Using Adhesiontext for Testing Arabic Spacing and Kerning

Vladimir Tamari's picture

I have reached the stage of testing the spacing in the font I am designing, and was happy to find an excellent tool to provide text for that purpose. Elsewhere on Typophile Miguel Sousa has described his adhesiontext online tool for generating words using a given set of letters in various languages including Arabic. It is a great resource and should be incorporated into Fontlab!

I installed my new font and set it as the default font in my Firefox browser. adhesiontext allowed me to see hundreds of Arabic dictionary words displayed in my font. My question here is whether there is a way to generate random nonsense words, also of paired combinations of chosen glyphs? A choice of point size would also be a great addition to this tool. Thanks Miguel. www.adhesiontext.com/

Vladimir

Miguel Sousa's picture

Thanks Vladimir.

> My question here is whether there is a way to generate random nonsense words, also of paired combinations of chosen glyphs?

For that I think you're better off using Tim Ahrens' Just Another Test Text Generator. Tim talks about it in this thread http://www.typophile.com/node/32756

I'm not sure that having fake words is ideal. I think that if you don't use real words you won't get the correct texture and the usual letter combinations for that given language. For example, if you're testing a typeface to be used mainly for setting German, you should make sure that the text content is German, otherwise you won't get the correct frequency of accented and unusual letters (ÄäÖöÜüß), and uppercase letters (in German, nouns -- such as 'dog', 'Hund' -- all start with a cap letter).

I must say that the text sample delivered by Tim's tool comes pretty close to that desired texture, but still. But then, adhesiontext isn't perfect either as its algorithm doesn't have any considerations regarding the language syntax. I guess that's the trade-off of having random and dynamic text, as opposed to "Lorem ipsum"-like text or text samples grabbed from Wikipedia, for example, that are static.

> A choice of point size would also be a great addition to this tool.

The tool was written to be a text generator, not a typesetting tool ;^) (Over time I've been happily surprised to hear about how people are using adhesiontext) In any case, the whole page will scale when you change the type size of your browser. I hope that works for you.

Miguel Sousa's picture

BTW, I'm not sure if you'll be able to test the kerning of your font this way.

Vladimir Tamari's picture

Thanks Miguel for your detailed and interesting reply. This is my first font, and was not sure what was the best procedure. I agree with you that seeing whole word-shapes is critical for judging a font, but I was worried that, as in proofreading text, the eye ignores a mistake because the sentence has meaning (reading the text backwards works against this tendency). That is why I suggested random words and the link you provided does that nicely. Oh if in the future you have time, perhaps you can find a way to incorporate diacritical marks (harakat) to the Arabic text in adhesiotext that would make it even more useful.

On the other hand,I do agree with you that real words must be used at some stage. By the way speaking of word-shapes, Arabic has a particular advantage over Latin here, because the majority of the letters have geometrically unique shapes (if the dot pattern is included as a shape element), serving legibility. At first glance a line of Arabic appears untidy compared to Latin type, but words have generally unique outlines. Latin tends to have a generally rectangular word-shape as high as the x with occasional bumps when capital letters, ascenders or descenders occur.

You are right about text size as a browser function.

To help me study kerning pairs I have devised an analog method to quickly test hundreds of pairs of glyphs systematically one against another, and hope to detail images of the method soon. Is there a digital method, apart from using ready-made class lists?

Vladimir

Syndicate content Syndicate content