The elided participant:
presenting an uncommonsense view of the researcher’s role
Geoff Thompson
[In A.-M. Simon-Vandenbergen, M. Taverniers and L. Ravelli (eds.) 2003. Grammatical Metaphor: Views
from systemic functional linguistics, Amsterdam: John Benjamins, 257-278.]
Abstract
In this paper I explore various non-congruent ways of writing about research which
allow writers to disguise their own role as observers and interpreters - a disguise
which is culturally accepted as a necessary aspect of objectivity. Some of these
resources are well known and well described: these include particularly
nominalisation and passivisation, which have tended to dominate discussions of
objectivity in scientific writing. However, I also examine other resources which need
to be brought into the picture, all of which contribute in different ways to eliding the
researchers and at the same time imbuing the text with a sense of their presence.
Nominalisation is, of course, a key type of grammatical metaphor; and I discuss the
extent to which the feature of participant eliding which it shares with the other
resources examined can be seen as inherently metaphorical.
1 Introduction
The main focus of my discussion in this paper will be an examination of what happens to the researcher who is a real-world participant in certain processes when those processes are talked about in the text. Broadly speaking, the processes fall into three overlapping groups (see Thompson & Ye, 1991, for this idealised view of research as construed in the transitivity of reporting verbs): material processes expressing research actions - performing an experiment, collecting data, etc.; mental processes expressing the act of observing and interpreting research findings; and verbal processes expressing the act of discussing these and other findings. The groups are inherently overlapping, in that collecting data involves observation, and interpreting data is frequently done through discussing them; but the basic model has practical heuristic value. I should point out that I am using the term „researcher‟ to include the writer(s), and/or other researchers, and/or the readers (since, in the texts that I am investigating, the readers are construed as peers who are, at least potentially, involved in similar research and are competent to participate in and evaluate the writer‟s argumentation).
2 Resources for eliding
2.1 An example
I would like to start the analysis of resources for eliding participants with a simple example from an article on economics (in the Guardian newspaper of 27th June 1992), in order to establish the kinds of
phenomena that will be focused on:
The north emerges from every statistical comparison that can be made as significantly
poorer than the south.
The researcher is elided from this sentence in (at least) three ways. The first resource that the writer draws on for this purpose is nominalisation. One of the consequences of construing a process as a „thing‟ is that mention of the participants involved in the process becomes structurally optional. Here
the nominal group „every statistical comparison‟ can be „paraclaused‟ (i.e. paraphrased as a clause) as „every time someone compares the statistics‟: the „comparer‟ has been elided from the nominalisation. The second resource is passivisation: the Agent by whom the comparison „can be made‟ is elided. With the
passive, it is easy to probe for the elided participant („Who by?‟); and even with the nominalisation the participant can be recovered without too much difficulty (particularly by expressing the process congruently as a verbal group - „Who compares the statistics?‟). In other words there are traces in the text of the participant, as inherent arguments of the verb.
In the case of the third resource, on the other hand, recovery is less straightforward: the participant is more thoroughly elided. The reconfiguration of meanings brought about by the nominalisation allows the interpretation of the results of the comparison to be represented as „emerging‟ from the comparison. Of
1
course, an interpretation like this depends on there being an interpreter - a more congruent wording would be: „every time someone compares the statistics they understand that the north is significantly poorer‟.
However, to recover the participant who interprets means recasting the wording more radically than with the nominalisation or passive. It involves reconstructing, at least partially, a plausible version of a physical and mental event (people looking at statistical data and drawing conclusions) which is referred to in the clause. The arguments of the verb „emerge‟ do not map at all onto the participants in this event, and it is only by referring „outwards‟ to that event that the analyst can see which potential participants have been elided.
This example suggests that, to explore eliding fully, it is useful to distinguish between what may be termed C-participants (where „C‟ stands for „clause‟) and W-participants (where „W‟ stands for „world‟). C-
participants are those entities which either are explicitly represented in the wording as involved in the process, or can be recovered as arguments of the verb. In the latter case, recovery may be necessary because they are not explicitly mentioned (as can occur with the Agent of a passive clause). It may also be necessary if the process is nominalised: here the participants may not be mentioned, or they may appear as various kinds of modifiers of the nominalisation. In such cases, paraclausing can bring the participants to the surface and show their relationship to the process. W-participants, on the other hand, are those entities which can be plausibly assumed to participate in the physical or mental event or state represented in the clause, whether or not the entities are also C-participants. The extent of the match between these two kinds of participants can vary in many ways, but there are three major possibilities:
; the W-participant and C-participant correspond more or less congruently: e.g. „someone compares the
statistics‟
; the W-participant corresponds to a recoverable C-participant: e.g. „every statistical comparison [by
someone]‟
; the W-participant has no corresponding C-participant: e.g. „the north emerges as poorer‟ [„someone
understands that the north is poorer‟]
It is the last two categories which are relevant to a discussion of the eliding of participants.
Before leaving the discussion of „emerges‟, it is worth relating this to the alternation identified by Halliday (1994: 114) in mental processes. He pointed out that these can be construed as generated in the sensing mind or as stimulated by external phenomena:
I understood the situation of the north
= mental cognition process construed as Senser apprehending (and thereby
construing/creating) Phenomenon
The situation of the north struck me
= mental cognition process construed as (pre-existing) Phenomenon affecting Senser -
note the common mental-as-material blend in the choice of process
The wording in the example can be seen as taking this a step further towards representing the Phenomenon as the dynamic entity in the process:
The situation of the north emerged
= mental cognition process construed as material action of (pre-existing) Phenomenon This type of dynamic representation of what is congruently a mental process in a sensing mind will be further explored in the discussion in 3.1 below of verbs such as „demonstrate‟.
2.2 Nominalisation
I have so far assumed without discussion that nominalisation allows participant eliding. However, it is, of course, possible with nominalisations to mention the participants as modifiers of various kinds - e.g. „student
self-assessment‟, „government of the people by the people‟. In a small-scale study of nominalisations in a 1 the distribution shown in Figure 1 range of randomly-chosen university textbooks and academic papers,
was found (the study covered only nominalisations functioning as Head of a nominal group - the total number was 489).
1 no participants mentioned 37%
2 „done-to‟ mentioned 26%
3 „doer‟ mentioned 22%
4 Carrier mentioned 14%
5 both „doer‟ and „done-to‟ mentioned 1%
Figure 1 Participants in nominalisations
2
Here are representative examples for each category, with paraclauses to show the congruent participant roles and with elided participant slots marked by „X‟ and „Y‟:
1 the illumination [X illuminates Y]
an acceleration equal to g [X accelerates at the speed of g]
quantitative analysis [X analyses Y quantitatively]
2 prolonged exposure of the body to [X exposes the body to less extreme heat for a
less extreme heat long time]
addition and subtraction of vectors [X adds and subtracts vectors]
increased electron absorption and[X absorbs electrons and generates X-rays more]
X-ray generation
3 the constant acceleration of a freely [a freely falling body accelerates constantly]
falling body
the body’s cooperation [the body cooperates]
increase in accelerating voltage [accelerating voltage increases]
4 the coherence of an electron source [an electron source is coherent]
the presence of a suitable potential [a suitable potential is present]
the instability in the sources [the sources are unstable]
5 ion bombardment of the filament by [gas ions bombard the filament with ions]
gas ions
These findings actually suggest that in academic writing it is the norm for C-participants in nominalised processes to be mentioned in some way: 63% of the instances have at least one participant mentioned. For reasons which will be made clear below, I am particularly interested in cases where what 2may be called the „natural Subject‟ of the more congruent clausal wording is expressed. Since categories
3, 4 and 5 all share this feature, they can be counted together in terms of looking at which participants are mentioned: they comprise 37% of the total. This makes it appear that nominalisations are as likely to occur with natural Subject expressed as they are to occur with no participants expressed. Even if, in order to concentrate on the relative frequency of mention of natural Subject in contrast to other options, we add categories 1 and 2 together, this still means that well over one-third of instances have natural Subject mentioned.
However, the natural Subject may be a range of different types of W-participants: as the examples above show, they are mainly the phenomena being studied (a freely falling body, gas ions, etc.). The processes are equally varied, but are mainly material and relational, referring to the actions and qualities of those phenomena. On the other hand, the picture alters dramatically when the focus is restricted to nominalisations that refer to processes in which the researcher is a W-participant. (The following figures are based on the same texts as above, but more extracts were examined in order to accumulate a comparable overall total: 447.) I found no cases in which the researcher, if mentioned, was anything but natural Subject, 3 so I have simply divided the relevant nominalisations into 2 categories - see Figure 2.
1 researcher not mentioned 91%
2 researcher mentioned 9%
Figure 2 Researcher as participant in nominalisations
Examples of each category are:
1 simultaneous collection of the whole [X collects the whole range of X-rays
range of X-rays simultaneously]
parameter measurements [X measures parameters]
a sophisticated understanding of the [X understands the nature of voice in a
nature of voice sophisticated way]
a metaphorical interpretation of the [X interprets the process metaphorically]
process
most discussions of these issues [usually, when X discusses these issues]
2 his analysis [he analyses Y]
our scrutiny [we scrutinised Y]
our preliminary speculations [we begin speculating about Y]
our discussion [we discuss Y]
3
Thus with regard to the particular case of the researcher, participant elision is very much the unmarked option in nominalisation in academic text (though other participants in those processes - e.g. the „natural Objects‟ - may well be mentioned).
2.3 Passivisation
As noted above, the example with which we began also draws on the other familiar resource for participant eliding in academic text, passivisation. In a survey of 438 finite passives in the same texts, the great majority were agentless, but there is again an interesting difference when the findings are divided into those with the researcher as understood or expressed Agent, and those with other Agent - see Figure 3.
Agent: researcher Agent: other
56% 31% no Agent mentioned
1% 12% Agent mentioned
57% 43% Total
Figure 3 Presence of Agents in passive clauses
The researcher is the Agent in slightly more than half the occurrences. However, whereas over one quarter of the cases with „other‟ Agents have the Agent mentioned, when the Agent is the researcher participant elision is overwhelmingly the unmarked option. Even the few cases in the agentive/researcher category that do occur are often marginal: for example, in one the researcher mentioned is not the researcher who wrote the paper: „The story was recorded by Eiríksson‟. (It is worth mentioning that „other: no Agent mentioned‟ includes one instance of the researcher as Subject of a passive form: „the investigator can be
forced to attend ...‟ - though this could equally be categorised as objective modulation rather than a true passive. It is also worth mentioning that 2% of cases included in „researcher: no Agent mentioned‟ are where the elided Agent refers specifically to the reader - e.g. „It will be observed ...‟.)
Representative examples of each category are:
the brightness of a scanning image can be scaled agentless/researcher:
different kinds of grammaticalization have to be distinguished
when the issue of context is raised, it is typically argued ... 4a term that will be used by most papers in this volume agentive/researcher:
X-rays will be generated agentless/other:
at the moment the utterance is produced
his adversary’s villainous tongue is eaten out by a shrimp agentive/other:
the vacancy can be filled by an electron
Thus, the traditional picture of nominalisation and passivisation as two important resources for „depersonalisation‟ of academic text - where I am interpreting „depersonalisation‟ as meaning participant
elision - is correct as long as it is made clear that it is one particular type of participant that is routinely elided (cf. Iedema, 1997, on similar strategies in administrative directives; and Coffin, 1997, on eliding the historian in school history textbooks). But that is not the whole story. There are a number of other resources which have not received as much attention but which also make a substantial contribution to avoiding the need to mention the researcher in the text.
3 Eliding in text
3.1 Eliding the Senser in methods and findings
The eliding of the W-Senser - the researcher as observer in a perception process and interpreter in a cognition process - is in fact a wider phenomenon (small „p‟!). In order to illustrate this point, it will
be useful to examine a stretch of text that is representative of one kind of academic writing, an extract from a medical case study reported in the Journal of Trauma (Deluca et al., 1996). This extract
focuses on the events which are later to be interpreted by the writers of the paper (i.e. it is oriented very much towards the research action/observation end).
On admission [to hospital], no fetal heart tones were detected by Doppler examination
and no fetal movement was seen with ultrasound evaluation, indicating fetal death. The
patient underwent an emergency laparotomy with splenectomy, and repair of the
diaphragmatic injury. A moderate sized, nonexpanding pelvic hematoma was noted.
The bladder rupture was treated by urethral catheterization and bladder
decompression. There was no evidence of uterine injury. The consultant obstetrician
chose not to perform a cesarean section at the time of surgery because the patient had
already sustained significant blood loss and had developed a coagulopathy.
4
Postoperatively, the patient was hemodynamically stable without clinical signs of
placental abruption or disseminated intravascular coagulation, and a decision was
made to electively induce a vaginal delivery at a later time.
On the first postoperative day, the patient developed sudden onset of oliguria with
progression to anuria over a 3-hour period. Her vital signs were stable, and a physical
examination was unremarkable. She demonstrated no signs of symptoms to suggest the
onset of labour. A repeat cystogram showed the urethral catheter within the bladder
without obstruction or contrast extravasation.
There is only one sentence here („The consultant obstetrician ...‟) in which the research participants - in the broad sense used here - come to the surface. How are they elided the rest of the time? There are clearly a number of nominalisations and passives which play a major role in this. In the following version I have italicised words and inserted [by R] wherever the researcher has been elided by either of these
two resources (note that „by‟ is used in the text on two occasions to encode Circumstance: Means rather than Agent; and two cases which are not unambiguously nominalisations have been marked as queries).
On admission, no fetal heart tones were detected [by R] by Doppler examination [by R]
and no fetal movement was seen [by R] with ultrasound evaluation [by R], indicating
fetal death. The patient underwent an emergency laparotomy with splenectomy, and
repair [by R] of the diaphragmatic injury. A moderate sized, nonexpanding pelvic
hematoma was noted [by R]. The bladder rupture was treated [by R] by urethral
catheterization [by R] and bladder decompression [by R]. There was no evidence of
uterine injury. The consultant obstetrician chose not to perform a cesarean section [?by
R] at the time of surgery [?by R] because the patient had already sustained significant
blood loss and had developed a coagulopathy. Postoperatively, the patient was
hemodynamically stable without clinical signs of placental abruption or disseminated
intravascular coagulation, and a decision [by R] was made [by R] to electively induce a
vaginal delivery [by R] at a later time.
On the first postoperative day, the patient developed sudden onset of oliguria with
progression to anuria over a 3-hour period. Her vital signs were stable, and a physical
examination [by R] was unremarkable. She demonstrated no signs or symptoms to
suggest the onset of labour. A repeat cystogram showed the urethral catheter within the
bladder without obstruction or contrast extravasation.
However, there are several sentences in which a further set of resources for the elision of Senser is drawn on. These are processes which imply the existence of a real-world observer; they are bolded and marked [to R].
On admission, no fetal heart tones were detected [by R] by Doppler examination [by R]
and no fetal movement was seen [by R] with ultrasound evaluation [by R], indicating [to
R] fetal death.
She demonstrated [to R] no signs or symptoms to suggest [to R] the onset of labour.
A repeat cystogram showed [to R] the urethral catheter within the bladder without
obstruction or contrast extravasation.
These processes, and a small set of others like them (show, demonstrate, suggest, indicate, reveal,
imply, signify, prove, display, manifest, etc.), frequently cause problems in doing a transitivity analysis since they can refer to a complex type of process that is not adequately captured by any one of the major categories. The simplest solution seems at first sight to be to label them as verbal processes, taking a very broad view of these as including any process which involves the „symbolic exchange of meaning‟ (Halliday, 1994: 140). In the instances above, there is no projection, so the labelling would be Sayer and Verbiage - e.g. „She [Sayer] demonstrated [Process: verbal] no signs or symptoms ... [Verbiage]‟. However, Halliday (1994: 142) argues that cases like these are best seen as identifying relational clauses (Token^Value). With instances like the first in the extract:
no fetal movement was seen with ultrasound evaluation, indicating fetal death
this is certainly more convincing than the verbal process analysis, since there is a clear relationship of identity: „no fetal movement‟ = „fetal death‟. However, the line between relational and verbal readings is
somewhat indeterminate. For instance, in the final sentence of the extract:
A repeat cystogram showed the urethral catheter within the bladder without obstruction
or contrast extravasation.
it is harder to see a relation of identity between the cystogram and the absence of obstruction that it showed. The difference here seems to arise from the fact that in the first „no fetal movement‟ is an observation, which can be equated with an interpretation since both are phenomena of the same kind (mental constructs); whereas in the second „the cystogram‟ is the means of observation (the messenger rather than the message), and the „no obstruction‟ is the observation. In this latter, there are close similarities to examples such as „my
5
watch says six o‟clock‟, and the arguments in favour of a verbal process reading seem very strong - „a repeat
cystogram [Sayer] showed [Process: verbal] the urethral catheter ...[Verbiage]‟.
One possible reason why these processes are on the borderline between verbal and relational (symbolising) is that they appear to construe the exchange of meaning from both ends, as it were: not only as something conveyed by the Sayer but simultaneously as something grasped by the Receiver - meaning extracted/constructed from phenomena rather than passed on. Relational processes are different from other process types in that „relating‟ is something done not by the participants but by the speaker (in transitivity terms, s/he might be called the Relator - except that s/he cannot normally appear in the clause except as Circumstance: Angle: „for me, this book was the clear winner‟). In a clause such as „the patient was
hemodynamically stable‟, the process „was‟ is a record of the speaker‟s attribution of the quality of stability
to the patient. Thus relational clauses have an inherent link with „extracting meaning from phenomena‟. Note, however, that a relational clause expresses a relation - i.e. the product of the relating. The demonstrate
type of process, on the other hand, might almost be seen as „relating‟ rather than „relational‟ - it expresses
the process of constructing the relation on the basis of evidence. For example, „no fetal movement indicat[ed] fetal death‟ construes the process of moving from an observation (of lack of movement) to a
conclusion as to what this represents (fetal death). At the same time the Relator/Receiver is typically not mentioned in such clauses: the meaning is construed as immanent in, or embodied by, the phenomenon. (It is worth noting that in all cases there is the structural possibility of making the Relator/Receiver explicit - „indicated fetal death to us‟; but in my data this does not occur even once.)
What both suggested analyses, as identifying and as verbal processes, miss is precisely what makes these processes so useful for, and frequent in, academic text: the fact that they simultaneously indicate that a process of interpretation by the researcher of the relation is going on (thus validating the Token-Value relation being represented by giving the grounds for the relation), and yet construe the interpretation as self-generated by the phenomenon and not as dependent on any particular Relator. The Relator role is implied but left open - anyone could fill it and would arrive at the same interpretation. The Relator‟s real-world role
is dual: perceiving the evidence (the Token) and interpreting the significance (the Value). These are, of course, roles that in other construals of the real-world events are attributed to the Senser in mental processes, which is why I have talked about the processes as eliding the Senser. Admittedly, all this does not actually help with the question of what transitivity label to give the processes which are not clearly verbal. Both perspectives on the process - as identifying and as reflecting the cognitive process involved in establishing the identification - seem equally important. For what it is worth, I therefore find it useful to label them as identifying/mental blends.
To look at the issue from a different perspective, it is significant that it seems natural to use „by R‟ to tag the nominalisations and passives, and „to R‟ for this group, even though both may involve the researcher as Senser. It corresponds in certain respects to the distinction mentioned above between Senser as „doer-to-Phenomenon‟ and Phenomenon as „doer-to-Senser‟. This suggests that this semantic distinction is
applicable beyond the basic structure of clauses with a mental process, and underlies other structural options - a view which, of course, fits in with general systemic functional theory (see especially the discussion of proportionalities in Martin, 1992). The „to R‟ tag also fits in with the idea of the observer merely receiving the interpretation from the phenomenon.
3.1.1 Indexing the W-participant
Before leaving this extract, it is worth noting a number of other wordings where the real-world participation of the researcher might be said to be pointed to, or „indexed‟, rather than elided: that is, wordings which refer to physical actions and states that inherently involve the researcher but where there is no straightforward way of paraclausing them to reinstate the researcher as C-participant. These include both the „by R‟ type, where nouns such as „laparotomy‟ are not strictly speaking nominalisations but refer to actions
by the researchers, and the „to R‟ type, where nouns such as „evidence‟ and adjectives such as „unremarkable‟ may or may not be related to mental processes („evidence‟ is not, but „unremarkable‟ is) but
nevertheless imply a W-participant as Senser. The justification for including these is that they contribute to the overall sense of the text as imbued by the unstated presence of the researcher: they resonate with the other eliding choices (see Thompson, 1998, on the concept of resonance in text). These have been added below, using the same conventions as above, to give a fairly full picture of the extent of researcher elision in the extract.
On admission, no fetal heart tones were detected [by R] by Doppler examination [by R]
and no fetal movement was seen [by R] with ultrasound evaluation [by R], indicating [to
R] fetal death. The patient underwent an emergency laparotomy [by R] with
splenectomy [by R], and repair [by R] of the diaphragmatic injury. A moderate sized,
6
nonexpanding pelvic hematoma was noted [by R]. The bladder rupture was treated [by
R] by urethral catheterization [by R] and bladder decompression [by R]. There was no
evidence [to R] of uterine injury. The consultant obstetrician chose not to perform a
cesarean section [?by R] at the time of surgery [?by R] because the patient had already
sustained significant [?to R] blood loss and had developed a coagulopathy.
Postoperatively, the patient was hemodynamically stable without clinical signs [to R] of
placental abruption or disseminated intravascular coagulation, and a decision [by R]
was made [by R] to electively induce a vaginal delivery [by R] at a later time.
On the first postoperative day, the patient developed sudden onset of oliguria with
progression to anuria over a 3-hour period. Her vital signs [to R] were stable, and a
physical examination [by R] was unremarkable [to R]. She demonstrated [to R] no signs [to
R] or symptoms [?to R] to suggest [to R] the onset of labour. A repeat cystogram [?by R]
showed [to R] the urethral catheter within the bladder without obstruction or contrast
extravasation.
3.2 Eliding the Senser in discussion and conclusions
As mentioned above, this extract focuses on the data collection stage of research (the fact that the events
were presumably originally seen by all concerned as treating a patient who unfortunately died is not relevant: in the text they are reconstrued as evidence in a research project). I want now to carry out the same
kind of analysis with an extract that focuses on interpretation and discussion, from a paper by Louise Ravelli, „A dynamic perspective: implications for metafunctional interaction and an understanding of Theme‟ (1995, slightly abridged). To save space I will immediately highlight the resources described in the
preceding analysis using the conventions already introduced.
In contrast [to the synoptic perspective], spoken language, which presents text as active
and on-going, encourages a view [by R] which is less totalising, and more process-like. To
account for this aspect of language, a dynamic perspective is needed [by R]. Yet it is
too easy to make a false equation [by R] between written language and a synoptic
perspective on the one hand, and spoken language and a dynamic perspective on the
other. .... It is therefore inappropriate to equate one mode with only one perspective: all
language is both product and process. Given that linguistics models have largely arisen
in conjunction with the study [by R] of written language, the relatively recent focus [by R]
on spoken language has revealed [to R] the inability of available models to account for
the process nature of language. However, both modes can - and should - be accounted
for [by R] from both perspectives. This suggests [to R] that a perspective on language
should be independent from any ‘intrinsic’ qualities of the data. A dynamic perspective,
then, is a way of modelling or explaining language as, or as if it were, unfolding.
Within the systemic functional approach the dynamic perspective has tended to be
most vigorously pursued [by R] in relation to certain linguistic phenomena. These
phenomena include the study [by R] of generic structure ... As these phenomena can be
seen [by R] to unfold in time, and to be dependent on the nature of the current
environment, they could be said [by R] to be ‘inherently’ dynamic. Certainly a full
account [by R] of them demands that they be considered [by R] as processes, and
doing so reveals [to R] inadequacies and weaknesses in synoptically oriented models.
However, the suggestion [by R] that some types of linguistic phenomena are ‘dynamic’,
and that by inference [by R] others are not, is an equation [by R] as false as the equation
[by R] of a dynamic perspective with a specific mode of discourse. A dynamic
perspective should not be reserved [by R] just for certain types of data, but should be
applied [by R] broadly to language as a whole.
This indicates that the same resources play an important part in both data reporting and discussion sections. However, it is clear that the analysis above by no means covers all the cases where the researcher is
elided: there are some resources which are perhaps more typically found in discussion sections. An important group of omissions are the non-finite verbal groups, which appear in the following sentences - the
groups are underlined, and [R] is inserted where the Subject is elided. I also include two extra „to R‟ instances - „too easy‟ and „inappropriate‟ - which will be discussed below.
[R] To account for this aspect of language, a dynamic perspective is needed [by R].
Yet it is too easy [to R] [R] to make a false equation [by R] between written language
and a synoptic perspective on the one hand, and spoken language and a dynamic
perspective on the other.
It is therefore inappropriate [to R] [R] to equate one mode with only one perspective: all
language is both product and process.
7
A dynamic perspective, then, is a way of [R] modelling or [R] explaining language as, or
as if it were, unfolding.
Certainly a full account [by R] of them demands that they be considered [by R] as
processes, and [R] doing so reveals [to R] inadequacies and weaknesses in synoptically
oriented models.
Non-finite verbal groups, by definition, have no Finite; and they also have no Subject, though the „doer‟ can of course be expressed elsewhere in the clause complex. Indeed, a crude rule-of-thumb that
language teachers often invoke is that a non-finite verb typically „takes‟ its Subject from a finite verb nearby
in the clause complex. One thing that is noticeable about the instances in the extract is that the „doer‟ is completely elided from the complex in each case. Three of these cases („modelling‟, „explaining‟, doing‟) are gerunds: they are therefore on the edge of nominalisation and do not depend on finite verbs elsewhere in the complex. One („to account for‟) occurs with an agentless passive verb in the clause on which it depends: the use of the non-finite form harmonises with the use of the agentless passive, precisely because they both elide the researcher.
The other two cases („to make‟, „to equate‟) are of particular interest. The embedded clauses in which they appear are extraposed Subjects of evaluative clauses. When such structures involve modality (e.g. „it is possible/necessary to...‟), they are described as explicit objective modality and recognized as interpersonal grammatical metaphor (Halliday, 1994: 355). Although Halliday does not explicitly extend the discussion to attitudinal expressions, it can be argued that, just as with modality, the expression of personal stance is being treated as if it were an attribute of the proposition - the similarity is underlined by the fact that „it is inappropriate to‟ could be viewed as objective modulation. What is being disguised in each case, of
course, is the outcome of an assessment by the researcher. If the researcher‟s role in this assessment is made explicit, the result is a projecting clause with a mental process in which the researcher is Senser („I think/believe/feel/etc. that ..‟). Thus it is again the researcher in their role as W-Senser that is being elided. I
have therefore included these two evaluations, though somewhat awkwardly, as falling into the „to R‟ category.
The usefulness of these anticipatory „it‟ structures in expressing writer stance in a depersonalised
way is already well established; but I am interested in the fact that the embedded clause is a non-finite „to‟-
infinitive clause rather than a finite „that‟-clause. Both these types of embedded clause can occur in the
structure (though usually with different types of evaluation in the matrix clause - see Thetela, 1997). Compare, for example:
It is inappropriate to equate one mode with only one perspective
It seems reasonable that learners should be helped to progress in the direction of some
variety of standard competence.
In one sense, the selection of an objective evaluative clause and of the type of embedded clause can be seen as two separate choices. However, the Subject in „that‟-clauses in this pattern is normally some participant
other than the researcher (as in the second example above, where the Subject is „learners‟). When the researcher is the „doer‟ of the process in the embedded clause, the normal option is the non-finite one (it is
the researcher who „equates‟ in the first example), which avoids the need to mention the doer. Thus, a good case can be made for viewing the choice as a single one, led by the writer‟s desire to elide themself from the message. The „it ... to ...‟ structure kills two participant birds with one stone: it elides the researcher as
evaluator in the text and as doer in the physical event.
This suggests that wordings in academic text may at times be selected in order to accommodate a non-finite form so that it is not necessary to have an expressed Subject. This is, of course, one of the reasons that has been claimed for choosing nominalisations. Non-finites and nominalisations can therefore be viewed as competing for the same ecological niche in this respect.
3.2.1 Metonymy
There is in fact yet another resource for not mentioning the W-participant that needs to be mentioned, one which appears to draw on the familiar category of metonymy. The word „perspective‟ could be seen as implying a Senser, but it does not fit easily into the two main categories established so far: it perhaps suggests a separate category - „of R‟. This could also be applied, though less obviously, to „models‟ and in fact to any terms which refer to the products of the researcher as thinker or writer when they are used as 5metonyms for the researcher. These can again be seen as indexing the researcher, in the sense in which I have used it above. Cases are included in the version below marked by bold italic, together with the other elidings identified above.
In contrast, spoken language, which presents text as active and on-going, encourages a
view [by R] which is less totalising, and more process-like. [R] To account for this aspect of
language, a dynamic perspective [of R] is needed [by R]. Yet it is too easy [R] to make a
8
false equation [by R] between written language and a synoptic perspective on the one
hand, and spoken language and a dynamic perspective on the other. .... It is therefore
inappropriate [R] to equate one mode with only one perspective: all language is both
product and process. Given that linguistics models [of R] have largely arisen in
conjunction with the study [by R] of written language, the relatively recent focus [by R]
on spoken language has revealed [to R] the inability of available models [of R] to
account for the process nature of language. However, both modes can - and should -
be accounted for [by R] from both perspectives [of R]. This suggests [to R] that a
perspective [of R] on language should be independent from any ‘intrinsic’ qualities of the
data. A dynamic perspective [of R], then, is a way of [R] modelling or [R] explaining
language as, or as if it were, unfolding.
Within the systemic functional approach the dynamic perspective [of R] has tended to
be most vigorously pursued [by R] in relation to certain linguistic phenomena. These
phenomena include the study [by R] of generic structure ... As these phenomena can be
seen [by R] to unfold in time, and to be dependent on the nature of the current
environment, they could be said [by R] to be ‘inherently’ dynamic. Certainly a full
account [by R] of them demands that they be considered [by R] as processes, and [R]
doing so reveals [to R] inadequacies and weaknesses in synoptically oriented models [of
R]. However, the suggestion [by R] that some types of linguistic phenomena are
‘dynamic’, and that by inference [by R] others are not, is an equation [by R] as false as
the equation [by R] of a dynamic perspective [of R] with a specific mode of discourse. A
dynamic perspective [of R] should not be reserved [by R] just for certain types of data,
but should be applied [by R] broadly to language as a whole.
3.3 Summary of environments for eliding participants
To summarise, I have discussed the following range of lexicogrammatical environments in which the researcher as participant may be elided:
„by R‟ (Agent): nominalisations; passives; „action‟ nouns
„to R‟ (Senser): identifying/mental processes; explicit objective assessment;
„mental‟ nouns and adjectives
„of R‟ (Carrier): metonymy
„R‟ (Subject): non-finites
At this stage the reader may have the impression that the study has turned into a manic „hunt-the-
participant‟ chase which has been stretched beyond the necessary limits. However, it should be emphasised that what the „full‟ analyses above are doing is showing the places where the participation of the researcher in the physical or mental activity is discernible in some way but is not explicitly encoded in the wording. In other words, at each of these places, it would have been possible in principle to mention the researcher. It is not being claimed that in a more congruent wording the researcher might be mentioned in every one of these slots - obviously that would not be feasible. However, the analyses do give some idea of just how absent the researcher is; and they also indicate the interweaving configurations of mutually supportive lexicogrammatical choices which contribute to that absence.
At the same time, there is a further crucial feature of all these resources, which has been touched on in relation to the identifying/mental blends discussed above. All the wordings trail diffuse clouds of researcher behind them; and what I have been doing in the analyses, I would claim, is to trace the sources of the vapour trail. Academic text is largely depersonalised, but it is also intended to come across as deeply imbued with the sense of an observing, interpreting mind. Indexing the researcher, leaving a trace of their presence even while eliding them from the explicit wording of the text, is a highly distinctive characteristic of this register. If I tell an anecdote about a funny event that I witnessed, I do not typically preface every statement with „I saw ...‟ or „I heard ...‟ even though I was the W-Senser: in casual conversation there is not
the conventional expectation that I will project a sense of everything being filtered through my interpreting mind. In academic writing, on the other hand, a sense of the filtering mind is essential; and the resources described above are designed to achieve the delicate balancing act of eliding the researcher and simultaneously infusing the text with their presence.
4 Issues remaining
The discussion so far raises a number of issues, of which three seem to me to be particularly interesting (though I cannot as yet see any way of resolving the last two).
I have so far presented the reasons for eliding the researcher as relatively unproblematic - essentially as a question of conforming to conventions. However, the possible motivations in any particular
9
6 In the first extract, the events described are clearly highly charged case are likely to be much more complex.
in emotional terms. The writers‟ main purpose is to present data to be interpreted rather than to tell an unhappy story of someone who died; but the sensitive nature of those data must presumably affect to a certain extent the way the writers present them and provide a further reason for construing them impersonally (potentially at least, omissions on the part of the medical staff could be seen as having contributed to the death). In relation to the second extract, Ravelli (personal communication) has pointed out that talking about „the researcher‟ as an undifferentiated entity obscures important distinctions, especially in texts putting forward views that could be controversial - as hers does. In particular, there is the question of alignment: whether „the researcher‟ represents the author and like-minded scholars, or researchers against
whose views s/he is arguing. The reasons for not mentioning the researcher in these two cases are presumably different. Similarly, although I have found it convenient to include the reader under the umbrella term of „the researcher‟, the reasons for not mentioning the reader are likely to be different again
(for some discussion of this, see Thompson, 2001). There is clearly room for more delicate description of the phenomenon of participant eliding, especially in terms of the factors that encourage it, within the framework set out here.
The second issue takes us into areas where I tread more warily. I argued earlier that what might be seen as separate choices are probably best viewed as different manifestations of the same choice. That can clearly be applied to complete texts (indeed, the whole register): the decision to present the uncommonsense view of the researcher‟s role that we associate with academic writing can be seen as opening up, or weighting the probabilities heavily in favour of, the syndromes of choices outlined above. (Of course, it is somewhat misleading to talk about it as a „decision‟: for most academic writers on most occasions it is a more-or-less automatic acceptance of working within the conventional probabilities, with just occasional Prufrockian hesitations - „Do I dare to eat a peach? Do I dare to mention me?‟).
However, such „signing up‟ to a particular register (which we might call the writer‟s „registration‟?) brings doubts about the picture that has been implied so far of the process of producing and understanding the wordings. In very simple terms, I have been writing as if the writer experiences the „real-
world‟ events and deliberately recasts them in uncommonsense terms in the text, selecting certain aspects to highlight and others (especially their own role) to background. In understanding the text, the reader then reconstitutes as far as possible the physical events, reinstating the researcher as a W-participant. See Figure 4.
writer „converts‟ reader „reconstitutes‟
; ； ; ；
real-world construal in text “real-world
event event”
Figure 4 The simple view of uncommonsense construal
However, it seems at least as likely that the process is shortened by familiarity, and that both writers and readers understand the uncommonsense construal without calling directly on experience of, or general familiarity with, the physical events. See Figure 5.
writer interprets reader
understands
; ； ; ；
real-world construal in text “event
event as construed”
Figure 5 Shortened uncommonsense construal
My own guess (and it is only that) is that both processes must occur, to differing degrees according to external factors. It seems unlikely that a researcher who has carried out an experiment can blank out that experience completely: the process of „converting‟ may play more of a role here than in, for example, the presentation of abstract argumentation. (Evidence from comparing spoken presentations with published versions of the same paper indicates that the „conversion‟ is typically much less pervasive in the former. For
example, presenters talking about their own research frequently intersperse short narratives of the physical sequence of events - see S. Thompson, 1997.) However, it seems plausible that the second process dominates
10