The interchange in Arabic of the consonants H and T in the feminine ending1. Evidence of the alternation also occurs in Hebrew, so it must be very ancient, going right back to their common ancestor in a branch of Semitic.

The letter is ة and its Unicode value is ة

In primitive Semitic the feminine ending was -at, and it was followed by case endings such as nominative u, accusative a. This was what we see in ancient languages like Akkadian, and is still the case in classical (Quranic) Arabic in connected text. However the case endings were lost very early in Hebrew, so that Biblical texts do not have them.

When spoken in isolation, or at the end of a sentence, the final short vowel of an Arabic word is dropped2, so that baytu "house" is pronounced bayt. With the feminine ending there is an additional change, with the now exposed final consonant T changing into an H. So malikatu "queen" is pronounced malikah, not malikat, when nothing follows it.

In the Arabic script this is written with the basic letter H, but with the two dots belonging to the letter T (ta) added to the H. It is written like this whether it's the continued-T or the pausal-H form. This compound H/T is called ta marbuta. The H is a closed loop in shape, and marbuta means 'joined'. It's regarded as a form of the letter ta. The ordinary T is called ta tawila.

Some speakers of good classical Arabic do pronounce the H: it's like the soft H in English "behind". (Don't give it a throaty rasp -- those are different Arabic letters.) But it's weak, so in modern Arabic it is usually silent. So you will see alternations like Fatima and Fatimah, hamza and hamzah. And in fact this node could be called ta marbutah.

Notes.

1. With some exceptions: it is used as a masculine ending in a few words like khalifah 'caliph' and in the numbers, e.g. thalathah 'three (m.)', thalath 'three (f.)'
It also turns a plural into a singular for some collective nouns: ward 'flowers', wardah 'flower'.

2. With some exceptions: this is a morphophonemic occurrence, not a strictly phonological one.


Not for the faint-hearted. Mercuryblues and Gone Jackal have both expressed a doubt about the -ah pronunciation. So what is the evidence that the ending usually pronounced -a and written with a form of H was ever pronounced with a consonant H? It's actually quite equivocal: I can't come to a firm conclusion myself.

First, I think I was taught to say it, with a voiced h, and my teachers of Classical Arabic had it. This was a long time ago, but I've always maintained that distinction in my pronunciation, different from words that end in plain -a.

My book on Arabic, A.S. Tritton's Teach Yourself Arabic, is not the last word but is fairly precise. About this sign it just says:

Even in old Arabic the feminine termination at was often pronounced ah and written with h. When they began to study their own grammar they rectified the existing spelling by putting the two dots of t over the h ة ... -- and this hybrid must be pronounced t.

I found nothing pertinent on the Web on the linguistic point, but learnt something of the history of the Quran, in which the explanation might lie. In the lifetime of the Prophet Muhammad the Quran was mainly oral, with many of his followers memorizing part or all of it and repeating it to him for his approval; the traditions of who recited what are handed down as evidence. It was first written down entire in the time of his successor Caliph Abu Bakr. It existed in seven ahruf (dialects or accents, plural of harf). Under Caliph Uthman the text in the harf of Muhammad's tribe, the Quraish of Mecca, was made the sole official text and the others destroyed.

These ahruf differed in minor matters of pronunciation.. This explains some of the consonant anomalies of the standard Quranic text: such as hamza not being represented by a consonant but by a diacritic; and the ta marbuta. There were no vowel signs yet, and no dots to distinguish different consonants. So the feminine ending was written with an H, pure and simple, at that stage.

Having got an H in the Uthmanic recension, it had to keep its written H forever after. That doesn't mean it was pronounced. It might have been silent (with the alternant -at-). There is another term in the history of the Quran, qira'ah (plural qira'at), literally 'reading', which is (I think) a more specific tradition of recitation passed from person to person, resolving fine details where the text can be read in more than one way. The transmission of pausal ta marbuta as -ah or as -a might count as variant qira'at.

History of the Quran discussed at:
http://www.quran.org.uk/ieb_quran_tawil.htm
http://salam.muslimsonline.com/~islamawi/Quran/Text/Qiraat/hafs.html
http://muslim-canada.org/quran_collation.html