The Construction Grammar model

Kay and Fillmore’s Construction Grammar model is monostratal. This means that it contains only one level of syntactic representation rather than a sequence of structures linked by transformations, a feature that characterises transformational generative models like Principles and Parameters Theory. Furthermore, the representations in Construction Grammar contain not only syntactic information but also semantic information relating to argument structure as well as pragmatic information.

Kay and Fillmore’s Construction Grammar contains a number of generalised constructions that underlie more specific constructions like the WXDY construction. Because this is a non-derivational monostratal model, it does not have any phrase structure rules that assemble words into phrases and sentences. Instead, it has constructions that represent syntactic patterns. For example, the model has a head-complement construction that represents the structural relationship between a lexical head (for example, a verb) and its complement(s) (for example, the object(s)). This construction captures the basic structural relationship that holds across different categories (for example, VP, AP, PP, NP). The model also has a subject-predicate construction, which captures relation ships between, for example, subject NP and predicate VP. In addition to various construction types, the model also contains a number of principles that ensure, for example, that categorial features of a lexical head will be shared with the constituent headed by that phrase (e.g. a verb heads a verb phrase), or that constituents local to a head, with the appropriate features, can be recognised as complements. We do not concern ourselves with the details of these principles here, beyond pointing out that readers familiar with HPSG will notice a number of striking similarities between these two models.

The various constructions that make up Kay and Fillmore’s Construction Grammar model are linked together via an inheritance relation. This means that more specific constructions inherit the properties of more general constructions. For example, the VP construction inherits all the information in the head-complement construction, but adds further information concerning the category of the head and the fact that the VP requires a subject in order to complete the valence requirements of the head. As we will see, the WXDY construction also inherits the properties of several more generalised constructions. To illustrate these properties, we will look in detail at the Construction Grammar representation of the WXDY construction, which is represented in Figure 19.3.

Although this diagram appears rather complex, we provide a ‘translation’ below, based on example (14).

According to Kay and Fillmore, this construction is headed by the verb be (the form is in our example), and the category (‘cat’) of the construction as a whole is therefore V. Like HPSG, and indeed like Langacker’s Cognitive Grammar, Kay and Fillmore’s Construction Grammar approach views the verb as the head of the sentence. This is the information that appears in the top set of brackets in Figure 19.3, marked ‘syn’. This is an abbreviation of ‘syntax’ and labels the construction as a whole in terms of its categorial status. Kay and Fillmore’s claim that the verb be heads the construction rests upon their view that this is not a progressive auxiliary but a copula, a discussion that we do not pursue here.

The next set of brackets, labelled ‘sem’, provides information about the semantic and pragmatic properties of the construction. The information ‘frame :incongruity judgement’ provides information about the pragmatics of the construction. The term ‘frame’ refers to the scene described by the sentence over which the pragmatic value ‘incongruity judgement’ is held by a ‘judge’. This ‘judge’ is labelled ‘prag’, which means that the identity of this judge has to be pragmatically resolved. In other words, the identity of the ‘judge’ may or may not be the speaker, depending on the context in which the construction is uttered. In our example, the ‘judge’ is likely to be the speaker, but if this example were a case of reported speech (e.g. . . . and then she said ‘What is Lily doing under my bed?’) the judge would be the person referred to by the speaker as she. The incongruity judgement is held by this ‘judge’ with respect to a situation labelled #1. In our example, we can paraphrase the situation over which the incongruity judgement is held as ‘Lily being under my bed’.

The largest set of curly brackets, labelled ‘val’ (valence), provides information about the structure of the construction. The first part, [rel [gf subj]], identifies a unit with the relation (rel) of grammatical function (gf) subject (subj). This is the X in the WXDY construction. In our example, the X corresponds to Lily, the subject of the verb be.

The largest set of square brackets, below the information about the subject, corresponds to the rest of the construction, headed by doing. In other words, Kay and Fillmore argue that doing Y forms a constituent. In our example, the string doing under my bed is the relevant part of the construction. Within these square brackets, the top brackets labelled ‘syn’ tell us that this part of the construction is headed by doing, which has the category V. Observe that this element is marked as having a negative value for negation (‘neg –’). This is how Kay and Fillmore capture the fact that the WXDY construction does not license negation of the verb doing, as we saw in example (11b). Indeed, we might ask why they be is not marked in the same way, given example (11a). The next set of brackets marked ‘sem’ tells us that the semantics of this part of the construction correspond to the situation over which the incongruity judgement holds. In other words, it is the Y part of doing Y that fills in the information that the ‘judge’ holds to be incongruous. The next set of brackets marked ‘rel’ tell us that this part of the construction, doing Y, is the complement of the verb be. Again, this rests on Kay and Fillmore’s views concerning the constituent structure of the construction which we do not pursue here.

The second largest set of curly brackets, labelled ‘val’, tells us about the structure of doing Y. One important aspect of the construction that this model needs to account for is how the main clause subject comes to be understood as the subject of the Y predicate. In other words, in our example, Lily is under stood not only as the subject of the verb be, but also as the subject of under my bed. This means that both instances of [gf subj] have to be linked in the construction. In Kay and Fillmore’s model, this linking is done by an independent construction that they call the coinstantiation construction (Kay and Fillmore 1999: 23). We do not concern ourselves here with the details of this.

A further point of interest relating to this part of the construction concerns the element what, which is not yet accounted for. Observe that the construction starts by telling us about the head (be), the subject (Lily) and the complement (doing Y), but has not so far told us anything about what, the first element in the construction. This is because what is part of a wh-dependency relation. The label ‘wh’ is shorthand for interrogative words like what, who, where and so on. The term ‘dependency’ relates to the idea that two positions in a structure are related. Consider example (15):

Although the question word what occurs in clause-initial position in the question, it is nevertheless interpreted as the object of doing, which is illustrated by the fact that the answer to the question, some dusting, occurs in the object position (after the verb). The question of how to account for dependencies like this is a recurring theme in models of grammar. Transformational generative models like Principles and Parameters Theory account for wh-dependencies by means of a movement analysis: the wh-phrase is moved by transformation from object position to clause-initial position. Non-transformational theories, which are monostratal, favour a different explanation. Like HPSG, Kay and Fillmore’s Construction Grammar model accounts for the wh-dependency by means of a filler-gap analysis. This means that the construction simultaneously represents the wh-phrase (filler) in its clause-initial position and the position in which the wh-phrase is interpreted (gap), and links the two together.

 Kay and Fillmore’s account of the wh-dependency rests on what they call the left isolation construction. The term ‘left isolation’ expresses the fact that (at least in languages like English) the wh-phrase occurs in the leftmost position in the clause, and is ‘isolated’ in the sense that it is separate from local constituents. The left isolation construction has two ‘daughters’, which means that it consists of two main constituents. (The term ‘daughter’ is inherited from syntactic theories that rely upon tree diagrams.) The left daughter is the wh-phrase, which is linked or unified with one of the arguments that is required by a predicate within the right daughter. In other words, the wh-phrase satisfies one of the valence requirements of a non-local predicate. In the WXDY construction, what is interpreted as the object of doing.

 Let’s look again at Figure 19.3 in order to see how this information is rep resented. Observe that the set of square brackets that tells us about the object of doing marks this part of the construction as [loc]. This means that the unit that satisfies the object requirement of doing is non-local. The information ‘inherit what’ tells us that the non-local constituent what is to be unified with this position in the structure, thus fulfilling the valence requirement of doing. This account of the filler-gap dependency can account not only for part of the syntax of the WXDY construction, but also for wh-interrogatives in general. This illustrates an important aspect of Kay and Fillmore’s theory, which is that ‘regular’ and ‘idiomatic’ constructions should in large part be accounted for by the same theoretical machinery.

However, the expression what that occurs in the WXDY construction, despite sharing syntactic properties with the expression what that occurs in ‘ordinary’ questions, does not share the referential features of the ‘ordinary’ question word what. In other words, observe that what in the ‘ordinary’ question in example (15Q) picks out some entity or event (the thing that Lily is doing) whose identity or value is questioned. This licenses an answer along the lines given in (15A), which fills in the required information about that entity or event. In contrast, given the pragmatic import of the WXDY construction, which identifies the function of the construction as the expression of an incongruity judgement, the what in this construction does not have referential value. In other words, it does not pick out an entity or event in the same way that the ‘ordinary’ question word what does, because it does not require an answer like ‘dusting’ or ‘hiding from George’. This explains why an answer like the waiter’s in example (9) is not licensed. This ‘special’ feature of what is marked in the WXDY construction by the information {[ref Ø]}, which tells us that what does not have referential value. Observe, however, that the position of what is not marked in the WXDY construction. This is because the left isolation construction is an independent construction from which the WXDY construction inherits certain properties. This illustrates how certain generalised properties of idiomatic constructions are accounted for in Kay and Fillmore’s model.

 The next set of square brackets provides information about the Y part of the construction. In our example, this part corresponds to under my bed. The representation tells us that Y is a complement of doing in Kay and Fillmore’s analysis. The representation also links the semantics of Y to the situation over which the incongruity judgement holds, which is tagged as #1 throughout the representation of this construction, as we have seen. The information about the semantics of the frame or situation is left blank in this diagram because this is a generalised representation of the WXDY construction rather than a representation of a specific example. In other words, one of the features that identifies the WXDY as a formal idiom is the fact that it provides a syntactic ‘template’ into which a potentially infinite set of specific lexical items can be inserted. The construction tells us that this unspecified Y constituent will contain at least one argument that corresponds to the subject requirement of the Y predicate. This is why the information about the argument of Y and its subject are linked by the tag #2.

 In sum, the WXDY construction has a number of ‘regular’ syntactic features, which it inherits from other less specific constructions. Firstly, the WXDY construction contains head-complement structures and subject-predicate structures, which means that it inherits the properties of these two fundamental constructions that underlie ‘regular’ as well as idiomatic constructions.

Furthermore, the properties of these basic constructions are inherited in turn by the specific categorial instantiations of these generalised constructions, such as VP, PP and so on. In these respects, the idiomatic construction shares much in common with all ‘regular’ constructions. Secondly, the WXDY construction inherits the syntactic properties of the left isolation construction, which is involved in ‘regular’ interrogatives as well as in this idiomatic construction. Thirdly, the WXDY construction inherits the properties of the coinstantiation construction, in order to link a single NP to the subject valence requirement of two distinct predicates.

 In addition to its ‘regular’ properties, the WXDY construction also has a number of features that identify it as idiomatic. As we have seen, these features not only involve the morphosyntax of the construction (in terms of the form of the verb doing, or the restriction on negation, for example), but also crucially involve the meaning of the construction, which brings with it a striking and idiosyncratic interpretation that cannot be straightforwardly predicted from the parts that make up the construction.

اخر الاخبار

اخبار العتبة العباسية المقدسة

قسم شؤون المعارف يصدر العدد 35 من مجلة الخطوة

بالتزامن مع حلول أولى ليالي القدر المباركة شعبة التوجيه الديني النسوي تنظم البرنامج التفاعلي أروقة رمضانية للزائرات

ضمن برنامج يبلغون.. المجمع العلمي ينظم 8 مجالس قرآنية رمضانية في محافظة البصرة

​استعدادًا لليالي القدر.. قسم الشؤون الخدمية يستنفر جهوده في تنظيف محيط مرقد أبي الفضل العباس (عليه السلام) وغسله

المزيد

الآخبار الصحية

مركب طبيعي في القهوة يحمينا من التهابات اللثة

داء السكري لا يرفع السكر فقط.. بل يزيد خطر الإصابة بأمراض القلب!

اكتشاف سبب غير متوقع لضعف الذاكرة والتركيز

نصائح علمية لتقوية دماغك وتقليل التدهور العقلي

المزيد

الآخبار التكنلوجية

باحثون يرصدون أثر تغيّر المناخ في دم البشر

الجميع يحلم.. لماذا يتذكر البعض أحلامهم بينما ينسى آخرون؟

حرق الأمونيا.. كيف يساعد محطات الكهرباء في اليابان على إزالة الكربون؟

الشمس تقذف نتوءا عملاقا باتجاه قطبها الشمالي

المزيد

مواضيع ذات صلة

Goldbergs construction grammar

The architecture of construction grammars

The Generalisation Commitment

Comparing Construction Grammar with Cognitive Grammar

Construction Grammar: a broadly generative model

Kenya Pidgin Swahili (KPS)

On pidgins and other restricted linguistic systems

Conclusion Some cognitive abilities of animals

Grammaticalization in animals?

Language-like abilities in animals

Discussion Some cognitive abilities of animals Problems

Taxonomic concepts