LINGUIST List 23.4714

Mon Nov 12 2012

Review: Text/Corpus Ling; Discourse Analysis: Prado-Alonso (2011)

Editor for this issue: Monica Macaulay <>

Date: 12-Nov-2012
From: Alan Huffman <>
Subject: Full-verb Inversion in Written and Spoken English
E-mail this message to a friend

Discuss this message

Announced at
AUTHOR: Carlos Prado-AlonsoTITLE: Full-verb Inversion in Written and Spoken EnglishSERIES TITLE: Linguistic Insights. Studies in Language and Communication. Vol. 127PUBLISHER: Peter LangYEAR: 2011

Alan Huffman, Program in Linguistics, Graduate Center of the City University ofNew York


Full-verb inversion (FVI) is one of the most oft-studied phenomena in Englishlinguistics, particularly from a functional perspective. There is substantialdisagreement among analysts as to just what is to be studied under this rubric;however, (1) would be included by most or all:

(1)On the table burned a candle.Adv. - Verb - Subject

Example (1) exemplifies a particular construction: Preposing of some constituentbefore the verb plus placing of the subject after the verb. It constitutes aproblem in linguistic analysis because of the non-canonical ordering of subjectand verb. Carlos Prado-Alonso’s (P-A) book reviews the history of the problem,details its many subcategories, giving reasons for either including them in hispurview or excluding them, and offers a review of the literature. This reviewis one of the most useful aspects of this book, being very thorough except forone omission (the significance of which is discussed below). P-A is concernedthat discussions of FVI have neglected two points: a) occurrences of FVI inspoken language, and b) in written language, the distribution of FVI infictional versus non-fictional texts. He introduces data from four largecorpora to elucidate these points. P-A refers to his own work as an“analysis”. However, it does not actually present an original approach tounderstanding FVI; rather, it introduces new kinds of data for thealready-existing approaches and evaluates their ability to incorporate this data.

In Chapter 1, P-A chooses what subcategories of inversion to include and whichto exclude. Broadly speaking, the excluded examples fall into two categories.

(I) The form of the inversion itself differs from that of FVI. There isanother form of inversion involving subject and only part of the verb, when theverb form has more than one word, which P-A calls “subject-operator inversion”(e.g. p. 19 ff.), or “subject-auxiliary inversion” (pp. 25, 137):

(2)How did the candle look on the table?Aux. - Subj. - Verb

There is an area of potential ambiguity between the two constructions: the verbs‘am/is/are/was/were’ may be inverted alone, without an auxiliary, in the latterconstruction, rendering such examples formally indistinguishable from FVI.However, as discussed below, there are both formal and semantic diagnosticswhich can usually disambiguate the two.

(II) The canonical form of the inversion construction, as in example (1), isnot present. This excludes such types as verb-first inversion (3) and quotationinversion (4):

(3) Came a terrific flash of lightning and clap of thunder.(4) “We always thought Perot would cause people to take a new look at the race,”said Charles Black.

Also excluded are existential-‘there’ examples (5), containing an additionalstructural element ‘there’, which P-A asserts is a “dummy”, i.e. devoid ofsemantic content; preposing (6) and left-dislocation (7), which in any case donot contain subject-verb inversion, and equatives (8), which P-A claims (p. 43)“do not allow us to distinguish between the subject and its complement.”

(5) There was a storm last night.(6) Three coins of gold I found yesterday.(7) Matthew, he is the one to be blamed for such bad results.(8) Dr. Jekyll is Mr. Hyde in Stevenson’s novel.

Included in P-A’s purview, then, are: adverb phrase inversion (9), adjectivephrase inversion (10), prepositional phrase inversion (11), noun phraseinversion (similar to and sometimes hard to distinguish from equatives) (12),verb phrase inversion (with a participial phrase placed first, similar toadjective phrase inversion) (13), and subordinator inversion (14):

(9) Therein lie the reasons for Clinton’s confidence that he can stave off anyBush comeback.(10) Prominent among inversions is full inversion.(11) Among them was the seriously injured driver of the Sprinter, Steve Carpenter.(12) A key connection was Frank O’Hara.(13) Gathered together are paintings that reveal his interest in linguistics.(14) Such were the practical results of the commissioner’s efforts to impose ascheme that no one in the locality had wanted.

Chapter 2 is a survey of modern research on FVI. Useful bibliography isprovided throughout the chapter. P-A justifiably gives short shrift tosyntactic accounts, since these largely ignore details of usage and because theflagship syntactic analysis - a “root transformation” - is contradicted bywell-known examples. He instead focuses on the numerous and diverse functionalaccounts. Green (1980, 1982) lumps FVI together with the formally distinctsubject-operator inversion (example 2), and, not surprisingly, discovers aheterogeneous variety of functions for “inversion”, making no serious attempt tounify them. Birner (1992, 1994, 1996), discussed in Huffman (2002), views FVIfrom the standpoint of the organization of the construction (example 1) as an“information-packaging” strategy, which allows the mentioning of older or morefamiliar information (the clause-initial constituent) before newer or lessfamiliar information (the inverted subject). However, Birner’s own data containnumerous examples in which the postposed information is not newer than thepreposed information (about a quarter of her corpus in Birner 1992). Inaddition, it is of course quite common for new information to be introducedwithout FVI, and Birner does not attempt to differentiate these cases.Furthermore, Birner’s information-packaging rationale does not answer thechallenge posed by example (15), in which the given¬-new relation betweeninitial constituent and following subject obtains as much as in example (1), yetthe canonical subject-verb order is used:

(15)On the table a candle burned brightly.Adv. - Subject - Verb

Dorgeloh (1997), like Green, pools FVI data with subject-auxiliary inversion;Dorgeloh, however, suggests that the use of non-canonical word order indicatesthe introduction by the writer of an element of subjectivity into the discourseand allows the writer to express a point of view and in some sense manage thereader’s attention. P A finds this rationale vague and untestable, in that itought to apply to any kind of non-canonical word order; P-A thus appears toimpute to Dorgeloh an iconic view of this use of word order. I would add that,like Birner, this analysis also fails to meet the challenge of example (15); andsince two distinct types of subject-verb inversion have been mingled here, oneis tempted to conclude that any consistent pragmatic effect is due only to theclause-initial preposing of a constituent that is neither subject nor verb, aconclusion that sits poorly with the idea of FVI as a “construction”.

Kreyer (2006) draws data from the British National Corpus. He finds that theconstituent postposed after the verb in FVI is likely to contain more words thanthe preposed constituent, which P-A, citing Green (1980), says is because thelonger constituent is more likely to contain new information. Kreyer also findsthat preposed information is relatively more retrievable from precedingdiscourse than postposed information, and that the postverbal position is usedto introduce or shift to a new topic or subtopic by putting in final positionthose elements that link to what is coming next. With all parts of theconstruction thus cognitively or communicatively motivated, one might concludethat the construction itself is an unnecessary entity, obscuring the fact thatthe same motivations would apply even when only one of these parts, but not theentire construction, is present. P-A himself, though, does not draw thisconclusion.

Chen (2003) offers an analysis of FVI in the framework of cognitive linguistics. Chen views FVI as an instantiation of the Ground-before-Figure model, theground being represented by the initial locative or other constituent, and thefigure by the postverbal subject. He takes examples with a preposed locativeand ‘am/is/are/was/were’ (16) as prototype and those with verbs of motion (17)and those with non-spatial preverbal constituents (18) as radial extensions ofthe prototype.

(16) On the table was a candle.(17) Into the room darted Lopez.(18) Of great concern to us is the shortage of qualified candidates.

Chen then attempts to relate these three types to discourse genres: description,narration, and exposition. P-A, being interested in the distribution of FVIacross genres, critiques Chen mainly on this point. However, there are a numberof more fundamental analytical questions not raised by P-A, which surely must beclarified first. 1) If Ground-before-Figure is the cognitively natural way torelate a subject to a spatial background, it is unexplained why the greatmajority of examples, even with locatives, use canonical subject-verb order, notFVI. 2) The challenge of example (15), which introduces constituents inGround-before-Figure order but does not invert subject and verb, remainsunaddressed. 3) The choice of the straight locative type (16) as prototype and(17) and (18) as derived is asserted but not proven. 4) If (16) does indeedrepresent the prototype, it is quite easy to see (17) as being derived from it;but the large category represented by (18), which includes most non-locatives,does not seem derivable from (16) without a strong appeal to metaphor (“This ...type facilitates the introduction of an item into a space of ideas andarguments”, P-A p. 100). Since Chen finds that this is the only type of FVIinversion used in exposition, such an exclusive appeal to metaphor seems highlyimplausible. Note also that Chen would apparently exclude example (1), becauseit does not contain ‘am/is/are/was/were’, and defines out of his purviewexamples like (3), verb-first inversion, and (4), quotation inversion, as well. Finally, P-A repeats from Chen without criticism a confusing and erroneousdiscussion regarding transitivity and inversion (P-A p. 102). Both authorsreject an alleged “transitivity constraint” on inversion, but create confusionwhen they present examples (example 103) of transitivity in the preposedconstituent, whereas the constraint is generally held to apply only to the mainverb itself. An analytical error is perpetuated by examples like (19) (example102 in P-A) purporting to demonstrate the lack of S-V inversion when there is anobject:

(19)a. Lopez pushed Davis through the revolving door.b. * Through the revolving door pushed Lopez Davis.Subj. - Obj.

Subject precedes object in both. However, in the rare but actually-occurringexamples of FVI with transitives, the object in fact precedes the subject(examples from Bolinger 1977:102):

(20)In the tower strikes the hour a clock of many chimes.Obj. - Subj.

(21)In that realm held sway a hated despot.Obj. - Subj.

In summary, P-A does a good job of presenting the positions of the variousanalyses in the literature, but comes up somewhat short in discerning andarticulating their weaknesses and analytical flaws. His interest lies mainly inpragmatic considerations and genre differences. But he ignores more fundamentalconsiderations of structural analysis, without which one cannot proceed to dealwith the pragmatics.

Chapter 3 describes the corpora used in this study, as well as the samplingtechniques and search methodology employed. The corpora are: For writtenEnglish the Freiburg-Lancaster-Oslo-Bergen Corpus of British English (1991), andthe Freiburg-Brown Corpus of American English (1992); for spoken English theInternational Corpus of English: the British Component (1990-1993), and theCorpus of Spoken Professional American English (1994-1998), for a total of 1.8million words. Both the written and the spoken corpora were subdivided intovarious genres, the written comprising both fiction and non-fiction. Analysisof the corpora was mainly done manually, but automated analysis was employed forthe spoken material.

In Chapter 4, P-A applies the analytical ideas discussed in Chapter 2 to thecorpora described in Chapter 3, with, again, a particular view to determiningwhether there are significant differences in occurrence patterns of FVI in a)fictional vs. non-fictional written discourse, and b) written vs. spokendiscourse. P-A leans very heavily on the view of the FVI construction asiconic, especially in seeing the location of the first constituent as helping topromote a text-structuring purpose. He applies a subcategorization of FVItypes, introducing first the categories “obligatory full inversion” and“non-obligatory full inversion”, and then reintroducing the subcategoriesenumerated in (9) - (14) above. P-A concludes that FVI is actively exploited infiction as well as non-fiction, and in speech as well as in writing. Thedifferences lie in favorings of the subcategories within each genre. Heconcludes that fiction and non-fiction differ greatly in their use of thevarious types of obligatory full inversion. As for non-obligatory fullinversion in written texts, he concludes that “fiction and non-fiction do notdiffer in the overall distribution of the construction but rather in thedifferent types of non-obligatory full inversion used, and in the differentfunctions that these inversions serve in both genres” (p. 183). Concerningspoken vs. written language, he concludes that writing makes more extensive useof non-obligatory full inversion than of obligatory full inversion, whereas inspeech the reverse is true. P-A offers explanations of these skewed frequenciesof FVI across the different genres as a function of the iconic nature of the FVIconstruction and in consideration of the appropriateness of FVI to meet thedifferent communicative demands of those genres.

In the evaluation to follow, two points in particular will be developed. 1) Theproblem of separating out FVI from “operator-(auxiliary-)subject inversion”,which has seriously plagued many previous analyses, is not completely squelchedhere either. Since it is unclear how many such examples are mixed in to P-A’sdata, it is unclear to what extent these data overall are reliable. 2) Thechallenge of example (15) has not been clearly addressed here either, and thischallenge reveals a serious underlying theoretical issue that cannot gounresolved if any collection of data is to be meaningful.


Full-verb inversion is a complicated object of analysis, both formally andfunctionally. Adequate and consistent controls need to be applied to its manyvariables. Without such controls, it is not even clear which data constitutethe object of study. P-A, on p. 26, deliberates whether syntactic-formalcriteria or pragmatic-semantic criteria are the appropriate ones to apply. Theanswer is, of course: both. This point recalls the Saussurean notion of thelinguistic sign, a unified object comporting both a formal and a semantic side -both a signal and a meaning. In recent times, the control of the sign has beenapplied most rigorously and consistently by William Diver and his successors(see Diver, Huffman, Davis 2012). P-A himself opts to favor syntactic criteria;but even here the analysis appears sometimes to stumble. As noted earlier, FVIcan be indistinguishable from auxiliary-subject inversion witham/is/are/was/were, which do not take auxiliaries in circumstances where otherverbs do and can be directly inverted with the subject:

(22)a. FVI:Near that table stood a candle. / Near that table was a candle.V - S / V - S

b. S-Aux Inv.:Never did a candle stand near that table. / Never was a candle near that table.Aux. - S - V / V - S

However, one can usually disambiguate these cases by switching to acompound-tense form:

(23)a. FVI:Near that table had been a candle.Aux. - V - S

b. S-Aux. Inv.:Never had a candle been near that table.Aux. - S - V

Subject-auxiliary inversion carries a distinct semantic content as well: Itintroduces some element of assessment of the probability or likelihood ofoccurrence of the event denoted by the verb. Either the event is presented asdirectly questioned, or only potentially occurring, or as non-occurring; evenfor an actually-occurring event, the occurrence is presented against thebackground of some question having been raised about its occurrence, or asoccurring contrary to some expectation. This is clearly quite different fromwhat is signaled by FVI. Applying these formal and semantic disambiguators,then, at least the following examples and tables classified by P A as FVI wouldappear to merit reevaluation:

Chapter 1: 37 (with ‘have’), 63.Chapter 4.1: 22, 23, 37, 51, 117, 118Chapter 4.2: 15, 17, 36Tables 24, 31; Figures 27, 29.

The identification of FVI requires evaluation of subtle semantic effects, andone must wonder how much of this can be left to a machine. The identificationof noun-phrase inversions, to take one salient example, clearly requires humanevaluation. Examples in the book sometimes seem inconsistent. For example, onp. 29, “Near the fire is colder.” is rejected because “the initial prepositionalphrase ... serves as a subject”; but on p. 147 “... now is the appropriate timeto make significant changes” is accepted, even though the adverbial ‘now’ alsoseems to serve as a subject. Examples (35) and (36) on pages 149-50 deal withinversion after ‘first’, ‘second’, ... ‘finally’. Example (35) contains thesentence “Finally, probably far more common than either of the other forms ofassault and harassment are the beatings of ...”, with ‘finally’ underlined, butthe preposed adjective phrase ‘far more common’, which is the more immediatecorrelate of the inversion, ignored; example (36) has: “Finally, fifth in thehierarchy is the lower class...”, with ‘finally’ underlined, but ‘fifth’ignored. Some of these questionable examples represent classes included in thedatabase (as in the first list given above, where eight of the eleven examplesare negative correlatives), and the implications for the validity of thedatabase as a whole are not immediately clear.

Our first point of evaluation has focused mainly on the formal side of theequation; the second point focuses more on the functional side: What are thefunctional units, actually? FVI consists of two parts: the initial placement ofsomething other than subject or verb, and the ordering of subject and verbrelative to each other. Functional analyses have emphasized thetext-structuring effect of first position, an iconic use of word order. P-A, wehave seen, maintains that the effect of the construction is due to iconicity.Word order is without a doubt functional; but it is not always iconic. It issometimes semiotic: an arbitrary signal of a meaning analogous to the plural -sor the past-tense -ed in English. The present author (Huffman 2002, notincluded in P-A’s literature review), has proposed that while the preposing isindeed iconic and serves a broad text-structuring purpose, the two orderings:subject-verb and verb-subject are signals of two meanings. According to thishypothesis, briefly, the two orderings signal two different degrees ofattention-worthiness attributed by the writer or speaker to the event andparticipant denoted by the subject and verb. Subject-verb ordering indicates agreater degree of attention-worthiness, verb-subject ordering a lesser degree.This choice is made separately from the choice of whether to prepose, althoughthe two may work hand in hand to achieve a particular type of communicativeeffect in a text. Thus, this work has shown that S-V tends to be used for majorcharacters and V-S for minor characters; V-S with verbs denoting low-key typesof events and S-V for those involving more activity; V-S to merely place acharacter on the scene without saying much about what he/she does, S-V for themore important actions of the character. Counts made on whole texts confirm, assuggested by examples (1) and (15) taken together, that the mere presence of anadverb draws enough attention for these events to favor S-V over V-S. Thescene-setting or text-structuring effected by preposing of the locative phrase,on the other hand, is an independent choice that may be equally appropriate forboth.

This deconstructing of the FVI construction redefines the object of study--thedata to be included--in a fundamental way. It brings in verb-first inversion(example 3), quotation inversion (example 4), and ‘there’-inversion (example 5). The failure of earlier analyses to distinguish between the effects of preposingand those of subject-verb ordering leads to a heterogeneous, evenself-contradictory list of “functions” for the FVI construction, such as we findin Green and Dorgeloh, and leaves unanswered questions such as that posed byexample (15) for Birner and Chen. P-A himself at times seems to be on the vergeof this solution (e.g. pp. 26, 184), but never takes the crucial step.

I would recommend that this book be on the bookshelf of any serious student ofEnglish full-verb inversion or English word order in general. It is especiallyvaluable for its review of current literature and its bibliography. (One couldsuggest including a more complete discussion of Bolinger 1977.) It does,however, fall short on critical evaluation of previous work, and its value as ananalysis in its own right is compromised by its encountering some of the sameobstacles as the works it surveys. The new data it presents are interesting,but their ultimate significance as part of the solution to the FVI problem mustbe held in abeyance until more fundamental questions are resolved.

Finally, the index is a bit thin. One annoying deficiency is the omission ofproper names from the index. It would be helpful to be able to turn quickly toall mentions of the various authors discussed in the text. I encounteredseveral minor typos and editorial slip-ups, but none that really impedeunderstanding.


Birner, Betty J. 1992. The Discourse Function of Inversion in English. Ph.D.dissertation: Northwestern University.

Birner, Betty J. 1994. “Information status and word order in English: ananalysis of English inversion.” Language 70 (2): 233-259.

Birner, Betty J. 1996. The Discourse Function of Inversion in English. NewYork and London: Garland.

Bolinger, Dwight. 1977. Meaning and Form. London: Longman.

Chen, Rong 2003. English Inversion: A Ground-Before-Figure Construction.Berlin: Mouton de Gruyter.

Diver, William, Alan Huffman, and Joseph Davis. 2012. Language: Communicationand Human Behavior. The Linguistic Essays of William Diver. Leiden and Boston:Brill.

Dorgeloh, Heidrun. 1997. Inversion in Modern English: Form and Function.Amsterdam and Philadelphia: John Benjamins.

Green, Georgia M. 1980. “Some wherefores of English inversions.” Language 56(3): 582-601.

Green, Georgia M. 1982. “Colloquial and literary uses of inversion.” InDeborah Tannen (ed.) Spoken and Written Language. Exploring Orality andLiteracy. Norwood NJ: Ablex, 119-154.

Huffman, Alan 1997. The Categories of Grammar: French lui and le. Amsterdamand Philadelphia: John Benjamins.

Huffman, Alan 2001. “The linguistics of William Diver and the ColumbiaSchool.” WORD 51 (1): 29-68.

Huffman, Alan 2002. “Cognitive and semiotic modes of explanation in functionalgrammar.” In Wallis Reid, Ricardo Otheguy and Nancy Stern (eds.) Signal,Meaning, and Message. Perspectives on sign-based linguistics. Amsterdam andPhiladelphia: John Benjamins.

Kreyer, Rolf 2006. Inversion in Modern Written English. Syntactic Complexity,Information Status and the Creative Writer. Language in Performance 32.Tuebingen: Gunter Narr.


Alan Huffman is Professor of linguistics at the Graduate Center of the City University of New York and of English and linguistics at the New York City College of Technology of CUNY. He is a specialist in Columbia-school linguistics. He has authored a book (Huffman 1997) dealing with clitic pronouns and case government in French, an article (Huffman 2001) giving an overview of Columbia-school linguistics, several shorter articles, and most recently, with Joseph Davis, a volume (Diver, Huffman, Davis 2012) presenting foundational works of the Columbia school. He is an active member of the University Seminar on Columbia School Linguistics at Columbia, where he received his master’s and doctoral degrees in linguistics.

Page Updated: 12-Nov-2012