Idafa is conceptual, not syntactic. Sib. makes this explicit in Article 100 باب الجر, where he explains the "particles" by saying that they make idafa between two nouns. But he gives the example of أخذته من عبد الله, explaining that "min" relates the taking to abdullah. Clearly he's using idafa to talk about how relations are expressed. So the idafa qua syntax is an expressive device rather than mere syntactic form; it serves to express a relation or "association" between one thing and another. Compare Brandom's account of logical vocabulary as expressive rather than epistemological, etc. Analogously we can say that the syntax of idafa is an expressive rather than a mere syntactic device. For Brandom, logic is about expressiveness rather than forms of proof; similarly, the idafa construct is about expressiveness rather than formal syntax. Thus it is about meaning before it is about form.

Not addition: addition is accumlation, combination. To add one thing to another is to produce a third thing greater than its parts: a combination or sum. This is emphaticallly not what Sībawayhi had in mind when he used the term idafa. He tells us explicitly what the term means: nisba, relation or association.

Thus an idafa construction encodes a kind of proposition whose force is to declare a relation between the two parts, not their sum (as the standard translation as "annexation" etc. implies).

see art 512: nisba may be the primary sense of idafa Caution: 318 is bab of idafa being nisba meaning: nisba expresses idafa conceptually, but it does not have jarr-majrur form, demonstrating that the concept idafa is not limited to jarr-majrur construct. So Arabic can express idafa in two ways: nisba and construct. Compare: زَاْدُ رَجُلُ ثَقِيْفٍ, ``Zayd [is] a man of Thaqîf'', and زَيْدٌ ثَقَفِيٌّ, ``Zayd [is] Thaqafy''. Sībawayhi uses the term idafa for both; see Articles 318 and 100. BUT ALSO: art 154 (nidā') uses idafa to characterize the relation between النِّدَاءُ, الْمَدْعُوُّ, and المَدْعُوُّ لَهُ (with a ref. to sabab as well).

Six (at least) formally distinct cases of idafa: 1) jarr-marjur; 2) nisba; 3) form فَعَّالٌ for صَاْحِبَ شَيْءٍ يُعَاْلِجُهُ e.g. لِصَاْحِبِ الثِّيَاْبِ ثَوَّاْبٌ, لَبَّاْنٌ etc. (art 341) 4) form فاعِلٌ for مَا يَكُوْنُ ذَا شَيْءٍ وَلَيْسَ بِصَنْعَةٍ يُعَاْلِجُهَا e.g. لِذِي الدِّرْعِ دُرَّاْعٌ وَلِذِي النَّبْلِ نَاْبِلٌ etc. (art 341) 5) lam (maftuh) of nida' to characterize the relation between النِّدَاءُ and الْمَدْعُوُّ 6) lam of the supplicated-for to characterize the relation between الْمَدْعُوُّ and المَدْعُوُّ لَهُ (art 154)

{\ar{إِضافَة}} \hspace{8pt}{\itshape\arxen{ʖiḍāfaħ}}: affiliation; {\ar{مُضاف}} \hspace{8pt}{\itshape\arxen{muḍāf}}: affiliandum; {\ar{مُضاف إِلَيهِ}} \hspace{8pt}{\itshape\arxen{muḍāf ʖilayhi}}: co-affiliandum better: association

WARNING: in art 146:2 uses في instead of إلى: اِعْلَمْ أَنَّ النِّدَاْءَ كُلُّ ٱسْمٍ مُضَاْفٍ فِيْهِ what is the referent of the pronoun in فيه? aha, it refers to النداء, not to the ism. he's referring to e.g. يا عبدَ اللهِ, where the first term is مضاف in the نداء expression.

contraries: مضاف and مُفْرَد (see art. 49, 146)

Article 100 هذا باب الجرّ is mostly about idafa. The task is to show how this term "pulling" corresponds naturally to simple, direct intuition.

Idafa is more than just jarr and majrur. Sib explicitly states وإذا قلتَ مررتُ بزيد فإنما أضفتَ المرورَ إلى زيد بالياء.

صفة - attribute /of/. "The tall man" - "tall" is an attribute /of/ the man. Tallness is attributed /to/ the man. Adjectives are attributes /of/ things; nouns are attributed /to/ things.

إضافة - attribution /to/. "Book of Zayd" - the book is attributed /to/ Zayd; it is not an attribute /of/ Zayd.

Kinship relation? Lisaan, article أمم: كلُّ قوم نُسِبُوا إِلى نبيّ فأُضيفوا إِليه فَهُمْ أُمَّتُه

Now the term "attribution" captures this notion (really a kind of theory about the world) simply, succinctly, and intuitively. There are many other English words which would serve just as well, such as assignation, affiliation, and so forth. Arabic also has many words that could be used to characterize this semantic situation. But the tradition settled on one, {\ar{إِضافَة}} \hspace{8pt}{\itshape\arxen{ʖiḍāfaħ}}, which turns out to be very difficult to translate into idiomatic English.

The English terms discussed do not serve well as translations of إضافة, although they capture the underlying idea well enough. "Attribute" comes from the Latin "tribuere", to give; to attribute one thing to another is to bestow the former on the latter. "Affiliation" comes from the Latin "filius", son; it uses a kinship relation metaphorically. "Assignation" (or "assignment") comes from "signum", mark or token, giving something like "to identify out by mark or sign". The Arabic إضافة does not involve any of these meanings. It a verbal noun based on the radicals ض ي ف , whose basic sense is "guest". In contemporary Arabic it may be used to mean "to host someone as a guest", but it has also come to mean something like addition or annexation. Most scholars writing about Arabic grammar in English use this sense to translate the technical term إضافة , rendering it as "annexation" or the like; but this concept of addtion is clearly not what Sibawayhi and his colleagues had in mind. The evidence for this is simply the semantic model: attribution of a book to Zayd clearly does not involve adding or annexing it to Zayd.

The Lisaan uses three notions to discuss the meaning of إضافة إلى. First, the core sense of guest/hospitality. It means something like "to lodge someone at/with someone else". It also makes clear that another core sense of the word is {\ar{إِمالَة}} \hspace{8pt}{\itshape\arxen{ʖimālaħ}}: inclination, tending, leaning toward. In this sense, إضافة means "inclining someone or thing toward another". Finally, it says relates it to the notion of proximity; it means something like "to draw someone or thing near to another".

أضافه إلى فلان : he presented him as a guest ( {\ar{أَنْزَلَهُ ضَيْفًا}} \hspace{8pt}{\itshape\arxen{Ɂanzalahu ḍayfᵃⁿ}}) to someone.

أضافه : he found him a guest ( {\ar{وَجَدَهُ ضَيْفًا}} \hspace{8pt}{\itshape\arxen{wajadahu ḍayfᵃⁿ}}). Or "deemed"?

So أضاف الكتابَ إلى زيدٍ : he presented the book as a guest to Zayd. (?)

NB: the form {\ar{أَفْعَلَهُ}} \hspace{8pt}{\itshape\arxen{Ɂafʕalahu}} has at least 10 different senses (per Hamlawy). The relevant sense here is مصادفة الشيء على صفة. E.g. أحْمَدْتُ زيدًا I found Zayd to be characterized by حمد (laudable?) (= صادفْتُه محمودًا. Another possibility: التعريض. So أرْهنْتُهُ المتاعَ وأبَعْتُه أي عرضْتُهُ للرهن وللبيع I offered it for wagering or sale. Accordingly for أضاف الكتابَ ألى زيدٍ may be read:

He found the book as guest to Zayd He offered the book to Zayd as a guest.

But more likely is simple transitivity? We go from قَعَدَ زيدٌ ``Zayd sat'' to أَقْعَدْتُ زَيْدًا I seated Zayd; or, in more traditional form, أَقْعَدْتُهُ فَقَعَدَ ``I seated him, so he sat.'' So ضاف زيدًا ``He landed at Zayd's as a guest'', أضَفْته زيدًا I made him a guest at Zayd's (or the other way around?) Introducing إلى removes the ambiguity.

So: المضاف would mean that which is made a guest, and المضاف إليه would mean that to whom/which something was made a guest. This is on the assumption that the fundamantal semantic of this verb strongly include the root notion of guest.

كتابُ زيدٍ interpreted as كتاب مضاف إلى زيد a book made a guest to Zayd.

NB: a guest lodged at Zayd's household is not essential to Zayd, any more than a book attributed to Zayd is essential to Zayd. By the same token, the role of guest is not essential to the guest. So this kind of attribution involves contingency, which is captured by the root ضيف. Etymology: guest, from: guest, enemy, stranger. host, from lord of strangers

So a core sense of إضافة qua guesting is that it describes a situation involving a stranger. "Attribution" does not capture this but it is a part of the core semantis; a thing attributed to another that is not of its essence is a stranger to it.

NB: this is a social metaphor, so we should expect its sense to include notions of social practice and obligation.

Key points:

1. The combination of two words in the إضافة construction is construed as a single اسم (token). This is also the case for the combination of a noun and adjective; رجل طويل is construed as a single token:

١٠١: فقولك «مَرَرْتُ برجلٍ ظريفٍ قَبْلُ» فصار النعْتُ مجرورًا مِثْلَ المنعوت لأنّهما كالاسم الواحد

باب ١٠١ ص ٤٩٣: وإذا قُلْتُ «مَرَرْتُ بِرَجُلٍ قائِمٍ» و «مَرَرْتُ بِرَجُلٍ قاعِدٍ» فهذا اسمٌ واحِدٌ.

Note the second passage states "this is one token", not "this is like one token".

BUT: not token, idafa is a semantic concept kitabu zaydin is not just a single token, its an ism that expresses a single thing

The same thing applies to e.g. «كِتابُ زَيْدٍ» [TODO: find the passage where this is explicitly claimed.]

This is entirely intuitive and obvious in both cases. For the N+Adj phrase, Sibawayhi explicitly lays out the semantics: when you say "I passed by a charming man previously" «مَرَرْتُ بِرَجُلٍ ظَريفٍ قبلُ» "you do not mean a single one of the men each of whom is "a man", but you mean a single one of the men each of whom is "a charming-man; for it is "unknown" (anonym) and is so because it is of a population (أُمَّة) each of whose elements is the like of its token. That is because each one of the men is "a man", and each of the charming men is "a charming-man". Thus its token mixes it with its population so that it is not known among them." (Art. 101)

فأما النَّعْت الذى جَرى على المَنْعُوتِ فقولك «مَرَرْتُ بِرَجُلٍ ظَريفٍ قَبْلُ» فَصار النَعْتُ مَجْرورًا مِثْلَ المَنْعوتِ لِأنَّهُما كَالاِسْمِ الواحِدِ وإِنَّمَا صَرَا كالاسمِ الواحدِ مِنْ قِبَلِ أنَّك لَمْ تُرِدْ الواحِدَ مِن الرِّجَالِ الَّذينَ كُلُّ واحِدٍ مِنْهُمْ رَجُلٌ ولَكِنَّكَ أَرَدْتَ الواحِدَ مِنْ الرِّجَالِ الَّذينَ كُلُّ واحِدٍ مِنْهُمْ رَجُلٌ ظَرِيفٌ فَهُوَ نَكِرَةٌ وإِنَّمَا كَانَ نَكِرَةً لِأَنَّهُ مِنْ أُمَّةٍ كُلُّهَا لَهُ مِثْلُ اسْمِهِ وذٰلِكَ أَنَّ الرِّجَالَ كُلَّ واحِدٍ مِنّهُمْ رَجُلٌ والرِّجَالُ الظُرَفَاءُ كُلُّ واحِدٍ مِنْهُمْ رَجُلٌ ظَريفٌ فاسْمُهُ يَخْلِطُهُ بأُمَّتِهِ حَتَّى لا يُعْرَفَ منها

The key point here is that the combination of N+Adj is construed as a single token (اسم) on strictly semantic grounds; the ("indefinite") combination denotes a class of individuals each of whom is characterized by both components of the combination.

NB: the N+Adj combination has compositional semantics; the whole is equivalent to the sum of its parts. The operation of combining does not introduce any specific semantics beyond combination. This is not the case for the idafa.

In the case of the idafa construction (assignation, attribution, affiliation, etc. but not "annexation"), the semantics are more complex but the combination is construed as a single token.

Such combinations are akin to lexical coinages; they add a (temporary) term to the lexicon; but since such terms are compositional and thus can be decomposed into more basic components, they are not considered part of the lexicon.

Now the Tractional disposition (حال الجرّ) is associated solely with the idafa. So we must explain both terms and show how they are both intuitive and how they are related.

The core idea is that idafa associates the first term to the second, producing a single combined concept (just as N+Adj does), and "pulling" is the metaphor that describes how the first term affects the second term morphosyntactically: it comes first and must "attract" a second component in order to form the combination. So "idafa" is a semantic term, and "jarr" is a term of morphosyntax.

ʔiḍaʌfah: the Lisaan says something like "place someone as a guest, incline it toward (the host), draw it near (to the host). To see why this metaphor is intuitive, consider the examples of «كتابُ زيدٍ» "the book of Zayd", which supports at least the following possible interpretations:

possession (the book that Zayd owns) authorship (the book that Zayd wrote) topic (the book that is about Zayd)

Note first of all that we have here a single term (idafa) that covers multiple possible meanings. But idafa is not the name of a syntactic case; it is rather a semantic term that covers the meanings associated with combinations in which the second term is in the jarr case. Thus it abstracts from the specific meanings like possession, authorship, etc. The idea seems to be that the meaning of the first term is modified by inclining it toward the meaning of the second term. Also, being a semantic concept, idafa refers to things, not words. We can express this abstractly in English with terms like assignment, affiliation, even attribution, and similar terms. To declare كتابُ زيدٍ is to relate/assign/affiliate/etc. one thing (a book) to another (a man named "Zayd"), without regard to the specific nature of the relationship. But the root of idafa (ض ي ف) provides a kind of minimal metaphor to capture the common semantics of all such combinations: the first element (thing, not word) is inclined toward, lodged at, hosted by, etc. the second element. [NB: it is not "annexed" to the second term (etymology: annexe: c.1386, from O.Fr. annexer "to join," from M.L. annexare, freq. of L. annecetere "to bind to," from ad- "to" + nectere "to tie, bind"; but idafa in no way conveys the sense of joining, binding or tying. Relation, yes, but not binding or annexing. Related things remain distinct, unlike annexed things.]

IOW, we're talking about the semantic structure of relation, not just syntax. Just as the terminology of verbs starts from analysis of the abstract structure of events. This is an essential point: we start from a kind of theory about the world, and then explain how language describes that structure. (Or: how linguistic expressions correspond to it.)

So the basic idea, that the first thing is assigned to the second in combination, is simple and intuitive. We see a book, and we know Zayd owns it, or authored it, or the like. The "thingness" of the book is inflected by our knowledge of its relation to Zayd; it isn't just any old book, but the one characterized by that relation. A caution is in order: this is not philosophy; we're not talking about the ultimate true nature of the book or of Zayd. Rather this is a kind of pragmatic phenomenology(?); we're constructing a mental model of the way the world works with just enough sophistication and formality to provide an explanation for the way language works in and with the world. Things in the world are a) characterized by a huge variety of attributes like color and size, and b) enmeshed in a complex network of mutual relations like possession, etc.; c) located in time and space; d) what else? N+Adj combinations address a), and idafa combinations address b.

Consider any object in the experienced world, and you can construct an infinite number of linguistic combinations to highlight the object from various perspectives: big book, red book, etc using adjectival combination, and similarly for relational combinations (Zayd's book, the book of swimming, etc.)

It's not by accident that Sibawayhi's arrangement puts articles about صفة in close proximity to articles about إضافة.

[NB: time/space location for nouns (excluding verbals). Compare big book, Zayd's book, and "a book behind Zayd" (خلفَ زيدٍ): here again idafa, but this one relates a location to Zayd, not the book; the book is then related to the location that is related to Zayd.

In constrast to N+Adj combination, the idafa combination does not have compositional semantics; the whole is greater than the sum of its parts, since it will always have some specific semantics (e.g. possession) that is not made explicit linguistically, but rather is captured abstractly by the term idafa. It is a specific kind of combination of things.

NB: Terminology and concepts are always based on canonical forms; the canonical idafa combines things, but the terms may also be adjectives or even verbal clauses. The abstract notion of assignation is common to all such combinations, however.

Another point: since these are construed as single tokens, they cannot be split without breaking the semantics. This also explains why عمل does not extend past its argument. An adjective is inflected by matching its object, not by the elaboration of the element that works on its object. e.g. رأيت رجلًا طويلًا - the noun rajul is elaborated by the verb, but the adjective is not; rather it follows the inflection of its noun. The combination N+Adj forms a single token because of the semantics, not the syntax.

In the case of idafa, the work of the first term only extends to the second term; if you interject a third element between the two, the idafa is broken since the work of attractation (and correlation) cannot span the interjected element to reach the assignee. The valence of attractation only extends to the neighboring word.

art 317: لأنَّ المضافَ من حدِّ التسمية also: المضاف إليه منتهى الاسم وكمالُهُ

lam-al-idafa: art. 176, 508

Ryding's article in Encyclopedia of Arabic Lanuage and Linguistics: “In the Kitàb Sìbawayhi, iḍāfah and cognate terms are very frequent (Mosel 1975:205–207): the verb ±a∂àfa occurs 233 times, iḍāfah 243 times, and mu∂àf/mu∂àf ±ilayhi 183 times (Troupeau 1976:132). They do not always indicate the same phenomenon, however. Troupeau distinguishes between the translations ‘annexer à’ and ‘relier (un individu) à’ for the verb aḍāfa, probably in order to differentiate between noun/noun constructions and preposition/noun constructions, but it is not quite clear which criteria he uses in assigning the loci to either sense. Talmon (2003:236–238) points out that aḍāfa may be used generally for any preposition linking a verb with a noun (e.g. Kitàb I, 177.11 yuḍāfu bihà ilà l-ism mà qablahu aw mà baʕdahu). Hence, all prepositions may be called ḥurūf al-iḍāfah (e.g. Kitàb II, 146.11; Owens 1990:14–17; art 397?). Elsewhere, Sìbawayhi restricts this use to the preposition li-, which is called làm al-iḍāfah (Kitàb II, 331.2; art 508).”

But this ignores the rest of article 508, which also lists مِنْ and باء as حروفُ الأضافة. It also overlooks the fact that art. 397 is explicitly about a restricted class: حُرُوْفِ الْإِضَاْفَةِ إلَى الْمَحْلُوْفِ بِهِ.

Wrong on several counts. \begin{itemize} \item “Talmon (2003:236–238) points out that iḍāfah may be used generally for any preposition linking a verb with a noun (e.g. Kitàb I, 177.11; art 100: yu∂àfu bihà ilà l-ism mà qablahu aw mà ba?dahu)”. Now “iḍāfah may be used generally for any preposition linking a verb with a noun” may be what Talmon points out, but it is not what Sib says (in article 100, هذا بابُ الجرّ). Most obviously, “preposition” is not one of Sib's words. He states quite explicitly that these huruf (neither nouns nor ẓarfs) serve as intermediaries in expressing an idafah relationship: وَأَمَّا الْبَاْءُ وَمَا أَشْبَهَهَا فَلَيْسَتْ بِظُرُوْفٍ وَلَا أَسْمَاْءٍ وَلٰكِنَّهَا يُضَاْفُ بِهَا إِلَى الِاسْمِ مَا قَبْلَهُ أَوْ مَا بَعْدَهُ The key phrase being بِهَا: iḍāfah may be expressed by (i.e. using) the ḥurūf al-jarr. For example: وَإِذَا قُلْتَ مَرَرْتُ بِزَيْدٍ فَإِنَّمَا أَضَفْتَ الْمُرُوْرَ إِلَى زَيْدٍ بِالْبَاءِ - that is, the mudaf is not the ba', but the (implicit) murur expressed by the verb. The ba' is a harf jarr by which the idafah is expressed; it is not a itself a harf of idafa. Furthermore in passage cited he refers to الباء وما أشبهها, by which he does not mean exclusing the huruf al-jarr. In fact he expressly lists examples that do not involve verbs or “prepositions”(?), e.g. يَا لَبَكْرٍ, رُبَّ رَجُلٍ (but: he says rubba is in the same category as bi etc.) and: وَإِذَا قُلْتَ أَنْتَ فِي الدَّاْرِ فَقَدْ أَضْفَتَ كَيْنُوْنَتَكَ فِي الدَّاْرِ إِلَى الدَّاِْرِ بِفِي Then there's the question of why he does not see idafah in something like أنت خَلْفَ عبدِ الله. Why doesn't he say أضفْتَ أنت إلى عبد الله بخلف? Perhaps because it is a zarf, so it expresses more than a mere relation of association? More likely (equivalently?), because xalfa is mansub and fiy is not. \item “Hence, all prepositions may be called ḥurūf al-iḍāfah (e.g. Kitàb II, 146.11; art 397; Owens 1990:14–17).” But in the article cited, 397, Sib clearly indicates that it he is talking explicitly about oaths, and the way ḥurūf al-jarr may be *used* to express idafah relations. He nowhere claims that “all prepositions may be called” ḥurūf al-iḍāfah. NB also his list includes تَاء القسم and واو القسم, which are not(?) huruf al-jarr. Main point being simply that using the term "preposition" just confuses things, since it is not one of Sībawayhi's terms. \item “Elsewhere, Sìbawayhi restricts this use to the preposition li-, which is called làm al-i∂àfa (Kitàb II, 331.2; art 508).” But in the passage cited (art. 508) he does no such thing. He does mention the lam of iḍāfah, but he clearly does not *restrict* such usage; he also lists "min" and "bi" as terms of iḍāfah. Again he is addressing a specific topic, namely عَدّة ما يكون عليه الكلم. He contrasts the lam with kaf, ba etc. calling the former lam al-idafah and the others kaf al-jarr etc. But that must not be construed as a restriction. After all he also mentions the waw and ta' of oaths, which in article 397 are expicitly called huruf al-idafah. Rather, his point is merely that lam, as a harf al-jarr, may be used to *express* iḍāfah. He is not offering a theoretical categorization of ḥurūf. \end{itemize}