In summer 2005, I was granted a scholarship under the auspices of the Faculty Enrichment Program provided by the Canadian government that enabled me to spend two weeks at the University of Toronto and another two weeks at Queen’s University in Kingston. In Toronto, Jack Chambers was a very inspiring source for my work, provided me with much food for thought as well as with a number of his published and at the time still unpublished contributions to the field of Canadian English. At Queen’s, I received outstanding support from Janice McAlpine, the director of the Strathy Language Unit at that time, who most kindly provided me with workspace within the Unit, with precious insight into her work on the 2nd edition of the Guide to Canadian English Usage which was then under preparation and also with the Strathy Corpus. A modified version of the corpus is now hosted at Kiel University and is frequently consulted by Uwe Vosberg who is currently working on his post-doctoral dissertation (‘Habilitationsschrift’) dedicated to a comparison of the syntax of Canadian English, by many students at Kiel University and by myself. I have found it extremely useful to study the competition of British and American spelling standards, issues of Canadian English usage and verb complement patterns and their versatility.
I am also very grateful to Anastasia Riehl, the current director of the Strathy Language Unit for kindly suggesting to me to make a contribution to the series of blogs published by the Unit. When she got in touch with me, I had just been looking at various verbal patterns from the Strathy corpus in order to put my new syntactic theory called Passivisable Object Theory (or PO Theory) to the test against Canadian data. While I had restricted my attention to British English in my first outline of the theory (Meyer, 2009), I have now taken Anastasia’s invitation to contribute a blog as a welcome encouragement to provide this alternative introduction to PO Theory using exclusively Canadian data from the Strathy corpus.
Passivisable-Object Theory (PO Theory) and an improved syntactic description of verbal complements in Canadian English
1 A short outline of POTheory
Around 2006 I started thinking about developing a new model of English verb complementation. The reason for this was a growing dissatisfaction with current non-transformational models such as those presented in the Comprehensive grammar of the English language (Quirk et al. 1995) or the Cambridge grammar of the English language(Huddleston & Pullum 2005). It seemed counter-intuitive to me, for instance, to class predicates such as lack courage, weigh 15 kilos, resemble one's aunt, have a sense of humour and other non-passivisable structures as being transitive and as involving an object. I found it improper to lump them together with classic transitive structures such as write a story, shoot the enemy, buy some sugar whose complements are easily passivisable. In varieties of Modern English where, unlike e.g. in Latin or Modern German, the morphological opposition between a dative and an accusative had broken down, the criterion of passivisation seemed to be the only one by which objects could reasonably be delimited from other verbal complements. Notice, for instance, that subject and object complements (in be a hero viz. in make Susan captain of the club) as well as adverbials can all be realised as noun phrases following the verb.
Thus many traditional objects are no longer seen as objects in any strict syntactic sense and passivisability is assumed to be mandatory for objects (though this is not a sufficient condition). The model advocated here also attempts to avoid the pitfalls of taking a semantic definition based on participant roles and semantic equivalences at the outset that obscure significant syntactic similarities and dissimilarities of verbal constructions. It also seems to be true that more functional differentiation is needed within the area of complements hitherto classed as direct or indirect objects so as to capture more subtle changes in grammatical behaviour on the one hand and grammatical differences within varieties of English and of other languages on the other.
Another early consideration for developing PO Theory was to make it suitable for corpus annotation. This meant that the theoretical overhead was to be kept to a minimum and that it should not invoke more complexity than is required to account for the facts. It thus requires no deep-structure level of analysis, no abstract/empty categories and no crash course in symbolic logic.
The model is described in greater detail in an article I published in 2009 (see the bibliography at the end of this blog) - I will only highlight a few basic issues here. All full-sentence examples are taken from the Strathy corpus (in the form hosted at Kiel University). The record identifier appended to each example used below provides the original file name of the text along with the year of publication plus a unique number identifying each sentence in each file.
“Objects, or more precisely slots holding objects, must be passivisable. Slots which are not regularly passivisable will not be classed as objects. A constituent will be classed as an object if it is either directly passivisable or if it can be replaced by a passivisable NP without changing the semantics of the verb. Notice, however, that passivisation does not guarantee object status, i.e. it may extract NPs from intransitive structures such as He must have slept in this bed. ~ This bed must have been slept in.” (Meyer, 2009:214)
This has far-reaching consequences and involves a reconsideration of a great many basic structures of verb complementation. Notice first that the traditional sequence indirect object (Oi) – direct object (Od) is typically reanalysed in PO Theory as ‘Object + Complement Extension’ as illustrated below. All examples marked '[fi=FILENAME.EXT nr=RECORD NUMBER]' below have been taken from the Strathy corpus and illustrate authentic Canadian usage.
This is because only the first NP in either example can be passivised and hence qualifies as an object here. Present-day Canadian English almost entirely avoids the so-called second passive which would have justified the assumption of a second object: ??Some of the videotaped information was shown our jurors. - ?All these special gifts were given us (? = ‘somewhat awkward or objectionable’; ?? = ‘very awkward or objectionable’). Instead, this second direct passive has largely been given up in favour of what I will call a prepositional passive, which however, must be seen as being derived exclusively from a corresponding prepositional active:
The new analyses above are based on a redefinition of objects along the following lines:
An object (O) is defined as a licenced constituent of the verb satisfying the following criteria (simplified):
realisation: NP or clause (finite or non-finite)
position: in unmarked word-order, almost always fills the first slot after the verb
co-referentiality: not co-referential with the subject (exception: reflexives)
passivisation: must be passivisable or must be substitutable by a passivisable NP
Furthermore a 'complement extension' (CE) is introduced as a new category which is defined as a licenced constituent of the verb satisfying the following criteria:
realisation: NP or clause (finite or non-finite) (but never a PP)
position: fills the second slot after the verb (context: V + NP + CE)
co-referentiality: not co-referential with the preceding object or the subject
passivisation: not passivisable
Yet another non-traditional sentence element in PO Theory is the so-called 'predicator complement' (PC) adopted from Aarts and Aarts (though it is not fully isomorphic with their category of the same name). It is defined as a licenced constituent of the verb satisfying the following criteria:
realisation: NP, PP or clause (finite or non-finite)
position: fills the first slot after the verb where it can be an NP/clause, or a PP, or the second slot where it is realised as a PP or a clause (which cannot be replaced by an NP)
co-referentiality: not co-referential with the subject or the object
passivisation: not passivisable
A PC is assumed – instead of a traditional non-passivisable object – in predicates such as resemble somebody, have short hair, weigh ten tons, lack certain features, as illustrated by means of the following examples:
A PC is also assumed in the second complement slot after the verb where it can be realised as a prepositional phrase or a clause:
Since objects in PO Theory are passivisable by definition, another new element must be introduced for the pattern 'V + NP1 + NP2' or 'V + NP1 + PP' (where V = 'verb', NP = 'noun phrase', PP = ‘prepositional phrase’) if NP1 is not passivisable. This is true of predicates such as buy sb a new coat, cook sb a nice dinner, write sb a letter and others which have no passive, as we can see from the unacceptability of *She was bought a new coat, *You have been cooked a nice dinner. Such NP-complements are therefore analysed as 'benefactive complements' (BCs). In traditional grammar, most BCs have been analysed as (non-passivisable) indirect objects. A BC is defined as follows:
realisation: NP or clause (finite or non-finite) (but never a PP)
position: occurs as NP1 in the context: V + NP1 + NP2/clause
co-referentiality: not co-referential with the subject or the following NP
passivisation: not passivisable
The BC is inspired by Aarts & Aarts who referred to this constituent as a 'benefactive object (BO)' but in PO Theory it is set apart from objects because it is not passivisable. A BC is illustrated by the following examples:
A BC need not be followed by a CE (though this is the most typical constellation), as is shown in the following example:
Apart from these newly defined sentence elements, PO Theory retains the traditional subject attribute (SA) and object attribute (OA). These are defined as follows:
Subject Attribute (SA):
position: immediately follows a copular verb such as be, seem, appear, look
realisation: NP or AP or clause
co-referentiality: co-referential with the subject (i.e. refers back to the subject)
passivisation: not passivisable
Object Attribute (OA):
position: found as the second complement marked C in the construction 'Verb - NP - C'
realisation: NP or AP or clause (the clause must be replaceable by an AP/NP)
co-referentiality: co-referential with the preceding NP (usually the object)
passivisation: not passivisable
2 A consideration of some points of criticism
Once students have been made sufficiently familiar with PO Theory, they are able to apply it rather confidently to basic structures of English. As only the foundations of it have been outlined so far, the theory still needs to be tested against a plethora of data. I have collected some potential objections that I have received from various sides or audiences. A short consideration of some of them will help the reader to better understand basic assumptions of PO Theory and to avoid potential misconceptions. Possible objections will be presented in the form of questions (marked ‘Q’ below) and my answer will be marked as ‘A’. The first objection sketched below relates to the fact that PO Theory makes the following distinction between predicates like buy sb sth and give sb sth, based on the definition of a benefactive complement (BC) and an object given earlier:
Q: While it is generally acknowledged that predicates like buy sb sth or cook sb a good dinner are not passivisable, it has been argued that intuitively the two complements of buy and give above appear to be very similar and that an inflectional language like German encodes the first complement in either case as dative and the second as accusative. The semantic constellations also seem to be very similar in that the first complement encodes a beneficiary or receiver and the second a transferred entity that may also be rather abstract. Isn’t it therefore ill-grounded to make this distinction despite the semantic similarity and despite parallel case assignment, e.g. in German?
A: First of all, PO Theory is a syntactic theory and considers syntactic properties of a construction independently of semantics (unlike e.g. Construction Grammar). This does not mean that it is hostile to semantics or that a close inspection of the matching of semantic roles and syntactic functions is unimportant. It merely means that syntactic behaviour cannot comprehensively be predicted by semantics and that therefore an unprejudiced definition of syntactic elements is necessary if we want to do justice to the grammar of English predicates. Secondly, the theory is in line with native speakers’ syntactic intuitions: They usually know with some confidence whether a given predicate can be passivised or not. If semantic equivalence were essential for assuming syntactic equivalence, we would (based on traditional grammar) have to say that to Mary in (i) We gave the book to Mary is an indirect object because it is also an indirect object in (ii) We gave Mary the book and because the two sentences are largely synonymous (leaving differences of focussing aside). This would be syntactically untenable as to Mary in (i) not only differs from Mary in (ii) in terms of phrase type but also occurs in a different syntactic slot. The fact that both these complements translate into a German dative is irrelevant since nothing follows for English from German (or any other highly inflectional language) because generally speaking translated sentences are (ideally) only semantically though not necessarily syntactically equivalent to their source language and because the loss of the dative-accusative distinction in late Middle English and Modern English makes formal criteria such as passivisation and position all the more decisive.
Q: Your theory distinguishes between two functions of the noun phrase ‘the bike’ in (i) buy [O a bike] and in (ii) buy [BC your friends] [CE a bike]. Here, the same verb is involved – so why should a bike not be an object in (ii) when it is one in (i)?
A: Again, we must not be misled by semantic equivalence and we should not fall into the trap of assuming that (i) is just an elliptical (shortened) version of (ii). Rather, buy occurs in a monotransitive (i) as well as in what I have called a benefactive-intransitive construction (ii), just like sing is intransitive in He was singing but transitive in He was singing a lullaby. Notice that a bike occurs in the first NP slot after the verb in (i) but in the second NP in (ii) and that there is as little reason to conclude that the two slots should be syntactically equivalent as in the case of (i) (to) Mary in give Mary the book and (ii) give the book to Mary. Here traditional grammar interestingly had analyzed (to) Mary as an adverbial or prepositional object (= a PO in PC Theory) in (i) and as a direct object (=O in PO Theory) in (ii) (against the semantic equivalence of the two slots!) Rather it seems to be an important grammatical property of Modern Standard English that – with very few exceptions – only the first slot after the verb is passivisable and that verbs like buy, write, cook that more typically occur in the simple frame ‘V – NP’ additionally disallow the passivisation of NP1 in their extended frame ‘Verb NP1 NP2’ (cook sb a nice dinner). Verbs like give and show, by contrast, prototypically occur with two participants after the verb (show the guests the way) and allow for the passivisation of the first.
Q: Are you aware of the fact that Huddleston & Pullum considered the possibility of distinguishing between passivisable objects as in own two yachts and non-passivisable complements such as have two yachts but rejected it? They preferred to analyse both as direct objects, stating the following:
“Whether or not there is an acceptable related passive for a given active clause depends on the interaction of pragmatic, semantic, syntactic, and lexical factors: it cannot satisfactorily be reduced to a simple matter of the presence or absence of O in the active. Example [iii] [=He has drunk out of this glass – This glass has been drunk out of] illustrates the second point, that it is not only transitive clauses that have related passives.” (2002:246)
A: Notice that in PO Theory passivisation was taken to be a necessary but not a sufficient condition for objects in English. This leads to a distinction between PC-intransitive predicates such as have [PC two yachts] and transitive ones such as own [O two yachts]. Interestingly, the two predicates are semantically fairly similar yet differ with respect to passivisation. This shows that the difference in grammatical behaviour cannot straightforwardly be derived from semantics. I still find it implausible to put the so-called ‘middle verbs’ that include non-passivisable predicates such as have two yachts, weigh a ton, resemble one’s father, lack courage, run two miles into the same category as transitive predicates like shoot the enemy, use the main road, send a postcard, believe the story, perceive the difference. While the two groups can be systematically distinguished with respect to passivisation, Huddleston & Pullum provide no single argument for lumping them together (as monotransitive constructions). In line with Aarts & Aarts, PO Theory prefers to see the middle verbs closer to verbs taking a licenced PP, such as complain [PC about the issue], dispose [PC of the right to appeal], steal [PC into the garage] (said of thieves), awake [PC from a restless sleep] without making it impossible, where necessary, to further distinguish between a PC-NP (= PC realised as an NP) and a PC-PP (PC in the form of a PP). While Huddleston & Pullum are right in assuming that drink out of this glass is an intransitive structure (analysed in PO Theory as drink [A out of this glass], along the lines of sit [A in this chair], where A = ‘adverbial’), their uniform treatment of intransitive verbs followed by a PP as in sit in this chair and transitive prepositional predicates such as [Vprep call for] [O drastic measures], [Vprep look after] [O Emily], [Vprep enlarge on] [O a topic] 1 (examples provided here with my analyses) seems ill-conceived. While passives are found in both groups, passives of predicates like sit in this chair are comparatively rare (given the overall frequency of ‘sit + adverbial’) and are subject to special pragmatic restrictions in a way that look after or enlarge on + NP are not. This correlates with sound lexicographical practice: While enlarge on, look after etc. are treated as functional units (e.g. in Cowie & Mackin 1993), no prepositional verbs such as sit in or drink out of have ever been posited.
1 Such analyses are actually oversimplifications. Transitive prepositional verbs like enlarge on a topic are systematically ambiguous (as shown in Quirk et al. 1985:1156) between the two structures [Vprep enlarge on] [O a topic] and enlarge [PC on a topic]. An elegant way to handle this in PO Theory is to say that whenever the syntactic frame ‘Vprep + O’ is specified, this automatically implies the alternative analysis ‘V + PC’ but not vice versa. The handling of phrasal and prepositional verbs in PO Theory will be addressed in a new paper in preparation.
Aarts, Jan & Flor Aarts. 1982. English syntactic structures. Functions and categories in sentence analysis.New York, London et al: Prentice Hall.
Cowie, A.P. & Ronald Mackin (eds.). 21993. Oxford dictionary of phrasal verbs. Oxford: OUP.
Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: CUP.
Meyer, Matthias L. G. 2009. “Revisiting the evidence for objects in English.” In: Ute Römer & Rainer Schulze (eds.). Exploring the lexis-grammar interface. Amsterdam: Benjamins, 211-227.
Quirk, Randolph et al. 1985. A comprehensive grammar of the English language. Harlow: Longman.