Theme in spoken language: when a tone group is not a clause

: While there is a clear tendency for a Theme not occur within its own tone group it is equally clear that clauses are frequently formed out of more than a single tone group. Hence it is precipitous to assume an a priori relationship between clause and tone group/information unit. Speakers in pursuit of their individual communicative choices manage their interactional and informational needs by producing prosodic and thematic choices appropriate to their goals. Such choices as we have seen are frequently mutually re-enforcing, but speakers may employ prosodic choices which foreground their interactional needs. Speci ﬁ cally, this entails that there is no direct relationship between lexico-grammatical and prosodic meanings; both redound with semantic meanings. Yet, these meanings may be constrained by the interactional demands of the communicative situation in which a speaker operates. In this paper, I examine the spoken realisation of Interpersonal Theme and ﬁ nd that Interpersonal Themes realized by Mood adjuncts and metaphors of modality tended to be realized with di ﬀ erent key. Similarly, the expectation that marked Theme was likely to attract high key was found to be subject to the speaker ’ s need to manage the interaction.


Introduction
Systemic Functional Linguistics (SFL) is unique among functional approaches to language in that it separates Theme and Information (Halliday 1967a). 1 This separation has led to much fruitful work on the study of Theme (e.g. the chapters in Hasan andFries 1995 andForey andThompson 2008).However, almost all work on Theme has concentrated on written texts with little attention given to spoken language (but see Arús-Hita 2022;Bäcklund 1992;Ping 2005;Taboada and Lavid 2003).Except for O'Grady (2017a, Forthcoming), there has been limited discussion of the inter-relationship between prosody and Theme.Here I will argue that neglect of the prosodic articulation of Theme results in analyses that do not fully capture the fact that speech happens as an unfolding continuous flow, and hence the interactional component of unscripted speech is discounted.Halliday (1967b) and Halliday and Matthiessen (2014) have described both the clause and the tone group as encoding a quantum of information.The tone group corresponds to a unit containing a single piece of information dubbed the information unit.Within SFL there has been an assumption that clauses and tone groups are coterminous, and that Given information typically precedes New information.This has led to the further assumption that Theme normally coincides with Given while New information is contained in the Rheme.The expected pattern is seen in Example (1) but not in Examples (2) or (3).All examples in the paper unless indicated by the asterisk (*) are taken from the dataset described in Section 3. The transcription symbols used in the examples can be found in the Appendix. (1) I read an article in the Guardian.In Example (1) the subject pronoun is Theme and also Given.The Focus falls on the final lexical item the Guardian which is presented as the main burden of the message.However, in Example (2) there are two Foci though both are Rhematic.The Thematic element the video is salient.It is unclear without examining the wider context whether it is Given or New, as this ultimately depends on whether the salient element is recoverable, either by being physically present or previously mentioned.In Example (3) though we find an unambiguous example of a Focal element being Thematic.
(3) the train line between here and home had closed |the train line | between / here and home | had \/ \/ closed Focus Given Focus New Focus Theme Rheme The speaker has articulated the clause as three information units; each of which must contain an information focus, with the first two coterminous with the Theme.These examples illustrate that in spoken discourse it is not correct to assume an a priori relationship between clause and tone group/information unit, and hence between Theme and Given.
In Section 2, I will briefly review (i) the literature on Theme focussing on its relationship with the interpersonal metafunction, and (ii) the literature on the prosodic encoding of information.I will discuss key choices in order to illustrate how tone groups are presented to hearers within the temporal flow of interactive speech (Brazil 1997; O'Grady 2010). 2 Section 3 will describe the data set and illustrate how it was coded.In Section 4, I will discuss the relationship between Theme and prosody by focussing on their coupling and decoupling in the creation of interpersonal meaning in interactive situated discourse.Finally, in Section 5, I will summarise the main findings.
2 State of the art Halliday and Matthiessen (2014: 89) define Theme as "the element which serves as the point of departure of the message; it is that which locates and orients the clause within its context".It is identified as extending from the beginning of the clause up to and including the first element in transitivity.This element is called the topical Theme (Halliday 1994: 43).The topical Theme may be preceded by optional initial textual and interpersonal elements.Examples (1)-( 3) are examples of Themes consisting of topical Themes only.Examples (4)-( 6) illustrate Themes with multiple elements.Thematicity is only exhausted through the realisation of an experiential element; a clause can only be located within its context when it orients towards what is going on in the context.4)-( 6) do more.The inherently Thematic conjunction so in Example (4) locates the clause as linked to the preceding one and functions logically to signal that the forthcoming clause will elaborate the proposition expressed in the previous clause.In Example (5), the Interpersonal orientation includes the speaker's assessment of the truth of her proposition as likely but not certain.In Example (6), the orientation includes the inherently thematic structural Theme because which extends the proposition in the previous clause by providing a reason for the speaker's action.The realisation of the modal Theme obviously presents the starting point as one where her hearers will not dispute the common sense of the proposition that is to come in the Rheme.
All the Theme examples so far consist of unmarked Theme which is defined as "the element which the speech function would determine for the point of departure of the clause" (Halliday 1967a: 213).To illustrate Declarative Mood comprises Subject Finite order and hence Subject represents the unmarked Theme choice in that Mood.The options are presented in Table 1.Halliday and Matthiessen (2014: 97) provides a semantic gloss that unmarked Theme represents the expected choice and is chosen as Theme in the absence of a good communicative reason.In a similar manner, I glossed marked Theme as signalling that the perspective being developed is approached from an unusual angle (O'Grady 2017a: 276).In Example (7), the clause is not orientated around the subject a lot of houses but rather around the temporality encoded in the Theme during like during the floods during that time.The angle being developed concerned what occurred at a time and not in a particular location.This is unexpected as the speaker's prior cotext had encoded information about the city Hull where her family lives.The following paragraphs briefly describe the tone group/information unit.For further information see (Halliday 1967b;Halliday and Greaves 2008;Tench 1996).Speech is parcelled out in tone groups each of which contains a tonic element.A tonic element is signalled by the presence of a tonic accent on the tonic syllable which is the site of the major pitch movement in the tone group.Minimally the information unit must contain a tonic syllable.It may contain additional syllables before and after the tonic.Phonetically the tonic focus is realized within a tonic syllable but informationally it is the grammatical item which contains the tonic focus that the speaker draws attention to.The elements found in the tail after the tonic syllable are non-salient and project Given information.Elements prior to the tonic syllable may be salient or non-salient.Informationally they may be Given or New.Theme in spoken language I have followed Brazil (1997) and classed tone groups as maximum or minimum depending on whether the tone group contained more than a single salience.The significance of this categorisation will become clear in the discussion of key below.Table 2 illustrates that tonic syllables present the lexical item they occur in as non-recoverable.While posttonic items are Given, the status of pretonic ones is somewhat ambiguous.In minimum tone groups, the lack of a pretonic salience means that the pretonic is presented as recoverable.However, in maximum tone groups, the position is not so clear.Halliday and Greaves (2008: 57) state that the Focal element may either represent the New or be the culmination of the New.The status of items between the head and the tonic depends on the context and whether the material in the pretonic is available.The same phonetic realisation may have a different information distribution depending on the context.The speaker has the additional choice of choosing to pitch the tonic syllable as higher or lower than the preceding onset.Following Brazil (1997) we describe such choices as high, mid, and low termination.A high termination signals that the speaker anticipates the hearer will adjudicate the proposition.A mid termination anticipates hearer concurrence with the proposition, while a low termination projects no expectations, signals closure, and resets the pitch.Returning to Example (10) the presence of the onset followed by the tonic allowed the speaker to realize independent key and termination choices.
As stated earlier, Halliday (1967b) and Halliday and Greaves (2008) argue that the unmarked realization for a single clause is to be articulated as a single tone group.However, this claim remains in need of empirical confirmation.Halliday and Greaves (2008) 5 do not base their argument for an unmarked relationship between clause and tone group solely on frequency.Rather, an unmarked choice is the choice made in the absence of any good reason for choosing anything else.Marked choices are therefore motivated by a good reason.However, the unmarked/marked choice contrasts in terms of "skew" (Halliday 1991: 35); allowing us to expect that a marked relationship would manifest frequency effects.Indeed without relying on frequency, our claims would amount to armchair linguistics!Hence, if there is an unmarked relationship we can expect that at least 70 % of tone groups will be conterminous with clauses.The sole study I know which examined the relationship between clause and tone group in English is Croft (1995) which despite being a rather small scale study offers some support for Halliday's claims.Croft showed that the relation between tone groups and grammatical units, though not necessarily clauses, was strong and occurred 91 % of the time.He further identified a number of tendencies, where the one-to-one relationship between clause and tone group did not hold: (i) parallelism, (ii) syntactic complexity in general, and (iii) distance. 6ivón (2017: 182) disagrees and argues that the equation of clause with tone group privileges a grammatical analysis over a pre-grammatical one where speech is produced as a sequence of subclausal units identified as "constituents falling under their own separate intonation contour" i.e. tone groups (Givón 2017: 157).He, like me (O'Grady 2010), argues that the sequencing of tone groups into larger communicative sequences is tightly governed by the conventions of the language.I further argue though, that irrespective of the relation between an individual tone group and an individual clause, tone groups form into sequences which equate with clause units.
To sum up this section, we can see that if Halliday is correct most Themes will co-occur with Rhemes in single tone groups.Where the Theme is longer or contains more syntactically distant material it may form into its own tone group.This leads us to expect that marked Themes and modal adjuncts will frequently occur within their own tone groups.We might also expect that as marked Theme signals an unusual angle in the development of the message, and as high key signals that the following proposition will be contrary to the previously created expectations that speakers may re-enforce marked Theme by simultaneously selecting high key.But we note that Taboada and Lavid (2003: 166) found in their study of elicited task-orientated conversation that while marked Theme was often used to signal the introduction of a new proposal, it was usually located turn medially.In our investigation of the relationship between Theme and interpersonal meaning, we predict: (1) Modal adjuncts and metaphorical expressions of modality may occur in independent tone groups.(2) Marked Themes are likely to contain a high key re-enforcing that something unexpected is about to occur in the conversation.
Nine volunteers, all undergraduate students, were provided with a video link which showed images of flooding in the UK and asked to watch the video at home.Eight of the volunteers were female.The video was silent so that there could be no possibility of language from the video priming responses. 7The volunteers were randomly assigned into four groups of three and asked to attend a meeting with their group members on the following day. 8The volunteers were not known to each other and hence given a few minutes to introduce themselves and break the ice.Then the video was played again with the participants encouraged to comment on it as it was playing.Once the video had finished playing, and the sound recorder had been turned on each volunteer was invited to speak in turn about what they had watched and any personal experiences they had of flooding.Each session concluded with a short conversation framed by the participants' reactions to what they had heard from their peers.
I used PRAAT software (Boersma and Weenick 2013) to transcribe the recordings into tone groups.Boundary cues such as pauses, changes in rhythm, and the presence of a major tone movement indicated a tone group boundary.Within each tone group, I transcribed the salient syllables with the final one being the tonic, and the initial one being the onset.In order to support my auditory analysis of high, mid, and low key and termination, I measured the F0 value for all onset and tonic syllables.Finally, I used the visual representation of the F0 (pitch) curve to establish the tone choice.The written transcripts of the recordings were coded as Theme ˆRheme following the principles set out in Section 2. Table 3 summarises the contributions by number of tone groups and Themes produced. 9 Across the whole corpus, there were 2.6 times more tone groups than Themes.Within the individual texts, this ranged from 2.2 times to 4.2 times more tone groups.It is clear there is no one-to-one relation between clauses and tone groups with tone groups mostly being formed out of sub-clausal elements.Examples ( 11)-( 15) illustrate the possibilities. 107 The choice of methodology was greatly influenced by the success of the silent Pear Stories film in generating speech.See Chafe (1980) for a full description.8 One female volunteer was unable to attend and hence one group comprised two people.9 As I consider the text to be the unit of analysis, I treat the dialogues and monologues as identical.My aim is to examine how the interaction between Theme and prosody realizes interpersonal meanings.I am not concerned with the choices of individual speakers and any possible variation between speakers.10 There were no examples of predicated Theme in the dataset.The sole possible equative theme is | Yeah I guess what you were \saying about like | \/Portland and stuff |.However, as the speaker produces no following material I discounted it as an abandoned Theme.
Theme in spoken language (11) I live near Hull.

Topical Theme | I \live near |\Hull | Theme Rheme
The Theme is realized by a pronoun and is nonsalient.The interpersonal Theme is realized by a metaphor of modality and, unlike the Topical Theme, is salient.The clause is produced as a single tone group.

Theme
In Example (15) as the Subject her house is not found within the Theme, the hypotactic clause where my friend was in Godlaming is the marked Theme of the clause complex. 11It is articulated in two tone groups and preceded by a non-salient textual Theme.
The five examples above illustrate that speakers articulate Themes in as many tone groups as appropriate to their message.Table 4 details the articulation of Theme in relation to tone group across the data set.
It is immediately obvious that a large majority (72 %) of Themes are not realized as one or more Tone groups.Excluding marked Themes the percentage rises to 79.7 % of Themes.It is also obvious that while only a small minority of unmarked Themes include an interpersonal component that the inclusion of an interpersonal element does not affect tonality with 88.2 % of interpersonal ˆtopical Themes being articulated within the same tone group as Rhematic elements.The number of interpersonal Themes in my data is much lower than recorded by previous studies 11 There was one ambiguous case where it was not immediately clear if the elements the following day was Thematic.Because of the presence of the extended pause, I classed it as an afterthought and as a NG which elaborates the previous tone group.However, in theory it could be a Marked Theme orienting the following clause.
Theme in spoken language e.g.Bäcklund (1992) and Taboada and Lavid (2003) who found that around 30 % of Themes contained an interpersonal element.In my dataset, only 14 % of unmarked Themes contained an interpersonal component.
Conversely, speakers' tonality decisions seem affected by the presence of a textual Theme with 22.1 % of textual ˆtopical Themes being realized as one or more tone groups.Examination of these cases revealed three patterns.The first and by far the most common was where the textual Theme was placed in its own tone group.This occurred on 70.6 % of the instances where textual ˆtopical Themes were articulated as one or more tone groups.The topical Theme was always realized by a non-salient pronominal element.This compares with the overall data set where a non-salient pronoun functioned as topical Theme 63.8 % of the time.In the second pattern, the tonic syllable co-occurred with the topical Theme and the textual Theme was nonsalient.In the third, there was a tonic on both the textual and topical Themes.In both these patterns the topical Theme was realized by a nominal group or an indefinite pronoun.These patterns occurred 13.8 % and 15.6 % of the time.Examples ( 16)-( 20  Examples ( 16)-( 18) illustrate that there are three sub-patterns, where textual Theme is found within an independent tone group.In 7.3 % of the cases, e.g.Example (16) the textual Theme was realized by a conjunctive adjunct.In the remainder of the cases, it was realized by either a conjunction (63.8 %), e.g.Example (17), or a continuative (28.9 %), e.g.Example (18).In Example ( 19), the textual element is not salient while the Topical Theme is tonic.In Example (20), both Textual and Topical elements are focal.What seems noteworthy is that all examples contain silent and filled pauses within or near the Theme.The speakers seem unsure as to the actual wording required to develop their message.They know how the proposition encoded within the unfolding message relates to the cotext and can signal the conjunctive relations.But in order to gain time to compose the most relevant and appropriate message, Themes especially textual Themes are made tonic. 12Hence these textual Themes seem to be as related to the management of the interaction as to the organisation of the message.In their work on the chunking of discourse, Sinclair and Mauranen (2006) propose that a message unfolds as series of M (message-oriented) elements and O (organisationaloriented) elements. 13The former increment experience while the latter are the 12 Unlike Example (10), there is no functional motivation for making the textual Themes tonic.For instance, in Example (17) the speaker signals an initial high key signalling that the hearers will find it contrary to the previous expectations that the speaker made it untroubled to Cardiff.But the speaker could have projected this expectation by pitching the first prominent syllable in the clause as high regardless of the tonality division.13 They do not explicitly equate their chunks with tone groups as they primarily work with written texts, but it would seem a logical thing to do in the study of the unfolding of speech.This is especially so when one considers that a tone group is an information unit, and Sinclair and Mauranen do, at least implicitly, claim a psychological reality for their chunking decisions.Though note that a tone group can contain a mixture of O and M elements.

# of TGs
Theme in spoken language remaining elements.O elements themselves are divided into those which are related to the text (OT) and those which manage the interaction (OI).The Textual Theme in Example ( 18) is an OI element while those in Examples ( 16) and ( 19) are OT elements.By making the textual Theme elements tonic the speakers have signalled that the conjunctive adjunct in Example (15) and the conjunction in Example ( 16) function simultaneously as OT and OI elements.Placement of a tonic textual Theme into its own tone group is related to the speaker's interactional needs, and not exclusively focused on signalling the development of the message.
Modal metaphors such as I think and I guess are difficult to disambiguate from projecting mental process clauses.For instance the string of words I think she is cool could be interpreted as either a projected clause with she is cool realising the projected phenomenon or as a monoclausal structure with I think coded as a modal element signalling probability.Four, not entirely convincing probes are found in the literature which claim to disambiguate the constructions: (i) the presence or absence of a complementizer, (ii) the subject mood tag probe, (iii) transparency to negation, and (iv) transparency to factive sentence adverbials14 (Kearns 2007;Thompson and Mulac 1991;Van Bogaert 2010).Table 5 illustrates.
I (O'Grady 2017b) pointed out that without considering context it is impossible to make definitive disambiguating claims.For instance, the negation and comment adjuncts examples could be read as equating to either: Negation I don't think she is cool or It is not that I think she is cool, it is that I know she is.Factive I think unfortunately she is cool or it is unfortunate that I think she is cool.
Similarly, the complementizer and mood tag probes may contradict, e.g.I think that she is cool isn't she?Boye and Harder (2007) argue that this construction has evolved a range of uses and that the actual value of a use can only be determined in context.With this in mind, I examined all potential instances of the modal metaphor Hence this example was coded as a projecting ˆprojected complex with two Theme components, but no interpersonal Theme.Turning our attention back to Themes with interpersonal elements we find that fully 50 % of the textual ˆinterpersonal topical Themes were realized as one or more tone groups.In other words, (at least) part of the Theme was presented as Focal, i.e. newly introduced to the discourse, e.g.Example (17).Overall, only 63 Themes 15 were identified which contained an interpersonal component.Of those 26 contained a key choice.Table 6 details the Key choices found in the Themes.
The majority of key choices for all types of Theme was mid.As noted earlier mid key carries the value of neither being contrastive nor equative.As such it signals that the speaker is continuing to talk.Thus, it seems likely that a mid key salience serves another function; either to enable an independent termination choice, or to signal that the salient item is part of the New.The high and low key choices explicitly signal an expectation that the hearer will find the Theme contrary to expectations or equivalent to the previous expectations.In Section 2, I argued that the existing 15 Ten of these were found prior to Marked Themes.
Theme in spoken language literature predicted a correlation between marked Theme and high key but that is not what Table 6 shows.Less than a quarter of the marked Themes contain a high key.The conflation of low key and marked Theme is surprising as it is hard to see how an unexpected angle of development can be equivalent to previously created expectations.
This brief survey of the data has thrown up a number of interesting issues in relation to Theme and interpersonal meaning which I will examine in the next section.First, I will look at the articulation of the interpersonal Themes in order to examine the communicative value realized by the accompanying prosodic choices.Secondly, I will examine the key choices that co-occur with interpersonal Themes in order to see whether such choices re-enforce or add to the interpersonal Themes.Finally, I will examine the realisations of high key, and especially the unexpected realisations of low key within marked Themes.

Discussion
In O'Grady (2020), I proposed a revision to the standard SFL architecture by arguing that prosodic systems belonged to the content and not the expression plane.The import of this suggestion is that there is no direct relationship between lexico-grammatical and prosodic meanings, though as they both redound with semantic meaning the coupling between the lexico-grammar and the prosody will be tight and predictable.Within the data set as noted above sixty-three Themes were identified which contained an interpersonal element. 16Structurally the interpersonal Themes comprised thirty adjuncts, and thirty three metaphors of modality.In addition, there were three finites all of which were non-salient.There were 10 interpersonal Theme elements which were followed by marked Themes which will be discussed below.Of the adjunct and metaphor of modality interpersonal Theme components, twenty-nine were nonsalient and thirty-three were produced with a tonic or non-tonic prominence.The following paragraphs will examine the interplay of the prosodic choices and the lexico-grammar in realising interpersonal meaning.Seventeen out of the thirty modal adjuncts received a prosodic prominence with four being tonic.Twelve of the salient modal adjuncts realized a mid key with the other being a high key.In Examples ( 22) and ( 23), the adjuncts that realize the interpersonal Themes are presented as Given information.But the communicative value is not the same.In Example ( 22), the speaker's assessment of her father's hypothetical behaviour is tentative.The lack of salience re-enforces her uncertainty as to her dad's potential actions.Had she made the adjunct salient, her claim would have been simply that her father was the type of person who might have made and posted such a video.She would not have signalled her doubt as to whether he was the type who might have posted such a video.In Example ( 23), the lack of salience signals that the "obviousness" of the speaker's location at the time of the flooding is to be taken for granted.A salience on the adjunct would suggest that the speaker felt the need to assert the obviousness of her location; her hearers either did not know it or for some reason disputed it.
(24) and usually people have time off In Examples ( 24) and ( 25), the adjunct is salient and hence potentially presented as part of the non-recoverable information.Yet, it is clear that the saliences realize different values.In Example (24), the speaker produces a message which conveys her estimation of the usuality of people not working over the Christmas period.This is clearly not news.The message itself sets the scene for the following clause which states that as a result people had some time to clean up/repair the flood damage.In Example (25), the salience on luckily presents her state of mind as not recoverable.
The hearers are presented as not being able to know that she is an inexperienced driver.The communicative effect is to highlight the danger she, as a novice driver, experienced.All three tone groups are articulated by fall-rising tone signalling that the speaker is presenting her information, not as a simple statement, but as a series of information units which convey implications.The hearers are invited to imagine themselves as occupying the driver's seat: a more dangerous and difficult experience than they might have imagined.
(26) but obviously it's an island that has been battered by storms for many years In Examples ( 26) and ( 27), the adjunct is tonic.In both examples, the fall presents the interpersonal Theme as major information.In Example (26), when discussing the isle of Portland the speaker presents the obviousness of her answer as information that one of her hearers needs to know.This hearer has previously admitted to not knowing where Portland is.The rest of the message consists of three tone groups.The first is a paratactic clause; presented as a piece of major information.The second and third form an elaborating hypotactic clause with the third tone group presented as information that does not tell.By making obviously tonic, the speaker is pointing out her reaction to the hearer's lack of knowledge. 17In Example ( 27), the adjunct really is tonic.While overly presenting an assurance that her message is truthful may seem like overkill in a description of damage caused by a previous flood in Hull, in fact, the interpersonal Theme heightens the effect of the flood disruption.She presents the actuality of the duration of suffering as not accessible to her hearers.They need to be explicitly told of the long term damage caused by flooding. (28) obviously it has been on the news and stuff | Hobviously it has been on the \news Interpersonal Topical Theme Rheme Given Focus and stuff

New
In Example (28), the adjunct is pitched with high key.The presence of the pronominal it referring to the storms is clearly recoverable.The salience on the adjunct does not present the element as part of the New.Instead, the functional motivation for the salience arises from the speaker's need to signal a high key choicesee also Example (10).She is transitioning from describing the video to a new topic namely television coverage of the floods.The introduction of the television coverage is not predictable.The choice of the site of the high key is constrained by the fact that, had the speaker placed it either on the non-salient pronoun or salient predicator, she would have introduced an unwarranted contrast either between the flood and something else or between the past and non-past.Such potential contrasts would realize a complaint that the flooding and not some other natural disaster has been on the news or that the coverage has been prematurely stopped.This was not her intention.
As noted above, I located 33 modal metaphors which Halliday and Matthiessen (2014) describe as explicit (incongruent) subjective adjuncts.Of these three were the construction I guess with the remainder being I think.Examples ( 29)-( 33) illustrate the prosodic patterns found in the data.
(29) but I think it is quite an interesting idea It is noticeable that high key was more commonly found with metaphors of modality than with adjuncts.While the data set is too small to extrapolate definite conclusions as to why this might be or indeed to know if this is anything other than an artefact, we can see that adjuncts tend to be more integrated into tone groups.This may imply that the speakers were less focused on managing the interactions and more focused on incrementing their messages.It is also possible that because modal metaphors are structurally identical to projecting clauses that their articulation signals the redundant expectation of a projection which signifies a possible transition in the narrative and attracts a key selection.Turning to the final issue of marked Theme, I noted in Section 3 that the literature predicted a correspondence between marked Theme and high key.However, Table 6 indicated that this was not the case.Only 23.8 % of marked Themes had an initial high key.Surprisingly there was almost an equal number of low keys.Mid key was the most frequent choice for marked Theme.In Example (36), I will examine the communicative value of the various key choices.The above discussion brings to mind Marx's oft-quoted claim: "Men make their own history, but not of their own free will; not under circumstances they themselves have chosen, but under the given and inherited circumstances with which they are directly confronted" (Marx 2019(Marx [1852]]: 480).In a similar manner, we can say that speakers make meanings but they do not make them as they please: they do not make them under self-selected circumstances, but within given social situations conegotiated with their interlocutors and subject to the contingent conventions of the language.Hence, while we have seen that high key and marked Theme both signal transitions; speakers are not free in all situations to couple them.It is precisely the uncoupling of lexico-grammar and prosody that allows (or constrains) speakers to produce unfolding meanings which simultaneously manage the interaction and advance their message.

Conclusions
To conclude, the data suggests that there is no direct relation between lexicogrammatical resources and prosodic ones.While speakers are at times constrained in coupling prosodic and lexico-grammatical choices, noticeable tendencies were found in the data.Despite there being more tone groups than clauses, the Theme is not usually found within its own tone group.This is the case irrespective of whether the Theme contains Interpersonal elements.However, textual Themes have a higher tendency to appear within their own tone groups.The expectation that adjuncts and mood metaphors as more syntactically distant elements would tend to occur in their own tone groups was not met.However, as with textual Themes, mood metaphors frequently co-occurred with filled and unfilled pauses indicating that the speakers were focussing on the management of the interaction, and not on the presentation of information.It was noticeable that adjuncts were more integrated into tone groups than mood metaphors and less likely to be the site of a high key.I interpreted this as indicating that the speakers choose high key to signal transitions in the discourse prior to finalising their wording.The expected convergence between high key and marked Theme was not found, as at times high key signalled unwarranted contrasts that did not accord with the speakers' message.Speakers in pursuit of their individual communicative choices manage their interactional and informational needs by producing prosodic and thematic choices.Such choices as we have seen are frequently mutually re-enforcing, but speakers may employ prosodic choices which foreground their interactional needs.As Halliday (1989: 97) said, "spoken language encodes them [phenomena] as things that happen in an outward continuous flow".But it is a flow with a co-constructed target.While speakers have their target firmly in mind, they are rather hazy as to how they and their hearers will get there.Thus, in ensuring their hearers navigate to the intended target, speakers co-deploy lexico-grammatical and prosodic resources in simultaneously managing perceived informational needs and interactional expectations.

Appendix: Transcription symbols
quite focused on uh the sea | the video is quite \focused on | \uh | the /sea | the floods during that time a lot of houses got flooded during like during the floods during that time a lot of houses got flooded abandoned Rheme Marked Theme (12) so um him and his family had to cook their Chris … like Christmas dinner and turkey at the local church and like Textual ^ Topical Theme |so \um | him and his family had \/to | | \cook their |Chris ... like Christmas dinner and /turkey | at the local /\church and like | Textual Topical Rheme ThemeThe Textual and Topical Themes are in different tone groups.The Topical Theme is realized by an NG and is salient.

|
-and | the Hkind of scenes like my \/dad would make | (0.63) | We were Hlucky that they /\opened up | (0.29) one of the \roads | (0.31) |the \sun | cause it was really a hot sunny \day | the L/following day | (0.93) they opened up the -roads | (0.18) |and erm … (0.35) H\one of the roads | and we were able with a \minor detour | [we] were able to … make it \/out to Cardiff | Table 2 schematises the options with examples from the corpus.Given elements are shaded in grey.Salient syllables are underlined and tonic syllables are underlined and in bold.

Table  :
Topical Theme and Mood.
a Note inHalliday (b)the Theme was finite only.A different approach is taken in Martin et al. (: ) who class the finite as the interpersonal Theme on the grounds that it is the element which signals that a response is required.They would similarly class the Wh-interrogative as a fusion of interpersonal and topical Theme.

Table  :
The possible realisations of tone groups with examples.

Table  :
The relation between tone groups and themes.aThesalient topical Theme is preceded by the salient adjunct obviously and the non-salient conjunction because.The Theme and the Rheme occur within a single tone group.
a As I have transcribed Theme in relation to T units and not clauses the figures are close but not identical to the ratio between tone group and clause.

Table  :
) illustrate.Number of tone groups contained within a Theme.
have excluded  Themes from consideration as they were abandoned, i.e. no Rheme was produced.

Table  :
Dehé and Wichmann (2010) explicit modal. in context: removed all negations, and used the Mood tag probe to test what the most natural subject was.There were no instances of complementizing that or comment adjuncts next to potential modal metaphors in the data.Dehé and Wichmann (2010)illustrate that in speech a prosodic prominence on the pronoun I rather than on the verbal element indicates a mental projecting process, e.g. construction

Table  :
Key and Theme.
The four tonics were all pitched as mid key/termination.The examples below illustrate: In both examples, the salient interpersonal Theme is presented as integrated with the message.The speaker presents Example (31) as having a lowered probability.Recognising that her argument is based on a single experienced example, she modulates her claim of a contrast between the attention paid to flood and river defences.In Example (32), the speaker's narrative cannot be challenged.She alone has privileged information as to how the flood affected her friends.Yet, she presents her claim as only probable.The marked tonality coupled with the presence of the filled pause and the filler uuuh signal some difficulty in assembling the words needed to complete her message.The choice of high key signals the transition in her narrative from telling the location of her friend's house to relating a particularly serious and traumatic flooding instance which probably occurred at night.The high key is located on the interpersonal Theme but does foreground her assessment of likelihood; it signals a change in the direction of her unfolding and unformed message.Examples (33) and (34) illustrate that the speaker's assessment is Focal, i.e. presented as non-recoverable.However, it is noticeable, within the dataset, that tonic modals of metaphor are always followed by a noticeable pause and frequently by filler material.This implies that this construction functions not only to signal the speaker's assessment, but also to help the speaker manage the interaction.The speaker in the pursuit of her message knows what she wishes to impart but has not sorted the words required to articulate her communicative intention.For instance, in Example (35), the speaker, who has previously related how the flooding negatively impacted upon her friend's Christmas holidays, signals through her selection of high key a change in the direction of her narrative.The message she develops is that perhaps if there has to be flooding, then flooding during a holiday period is more manageable than at other times.The interpersonal Theme signals her uncertainty that what she says is actually the case.She presents her claim as tentative; one that her hearers are free to disagree with.(35) | I H/guess | (.51) uh … like … it did come out a /bad time like | /Christmas like | (.69) | It was like \probably | (.51) ruined a lot of \people's | (.39) Christmas \spirit but | then at the \same time | it's -kind of | (.75) | if it's going to H\happen | (.29) it it's (.35) a Hgood time in a -sense | that everyone's -there like | (.81) \rather than | (.31)| cause they don't have L/work | and \stuff | Lum | the whole -Ltown | got \flooded | (.44) | but my house somehow /didn't | (.5) | so \like | (.42) \we | like /\my houseThere are two marked Themes notated in bold.Both of them contain a mid key choice.It is clear that a high key choice, which while re-enforcing the speaker's decision to commence her message from an unexpected angle: notably specifying where she lives and its relevance to inland flooding, would have signalled a contrast between sea and river flooding.The prior cotext focused only on sea flooding and the transition to inland flooding was unexpected.But the speaker's point was that the damage caused by inland flooding was as severe as that caused by the sea.Hence a high key would have introduced an unwarranted contrast whereas the speaker wanted to signal equivalence.The marked Theme is repeated to specify the name of the town where the speaker lives.Once again it is pitched as a mid key; to do otherwise would be to introduce the unwelcome notion that somehow it is surprising that the town she lives in is Guilford.Without the marked Theme a hearer would have surmised incorrectly that it was her friend and not the speaker who lived there.It is noticeable that the abandoned Theme notated by italics was pitched with a high key.It is clearly the speaker's intention to explicitly contrast her experience of the flood with that of her friend.In Example (37), the speaker produces a marked Theme illustrating an example of extreme weather.It is the setting for her point that everything goes into meltdown when the weather exceeds the norm.While the marked Theme signals an unexpected angle of development it does not signal a development that is contrary to the discourse expectations; the UK's lack of resilience to weather is widely commented on.The low key signals that the proposition is available to the hearers should they wish to search their memories for instances of Britain failing to deal with extreme weather.The modal adjunct obviously is salient.It is not only the condition of the snow falling but also the obviousness of the condition that is presented as being equivalent to the discourse expectations.Example (38) is another example of marked Theme coinciding with low key.The marked Theme signals a boundary between her description of the scenes shown and not shown in the clip.Her point that the video likely exaggerated the impact of the storm is inferable from what she has previously said.The low key signals that her following message elaborates by providing an illustration of what she claims is a predictable example of hyperbole.The Marked theme adds to the message by expressly linking the verbal projection with the cotext.L\bad | as L/other places | | whichthey | (.26) probably didn't they didn't really L\/show in the clip | (.21) | whereas L /where they were like | this is this L\inn | that's been like at\tacked | and it's Lstill L\standing | (.45) Example (39) provides an example which is more in line with what literature had led us to expect.The marked Theme signals that the message is being developed by the introduction of an unexpected angle and the high key re-enforces the change of direction by signalling the transition to a new unexpected topic.As there has been no mention of any particular year prior to the articulation of the high key, no unwarranted contrast between years is made.The speaker is free to use both high key and marked Theme to signal a transition in the angle of the development of the narrative.