Navigation – Plan du site
L'ordre des mots dans l'espace de la phrase

Word order in utterances of direct speech in English: a subtle balance between conventions and innovation

Grégoire LACAZE


L’ordre des mots dans un énoncé n’est pas censé être un processus chaotique. Même si l’ordre des mots diffère d’une langue à l’autre, un locuteur qui s’exprime dans sa langue maternelle est en toute logique capable de déterminer si un énoncé est grammatical ou agrammatical. Cette capacité de distinguer ce qui est correct de ce qui ne l’est pas est une compétence que possède un locuteur parlant sa langue maternelle.
Nous nous intéressons principalement au choix de l’ordre des mots retenu par un producteur de discours rapporté, autrement dit, un locuteur rapporteur, qui construit une occurrence de discours direct en anglais. Ce choix n’étant parfois pas unique, un ordre des mots différent aurait également pu produire un autre énoncé grammatical. Ainsi, nous tentons d’étudier les raisons qui encouragent un locuteur rapporteur à retenir un ordre donné. Cette analyse essaie tout particulièrement de mettre en relief les enjeux qui conditionnent ce choix.

Haut de page


Haut de page

Texte intégral


1Word order in an utterance produced by a speaker in connected speech is not supposed to be chaotic. Even though word order differs from one language to another, a given person who speaks his/her own mother tongue is normally able to say whether a given utterance is grammatical or not. This proficiency to point out what is correct and what is erroneous is a characteristic of an individual’s speaking his own language as a native speaker, hence the concept of “a native speaker’s pragmatic competence” (Birner and Ward 1998 1).

2The main focus of this paper is on how a producer of reported speech, namely a reporter, when led to build an occurrence of direct speech in English, will choose one word order for a sequence of words. In some cases, another word order would have given another fully grammatical utterance. What are the reasons behind a reporter favouring one particular word order? This analysis will particularly try to highlight what is at stake in this choice.

I. Direct speech and the problematics of word order

A. Direct speech and the concept of the contextualizing sequence

3Most linguists agree that an utterance of direct speech (namely DS) usually includes a reporting clause (henceforth RC) and a reported speech. The RC contains the “Narrator’s Report of Speech” (Semino and Short 30) whereas the reported speech “purports to give a verbatim rendition of the words that were spoken” (Coulmas 6). This traditional theoretical approach to any utterance of direct speech is far from being fully adequate to describe all varieties of utterances. This has been highlighted in an article about the French terminological term incise (Lacaze 2011). This contribution favours the analysis of a given utterance within a larger framework and this concept is fundamentally based on a close analysis of the contextual environment. The contextual environment refers to the textual fragments that are located in the vicinity of the utterance of DS that is being examined. This new approach has led to the coining of the expression contextualizing sequence (hereafter CS) that can be roughly regarded as an extended version of an RC. This extension goes beyond the boundaries of a given sentence because the semantic value does not necessarily coincide with the physical limits of a sentence. In this respect, contextualizing sequences include what Helmut Bonheim names crypto-inquits and that is defined thereafter: “the description of the speaker or of his manner of speaking has been separated out of the inquit and now often stands by itself, becoming a substitute for the older style of an inquit: a crypto-inquit” (84). He adds the following statement: “Frequently the crypto-inquit takes the form of a stage direction, followed immediately by direct speech” (84). This can be illustrated with the following utterance studied previously (Lacaze 2011 34):

“What have I to fear but starvation?” Kino asked.
But Juan Tomás shook his head slowly. “That we must all fear. But suppose you are correct — suppose your pearl is of great value — do you think then the game is over?” (Steinbeck The Pearl 58, my emphasis)

4Despite the syntactic independence between the CS and the quotation itself, there exists a semantic dependence between them.

5Thanks to this methodological approach rooted in the concepts of textual coherence and cohesion, two concepts that have been studied by Michael Halliday and Ruqaiya Hasan (1976), both the content of the CS and the reported speech itself can be analysed more easily.

B. Word order in direct quotes: two levels of study

6Word order in direct quotes depends directly on two hierarchical and syntactic levels: word order primarily depends on the location of the CS and the choice of the position of the CS has a strong influence on the composition of the textual sequence announcing DS. As far as the content of the CS is concerned, the focus will be on the utterer’s identity and the reporting verb that introduces DS.

1. The three possible positions of a contextualizing sequence

7Sidney Greenbaum and Gerald Nelson (1999, 2009) have identified the three possible positions for an RC within an utterance of DS (An Introduction to English Grammar 210). The “initial reporting clause” (210) is an RC located in initial position, that is to say in a position preceding the reported speech. An RC can also be inserted in “medial” position (211): there are fragments of reported speech on either side of the RC. The third position is the “final” position (211). According to Georgia Green, this position corresponds to “quote-preposing” (591): the RC is located after the reported speech.

8Helmut Bonheim had previously identified the different positions “of the inquit relative to the speech to which it is attached” (1982 75):

[…] the inquit sometimes precedes the speech, sometimes interrupts the speech, sometimes follows. That is, it occurs in initial, medial or final position:
1. Initial position: “The same steady voice answered him. ‘Yes, keep it about two points off the port bow.’” Stephen Crane
2. Medial position: “‘I don’t know,’ the nigger said. ‘I hurt bad’.” Ernest Hemingway
3. Final position: “‘What were you doing on the floor?’ I asked.” Graham Greene

  • 1 Even though most linguists quoted in this paper do not use the concept of contextualizing sequence, (...)

9Since the contextualizing sequence tends to be a generalization of the reporting clause1, some fragments of a contextualizing sequence can consequently appear in the three aforementioned positions.

2. The various combinations of word order in utterances of direct speech

10Betty Birner lists the various combinations of word order in contextualizing sequences (namely CSs) announcing DS: “In addition to the ‘canonical’, but not necessarily most common, NP-verb-quotation (NP-V-Q) word order, we find V-NP-Q […], Q-NP-V and Q-V-NP […], and Q-NP-V-Q and Q-V-NP-Q […]” (1996 22). Altogether, six different word orders have been listed. In other words, an inversion involving the noun phrase and the verb may a priori appear in the three possible locations of a CS. Nonetheless, some parameters will reduce the possibilities of inversion and they will be mentioned below.

3. Inversion and its functions

11Before analysing the various combinations, the terminological terms that will be used need defining. First of all, the problematics of inversion in CSs is particularly relevant for Subject-Verb-Object (SVO) languages such as English. Hence the distinction between canonical word order (hereafter CWO) and noncanonical word order (henceforth NWO). Canonical word order is S-V-O in positive assertions whereas noncanonical word order implies an inversion subject-verb.

  • 2 The term inversion used here will be as defined by Betty Birner, unless otherwise stated.

12In this paper, the interest in word order in CSs focuses on inversion, not on preposing. Indeed, according to Birner, there exists a major difference between inversion and preposing: “An INVERSION2 is a sentence in which the logical subject appears in post-verbal position while some other, canonically post-verbal, constituent appears in clause-initial position” (12). “Preposing involves only the fronting of some constituent, while the verb and its auxiliaries remain to the right of the subject” (16).

  • 3 The term givenness is to be understood “in the sense of "shared knowledge"” (Prince 1981 232).

13It is generally admitted that “inversion serves an information-packaging function (Chafe 1976), linking relatively unfamiliar information to the prior context via the clause-initial placement of information which is relatively familiar” (Birner 1996 77). This is why inversion is often regarded in terms of “givenness3” (Prince 1981) and “familiarity” (Prince 1992). According to Green, inversion has a “connective function” (586). This function relates “the information represented by the inversion to the surrounding context in a coherent manner” (Birner 1996 5).

  • 4 The initials S, V and Q respectively stand for Subject, Verb and Quotation.

14Moreover, it is often mentioned that the exploitation of an “alternate word order” is likely to create “many specific rhetorical effects” and serve “pragmatic functions” (Green 582). Those pragmatic functions are likely to favour the “alternation […] among S V Q, Q S V, and Q V S”4 (594). Georgia Green, who has studied inversion in DS, gives the following example:

‘She’s too young to play,’ said Mrs. Rabbit. (Inversion after Preposed Quote) (Green 583)

15Georgia Green assumes that the “Old information first, new information last” theory supported by the Prague School linguists is not fully adequate “as a general characterization of inversion” (583). Indeed, “what is new is the quotation, what Mrs. Rabbit said – not Mrs. Rabbit” (583).

16The theory supported by the Prague School linguists is “based on the communicative function of word order” (Breivik 9). According to Firbas (1966 240), an utterance can be analysed in terms of “communicative dynamism (= CD)”:

The elements carrying the lowest degrees of CD constitute the theme, those carrying the highest degrees, the rheme ... In addition to the theme and the rheme, there is the transition, which, in regard to CD carried, ranks above the former on the one hand, and below the latter on the other. The basic distribution of CD is a consistent theme-transition-rheme sequence. (Firbas 240, qtd. in Breivik 9)

C. How to choose a location for a CS

1. Structuring information and information processing

17When a reporter is led to build an occurrence of DS, the two notions of structuring information and information processing prevail. Word order is closely linked to the notion of information structure, i.e. word order influences the way information is distributed among the various elements forming a sentence. Indeed, a textual production abides by “an information-packaging mechanism” (Birner 5). This basic principle usually serves as a guideline for the writer who builds a representation of speech in a novel or a short story. The reporter constantly takes decisions about the positions of CSs and their contents when using DS.

2. The CS in initial position

18The archetypal structure of DS is represented by a CS in initial position such as:

He said: “I won’t come tomorrow.”

19According to Bonheim, this position seems a bit old-fashioned: “The initial position was dominant in narratives of the renaissance, but today’s writers, by and large, avoid it” (75). Notwithstanding Bonheim’s assertion, initial position is still largely used in modern fiction and in the press. It also remains the epitome of DS in English grammar books used for pedagogical purposes:

Paul said, ‘I am feeling ill.’ (Murphy 94)

20Helmut Bonheim tries to justify the choice made by an author to include a CS in initial position: “If […] he likes to interlard his inquits with extended descriptions of the speaker and his situation, he will prefer the inquit in initial position. For only an initial inquit normally allows a heavy admixture of material from the modes of description and comment” (Bonheim 1982 76). The length of the reported speech is a strong argument favouring initial position, as Helmut Bonheim puts it: “The longer the speech, however, the more awkward does the final position of the inquit become. […] A speech long enough to require indentation if quoted in a scholarly article requires an inquit in initial or medial position” (78). The commonplace argument in favour of an initial CS is: “The longer the inquit, in fact, the more likely that it precede rather than follow the speech” (Bonheim 80).

21Georgia Green pretends that the initial position of the RC is more likely to appear when some important information is given by the reporter about the utterer’s identity and the way the enunciative act is performed:

In fact, when more information is (e.g. adverbs, coordinate verbs or clauses, mention of the addressee) is included in the non-quote part of a sentence—besides the name of the speaker and the fact and/or mode of saying—the quote generally does not precede. (Green 591-2)

  • 5 The theory of “turn-taking organization for conversation” has been studied by Emanuel Schegloff (20 (...)

22In fact, medial and final positions are also perfectly fitted to convey details about the utterer’s identity and about the enunciative situation in which a speech act takes place. I would suggest that quote-preposing gives prominence to the content of the reported speech, and the words that are reported are then attributed to a speaker whose identity is revealed afterwards. In a dialogue involving at least two characters, the regular alternation of “turns-at-talk”5 between the speakers is an element that the reader should remember. Obviously, what constitutes new information is the very words pronounced by each speaker: “The content of the quote is usually unpredictable—and, if it can be called ‘information’, it is new information” (Green 592).

23Moreover, the reporter is encouraged to use a CS in initial position if a description precedes a new turn-at-talk:

There was a moment of slight awkwardness as Mr Lewis glanced around the table. Then Lord Darlington said:
‘Naturally, some bitterness is inevitable. But then, of course, we English also fought the Germans long and hard.” (Ishiguro 
The Remains of the Day 90, my emphasis)

24This position tends to emphasize authorial intervention in the text. The speech act that is reported is strongly highlighted. The initial CS is particularly relevant when the reporter wants to characterize the speaker’s identity or the way the speech act is performed: “Where report in the form of an inquit is weighted heavily with embedded description, it naturally grows longer and tends to the initial position” (Bonheim 1982 79-80). A detailed physical description of the speaker is another argument supporting a CS in initial position: “Description of the speaker is apparently felt to be more appropriate in the initial inquit” (Bonheim 80).

25The absence of inversion in an initial CS means that the verb of saying follows the subject NP. The RC Lord Darlington said can be regarded as a prototypical CS containing a subject NP and an introductory verb. The previous utterance could be altered this way:

There was a moment of slight awkwardness as Mr Lewis glanced around the table. Then it was Lord Darlington who said:
‘Naturally, some bitterness is inevitable. But then, of course, we English also fought the Germans long and hard.”

26The insertion of a cleft sentence within the CS gives more prominence to the speaker by emphasizing the NP. This syntactic transformation that can often be added to initial CSs is in keeping with the idea that initial position tends to strengthen the illocutionary force of the speech act by highlighting the speaker’s identity.

3. The CS in medial position

27In its very definition, medial position presupposes that two fragments of the reported speech are present on either side of the CS. A medial position often implies an extended reported speech with more than one sentence. In this respect, the CS can be located after the first sentence. It can also introduce some light break in connected speech. It may also coincide with pauses for breath, when the dialogue is read aloud. What seems a recurrent feature of a medial position is the fact that this particular location is often chosen for rhythmic purposes.

28Helmut Bonheim reckons that “an inquit in medial position is one way of allowing speech to follow on speech” (1982 75). He also assumes that “the inquit in medial position as it appears in the eighteenth and nineteenth centuries has a somewhat formal or stilted look because it generally appeared in inverted form: said he instead of he said” (81). Moreover, he makes the following assumption: “In the modern novel the habit of syntactic inversion applied to the medial inquit becomes optional, and is thus a matter of rhetoric (where an optional inversion is called an anastrophe). Such anastrophes tend to be retained in the medial, sometimes also in the final inquit where the speaker is named” (81).

29According to Bonheim, there has been a gradual trend to shift the inquit from medial position to final position: “The development of modern narrative […] pushed the verb to the rear” (81). Notwithstanding this assumption, it is worth noticing that medial position is a very subtle tool that a writer can use whenever he/she aims at trying to provide readers with a faithful description of the voice and attitudes of the speaker as the verbal production is being reproduced in the fragments of reported speech.

30Medial position is also fitted for the reporter to give an accurate description of simultaneous actions or activities made by various characters while one speaker’s speech is being reported. This can be illustrated by an excerpt from Ishiguro’s The Remains of the Day:

‘Just wondering, sir,’ he said as I approached, ‘how fit your legs were.’ (Ishiguro 25, my emphasis)

4. The CS in final position or “quote-preposing”

31The third possible location for a CS is final position in an utterance of DS: the CS follows the quotation. As far as DS is concerned, Geoffrey Leech and Mick Short state that “it is quite normal to put the reported speech before the locutionary clause” (333).

32When the reporter is inclined to express a personal comment on the content of the reported speech, final position seems to be more suited to the mentioning of an opinion or a point of view regarding the very words that are reported in DS.

33The extension of a CS in final position is somewhat variable and does not constitute a mandatory constraint for the reporter. The shortest final CS in its minimal version includes the subject NP and an introductory verb, which is not necessarily a verb of saying.

34Whenever a careful description of the conditions of the speech act is contained within a CS in final position, this means that readers are invited to reread the preposed quote so that they might have a more accurate mental representation of the speech act than they had after the first reading of the text. This idea is illustrated by Helmut Bonheim’s remark: “All the detail devoted to showing us how the speaker looks, his frame of mind and the quality of his voice, would be wasted on us if it came after instead of before the speech” (1980 80).

35Far from being useless, the use of such a description in final position is a deliberate choice made by the reporter. Stylistically speaking, it may appear as an intrusion of the reporter, and consequently, as a testimony of authorial intervention in the representation of DS.

36Furthermore, a CS in final position can expand the textual boundaries of reported speech as is illustrated in the following utterance drawn from Mrs Dalloway:

“There it is,” said Rezia, twirling Mrs. Peters’ hat on the tips of her fingers. “That’ll do for the moment. Later …” her sentence bubbled away drip, drip, drip, like a contented tap left running. (Woolf 159, my emphasis)

37In this utterance, the long CS enhances the expression of sound perceptions and plays an active role in the characterisation of the speech act by laying emphasis on its musicality, thanks to the repetition of drip, often called “epizeuxis” (Howard 95).

5. A CS acting as a transition between narrative and dialogue

38It is noticeable that initial and final CSs share one major feature: both stand for subtle transitions between a narrative sequence and an occurrence of DS because they are located on the textual boundaries between narrative and dialogue: “The inquit acts as a hinge between passages of speech and the adjacent narrative modes” (Bonheim 75). Furthermore, a CS usually contains a subject NP and a verb of saying that bears tense. Thus, when the CS contains a Simple Past verb, the speech act is inserted within a set of consecutive events and it is viewed as a punctual event usually implying telicity.

39The following excerpt from Pamela gives an illustration of a very long initial CS that serves as a transition between the narrative and the reported speech:

Indeed I wanted to be out of his sight, as much as he did to have me gone: but I trembled so, that I was forced to go holding by the wainscot all the way with both my hands, and thought I should hardly have got to the door: but when I did, as I hoped this would be my last interview with this terrible hard-hearted master, I recovered presence of mind enough to turn about, and with a low court’sy, and my hands clasped, to say, ‘God bless you, sir! God bless you, Mr Longman!’ (Richardson Pamela 107, my emphasis)

II. The study of word order within contextualizing sequences

A. Investigating the possibilities of inversion within a CS

40According to Betty Birner (1996), unlike other types of inversion, “quotation inversion” seems to be quite different. As far as Q-V-NP word order is concerned, she makes the following statement: “the postposed NP denoting the speaker very often represents familiar information (in extended dialogues, for example), while the information in the preposed quotation represents new information” (1996 23).

41Indeed, the reported words themselves are the very essence of a dialogue but the information about the conditions of the original speech act are of paramount importance as well. For instance, this information includes the identities of the speakers who are thus characterized by the reporter and, sometimes, the way those words are uttered. That is why word order within CSs deserves attention.

42I have previously dealt with the choice of the location of a CS. This time, the focus will be on word order within a CS for each of the three positions. This analysis aims at investigating the use of the possible degrees of liberty facing the reporter.

1. The motivations for inversion

43The operation of inversion, as it considerably alters syntax, is one privileged choice among other syntactic structures to fulfil “specific pragmatic ends” (Birner and Ward 1). Stylistic motivations seem to encourage a given speaker to use NWO: hence the insertion of “some marked, or noncanonical, syntactic construction in a particular context” (Birner and Ward 2).

44Inversion in final CSs seems to be quite influenced by the two theories of “end-focus” and “end-weight” that are often combined to justify inversion. The end-focus theory can be thus defined: “It is normal to arrange the information in our message so that the most important information comes at the end. We follow this principle of end-focus when we put such information at the end of the sentence or clause” (Greenbaum and Nelson 149). The end-weight theory is based on the “relative syntactic and informational ‘weight’ of the subject and verb,” according to Betty Birner (1996 18). After Hartvigson and Jakobsen (1974), Birner mentions that the end-weight theory consists of “postposing the relatively ‘heavy’ subject” (18).

2. The CS in initial position

45In contemporary fiction, it is CWO that is very likely to appear in an initial CS including a positive assertion. Subject-auxiliary inversion (hereafter SAI) can also be present when the CS itself is a question:

Did she say, ‘It is against my religious principles’? (Greenbaum and Nelson 211, my emphasis)

46Such an utterance does not contain what Betty Birner calls inversion but it is an occurrence of “subject-auxiliary inversion” (1996 19). Even though it is not an inversion, word order is definitely altered in such an utterance.

47Unlike present-day English, in 18th century English fiction, NWO, that is to say inversion in an initial CS, was part of the norm as is illustrated in the following excerpts from Richardson’s Pamela:

Poor Mr Longman faltered in his speech, and was ready to cry. Said my insulting master to me, ‘Why pr’ythee, Pamela, now shew thyself as thou art, before Mr Longman.’ (105, my emphasis)
Said my cruel master, with an imperious tone, ‘Get out of my presence, girl! I can’t bear you in my sight.’ (107, my emphasis)

48The fronting of the verb of saying definitely emphasizes the speech act that follows the CS. Such inversion in an initial CS is also likely to appear in journalistic discourse, as Sylvie Hanote and Hélène Chuquet put it: “Il existe […] un cas très particulier du style journalistique que l’on appelle parfois "Timese" (du nom de l’hebdomadaire américain Time), qui a recours aux déclaratives antéposées de type : Said Senator John Smith: "…"(2004 26). This assertion is illustrated by an excerpt from Time:

  • 6 The acronym DCCC stands for Democratic Congressional Campaign Committee.

In fact, you are far more likely to see Bush in a Democratic ad this year than a Republican one. The President is now being featured in 89 separate TV commercials for Democratic House candidates nationwide. Says DCCC6 spokesman Bill Burton: “No other issue is that dominant.” (Tumulty and Allen, my emphasis)

49Randolph Quirk et al. mention the authorized use of this inversion: “In journalistic writing, a reporting clause with inversion sometimes occurs even in initial position: Declared tall, nineteen-year-old Napier: ‘The show will go on.’” (1024, my emphasis).

3. The common features between a medial CS and a final CS

50As far as word order is concerned, a great number of similarities exist between medial CSs and final CSs, essentially for syntactic concerns. When a CS appears in medial or final position, the emphasis is on the content of the reported speech itself, as Georgia Green puts it: “having the non-quote parts of the sentence at the end of each sentence, in reporting dialog, makes the dialog appear much closer to natural conversation than it would if it consisted of a sequence of Speaker-Verb-Quote constructions” (591). The medial and final positions of a CS favour the fluidity of the reported act in that they tend to signal that the CS is similar to a parenthetical clause that can easily be removed, even though the reader’s understanding of the speech act can be impaired.

51Jean-Rémi Lapaire and Wilfrid Rotgé justify inversion in fiction in medial and final CSs: “dans un contexte littéraire, on trouve assez souvent l’ordre verbe + sujet pour attribuer des paroles à un personnage : said the ostler / asked the policeman. C’est le sujet qui est important ici, plus que le verbe, qui est sémantiquement peu marqué” (Réussir le commentaire grammatical de textes 257).

4. The CS in medial position

52Here are the very first lines of Doris Lessing’s The Golden Notebook:

Anna meets her friend Molly in the summer of 1957 after a separation…
The two women were alone in the London flat.
“The point is,”
said Anna, as her friend came back from the telephone on the landing, “the point is, that as far as I can see, everything’s cracking up.” (Lessing 25, my emphasis)

53The incipit in medias res begins with the introduction of two female characters, immediately followed by a dialogue involving Anna and Molly. NWO has been chosen by the reporter for the CS in medial position. This medial position enables a description of simultaneous actions accomplished by the speaker and the other participants of the speech act. CWO would have given a fully grammatical utterance as well. The other turns-at-talk located in the vicinity of this utterance also contain inversions in medial CSs. This seems to be a recurrent and deliberate choice of the narrator in order to introduce DS in this novel.

5. The CS in final position

  • 7 This issue has been dealt with in a study of Hemingway’s short story “Hills Like White Elephants” ( (...)

54Inversion in a final CS means that “the logical subject appears post-verbally, with the [direct] quotation in initial position” (Birner 20). A final CS is the choice that is often adopted by most writers of contemporary fiction. The subject NP is located at the end of the sentence, catching the attention of the reader who tries to identify the speaker in every turn-at-talk. The final subject NP thus entails a delayed unveiling of the speaker’s identity. This phenomenon is particularly relevant in polylogues when at least three speakers interact. In dialogues involving only two speakers, the logical alternation of turns-at-talk between the speakers makes it easier for the reader to guess who speaks in every turn-at-talk7. Whereas some authors are more apt to use inversions, others resort to NWO in a limited number of occurrences. For instance, there is a small number of occurrences of NWO in John Steinbeck’s The Pearl. In this novella, CWO is the norm whereas there are only 28 inversions, most of which involve Kino’s brother whose name is Juan Tomás. This can be illustrated by the following excerpt:

“To the north,” said Kino. “I have heard that there are cities in the north.”
“Avoid the shore,”
said Juan Tomás. […]
“I have it,”
said Kino.
“The wind is good,”
said Juan Tomás. (Steinbeck 73, my emphasis)

55As a previous analysis of this excerpt has shown (Lacaze 2013), whereas the canonical RC (Kino said) is widely present in the book, the choice of the noncanonical RC (said Kino) seems to be determined by the presence of the noncanonical RC (said Juan Tomás) that may have been chosen according to the end-weight principle. The extended subject NP composed of two first names is more likely to appear post-verbally than in canonical position.

B. The linguistic elements limiting the number of degrees of liberty

56The complexification of the basic structure of a CS including the subject NP and the verb of saying may influence the choice of the reporter by reducing his degrees of liberty. The various elements mentioned below are far from being exhaustive but they show the diversity and hence the complexity of the optional parameters restraining the choice of word order.

1. The subject NP reduced to a personal pronoun

57In her corpus of contemporary fiction, the French linguist Lucie Gournay notes that whenever the subject NP is a personal pronoun, word order is S-V in medial and final reporting clauses: “Dans notre corpus de textes contemporains, lorsque le sujet est de forme pronominale, l’agencement de l’énoncé en incise est S-V” (2000 38). Yet, inversion would generate a grammatical utterance but this utterance would seem somewhat archaic: “On peut donc [y] voir [...] l’expression d’un archaïsme de convention littéraire” (Gournay 39). It is a matter of convention that seems to justify the absence of inversion in a CS containing a personal pronoun. Inversion in 20th century novels does not question the grammaticality of such utterances, as is illustrated in E. M. Forster’s A Room with a View:

The son, who took every little contretemps as if it were a tragedy, was listening also.
‘My father has that effect on nearly everyone,’ he informed her. ‘He will try to be kind.’
‘I hope we all try,’
said she, smiling nervously. (45, my emphasis)

58Lucy’s reply to George contains an inversion in the final CS, even though the subject NP is the personal pronoun she.

59Thus, while inversion including the verb say is still met in early 20th century fiction, it is unlikely with other introductory verbs in combination with personal pronouns as subject NPs, according to Otto Jespersen:

As Jespersen pointed out, the anastrophe is less likely with verbs other than to say, so that a writer who used “said he” would nevertheless use “he whispered” and “he suggested”. This may be because the focus in the simple “said he” is on he, and if we want to shift the focus to the verb, using a more specific, as it were “marked” form like whispered, it is only natural to place it in the second position, where the stress and focus seem naturally to belong. (Bonheim 82)

2. The passive voice

60The presence of the passive voice in an RC is likely to forbid inversion:

I didn’t show any real emotion. That wasn’t allowed. All I said was, ‘I can’t pay you back right away.’
‘You don’t have to pay me back at all,’
I was told. (Carcaterra Sleepers 95, my emphasis)

61In point of fact, the passive voice is hardly inserted into CSs. Indeed, the speaker’s name usually constitutes the subject NP in a CS. Consequently, the subject NP is in collocation with the verb of saying that introduces direct speech. In the utterance above, the speaker’s name is voluntarily unmentioned. Since the reporter is the novel’s homodiegetic narrator, he can use a CS including a deictic marker associated with the passive voice. Involving a narrator-centred dialogue, this use is remarkably rare in fiction. It is the narrator who tells his own story and who rebuilds the dialogue he is supposed to have experienced.

3. Modality and negation: canonical word order only

62The presence of a touch of modality in a CS, thanks to a modal auxiliary or a periphrastic form such as USED TO, has repercussions on word order.

63This is illustrated by an excerpt from Alison Lurie’s The Last Resort:

In the past Wilkie had always been lavish with praise and compliments for everything from her creamy scrambled eggs to her discovery of a lost footnote. “Darling, you are a wonder,” he used to say, sometimes more than once a day. But now he was withdrawn and unappreciative. (104-5, my emphasis)

64Word order is not flexible in the utterance and CWO is the compulsory choice.

65Moreover, the presence of negation in a CS reduces the possibility of inversion. The following example is drawn from an article published in The Guardian:

“Having thoroughly investigated this matter, we do not believe we have enough evidence to prosecute anyone,” said Sheriff Leon Lott of Richland County, South Carolina, his words etched with an unmistakable tinge of regret. “We had a photo and him saying he was sorry for inappropriate behaviour. He never said, ‘I smoked marijuana.’ We didn’t have physical evidence.” (Donegan, my emphasis)

66No inversion seems allowed in such an utterance and initial position is the favourite location of the CS, which drastically restrains the reporter’s choice.

4. Aspectuality

67The insertion of aspectuality can also diminish the degrees of liberty. The presence of the BE+-ING form in a CS seems to prevent inversion. This is illustrated by an excerpt from Forster’s A Room with a View:

She was talking to Mr Beebe, and as she spoke her long narrow head drove backwards and forwards, slowly, regularly, as though she were demolishing some invisible obstacle. ‘We are most grateful to you,’ she was saying. ‘The first evening means so much. […]’ (28, my emphasis)

68The utterance reproduces Charlotte Bartlett’s turn-at-talk when she addresses Mr Beebe. With the use of the BE+-ING form, CWO is the norm, annihilating any attempts from the reporter to escape from conventional use.

69Conversely, the perfective use of HAVE does not most often contribute to a limitation of the choices available for the reporter, as is illustrated in this excerpt from The Remains of the Day:

When the butler reappeared in the drawing-room some time afterwards to refresh the teapots, the employer had inquired if all was well.
‘Perfectly fine, thank you, sir,’
had come the reply. ‘Dinner will be served at the usual time and I am pleased to say there will be no discernible traces left of the recent occurrence by that time.’ (Ishiguro 37, my emphasis)

70NWO emerges in the RC and it is combined with the fronting of the auxiliary.

C. Beyond the “old information” / “new information” theory: towards a new approach on word order

71As the examples mentioned in this study have shown, the justification of word order in terms of information processing based on the concepts of givenness or the “old information” / “new information” theory does not prove to be an entirely adequate or accurate description of word order in every CS. Indeed, some parameters limiting the choice of the reporter have been identified.

1. The limitations of the concept of information processing

72Sometimes, inversion is very unlikely to happen even though it would have been logical for an inversion to be present in a CS, if the concept of information processing were the single fundamental rule conditioning the reporter’s choice.

73In final CSs, Georgia Green notices that “inversion of subject and verb puts the subject in clause-final position” (592). She adds: “If this is really a position which emphasizes (or focuses on) ‘new information’, then we should not find Q V S forms when the subject is predictable” (592). Along with this idea, she makes the following assumption: “If clause-final position is reserved for focusing on significant ‘new information’, then we should expect Q S V forms, rather than Q V S, when the verb of saying describes more than the fact of speaking — e.g. with so-called ‘manner-of-speaking’ verbs like retorted, mumbled, and confessed” (593). She illustrates this with two examples:

‘Meow, meow, meow,’ the kittens protest.
‘We could have another fair,’ suggested Arthur. (593)

74As far as new information is concerned, she assumes that those utterances are more likely to be used than their counterparts:

?‘Meow, meow, meow,’ protest the kittens.
‘We could have another fair,’ Arthur suggested.  

75Commenting upon these examples, Béatrice Vautherin shares Georgia Green’s point of view: “L’ordre des éléments est Sujet + Verbe dans le premier de ces exemples et Verbe + Sujet dans le deuxième. Il semble que pour chacun d’eux l’ordre inverse soit moins naturel. Ceci peut s’expliquer si l’on considère la position finale comme privilégiée. C’est là que l’on placera donc l’information la plus importante” (1987 29). Yet, such reasoning is not always entirely adequate when interpreting the choice of word order in reporting clauses, as is illustrated in the following excerpt from Hemingway’s Hills Like White Elephants:

The hills across the valley of the Ebro were long and white. [...] The American and the girl with him sat at a table in the shade, outside the building. It was very hot and the express from Barcelona would come in forty minutes. It stopped at this junction for two minutes and went on to Madrid.
“What should we drink?”
the girl asked. (Hemingway “Hills Like White Elephants” 211, my emphasis)

76This turn-at-talk is located just after the incipit of the short story. There is no clue in the preceding context inferring that the girl is supposed to speak first in the dialogue. It is noticeable that an inversion in the CS would have produced a grammatical utterance as well:

“What should we drink?” asked the girl.

77Since the content of the quotation is a question, the verb ask seems redundant in the CS and does not offer more information about the speech act. That is why it could be regarded as “old information” whereas the main point in this turn-at-talk is the identity of the first speaker. Nevertheless and paradoxically enough, the subject NP the girl is not in clause-final position, which is often identified as the “end-focus” position.

78Trying to subvert the “old information / new information” theory seems to be in keeping with the technique of opening in medias res that has often been adopted by Hemingway as part of his style, as Robert Paul Lamb points out: “Opening in medias res without later exposition is familiar now but in Hemingway’s day was still somewhat unconventional” (139).

79My own approach favours the analysis of a given CS within its contextual environment. This analysis will also take into account the speech genre in which the occurrence of reported speech emerges.

2. The contextualizing approach to inversion

80As cohesion and coherence are particularly relevant in discourse analysis, the compositional choice of a CS is directly influenced by its contextual environment. My assumption here is that word order in a given quotation is not only determined by the “old information” / “new information” theory but it is closely dependent on the contextual environment of this quotation. A previous article (Lacaze 2011) has shown that the content of an NP within an RC is strongly influenced by the content of the referential chain which the NP belongs to.

81When two competing word orders are allowed in an utterance, the reporter indubitably has some degree of liberty that he will use according to the contextual environment of a given utterance. This can be illustrated by the following excerpt from Ishiguro’s The Remains of the Day:

Lord Darlington called me into his study […]. He seated himself at his desk, and as usual, resorted to holding open a book […] turning a page to and fro.
‘Oh, Stevens,’
he began with a false air of nonchalance […]. His lordship went on fingering his page for a moment, leaned forward to scrutinize an entry, then said: ‘Stevens, I realize this is a somewhat irregular thing to ask you to do.’ (Ishiguro 84-5, my emphasis)

82In the utterance above, the narrator exploits various denominations for the original speaker. By choosing three different NPs belonging to the paradigm of the speaker’s identity, the reporter offers alternations in the referential chain of the speaker. Hence the composition of the RC he began depends on the composition of its contextual environment. The choice of the personal pronoun he reduces the number of degrees of liberty. Inversion in the first RC does not seem to be acceptable. If the reporter had not chosen a personal pronoun, he would have been faced with two possible competing word orders. NWO would have been perfectly acceptable, as much as CWO:

‘Oh, Stevens,’ began Lord Darlington with a false air of nonchalance […].

‘Oh, Stevens,’ Lord Darlington began with a false air of nonchalance […].

3. The quasi-mandatory quote-preposing in headlines in the press

83As far as word order is concerned in occurrences of DS, the speech genre of the verbal production in which the reported speech occurs has received scant attention in this study. Nonetheless, there are some specificities that encourage the use of certain structures for the CS.

  • 8 As the journalist Allan Bell (1993 40-1) points out, “material other than ‘body copy’ (the continuo (...)

84Remarkably enough, CSs in final position are favoured in journalistic discourse. When a quotation is reported in the headline of an article, the journalist or the editor8 most frequently chooses a CS in final position when the CS contains a subject NP and a verb of saying. In a given quotation, the reported speech itself constitutes new information and what comes next, that is to say, the speaker’s identity, is then unveiled to the reader. Even though the speaker is usually known by the readership, the new information here is the words pronounced by this speaker. Thanks to quote-preposing, the mention of the speaker’s name in final position necessarily catches readers’ attention. This is in keeping with the “end-focus” principle. Here are some examples of headlines published in recent years:

Internet has changed foreign policy for ever, says Gordon Brown (Viner)

No new spending plans before election, Mandelson confirms (Summers)

Diffuser ruling may determine formula one title, fears Alonso (

Wind farms are useless, says Duke of Edinburgh (Wynne-Jones)

85Inversion is very frequent in final CSs, even though it is not compulsory. Thanks to the fronting of the quotation, the content of quoted speech is highlighted and the relationship between the quoted words and the speaker who has pronounced them is unveiled at the very end of the utterance in most cases when the verb say is used. Hence an utterance in a headline seems well-balanced with two important poles of information separated by the verb of saying which is usually semantically poor. When the verb of saying is “enriched semantically” such as confirm, it is more likely for a shift of the verb to the right to occur. This seems to be in accordance with the theories of “end-weight” and “end-focus”. Ultimately, it is the reporter who keeps the control over the location of the verb; for instance, the verb fear in the utterance above has a “heavy” semantic weight but the utterer’s identity seems here even more important than the introductory verb.

4. The propensity for innovation: the use of degrees of liberty for stylistic purposes

86The main interest in this paper is to try to assess the innovative contribution of the reporter in his/her verbal production, as far as word order is concerned.

87When the subject NP is not a personal pronoun, CWO and NWO are usually both grammatical, the prevalence of each word order depending on the verb of saying and on the style of the texts in which they are inserted (Gournay 39-40). Breaking up conventional rules may be the target sought by the reporter. Including inversion where there is not supposed to be one can be regarded as an engaged response from the reporter trying to affirm his/her authorial style.

88In the case of an anastrophe, a reporter choosing inversion intends to emphasize stylistic purposes. This phenomenon is justified by Betty Birner for whom inversion can exemplify “degrees of salience” (78). This can be illustrated by the following utterance drawn from E. M. Forster’s A Room with a View:

Lucy screwed up her mouth and said: ‘Perhaps I spoke hastily.’
‘Oh, goodness!’
her mother flashed. ‘How you do remind me of Charlotte Bartlett!’
Charlotte?flashed Lucy in her turn, pierced at last by a vivid pain. (214, my emphasis)

89Inversion in Lucy’s turn-at-talk is optional but the reporter chose to introduce it in the dialogue between Mrs Honeychurch and her daughter Lucy. This is the reporter’s deliberate desire to produce a stylistic effect as a mirrored view, often called “chiasmus” (Howard 64). If a notional domain of the degrees of salience of inversion were to be mapped, the previous utterance would illustrate a high degree of salience. Indeed, whereas Mrs Honeychurch’s turn-at-talk is built according to CWO, her daughter’s reply is strongly marked thanks to NWO.

90This example clearly shows that word order is strongly influenced by the context of the utterance. While CWO can be regarded as the norm, NWO seems synonymous with a “deviation from that norm for rhetorical purposes” (Bonheim 89).


91Word order in contextualizing sequences announcing direct speech is obviously a complex and subtle phenomenon. Various competing parameters have to be taken into account to try and analyse the choices made by a reporter who produces an occurrence of direct speech. Despite certain normative rules that are more or less admitted by the speakers of a given language, creation and innovation are strong incentives that encourage a reporter to eschew the use of canonical word order. By doing so, the reporter gives birth to a textual production including various degrees of salience.

92Other studies on the justification of the use of noncanonical word order should be carried out in order to give a more accurate description of the different contextual parameters influencing the choices made by a producer of direct speech.

93As far as the author’s style is concerned, inversion is part and parcel of the strategies implemented by the author. Some authors are attached to inversion in medial and final CSs in some of their pieces of fiction, whereas others prefer to use inversion parsimoniously. Even though global trends can be identified, it is the author’s choice that ultimately legitimates the use or the absence of inversion in a given CS.

94To conclude, if only for the time being, the following quotation from Helmut Bonheim convincingly encapsulates what is at stake for word order in contextualizing sequences: “the inquit conventions are not "mere conventions" […], but a set of interrelated norms subject at least in part to authorial manipulation and, therefore, to critical analysis” (1982 90).

Haut de page


1) Corpus

- Fiction

CARCATERRA, Lorenzo. Sleepers. 1995. London: Arrow Books, 1996.

FORSTER, Edward Morgan. A Room with a View. 1908. London: Penguin Books, 1960.

HEMINGWAY, Ernest. “Hills Like White Elephants.” 1927. The Complete Short Stories of Ernest Hemingway. The Finca Vigía Edition. New York: Scribner, 2003. 211-214.

ISHIGURO, Kazuo. The Remains of the Day. 1989. London: Faber and Faber, 1996.

LESSING, Doris. The Golden Notebook. 1962. London: Harper Perennial, 2007.

LURIE, Alison. The Last Resort. 1998. London: Vintage, 1999.

RICHARDSON, Samuel. Pamela or Virtue Rewarded. 1740. London: Penguin Classics, 1980.

STEINBECK, John. The Pearl. 1947. London: Penguin Books, 1994.

WOOLF, Virginia. Mrs Dalloway. 1925. London: Penguin Books, 1966.

- Press

“Diffuser ruling may determine formula one title, fears Alonso.” 14 Apr. 2009. 25 Apr. 2009

DONEGAN, Lawrence. “Michael Phelps finds redemption in return to the pool.” The Guardian. 14 May 2009: 9. 16 July 2009

SUMMERS, Deborah. “No new spending plans before election, Mandelson confirms.” 29 June 2009

TUMULTY, Karen, and Mike ALLEN. “It’s Lonely At the Top.” Time 6 Nov. 2006.

VINER, Katharine. “Internet has changed foreign policy for ever, says Gordon Brown.” 19 Jun. 2009. 17 Aug. 2009

WYNNE-JONES, Jonathan. “Wind farms are useless, says Duke of Edinburgh.” 19 Nov. 2011. 7 Dec. 2011

2) Research books and papers

BELL, Allan. The Language of News Media. 1991. Cambridge (Massachusetts): Blackwell Publishers, 1993.

BIRNER, Betty J. The Discourse Function of Inversion in English. New York: Garland Publishing, 1996.

BIRNER, Betty J., and Gregory WARD. Information Status and Noncanonical Word Order in English. Amsterdam: John Benjamins Publishing Company, 1998.

BONHEIM, Helmut. The Narrative Modes: Techniques of the Short Story. Cambridge: D. S. Brewer, 1982.

BREIVIK, Leiv Egil. “On the Interpretation of Existential There.” Language 57.1 (1981) : 1‑25.

CHAFE, Wallace. “Givenness, contrastiveness, definiteness, subjects, topics and point of view.” Subject and Topic. Ed. Charles Li. New York: Academic Press, 1976.

COULMAS, Florian. “Reported Speech: Some general issues.” Direct and Indirect Speech. Ed. Florian Coulmas. Trends in Linguistics. Studies and Monographs 31. Berlin: Mouton de Gruyter, 1986. 1‑28.

FIRBAS, Jan. “Non-thematic subjects in contemporary English.” Travaux Linguistiques de Prague 2 (1966) : 239-256.

GREEN, Georgia M. “Some Wherefores of English Inversion.” Language 56.3 (1980): 582601.

GOURNAY, Lucie. « Linguistique contrastive et narratologie : dit-il, he said… Ce que nous apprend l’analyse linguistique des énoncés en incise dans le discours direct sur la relation entre récit et discours direct », Linguistique contrastive et traduction 5, Jacqueline Guillemin-Fischer (éd), Gap : Ophrys, 2000. 33‑64.

HALLIDAY, Michael A. K., and Ruqaiya HASAN. Cohesion in English. English Language Series. New York: Longman, 1976.

HANOTE, Sylvie et Hélène CHUQUET. ‘Who’s speaking, please?’ : le discours rapporté. Paris : Ophrys, 2004.

HARTVIGSON, Hans H., and Leif Kvistgaard JAKOBSEN. Inversion in Present-Day English. Odense (Denmark): Odense University Press, 1974.

ISANI, Shaeda. “Of headlines & headlinese: Towards distinctive linguistic and pragmatic genericity.” ASp 60 (2011): 81102.

LACAZE, Grégoire. « Une analyse d’une production de discours rapporté : l’introduction des tours de parole dans "Hills Like White Elephants" d’Ernest Hemingway ». E-rea 8.1. 2010. <>.

——. « De l’incise au segment contextualisant : un changement d’horizon dans l’introduction du discours direct ». Études de Stylistique Anglaise N°1 (2011) : 2544.

——. « Regards croisés sur l’introduction du discours direct en anglais et en français : mise en contexte dans The Pearl et La perle de John Steinbeck ». Études de Stylistique Anglaise N°3 (2013) : 89106.

LAMB, Robert Paul. Art Matters: Hemingway, Craft and the Creation of the Modern Short Story. Baton Rouge (Louisiana): LSU Press, 2010.

LEECH, Geoffrey N., and Michael H. SHORT. Style in Fiction: a linguistic introduction to English fictional prose. London: Longman, 1981.

PRINCE, Ellen F. “Toward a Taxonomy of Given-New Information.” Radical Pragmatics. Ed. Peter Cole. New York: Academic Press, 1981. 223254.

——. “The ZPG Letter: Subjects, Definiteness, and Information-status.” Discourse Description: Diverse linguistic analyses of a fund-raising text. Eds. William C. Mann and Sandra A. Thompson. Amsterdam: John Benjamins Publishing Company, 1992. 295326.

SCHEGLOFF, Emanuel A. Sequence Organization in Interaction: A Primer In Conversation Analysis. Volume 1. Cambridge: Cambridge University Press, 2007.

SEMINO, Elena, and Mick SHORT. Corpus Stylistics: Speech, Writing and Thought Presentation in a Corpus of English Writing. Abingdon: Routledge, 2004.

VAUTHERIN, Béatrice. « Les formes spécifiques du discours direct dans “Hills Like White Elephants” ». Palimpsestes 1 (1987) : 27‑39.

3) Dictionaries and grammar books

GREENBAUM, Sidney, and Gerald NELSON. An Introduction to English Grammar. 1999. Third Edition. Harlow: Pearson Education Limited, 2009.

HOWARD, Gregory T. Dictionary of Rhetorical Terms. Thorofare (New Jersey): Xlibris Corporation, 2010.

JESPERSEN, Otto. Essentials of English Grammar. London: Allen & Unwin, 1933.

LAPAIRE, Jean-Rémi et Wilfrid ROTGÉ. Réussir le commentaire grammatical de textes. 1992. 2e édition. Paris : Ellipses, 2004.

MURPHY, Raymond. English Grammar in Use: a self-study reference and practice book for intermediate students of English. Third edition. Cambridge: Cambridge University Press, 2004.

QUIRK, Randolph, Sidney GREENBAUM, Geoffrey LEECH, and Jan SVARTVIK. A Comprehensive Grammar of the English Language. London: Longman, 2005.

Haut de page


1 Even though most linguists quoted in this paper do not use the concept of contextualizing sequence, it will be adopted here for practical reasons.

2 The term inversion used here will be as defined by Betty Birner, unless otherwise stated.

3 The term givenness is to be understood “in the sense of "shared knowledge"” (Prince 1981 232).

4 The initials S, V and Q respectively stand for Subject, Verb and Quotation.

5 The theory of “turn-taking organization for conversation” has been studied by Emanuel Schegloff (2007 1).

6 The acronym DCCC stands for Democratic Congressional Campaign Committee.

7 This issue has been dealt with in a study of Hemingway’s short story “Hills Like White Elephants” (Lacaze 2010).

8 As the journalist Allan Bell (1993 40-1) points out, “material other than ‘body copy’ (the continuous text of a news story) is usually generated by non-journalists: illustration editors write captions to photographs, and subeditors write headlines” (qtd. in Isani 2011 86).

Haut de page

Pour citer cet article

Référence électronique

Grégoire LACAZE, « Word order in utterances of direct speech in English: a subtle balance between conventions and innovation », E-rea [En ligne], 11.1 | 2013, mis en ligne le 15 décembre 2013, consulté le 24 mars 2017. URL : ; DOI : 10.4000/erea.3406

Haut de page


Grégoire LACAZE

Aix-Marseille Université, LERMA EA 853
Grégoire Lacaze is Lecturer in English linguistics at Aix-Marseille Université. He wrote his doctoral thesis on the introduction of direct speech in present-day English. His research deals with linguistics, stylistics, semantics, and discourse analysis applied to the study of reported speech.

Articles du même auteur

Haut de page
  • Logo Laboratoire d’Études et de Recherche sur le Monde Anglophone
  • Logo DOAJ - Directory of Open Access Journals