1. Know that there are nouns from this class that appear without any added letter, and the final root he is lacking; therefore they consist of only two letters. Some of them are always pointed with qamats, some always with pataḥ, and some sometimes with qamats and sometimes with pataḥ; they are determined according to the received tradition, such as ʾav, yad, dag, qav (“line”), as in qav la-qav (Isa. 28:10), and tsav, as in tsav la-tsav (ibid.). All of these overlap with the classes of verbs whose middle radical is quiescent and with the geminate roots; in the following principle I will give signs and rules by which you may distinguish these from those. Some are also pointed with ḥiriq and a quiescent yod; these are nouns derived from roots whose middle radical is one of the letters vav, yod, or he. Thus from the root ravah one says ri, as in af beri (Job 37:11); from nahah, ni, as in qinah (“lamentation,” Ezek. 32:16); from tsayah, tsi, as in ve-tsi addir (Isa. 33:21); and from kavah, ki, as in ki taḥat yofi (Isa. 3:24). In all of these the yod stands in place of the root he, and the middle radical is absent.

2. However, pi (“mouth of”) does not belong to this class, for its root is peh, consisting of only two letters, following the pattern seh. The form pi is the same both in the construct state and with the first-person singular suffix; that is, there is no difference between pi YHWH (“the mouth of the LORD,” Num. 14:41) and pi ha-medabber ʾalekhem (“my mouth that speaks to you,” Gen. 45:12). Properly it should have been piyyi, but because of the heaviness of the two yods, one of them was dropped. From the context you can recognize in each case whether it is a construct form or a pronominal suffix.

3. As an example with zakh: if you wish to know from which class it comes, add at the end either a plural ending, a pronominal suffix, or the feminine he, and you will find zakhim, zakhyo, zakha. Since all of these have a dagesh, they belong to the class of geminate roots. Likewise from ḥag one says ḥagim, ḥago. But from dag one says dagim, dagah, and the gimel is soft; therefore it belongs to the class of weak roots. I have already given you a sign in this treatise, in the Eighth Principle, by which you may distinguish verbs whose middle radical is weak from those whose final radical is weak. I said there that verbs whose middle radical is weak never change. Therefore, since one says dag but dege ha-yam (“the fish of the sea,” Gen. 9:2), where the qamats changes to sheva, it belongs to the class whose final radical is weak, and its root is gadhah. But from ger one says gerim, gero, and the tsere does not change; therefore it belongs to the class whose middle radical is weak, and its root is gur.