page was translated from en:Page using Apertium. Please fix the spelling and grammar where appropriate". (You can decide what to put). - Francis Tyers 12:39, 17 Mei 2007 (UTC)
Unless the program is directly affiliated with Wikimedia, I'm afraid that mentioning it in the template will constitute advertising, which isn't allowed. There's nothing that prevents you from mentioning it on the talk page though.
If you don't mind, I'd like to let this discussion set a bit and allow other users to read it and voice their opinions, if they have any. If there aren't any objections to your method in the next day or two, I'll create the tag. Anrie 12:43, 17 Mei 2007 (UTC)
Ok, no problem, I understand. — also no problem with letting this stand for a few days. The more input the better. - Francis Tyers 12:48, 17 Mei 2007 (UTC)
Yeah, totally, my vote goes for this project. Praat van projek, dalk moet ons 'n projek of 'n ding stig? Ek't al baie op die Engelse Wikipedia gesien dat hulle groot projekte het soos "Wiki Bible" en "Music Articles" en goed soos dit; dalk kan ons 'n projek maak wat soos byvoorbeeld al die getagde artikels wat direk vertaal is wys en sê watter moet nog vertaal word, ensovoorts... dis maar net 'n oop idee, maar vir tyd en wyl is ek vir die vertaling van die artikels vanuit Engels na Afrikaans deur middel van 'n rekenaar. Ek dink hier is genoeg aktiewe mense op Wikipedia, en dit kan net meer word, plus Wikipedia is hier om ge-edit te word, so as daar so hier en daar probleme opduik met betrekking tot die verkeerdvertaalde woorde wat sal te voorskyn kom, sal dit vir ons rede gee om so 'n bietjie meer aktief te wees en edits te maak, al mag dit dalk hoe min wees... maar watookal. Ek praat seker nou te veel. Al wat ek sê is dat ek hierdie idee van Francis steun. Adriaan ( SpreekBydraes ) ♪♫ 13:32, 17 Mei 2007 (UTC)
Ons moet darem net seker maak dat ons werk aan belangrike goeters, en werk vanaf hoë kwaliteit artikels. Ons het vir eers genoeg 5-sin artikels oor obskure onderwerpe. Thank you for your interest, Francis. I think we should ensure that it actually benefits our project in tangible ways. We had some bad experience with MT-people before. Thanks for at least asking first :-) I think we should just stay aware that most of our editors don't just translate the English, but tries to compile the best out of all languages they understand, often at least en and nl, but for some of us also de and fr, apart from other external sources. We try to do better than the English where possible, and this will make it very easy to get lazy in doing real encyclopaedic writing. But if this proves good enough, it can be a great help. Ideally I think it should rather be a tool used by the each editor, rather than the editors becoming a part in an assembly line, if you understand what I mean. --Alias 16:21, 17 Mei 2007 (UTC)
Yes, I understand. Actually, an nl-af system would make sense and would probably give even higher quality... but it would be harder for me to evaluate as I don't even speak one of the languages. There is a link from my page to where you can try it out, but I think that at least for now, the amount of post-editing I need to do (to fix some inflections and remove some debugging symbols) is more than it would take you to write the article from scratch. I hope in the future that it will become a useful tool for you guys (and girls), but for the moment it is definitely in the 'testing and development' phase. - Francis Tyers 23:10, 17 Mei 2007 (UTC)
Baie goed saamgevat, Alias. Ek het vandag eers hier aangekom en was baie bekommerd oor die outomatiese vertalery. Ons kan nie eksklusief wees nie, maar Wikipedia in Afrikaans moet daarop konsentreer om 'n goeie versameling van relevante, bruikbare artikels in goeie Afrikaans op te bou. Pendoring 16:36, 17 Mei 2007 (UTC)
[Very good *saamgevat, *Alias. I have today first here *aangekom and *was very anxious *oor the automatic *vertalery. We can not exclusive be, but Wikipedia in Afrikaans must *daarop concentrate *om a good collection of relevant, useful articles in good Afrikaans on too *bou.]
I understand your sentiment (actually I machine translated it to get the 'gist' — see above for my benefit) :) — it is important to me that unedited machine translation output is not allowed to be viewed by the public at large — it would bring disrepute to the encyclopaedia. - Francis Tyers 23:10, 17 Mei 2007 (UTC)
Vanaf Bespreking:Soedanese literatuur.

Hierdie vertaalmasjien is 'n onding wat baie meer moeite gaan meebring as wat dit werd is. As die ding dan vervolmaak word, wat is nog die nut van Afrikaanse gebruikers, buiten om alles wat reeds in ander tale verskyn het, klakkeloos te láát vertaal. 2 000 000 artikels hier ons kom - ek bedoel "hier kom ons", nie dat die vertaalmasjien die verskil sal weet, ek bedoel "ken" nie. Morne 15:42, 18 Mei 2007 (UTC)

Gebruiker:Francis Tyers is besig om 'n paar toets-artikels te probeer. As jy nie van die resultate hou nie, moet jy dit asseblief in Engels in die gebruikershoekie gaan sê. Hy wil graag terugvoering van die gebruikers hier hê. As te veel gebruikers hierteen gekant is sal hy ophou. Anrie 15:50, 18 Mei 2007 (UTC)
Hy moet sommer sy vertaalmasjien gebruik om my woorde te interpreteer. Ek sien hy het waarskynlik reeds, want hy het (op my besprekingsblad) afgelei ek hou nie daarvan nie. Morne 16:11, 18 Mei 2007 (UTC)
Ek dink nie dit gaan regtig oor wie hiervan hou nie. Indien artikels nie aanvaarbaar is nie, sal dit deur admins verwyder word. Ek het nou sopas 'n paar vertalings gedoen in die artikel, maar ek dink nie dis perfek reg nie, want my kennis van Engels is ook nie te goed nie. Wikipedia is hier om ge-edit te word. Edits wat nie deug nie, word deur iemand anders ge-edit of verwyder. Ek gee glad nie om om direkte vertalings te hervertaal nie. Dis tog wat ek in elk geval sou doen as ek van die Engelse wiki sou vertaal - en dan sou dit moeiliker wees omdat ek nie Engelssprekend is nie. Adriaan ( SpreekBydraes ) ♪♫ 16:41, 18 Mei 2007 (UTC)
Vanaf Gebruikerbespreking:Francis Tyers.

I tried out your MT system and it looks like it might have potential. How would one go about helping to impove the system or develop translation in new language pairs? You mention in the geselshoekie that Dutch-Afrikaans should have potential. I would also be ionetrested in German-Afrikaans, particularly as the German Wikipedia is generally of good qualitry (often better than the Engelish equivalents). Does the online version have the latest language files?

Laurens 07:06, 20 Mei 2007 (UTC)

BTW - why did your project choose shallow-transfer machine translation? There seems to be a lot of energy on Statitical Machine tranlation, could one benefit from some of that work in your approach? Laurens 07:06, 20 Mei 2007 (UTC)

Opa! Sorry for the late replies... I chose shallow transfer because 1) It is easy to get off the ground, 2) It is fast, 3) It is open source. There are open source statistical systems (e.g. Moses), but this seemed like the best option at the time. 4) I've been involved with the group producing the engine (at Alicante) in packaging their software for Debian. 5) Statistical systems require huge parallel corpora, and even if these were available for Afrikaans<->English, the chances are they wouldn't be free (as in libre). I'll respond to your other points below. I'll email you on your emailrespond also on your talk page. - Francis Tyers 16:13, 20 Mei 2007 (UTC)

Responding to responses[wysig bron]

It would seem that, while most users think Francis' system is a potentially useful one, other are worried that:

  1. It would amass to a flood of thousands of articles that:
    1. Are poorly written (spelling and syntactical errors)
    2. Would require a tremendous amount of effort to fix.
    3. Aren't especially good prose
    4. Concentrates on obscure objects.

There would also seem to be a slight man vs. machine argument.

It would indeed seem that some articles lend themselves better to automatic translation: Khujand had relatively few errors, while Soedanese literatuur had significantly more, so it's a bit of a gamble.

I am surprised that I feel like supporting this project, even though I would normally support all of the above arguments.

I suppose that I can see the potential of something like this if it were to concentrate on, perhaps, the subjects listed in en:Wikipedia:Core topics - 1,000. Naturally I wouldn't dream of telling others what they should write about - that is not Wikipedia's policy. People are allowed to write on whichever subject they wish, no matter how obscure some might think it is.

I don't know how you go about choosing your topics - your past contributions have been so eclectic that it's been impossible for me to recognise a trend or a pattern. I do however feel that this Wikipedia is lacking articles on so many of the basic subjects, that a one or two paragraph stub on them would be invaluable, even if I had to go afterwards and "fix" everyone myself.

Regarding the other arguments listed above, I would like to say:

  1. Francis has agreed to only do one stub/article a day, which would hardly equal a flood of articles.
    1. The articles would be naturally short, so the amount or errors would be minimal, taking no more than a few minutes to corrent.
    2. See above.
    3. Few of the stubs here are, but if they provide the relevant information, that should suffice until they are expanded.
    4. Since when are we allowed to tell people what they are allowed to write about?

Even though I myself would not prefer to have articles on "obscure" subjects (at this moment in time), I realise that this is not the way Wikipedia works and that it would amount to a type of censorship to not allow certain articles. Even so, I find that many of his stubs aren't obscure - some people might find Soedanese literatuur obscure, but I think it's important. I suppose it depends on your field of interest and that everyone have different opinions on what is important and what not.

Francis, I would really like to hear why you choose the articles you do. If you choose specific ones for your own personal reasons, then I am fine with it. If, however, it is somewhat random, I would again like to point out the en:Wikipedia:Core topics - 1,000, which I myself try to use (now and then, along with the featured articles) when trying to decide which article to translate next.

Anrie 08:12, 20 Mei 2007 (UTC)

Ok, I chose the articles partly because I worked on them on the English Wikipedia (Sudanese literature), and partly because I'm interested in the area (Tajikistan). The only pattern is my interests. I would be happy to adjust my translating to articles that you direct. Actually I think the 'Core topics' would be a good way of improving the quality of the dictionaries. I hadn't thought of that before, but it is a good idea. - Francis Tyers 16:28, 20 Mei 2007 (UTC)
Ek dink alles wat almal nog oor die Afrikaanse Wikipedia gesê het, kom neer op gehalte bo getalle. Wikipedia bied 'n platform aan mense om kreatief mee te werk aan 'n ensiklopedie deur mense, oor mense (en visse, voëls, diere, berge, mere ens.), vir mense. As vertaalmasjiene (VM'S) in so 'n mate verfyn word dat hulle na aan volmaak is, skakel dit amper geheel en al die rol van menslike bydraers uit (buiten hier en daar regstellings), en sal al die tale wat Wiki's en VM's het, oombliklik 1,7 miljoen artikels kan hê (na die Engelse Wiki).
'n Beter oplossing sou wees dat gebruikers wat belang stel, toegang tot die VM verkry en dit dan as hulpmiddel by die skep van artikels gebruik, waar nodig. My Duits is byvoorbeeld verroes, dus met die skep van die artikel oor Curt von François sou dit gaaf wees om 'n VM te gehad het (selfs uit Engels in Duits), wat met enkele moeilike, onbekende woorde kon help. Anrie het eenmaal gesê ek moet veral let op my taal, juis omdat ek so baie artikels skep. Die VM kan van die Net afgelaai word. As ons dié werkswyse toelaat, kán daar voorkant toe 'n stortvloed geradbraakte artikels van ander gebruikers verskyn.
Boonop is dit meer dikwels meer tydrowend om ander se twak reg te maak, as om sommer gou self 'n saadjie te skep want dikwels lei dit weer tot meningsverskille en debatte soos dié een. (Ek kon in die tyd dat die reaksie my geneem het, reeds nog twéé Duitse kommissarisse van Suidwes-Afrika ingesit het. Miskien moes ek . . . Morne 10:11, 20 Mei 2007 (UTC)
The above, translated for Francis, who doesn't speak a word of Afrikaans... Anrie 10:28, 20 Mei 2007 (UTC)
I think that what most have said about the Afrikaans Wikipedia, rests on quality above quantity. Wikipedia provides a platform for people to work together creatively on an encyclopaedia by people, about people (and fish, birds, animals, mountains, lakes, etc.), for people. If translation machines ("VMs") are refined to such a point that they are near perfection, it would make the role of all the human contributors redundant (except for a few corrections here and there), and all the languages which have Wikis and VMs could have 1.7 million articles instantaneously (after the English Wiki).
A better solution would be for those users who are interested, to have access to the VM and use it as an aid when creating articles, where necessary. My German, for example, is quite rusty, thus, with the creation of an article on Curt von François it would have been nice to have a VM (even from English to German), which could have helped with 'n few difficult, unfamiliar words. Anrie has once said that I should concentrate om my language, especially since I create so many articles. The VM can be downloaded from the Net. If we allow this method, it can lead to a flood of messy articles from other users.
Notwithstanding that it is often more time-consuming to fix another's bad work, as to quickly create a proper stub, since it often leads to differing opinions and debates like this one. (In the time this reaction took, I could have added two more German commissaries from Southwest Africa. Maybe I should have...) Signed, Morné.
I have already responded to some of the points mentioned by Morné here, but he does make one very valid one: I have mentioned a few times that when a person continually contributes articles that are sub-par, they should be made aware of that and that we then expect of them to follow the guidelines on style, syntax, spelling, etc.
My understanding was that this machine would be able to learn from it's mistakes as it went along, and that, after a while, the errors would be few and far in between. Certainly we can't go deliberately correcting an automatically translated article every day for the next year, since this would have been unacceptable if it was a human user who transgressed.
Do you anticipate that your system will get better as time goes by, Francis, or what are your end goals with it? Anrie 10:35, 20 Mei 2007 (UTC)
Yes, it will get better, if it doesn't I hope you'd tell me so I know where I'm going wrong, and if I come up against an impass then I'll have to re-evaluate what I'm saying now. Currently the software is in extreme pre-alpha. For each article I translate from English, I first tweak the wording slightly in order for the machine to be able to translate it better, then translate it, then fix the formatting... as time goes on, and it improves then I hope that this pre-editing becomes unnecessary, and the end result improves.
A much better pair using this system (Spanish - Catalan), which has been in development for a number of years produces very good translations. I don't expect to get to that quality with English - Afrikaans for a few months yet, but it should improve dramatically as I add both vocabulary and transfer rules.
Just a short note that it won't learn automatically from the mistakes, I have to program it. So I submit and article, see what you change, and make the necessary changes to the machine. If I have any questions I'll ask you (if thats ok). - Francis Tyers 16:28, 20 Mei 2007 (UTC)
I would also like to point out that calling someone's work "twak" is derogatory and uncalled for and that users should refrain from deliberately using derogatory words when discussing another user or his/her work. Anrie 11:02, 20 Mei 2007 (UTC)
Ek moet om vergifnis vra dat ek my aan die onvergeeflike sonde van die eufemisme skuldig gemaak het. Wat ek eintlik wou gesê het, is dat dit tydrowend is om ander se "kak" reg te maak. There are, daar is, many ways, baie maniere, to make others' work, om andere se werk, sound like, na kak, shit, te laat klink. Morne 21:05, 20 Mei 2007 (UTC)

Ek sal probeer, I shall endeavour, om voortaan, to hence forth, meer direk, be more direct, te wees. Morne 21:08, 20 Mei 2007 (UTC)

As die Wikipedia, if the Wikipedia's, se hoofdoel, main aim is, om kak, to improve, te verbeter, other's shit, wil ek geen deel daaraan, then I want no part, hê nie, in it. Morne 21:11, 20 Mei 2007 (UTC)

Ek dink jy moet bietjie gaan afkoel en miskien môre weer na die gesprek toe terugkom, Morné. Dit is heel duidelik dat iets jou omgekrap het, maar ek dink nie dat jou huidige gemoedstoestand enigsins jou vriend is op die oomblik nie. Anrie 21:18, 20 Mei 2007 (UTC)
  • Sal dit nie miskien 'n goeie idee wees om hierdie gesprek na Wikipedia:Geselshoekie/MT te skuif nie? Dit is darem nou al baie lank. Anrie 21:42, 20 Mei 2007 (UTC)
Dankie vir die skuif. Morné, Francis het sy idee hier gegee en eers begin toe mense hier aangedui het dat hulle dit sal regmaak en hy kan voortgaan - ek is seker hy sou daarop ag geslaan het as daar groot teenkanting was. Ek glo dus sy optrede was redelik alhoewel ek ook nie mal is oor die idee van masjienvertaling vir die Afrikaanse Wikipedia nie (ten minste nie op die huidige standaard nie). Verder, soos wat sommige van ons nie hou van masjienvertaling nie, hou sommige van ons nou weer nie van sekere registers van Afrikaans nie. Ek hoop ons kan maniere kry om mekaar te akkommodeer. --Alias 09:35, 21 Mei 2007 (UTC)
Ek vra om verskoning dat ek my humeur verloor het. Ek wil net nie sien dat dubbele standaarde toegepas word nie. As ek nie by die letter van die reëls hou nie, word ek gou tereggewys (met woorde wat daarop neerkom dat ek nie moet twak aanjaag nie, hoewel die T-woord nie gebruik word nie), maar nou is daar 'n 4 000-woorde-debat oor die wenslikheid van masjienvertalings, wat 'n klomp ekstra en onnodige werk gaan skep. Selfs Google in al sy wysheid kan nie masjienvertaal dat dit na mensetaal klink nie. Vgl. hierdie reëls uit die Duitse treffer Ich weiß, es wird ein mal ein Wunder gescheh'n:

Die Duits: Ich weiß es wird einmal ein Wunder gescheh'n und dann werden tausend Märchen wahr. Ich weiß so schnell kann keine Liebe vergeh'n die so groß ist und so wunderbar! Wir haben beide den selben Stern und dein Schicksal ist auch meins. Du bist mir fern und doch nicht fern denn uns're Seelen sind eins. Die Engels (met hulp deur afkappingstekens uit te haal): I white it once a miracle will happen and then will become thousand fairy tales truely. I white so fast cannot offense love so largely am so marvelous and! We do not have both same star and your fate are meins you are far and nevertheless also me far from our souls are one. Wat moes gewees het: I know one day a miracle will happen and then a thousand fairy tales will come true. I know a love that is so big and wonderful cannot fade so fast. We both have the same star and your fate is also mine. You are far from me, and yet not far because our souls are one. (Ek ek is seker hierop kan nog baie verbeter word.) Morne 09:58, 21 Mei 2007 (UTC)


Core topics[wysig bron]

It might be a good idea to create a subpage -- also, do you have on Afrikaans Wikipedia a translation of the 'core topics' page? It might be a good idea to create it (with links), so people (myself included) can see what is done and what is not. Just an idea though... - Francis Tyers 08:38, 21 Mei 2007 (UTC)
It's on my to-do list, you bet. I'm a bit busy with Real Life stuff at the moment, but I should have more time starting July. Even so, I'll see if I can translate it a little bit at a time during off periods. It'll have to be aproved afterwards by the community - I've seen the English one has been accused in the past of being too subjective or partial to certain topics. Even so, I think it provides an extremely good base to work from.
I am curious as to why you have chosen Eng-Afr for this project. You've mentioned the Spanish to Catalan system, which sounds very promising. English differs so much from Afrikaans regarding syntax and vocabulary, it seems like a monumental project trying to develop a translating system for such a little-spoken language. (Not that I'm dissing my own language at all, just curious as to your reasons). Anrie 09:30, 21 Mei 2007 (UTC)
Ok thanks (on the core topics part).
I've been working with the Apertium project for a while, doing packaging for Debian. I started English-Afrikaans after seeing a post on Meta [1] from User:Greenman. I was interested in starting up a pair of my own and decided to ask him if he was interested in joining me. I made a basic system to demonstrate it to him, and have since been improving it in my spare time. Unfortunately he is too busy to help me.
English<->Afrikaans is not too bad a pair actually. The vocabulary is similar in many cases, although there are amiguities (prepositions as always are tricky). For Germanic languages, I'd say that they are the two that are closest together that are "widely spoken" (e.g. not Frisian, Plaatdutsch etc.). The morphology is greatly reduced in both in comparison to other Germanic languages, e.g. no cases, simple verb conjugation, favouring analytic over synthetic constructions.
This reduced morphology makes writing the dictionaries easier, and although I think that there will be issues regarding word-order (SVO vs. SOV), it is a worthwhile project. It could (in the future) help speed up the localisation of free software into Afrikaans.
Its worth remembering that almost any translation project is worthwhile, they can just have different focusses. For example, the Spanish-Catalan system focusses on 'assimilation and dissemination', whereas other more distance pairs might just focus on 'assimilation', e.g. understanding rather than producing high quality translations for publication. This is why we have a Welsh-English and Catalan-English projects in the pipeline too (actually the Catalan one is quite good too -- but not publishable quality). The aim is always for publishable quality, which is what I'm looking for, but 'assimilation' is a nice by-product. - Francis Tyers 12:40, 21 Mei 2007 (UTC)
Wikipedia:Kernartikels <-- guess this is is. - Francis Tyers 21:19, 21 Mei 2007 (UTC)

Hersiening[wysig bron]

Francis het die afgelope maande, in plaas daarvan om meer artikels by te dra, eerder daarop gekonsentreer om sy program te verbeter.

Hy het onderneem om in die toekoms:

  1. Op belangrike onderwerpe te konsentreer, bv. die by Wikipedia:Kernartikels. Saadjies oor hierdie onderwerpe het 'n baie goeie kans om deur ander gebruikers uitgebrei te word (anders as sommige van die menigte 1-sin saadjies wat ons het).
  2. Die saadjies sal van 'n aanvaarbare lengte wees, m.a.w. 'n paar paragrawe i.p.v. net 'n sin of twee.
  3. Daar sal slegs een so 'n artikel geskep word per week, indien so baie.
  4. Ek (of iemand anders) sal altyd byderhand wees om die artikel te wysig. Daar sal twee dadelike wysigings wees: die eerste wysiging is 'n wysiging wat bloot sal verseker dat die artikel 'n getroue vertaling van die bronteks is, ten einde terugvoering aan Francis te lewer waarvan hy sy vertaling kan evalueer en sy program verander. Die tweede wysiging sal die taal en styl verbeter.

Ek plaas dit maar net hier vir diegene wat wonder wat van die saak geword het op die ou einde. Anrie 21:13, 17 Augustus 2007 (UTC)