"The complete consort dancing together, [...]" - 
interaction in e-mail

1 Introduction

In the case of Aristotle, the description takes the form of a manual for producing good rhetorical style. The text is prescriptive (example 1) and, as an authority on the subject, the author chooses the elements which he feels are correct and necessary.

(1)
There are, then, these three means of effecting persuasion. The man who is to be in command of them must, it is clear, be able (1) to reason logically, (2) to understand human character and goodness in their various forms, and (3) to understand the emotions- that is, to name them and describe them, to know their causes and the way in which they are excited (Rhetoric, 1356a l. 22).

(2)
On Barbarism
Barbarism is a bad part of speech in ordinary speech, in poetic discourse it is called metaplasm. In our language it is called barbarism, in the speech of foreigners it is called barbarolexis, as if someone were to say mastruga `sheepskin' (a Sardinian word), cateia `club' (a Celto- Germanic word), magalia `hut' (a Punic word). Barbarism occurs in two ways: in pronunciation and in writing. These have four types each: addition, subtraction, changing and transposing of letters, syllables, tones and aspiration. [...] There are also poor transitions, that is cacosynthesis, which some consider to be barbarism; in it we find mytacism, labdacism, iotacism, hiatus, collisions and other utterances which more or less are rejected by educated ears. Having indicated that one should avoid these errors, we give up the argument about what to call them (Donatus, "de barbarismo.").

(3)
[...] the subject I have chosen for my discourse is one of general interest and great potential benefit to mankind. It is this. Who are the most important of the ancient orators? What manner of life and style of writing did they adopt? Which characteristics of them should we imitate, and which should we avoid? [...] I [...] shall select the most elegant of them and examine them chronologically [...] (The Ancient Orators, chap. 4)

(4)
"Hence young people need, at the beginning, much prudent supervision and guidance, [...] to select words which are both pure and refined and to arrange them in a combination which unites grace and dignity. So it is to supply this latter faculty, the first to which the young should apply themselves, that '. . . for the sake of love I offer you a song', in the form of this work on literary composition (On Literary Composition, chap. 1).

Of particular interest here is the approach used by Dionysius. His reliance on the use of authentic material to identify those language items and characteristics that should be taught is vaguely similar to methods used today. His motivation, the desire to improve the quality of oratory and restore the style of the Ancient Attic Orators, is only important, here, in relation to the method he chose to reach his objective and that method was a form of discourse analysis aimed at revealing the salient characteristics of oratory as practiced by great orators.

The methods of LD have evolved over time, and today, if we look at the variety and quantity of grammars, manuals and analyses available, we see that LD is applied to nearly any identifiable aspect of a language. Table 1.1 shows a few relevant examples.

The works under the "Manuals" heading are similar in that they all take the form of a series of rules and/or activities which, when implemented successfully, produce appropriate use of specific language items. Manuals, by nature, are prescriptive. Their principle purpose is to produce uniformity of activity and, consequently, results.

Grammars are lists of rules or relationships in a broader sense. While they contain rules for the use of a language, they don't often contain learning activities directed at developing specific skills. Until recently, grammars tended to be primarily prescriptive. Their main purpose was, and still is in many cases, normative. A normative grammar is the official word on what is correct or incorrect in language usage. There is another type of grammar which is intended to be descriptive and, as such, is based on the observation of language use. Descriptive grammars show how people actually use language, not how they should use it. A recent example of this type is Halliday's Introduction to Functional Grammar (1994).

The works under the heading "Analyses" are of a different nature than those in the first two categories: they are uniformly descriptive. Like descriptive grammars, no claims are made in them about how language should be used. Another feature of analyses is that they normally deal with specific aspects of language like lexis, syntax or function. In this sense, all of the analyses cited in Table 1.1 are distant cousins of Dionysius' Critical Essays. They all share the general characteristics of discourse analysis, as it is described by Brown and Yule (1983: 1):

The analysis of discourse is, necessarily, the analysis of language in use. As such, it cannot be restricted to the description of linguistic forms independent of the purposes or functions which those forms are designed to serve in human affairs .

Furthermore, a humanist outlook demands that, whatever the method, it must take into consideration the social aspect of language, or language use as a human activity. Thus, it will also be the objective of this dissertation to examine and describe that part of verbal communication which deals with the relationship between the participants and how the language they use reflects that relationship.

As with the works of Dionysius of Halicarnasus mentioned earlier, a target language, or language set, must be defined. Just as oratory was an important form of expression in antiquity, telecommunication has assumed a role of growing importance today. Although there are various forms of telecommunication available (telephone, telegraph, teletype, facsimile, etc.), the focus here will be on the language used in asynchronous computer networking through the Internet.

The explosive growth of the Internet over the past two decades has opened a new chapter in human communication. The Internet, and its principal tool, the World Wide Web (WWW), have "virtually" defeated the boundaries of space and time by permitting users to communicate freely and, at times, almost instantaneously. From its modest beginnings as ARPANET, with less than ten hosts connected in 1968, the present system has grown to connect nearly 10,000,000 hosts, worldwide (appendix I.). Designed by the US. Department of Defense, the ARPANET's purpose was to connect distant centers of information and command in case of nuclear attack. Once American universities were connected, the net quickly became an important means of communication between academics and institutions.

Today, the Internet offers a variety of resources which fall into two broad categories: Information Services and Tools. There are three Basic Information Services: E-mail, Remote Execution (Telnet) and File Transfer (FTP). The two principal Internet Tools are Gopher, a text based information retrieval system whose main subsystems are Veronica and Jughead, and the World Wide Web, a graphic interfaced, multi-media information retrieval system whose main subsystems consist of Spiders (search engines) and Guides (indexed search sites such as, Yahoo, AltaVista, Infoseek, etc.).

The oldest of these, e-mail, is still very much with us. It is the principal means of contact between users. E-mail systems are based on the ability of computers to store and transmit verbal and numerical information. Messages are usually entered via computer keyboard and then stored in data files which are sequences of digital code delineated by a "head", which contains the name of the sender, date, subject and message size, and a "tail", which marks the end of the sequence. The file is then transmitted by a combination of cable, telephone line and micro-wave, to sites around the world. Usually, messages pass through several intermediate sites until they reach their destination, where they are stored, once again, and the recipient notified.

E-mail has two subsystems: Netnews (Usenet) and Discussion Lists. These allow users to organize and group themselves according to their interests. The systems permit users to post e-mail to individuals or all the members of a group at once. Users can read all public messages posted to the group and, likewise, make their comments and contributions available to all. These characteristics make e-mail an attractive research universe. The sheer volume of communication, the availability of archived message bases and the existence of delimited discourse communities offer ample opportunities for description of actual language use by discrete groups.

In its early period, the Internet was an entirely American affair and, as a consequence, the large volumes of verbal information that began to accumulate were in English. Even today, as more and more countries come on-line, the overwhelming majority of hosts (access providers) are found in English speaking cultures. A 1996 press release from ISOC (the Internet Society) places the total number of hosts at 9,472,000, of which 6,975,888 are in English speaking countries (US., U.K., Canada, Australia, New Zealand) (Rutkowski). Furthermore, Unix, the software operating system which is the basis of the Internet, and its accompanying assortment of manuals, are written in English. Add to this the fact that all the principal software networking applications were developed in the U.S. and it's easy to understand the place that English holds in Internet communication.

That is not to say, however, that English is the official language nor the only language used on the net. Germany alone has 452,997 hosts, the majority of which use German as their principal language. Nevertheless, the preponderance of English language exchanges on the Internet, linked with the position English holds as an international academic and business language limit language options when participating in international or multi-cultural exchanges. Consequently, international users, more often than not, find it necessary to write or speak English.

For the corpus researcher, the Internet provides a rich source of written texts. Although e-mail forms the vast majority of this body, many other types of texts may be found, ranging from reports to religious dogma, from game rules to jokes, from literature to business reports, and so on, representing a large slice of English language culture. Most of this material is readily accessible. For corpus linguists, it may be the "feast" of the century. The "dish" chosen for the present work consists of e-mail exchanged in an academic or professional setting. The criteria used in forming a corpus from these texts is described in detail in chapter 3.

The theoretical model used here is that of Systemic-functional Linguistics (SFL) because of its pragmatic and humanist characteristics. As we will see in more detail in the next chapter, SFL values the observation of language in use and provides a systemic view of how it functions. Furthermore, it recognizes the importance of a text's social context and provides a three part conceptual framework for interpreting this context: FIELD OF DISCOURSE, TENOR OF DISCOURSE and MODE OF DISCOURSE. This work will not attempt to deal with all of these. Here, the focus will be on TENOR, which "refers to who is taking part" in the exchange and how they relate to each other (Halliday and Hasan 1989). Language exchanges (moves) will be analyzed on the basis of their interpersonal content. "In the clause, the interpersonal element is represented by mood and modality: the selection by the speaker of a particular role in the speech situation, and his determination of the choice of roles for the addressee (mood), and the expression of his judgments and predictions (modality) (Halliday, 1973, 41)." Halliday (1994) calls the grammar of interpersonal meaning the Mood system.

This dissertation is concerned with investigating the Mood system, as it is presented in Halliday (1994) and Eggins (1994). The question raised is whether the MOOD system provides adequate and appropriate means to accurately describe the interpersonal aspects of verbal communication in e-mail messages. In order to answer this question, four aims have been set:

1. to create an electronic corpus from representative e-mail messages
2. to create a functional tagging system and a unit of analysis based on the Mood system.
3. to establish a statistical basis for use of speaker intuition in function classification
4. to classify, tag and analyze both whole messages and their constituent clauses by function.

Chapter 3 presents the project's methodology. There, the research universe is defined, data collection described and a unit of analysis is created. A tagging system is then devised and a description of the analytical method that is used in chapter 4 is made.

In chapter 4 the results are analyzed and discussed. Survey results are compared to the results of message type analysis. Individual messages are analyzed for content and SFG shortcomings are discussed.

Chapter 5 presents conclusions and possible applications of the analysis. The relevance of ESP (English for Specific Purposes) to the application of this project's results is discussed and suggestions are offered for future research.

2 Background

To understand the advantages of SFL as a tool for language description, it is important to understand SFL's four basic concepts: "language use is functional", "its function is to make meanings", "meanings are influenced by the social and cultural context in which they are exchanged" and "the process of using language is a semiotic process" (Eggins 1994: 2).

2.1 Language use is functional.

SFL proposes that all language use is motivated by social needs and that it is functional in as much as it fulfills those needs. "The systemic approach is functional in two main respects, [...] systemicists ask how do people use language? [and] how is language structured for use (Eggins 1994: 2)?" Answering these questions involves observing real life interaction, and how people make meanings through that interaction.

Furthermore, SFL is built on functions, each of which has a specific communicative purpose, enabling people to exchange goods, services or information. In doing so, functions become the "fundamental components of meaning" (Halliday 1994: xiii), or, in other words, meaning is use. Moreover, one could go so far as to say that "the form of language can be substantially explained by examining its functions (Thompson 1996: 1.1.2)."

Underlying these functions, and, consequently, all verbal communication, are what Halliday calls Speech Roles. They indicate the fundamental purpose of each exchange, which can either be "giving" or "demanding". In every exchange there is a commodity which falls into one of the categories of "goods&services" or "information". By combining "Speech Roles" and "Commodities" we create the basis for all verbal expression (Figure 2.1).
 

These combinations, in turn, can be expressed as four primary Speech Functions and their expected responses: an offer can be accepted or rejected, a command can be undertaken (obeyed) or refused (disobeyed), a statement can be acknowledged or contradicted, and a question can be answered or disclaimed. In Table 2.1 each Speech Role + Commodity combination seems to have an exclusive relationship with a corresponding Speech Function. We shall see later that this is not the case. In fact, there is a great deal of flexibility in the way that Speech Role + Commodity combinations are expressed.

2.2 Language use is semantic.

The function of language is to make meanings. Halliday identifies three "metafunctions" or types of simultaneous meaning: Experiential (the clause as representation: the construal of some process in ongoing human experience), Textual (the clause as message: the thematic element ["quantum of information"] the speaker selects for "grounding" what he is going on to say) and Interpersonal (the clause as exchange: the element the speaker makes responsible for the validity what he is saying) (Halliday 1994: 34).

Each metafunction has an associated grammar system. Experiential meaning is described in the system of Transitivity "which construes the world of experience into a manageable set of PROCESS TYPES (Halliday 1994: 106)." Processes fall into six categories: Three main types -

MENTAL ("processes of consciousness"),
MATERIAL ("processes of the external world"),
RELATIONAL (processes of "classifying and identifying"),

BEHAVIORAL ("manifestations of inner workings"),
EXISTENTIAL (processes of being) and
VERBAL (processes of "saying and meaning") (Halliday 1994:107).

"(i) the process itself;
(ii) participants in the process;
(iii) circumstances associated with the process.

Textual meaning is described through the system of Theme. Thematic structure "gives the clause its character as a message (Halliday 1994: 37)" and thus, creates relevance to the context. The descriptive elements used to show this are called THEME and RHEME.

"The THEME is the element which serves as the point of departure of the message; it is that with which the clause is concerned. The remainder of the message, the part in which the THEME is developed, in Prague school terminology, is the RHEME. As a message structure, therefore, a clause consists of a THEME accompanied by a RHEME; and the structure is expressed by the order - whatever is chosen as the THEME is put first. (Halliday 1994: 37)."

Interpersonal meaning, or meaning as exchange, is described through the system of Mood. The term "mood" has a history of use in linguistics and grammar. It is important to differentiate between the two, here. Kittredge and Farley (1913)¹ define "mood" as "that property of verbs which shows the manner in which the action or state is expressed (115)." The authors go on to define three moods, the indicative, the imperative and the subjunctive (1913: 115). Whenever Mood as a system of meaning is mentioned, it appears in bold type. MOOD as an element of the system, or constituent, appears in capitals. Finally, mood, in its traditional grammatical sense, is indicated in normal type.

Halliday divides Mood into two parts: MOOD and RESIDUE. The MOOD element is likewise divided into two elements: SUBJECT and FINITE. Finiteness combines the specification of polarity with the specification of temporal or modal reference to the speech event. It constitutes the verbal component in the Mood. But there also has to be a nominal component; and this is the function of the Subject. [...] The Subject supplies the rest of what it takes to form a proposition²; namely, something by reference to which the proposition can be affirmed or denied (Halliday 1994: 76-7).

Thompson (1996) explains "that the Subject expresses the entity that the speaker wants to make responsible for the validity of the proposition being advanced in the clause" and "the Finite makes it possible to negotiate about the validity of the proposition (section 4.3.3)." MOOD, therefore, is the principal indicator of the relationship of the speaker to what is said and to the listener. In Figure 2.4 we can see that the exchanges revolve around the Subjects: "The shuttle" and "The King", respectively. Both statements are polar and, consequently, can be affirmed or denied: is/isnt, has/hasn't. Furthermore, both reflect the speaker's desire that the listener accept them as the truth, since no doubt is expressed.
 

Although all three types of meaning occur simultaneously in every text, this dissertation will focus on interpersonal meaning. While one type of meaning is not inherently more important than the others, MOOD has the power to reveal the "warp" and "weft" of society's "fabric", that is, how speakers and listeners, or writers and readers, weave the "thread" of dialogues, how they treat each other as members of a community of communicators. By permitting the breakdown of language into Speech Functions and, ultimately, into Speech Acts, analysis for MOOD permits us to examine what speakers hope to accomplish with language and how they accomplish it.

2.3 Language use is contextual.

Halliday (Halliday and Hasan 1989: 5) traces the notion of context to Malinowski (1923, 1935) who perceived the impact of environment on the meaning of expressions. Malinowki's experiences while studying Trobriand islanders led him to include detailed descriptions of the activity which surrounded language uses in order to make meanings accessible, later, to the readers of his study. For this purpose, he coined the term "context of situation". Context of situation is a social aspect of language, it is the knowledge of who is participating, what is taking place and what role language is playing in the activity. Halliday describes these three elements as:

1. The field of discourse - the nature of the social action taking place. This element is described by the TRANSITIVITY system, which deals with processes and experiential meaning.
2. The tenor of discourse - who is taking part and what their relationship to each other is. This is interpersonal meaning which is described through the MOOD system.
3. The mode of discourse - "what the participants are expecting language to do for them." This element is described by the THEME system which describes the clause's character as a message or communicative event.

Although this dissertation does not attempt to identify a REGISTER for the texts which were analyzed, it is important to say that the idea that there are typical ways of performing Speech Functions in those texts, is central to the analysis performed here. The belief that the writers of the texts comprise distinct discourse communities is discussed in chapter 3.

2.4 Language use is semiotic.

"The speaker of a language, like a person engaging in any kind of culturally determined behaviour, can be regarded as carrying out, simultaneously and successively, a number of distinct choices. (Halliday and Hasan 1989: 3)" Language presents us with a range of possibilities from which we select combinations that best express our meaning. These choices are arranged in systems which are sets of options with entry conditions, or a set of things from which one must be chosen, together with a statement of the conditions under which each choice is available (Halliday and Hasan 1989). Halliday goes on to say that "the system is the grammar (Halliday and Hasan 1989: 5)."

The choices a speaker makes reflects movement, in increasing delicacy, through the system to arrive at an appropriate speech act. The formation of the utterance takes into consideration many elements of the grammar systems involved. Situation, participants and topic have all been processed to ensure that what the speaker says is pertinent, shows the proper amount of respect for the listener and conforms to expectations raised by the situation.

The model for analysis adopted in this thesis is taken from Eggins (1994: 216). The author presents a diagram (Figure 2.5) of a "Speech Function system (discourse-semantic stratum)" based on Mood system choices which, as was described at the end of section 2.2, are the focus of this dissertation. The diagram presents Eggins' view of the essential choices for creating interpersonal meaning and is extended to one further level of delicacy or refinement.

The system (Figure 2.5) reads from left to right and shows the three initial choices, called "first-level choices" hereafter, in vertical alignment on the left side. The next set of choices, called "second-level choices", occupies the right side and represents an increase in system delicacy. Eggins (1994) suggests that the system could be extended through increasing levels of delicacy to finally arrive at individual speech acts. Figure 2.6 shows the two levels of choices, in the large rectangles, along with the names, in small rounded boxes, of the first-level choices.
 

In addition to Halliday's Speech Roles (giving OR demanding) and COMMODITIES (goods & services OR information) (see section 2.1). Eggins includes one more first-level choice, initiating OR responding, referred to as Moves, here. Eggins explains that initiating moves tend to be longer and more complete than responding moves, which undergo ellipsis and become minor clauses (Halliday 1994: 152). Responding moves may either be supporting (positive polarity) or confronting (negative polarity). As the diagram illustrates, each entry level choice is connected to selected second-level choices indicating that every time a choice is made, it determines what choices are available on the next level of delicacy. This characteristic is referred to as "pre-selection" (Eggins 1994: 217). Observe that the choice giving+goods & services is connected only to the choice "subjective orientation OR addressee orientation". This indicates that, according to the diagram, the Speech Function for giving + goods&services uses a modulated interrogative taking the subject "I" or "you". The Speech Function, demanding + goods&services connects to the "inclusive OR exclusive" choice which is an imperative or an imperative plus "let's" in the case of "inclusive". The role giving and the Commodity information do not have a common connection. This may be an attempt to keep the diagram from being too confusing. If they did meet, they would meet at the second-level choice, fact OR opinion. According to the diagram, an opinion is a clause plus modulation. Thus, by default, a fact is considered an unmodulated clause.

As we shall see in chapter 4, the analysis of the data brought to light a number of systemic inadequacies. Among these are the system's inability to accommodate the use of non-typical clause types to perform Speech Functions and the ambiguous nature of exchange Commodities in some circumstances. These problems can be described and explained, in part, by way of Brown and Levinson's theory of politeness (1987). Central to that theory is the idea of "face", which in its simplified form says that:

[...] all adult members of a society have (and know each other to have)

(i) face, the public self-image that every member wants to claim for himself, consisting in two related aspects:

(a) negative face: the basic claim to territories, personal preserves, rights to non-distraction - i.e. to freedom of action and freedom from imposition
(b) positive face: the positive consistent self-image or 'personality' (crucially including the desire that this self-image be appreciated and approved of) claimed by interactants

(ii) certain rational capacities, in particular consistent modes of reasoning from ends to the means that will achieve those ends (Brown and Levinson 1987: 61)

The authors treat "aspects of face as basic wants" (1987: 62), thus providing a motivational basis for choices of expression. This basis will be used later to explain some non-typical clause mood choices for Speech Functions and some rhetorical uses of Speech Functions.

It will be argued that the system as presented in Eggins (1994: 216) is inadequate to explain some common language usage, such as indirect speech acts, and that politeness strategies can be incorporated into the system as a means of increasing its delicacy.

3 Research Methodology

3.1 Research Universe

The groups formed by Internet users with like interests are frequently called lists. Their principal medium of exchange is the e-mail message. These groups exhibit the "six defining characteristics" necessary for being identified as discourse communities, according to Swales (1990: 24-7):

1. A discourse community has a broadly agreed set of common public goals.
2. A discourse community has mechanisms of intercommunication between its members.
3. A discourse community uses its participatory mechanisms to provide information and feedback.
4. A discourse community utilizes and hence posses one or more genres in the communicative furtherance of its aims.
5. In addition to owning genres, a discourse community has acquired some specific lexis.
6. A discourse has a threshold level of members with suitable degree of relevant content and discoursal expertise .

There are several types of lists with differences in purpose, amounts of moderation and freedom of access. On the liberal side of the spectrum one finds Usenet, which is made up mostly of unmoderated, general interest lists called newsgroups. Usually, anyone can participate. The free-spirited users of many of these lists zealously guard their First Amendment rights (under the U.S. Constitution) to say whatever they please in whatever manner they please. This contributes greatly to the Usenet's popularity, or as some think, to its infamy. The Internet is also home to a large number of lists dedicated to academic and professional interests. These lists are usually moderated and sometimes restricted in membership. Many academic lists were begun in the Internet's precursor, the Bitnet.

The present project focuses on the latter group, academic and professional lists. Consequently, three academic/professional Internet lists were chosen as the objects of inquiry: TESL-L, a list for teachers of English as second or foreign language, LINGUIST, a list for researchers in any facet of linguistics, and CHEM, a list for the use of computers in chemistry. These lists were chosen partly for their importance (large membership) and their volume of messages. Furthermore, all three have extensive archives of previous messages, which allow access to large quantities of raw data, and, finally, all are conducted in English.

3.2 Classification

3.2.1 Unit of Analysis

In order to analyze speech functions, a unit of analysis must be defined. Within the messages, one of the most easily identifiable structure is the sentence which, as Hoey says, "begins with a capital letter and ends with a full stop [...] (1991: 268)." However, since Halliday states that "[...] a sentence can be interpreted as a CLAUSE COMPLEX: a Head clause together with other clauses that modify it", and "that the notion of 'clause complex' enables us to account in full for the functional organization of sentences (1994: 215).", this latter, more functional view, will be used as the basis here.

Halliday goes on to say:

We shall interpret the relations between clauses in terms of the 'logical' component of the linguistic system: the functional-semantic relations that make up the logic of natural language. There are two systemic dimensions in the interpretation. One is the system of interdependency, or 'tactic' system, parataxis and hypotaxis,[...]. The other is the logico-semantic system of expansion and projection, which is specifically an inter-clausal relation [...]. These two together will provide the functional framework for describing the clause complex (1994: 216).

The clauses making up such a nexus are PRIMARY and SECONDARY. The primary is the initiating clause in a paratactic nexus, and the dominant clause in a hypotactic; the secondary is the continuing clause in a paratactic nexus and a dependent clause in a hypotactic (Halliday 1994: 218).

The Clause Complex and the system of taxis offer precedents within SFG for Head + Modifier type organization. By applying it as a paradigm, a unit of analysis, henceforth called Function Complex (FC), was created. "Occam's razor"³ warns against the unnecessary creation of "entities"; however, the invention of the FC can be justified because it, hopefully, describes a distinct SFG entity and avoids the possible confusion inherent in using terms like "clause" and "group", which have their own specific meanings in Halliday (1994). While all meanings (textual, experiential and interpersonal) have equal weight in a Clause Complex, an FC is intended to represent only interpersonal meaning, as described by the Mood system, and just as the logical component of Clause Complexes is interpreted by the "primary + secondary" relationship of taxis, FCs demonstrate a similar, parallel organization within Mood. An FC has a "head (primary) function + modifying (secondary) function(s)" structure. In the analysis, each clause complex is broken into its constituent clauses. The Head of the "primary clause" and then the MOOD constituent of that Head are identified. Then, the MOOD constituent is treated as the Head Function and its FC type classification is applied to the whole complex. FC types are explained in the next section (3.2.2) and section 3.4 shows how the identification and classification process works.

3.2.2 The System

In SFL, networks of choices are usually illustrated as tree diagrams which, when read from left to right, move from general to specific language options. Figure 2.5 serves as the basis for an analysis which breaks sample texts into their respective Speech Functions with enough delicacy to reveal functional-semantic preferences and tendencies. By following the system from first-level to second-level choices, it is possible to identify speech functions step-by-step.

To further facilitate the analysis, a speech-function tagging system was developed from Eggins' diagram (Figure 2.5). The use of tags offers increased control over the data once it is entered into a computer database. Through them, all the varieties of language employed for a particular Speech Function can be gathered, alphabetized and compared quickly and efficiently. Database searches were performed combining both tags and specific language items. This method permits detailed searches and cross comparisons between functions too. The program used for this was Lotus Approach, a full featured relational database.

A tag consists of eight digits: (00000000). Each digit represents a first-level or second-level choice (Figure 3.1).

The first digit (00000000) represents the first-level choice, giving OR demanding. This choice indicates the direction in which a Commodity will flow. In giving, the direction is away from the writer/speaker (Example 5a), while in demanding an appropriate response is expected in the form of a Commodity flowing toward the writer/speaker (Example 5b).

(5)
a) "The book you ordered was mailed on Jan 3."
b) "Please, send me a copy of ...."

The third digit (00000000) represents the first-level choice: initiating OR responding. Initiating a verbal exchange entails performing one of the "four primary Speech Functions": "offer", "command", "statement", "question" (Halliday 1994: 69). Responding requires performing one of the "desired responses: accepting or refusing an offer, carrying out a command, acknowledging a statement [...] answering a question", or one of the "discretionary responses": "rejection", "refusal", "contradiction" or "disclaimer" (Halliday 1994: 69).

The remaining five second-level choices determine the manner in which a Speech Function is realized and are dependent upon the first-level choices. No more than two of the second-level choices may be used at one time. For instance, in the case of giving + information + initiating (121.....), the only second-level choice to make is polarized OR marked-modal, digit seven. None of the other second-level choices is applicable. So, the resulting tag for giving + information + initiating + polarized is 12100010.

The fourth digit (00000000) is used only when the function is giving + goods & services. It accepts one of three choices: subjective orientation OR addressee orientation OR third person orientation. Subjective orientation is identified by the use of the modulated interrogative mood with the subject "I" (Eggins 1994) (example 6a). Addressee orientation can be identified by its use of the modulated interrogative mood with the subject "you" (Eggins 1994) (example 6b). To these a third possibility was added: third person, or indirect orientation, which takes the declarative mood (example 6c).

(6)
a) "Can I offer you some coffee?" (tag: 11110000)
b) "Would you like some coffee?" (tag: 11120000)
c) "There's a pot of hot coffee on the stove, if you want some." (tag: 11130000)

(7)
a) "Let's have some coffee." (tag: 21101000)
b) "Have some coffee." (tag: 21102000)
c) "I would appreciate it if you to tried some of my coffee." (tag: 21103000)

(8)
a) "What brand of coffee is this?" (tag: 22100100)
b) "Is this Brazilian coffee?" (tag: 22100200)
c) "I would love to know what brand of coffee this is." (tag: 22100300)
d) "Could you tell me what brand of coffee this is?" (tag: 22100040)

Hodge and Kress hold that when a statement is made "emphatically, without qualifications, we are being asked to believe that it is true" (1988: 121). They go on to say that their "own solution to this problem, for the purposes of social semiotics, is to posit 'truth' and 'reality' as integral relations and places in the basic model of semiosis. [...] 'Truth' is therefore a description of the state when social participants in the semiotic process accept the system of classifications of the mimetic plane" (Hodge and Kress 1988: 122). The assertion is that factuality is a matter of convention between speaker and hearer, i.e., if both parts accept a statement as 'true' then it is a 'fact' for them. Moreover, "it is reasonable to assume that in uttering a declarative sentence the speaker is expressing his opinion, that he is making the modal judgment that what he says is true" (Palmer 1986: 27). For these reasons, categorical statements will be treated as a part of modality, not a separate case.

Halliday provides for this type of classification under the heading of "Mood Adjuncts" (1994: 82). He groups "adjuncts of polarity and modality" together, where "adjuncts of polarity" are described as categorical choices of is OR isn't or do OR don't. Consequently, for the purposes of this dissertation, both categorical statements and modal statements will treated as members of the category, modality, and will be labeled as the choice: polarized OR marked-modal. Thus, the problem of whether a statement is fact OR opinion disappears when the target language is put into non-subjective, functional categories.

The eighth, and final, digit (00000000) contains the choice, supporting OR confronting, which is related only to cases where the writer is responding. Supporting can be seen as acceptance, compliance, acknowledgment or answer, and confronting as rejection, refusal, contradiction or disclaimer. This classification deviates from Eggins' (1994) diagram where supporting responses exhibit positive polarity and confronting responses exhibit negative polarity. The problem here is terminological, once again.

The difficulty arises in the definitions of answer and disclaimer. Halliday (1994) describes the "discretionary responses" (see Figure 2.2) in a way that suggests that they are opposites. Thus a supporting move for a question is an answer and a confronting move a disclaimer, i.e. a disclaimer is not an answer. When Haliday's responses are carried over to Eggins' diagram, we see that answer is put into a category of positive polarity. Since everyone knows that questions can receive negative answers and not all negative responses are disclaimers, the categories become ambiguous, or worse yet, contradictory. Although disclaimers are negative by nature, they are not answers as such. They fall into the realm of non-answers, or more accurately, non-cooperation.

This conflict becomes clear in the case of polar interrogatives. The question, "Is it raining?", will expect a limited range of responses. Some possibilities are: "yes", "no", "I don't know" and "who cares?". The first two are answers because they provide information relevant to the question while the latter two are disclaimers because they both deny responsibility for providing an answer. "I don't know." tells the listener that the speaker is not cooperating because of a lack of ability, while "Who cares?" indicates a lack of willingness to cooperate. Simply put, disclaimers deny cooperation and answers provide it. The question then arises as to when a negative answer is a supporting response. One fairly obvious case is that of negative declaratives with question tags. The example, "You don't really want to go out tonight, do you?", expects a negative answer which is, in this case, a supporting response.

Having said all that, a disclaimer will be understood as any non-cooperative response to a clause in the interrogative mood and an answer as any cooperative response, regardless of polarity. While this stretches the meaning of disclaimer^{4 &5}a bit, it allows the discretionary responses to all function in the same manner. The requirements in Eggins' system for negative or positive polarity are eliminated and the pairs of responses in Figure 2.5 can be considered opposites. Figure 3.2 shows the complete tagging system.

3.3 Data Collection

3.3.1 The Corpus

In order to put together a sample, the 40 most recently archived messages were extracted from each of the three lists and combined in a single ascii text file. From the sample of 120 messages 24 were eliminated because they were either garbled, incomplete or otherwise unintelligible. The remaining 96 were then prepared for analysis by removing headers and footers (i.e. names, addresses and message route). From the headers only the subject field was kept. Then, each message was given an identification number (e.g. M#001). Finally, the file was then named SAMPLE.TXT and saved to disk.

3.3.2 Intersubjectivity Test

The intersubjectivity test was used to validate native speaker ability to determine the main objective of the texts in the corpus. For the purpose of this project, simple majority agreement on text classification is considered sufficient justification for accepting native speaker intuition as an authoritative and consistent measure of objectivity.

Before the survey was performed, all the messages were evaluated for function and assigned a tag. For this purpose, only the first three digits in the tagging system were used since they represent the system's first-level choices: giving OR demanding, goods & services OR information, initiating OR responding. For instance, a "call for papers" was tagged as demanding + goods & services + initiating (211) and a message which continued a discussion thread was tagged as giving + information + responding.

The survey used 17 messages, from the total, for analysis by native speaker volunteers. A rotation system was set up where the texts were divided into three groups and the groups put into 18 different orderings to avoid biasing the results. After the message groups were ready, volunteers were solicited on several lists. The lists from which the samples originally came were excluded. The instructions sent with the survey (appendix II) do not give bibliographic references or any other indications of research objectives. Volunteers were given only terse instructions and had to rely on their own language intuition. Once the results were received, they were put into a spread sheet for further analysis. Survey results were then compared to the message classifications made earlier.

3.4 Analysis of constituent Function Complexes

After the survey, attention was focused on message content. The messages were broken down into Function Complexes, which coincided exactly with sentence boundaries. The resulting FCs were entered into a data file as individual records. Each record contained an identification number (which indicated the source message and the FC's position within it), a tag and the FC itself (Figure 3.4). The identification number indicates the quantity and relative position of each FC inside that message. The numbers have a five digit decimal like format. The first three digits contain the message number and the two digits after the separator ('.') contain the FC number. This produced 803 FCs which were then analyzed one-by-one.
 

In each case the HEAD (Halliday 1994) of the primary clause was used as the identifying element of the FC. The process of identifying and tagging an FC is described below.

(5) Example of tagging
FC: "Somebody asked (sorry, but the delete finger is fast) why _The Pearl_ was banned in the orient."
This FC happens to be a clause complex consisting of four distinct clauses:

1. a primary clause (with HEAD underlined) Somebody asked why _The Pearl_ was banned in the orient. whose MODIFIER is a
2. secondary clause why _The Pearl_ was banned in the orient. whose modifier is also a
3. secondary clause banned in the orient. separated from the head by an interjection which is, likewise, a
4. secondary clause [I am] Sorry, but the delete finger is fast. whose modifier is, likewise, another
5. secondary clause but the delete finger is fast.

For the purpose of identifying FCs, only the HEAD of the primary clause is tagged.

• Somebody asked (something)
• polar declarative: no response required
• giving information
• initiating exchange by restating question
• tag = 12100010

As was demonstrated above, not all of the second-level choices can apply to any one FC. Therefore, the maximum number of non-zero digits in any tag is six. For example, in the case of an FC which is giving + information + responding in a supporting categorical manner, five non-zero digits are needed and the tag would be: 12200011 (Figure 3.5).
 

After all of the data was tagged, the resultant FCs were grouped by tag, alphabetized and written to text files. Each file  represents what can be thought of as a family of clause complexes which serve the same function. An analysis of the files was performed to reveal common lexico-grammatical content shared by the FCs. The results were then used to create a description or functional profile of the language used by list participants. As will be shown in chapter 4, the tags permitted a detailed look of writer language preferences, some of which raised questions as to the flexibility and efficiency of the classification system.

4 Results, analysis and discussion

4.1 Results

Results are presented in accordance with the three procedural stages in which the analysis process was divided. These stages correspond to an increase in detail of analysis, moving from message types to their constituent FCs.

4.1.1 Stage 1: message types

The reliance on speaker competence in SFL is exemplified by Halliday's question tag test (1994: 69) to determine the Subject element in MOOD. The test is deceptively simple in that it presupposes that the reader can form the correct question tag, such as:

(9)
"He was born in Illinois, wasn't he?" or
"You'll pick up some milk, won't you?"

When considered separately, the first-level choices found in the three right-most columns show a much higher degree of agreement. The high average (87% for giving OR demanding, 80% for goods&services OR information and 90% for initiating OR responding) tends to support the premise that competent speakers of English have the ability to judge the purpose of an e-mail text with little difficulty.

The results of the survey supported the analysis of all 96 messages by the same intuitive method, i.e generalize first-level choices and apply them to each whole message. Of the total, 94 (98%) were dedicated to the exchange of information, leaving only 2 (2%) as the sole examples of exchanging goods&services.

The largest group of messages (35.45%) was of type: demanding + information + initiating (221). The surplus of type: demanding + information + initiating (221) in relation to type: giving + information + responding (122) is the result of several factors, among which three stand out. First, some (or many) requests may go unresponded. Second, some (or many) responses may be sent to private mailboxes instead of the list. This latter may happen by request or simply because the respondent may believe that there is not enough general interest to warrant posting to everyone. Finally, a large number of requests can often be dealt with by one response. For example, requests for e-mail addresses or other specific and non-controversial information are not necessarily cause for discussion. All this may result in a preponderance of requests for information in the lists examined.
 

Next in frequency is the message type: giving + information + responding (122) which accounted for 33.33% of the messages. Even though there are fewer of them than there are requests, these messages are the staple item of Internet dialogue. If the principal activity of the Internet is the unimpeded and democratic exchange of information, then this format is the main vehicle of that exchange in academic contexts. Whenever a topic is of common interest, these message types are found in abundance.

Messages offering unsolicited information (type: giving + information + initiating) comprise 23.9% of the corpus. These messages include: calls for papers, announcements for conferences, virus warnings, job offers, notices of publication for books and periodicals, and the like.

The last group to be given consideration is fairly uncommon. Messages of type: demanding + information + responding, can only occur in contexts where clarification of a prior request is needed. A writer may find it necessary to have additional information in order to offer a satisfactory response to a request. This group will not be given much attention due to its small and exclusive representation in the corpus.

4.1.2 Stage 2: number and types of FCs

In the second stage, the corpus was treated as a collection of FCs, not messages. The 96 messages analyzed produced 803 FCs. 39 of the FCs were discarded because they were garbled, unintelligible or contained non-vervbal information like dividing lines or other visual cues. This left a base of 764 FCs to work with. It was found helpful to combine FCs which were type initiating with similar FCs of type responding, since they tended to have more similarities than differences. For example, FCs of type: giving + information + initiating were combined with type: giving + information + responding to form a class of type: giving + information. Of the 764, 569 FCs (74.5%) were of type: giving + information, 122 FCs (16%) were of type: demanding + goods&services, 56 (7%) were of type: demanding + information and 6 (0.7%) were of type: giving + goods&services (Figure 4.2).
 
 

4.1.3 Stage 3: correlation of message types with FCs

The third stage combined the results of the first two; message type was compared to message content and the individual structure of each message was examined.

Messages classified as giving + information accounted for 560 (74%) of the 764 FCs, while messages classified as demanding + information accounted for another 144 (19%) of the total. The remaining 22 (2.9%) FCs were contained in the three messages which offered or demanded + goods&services .

Table 4.2 shows a detailed breakdown of the message types in the corpus. The top row shows the type of Commodity being exchanged. The second row contains the message types for each column. They are organized by Commodity to show the striking difference between the number of messages which exchange information and the messages which exchange goods & services. Below each message type is a total and a percentage. For example in the case of message type: 121 (giving + information + initiating), there were 24 messages which made up 25% of the corpus.
 

Table 4.3 correlates the message types with FC types. For example, under the heading for message type 121 (giving + information + initiating), the first entry shows that FCs of type: giving + information + initiating make up 73% of the content of those messages. Messages which exchanged goods & services are not shown because they only account for 4% of the total corpus.

The corpus contains no examples of messages of the type giving + goods&services + responding and only one of demanding + goods&services (whether initiating OR receiving). This is, in large part, due to the general prohibition of commerce on the academic side of the Internet. The two messages which offered goods&services + initiating were announcements for free, on-line learning resources. Even though all these message types are possible, they are not widely used in the context of academic lists. Accordingly, the corpus demonstrated a low incidence of the FC types: giving + goods&services, 6 occurrences (0.7%). However, in contrast to this, the FC types: demanding + goods&services appear frequently and in a variety of contexts. In fact, all the message types contained instances where this FC type was used.

Table 4.4 combines data from tables 4.2 and 4.3, and clearly shows the overwhelming preponderance of messages and FCs which give information. On the average, FCs of type: giving + information make up 73% of the content of messages which are intended to give information (types 121 and 122). In the case of messages which demand information (types 221 and 222) more than half (58%) of the FCs they contain are of the type: giving + information.

Since the quantity of messages which exchange goods&services is too small to be of interest or to give a clear idea of their usual composition, attention will be focused on the majority (94%), messages in which the exchange of information is the main objective. Thus, the exchange of information will be treated as the principal use of language in the target situation.

In messages of the type giving + information (121 and 122), we find that 74% of the FCs are of type: giving + information, 2% are of type: demanding + information and 24% are of type: demanding + goods&services. When we look at messages of type: demanding + information, the number of FCs which are type: giving + information drops to 54%, but still constitutes the majority, the number of FCs which are type: demanding + information increases to 26% and the number of FCs which are type: demanding + goods&services remains nearly the same at 22%. These results are discussed in the next section (4.2).

4.2 Analysis of e-mail message types

The objective of this section is to offer a more detailed description of the analysis process and to act as a starting point for a discussion of the results presented above. Five message texts were selected which are representative of the most common types of messages found in the corpus: giving + information and demanding + information.

Analysis #1: message type - 221 (demanding + information + initiating)

M#04 Subject: help!!
My name is [NAME], and I am currently writing an article focusing on the various intricacies in language acquisition encountered by ESL students. My problem is that I've used a Robert Kaplan article entitled "Cultural Thought Patterns in Inter-Cultural Education" as a reference and I can't find where this article appeared, whether in a journal, book . . .? I received a copy of the article at a conference, and the person who hosted the conference now cannot find the source of the article. Has anyone heard of this article? I'm getting desperate here; I've tried every avenue I can think of, and you're my last hope!! I need the name of the journal or book in which it appeared, date of publication, city of publication (if book), volume & number (if journal), etc., etc.--all the usual bibliography requirements. And, unfortunately, I need these requirements ASAP. Can you help?

1. My name is [NAME], and I am currently writing an article focusing on the various intricacies in language acquisition encountered by ESL students. [12100010] Proposition giving + information + initiating + polarized declarative
2. My problem is that I've used a Robert Kaplan article entitled "Cultural Thought Patterns in Inter-Cultural Education" as a reference and I can't find where this article appeared, whether in a journal, book . . .? [12100010] Proposition giving + information + initiating + polarized declarative
3. I received a copy of the article at a conference, and the person who hosted the conference now cannot find the source of the article. [12100010] Proposition giving + information + initiating + polarized declarative
4. Has anyone heard of this article? [22100200] Proposition demanding + information + initiating + questioning
5. I'm getting desperate here; I've tried every avenue I can think of, and you're my last hope!! [12100010] Proposition giving + information + initiating + polarized declarative
6. I need the name of the journal or book in which it appeared, date of publication, city of publication (if book), volume & number (if journal), etc., etc.--all the usual bibliography requirements. [12100010] Proposition giving + information + initiating + polarized declarative
7. And, unfortunately, I need these requirements ASAP. [12100010] Proposition giving + information + initiating + polarized declarative
8. Can you help? [21100200] Proposition demanding + goods&services + initiating + questioning

Analysis #2: message type - 221 (demanding + information + initiating)

M#03 Subject: AAIP Recommendations?
Does anyone have any information on the results of the TESOL AAIP meetings/ deliberations/ whatever concerning accreditation of ESOL schools? If not, maybe someone can at least tell me who to ask! Thanks.

However, the example used here is not the most common type, but is made up entirely of FCs type: demanding + information, instead of containing the usual complement of FCs type: giving + information. It represents a less used but simpler, more direct form of the message type.

The main reason for choosing this somewhat less frequent example is that it illustrates the use of indirect speech acts in the corpus. Classification of this type of speech act became a problem almost from the beginning of the research. A brief description of the message will lay the groundwork for a discussion of the problem in greater detail.

1. Does anyone have any information on the results of the TESOL AAIP meetings / deliberations / whatever concerning accreditation of ESOL schools? [22100100] Proposition demanding + information + initiating + questioning
2. If not, maybe someone can at least tell me who to ask! [22100320] Proposition demanding + information + initiating + declarative + marked-modal
3. Thanks. [12100010] Proposition giving + information + initiating + polarized declarative

In its original form, there is no place provided for either of these FCs in the system, as described in Eggins (1994: 216), even though both Halliday (1994: 69) and Eggins (1994: 151) recognize the possibility for non-typical responses. As Figures 2.5 and 2.6 show, all of the first-level systemic choices in the original system are binary. Consequently, the second-level choices for FC type demanding + information + initiating are limited to querying (clause: wh/interrog.) and questioning (clause: polar interrog.). No other Speech Role + Commodity combination connects to those second-level choices. An advantage of this organization is that each function has a unique and separate grammatical form. The idea of mutually exclusive grammatical expressions for each function is attractive, but it is unrealistic. The corpus provides examples that clearly show extensive use of non-typical responses, and, as will be demonstrated, there is considerable overlap.

Returning to FC1, the problem arises when we realize that, although it is typical, according to the system, to use polar interrogatives to get information, the information being requested is not in polarized, "yes/no", "do/don't", "is/isn't" form. The typical choice would be to use a wh/ interrogative form. Nevertheless, the is no doubt about what kind response is expected and a literal "yes/no" answer would not be appropriate.

The situation is more complicated still when we consider that FC2, a declarative performing the function of demanding + information, has no place at all in the diagram. As indicated before, both Halliday and Eggins hold that statements are the typical initiating move for giving + information^7, yet the pragmatic reality of the example is that FC2 is unquestionably a request. That reality is supported by the fundamental SFL belief that language use is social and motivated, humans have reasons for communicating. Consequently, little imagination is required to understand what the purpose of FC2 is. Of the 56 FCs of type: demanding + information, 32% were declaratives.

If the four Speech Functions are examined together with both typical and non-typical clause moods, it is possible to see more clearly the areas of overlap between the functions (Table 4.5). Interrogatives appear as options in three of the four Speech Functions (twice as a typical choice and once as non-typical) and declaratives appear in all four Speech Functions for a total of five options (once as a typical choice and four times as non-typical).

Instances a to i in example 10 show all the cases in which information is requested by using a modulated interrogative, beginning with "can". Furthermore, except for the last ("i"), they all have an indefinite pronoun as SUBJECT. All were classified as type: demanding + information + initiating + questioning. None of them actually demands a polarized answer. What is expected is always implied, not explicit. In the case of example 10a, the writer expects the actual help, not a simple "yes" or "no". In example 10b, what is expected is the explanation, not "yes" or "no". In fact, all of the examples below are the same in this respect.

(10)
a) "22100200"\"Can anybody help?"\
b) "22100200"\"Can anyone explain why fluent English speakers feel uncomfortable if the order of adjectives in either of these examples is reversed to fit a more ""logical" "pattern, and supply further examples of adjectival ordering, where parallel machine translation could be affected?"\
c) "22100200"\"Can anyone out there point me toward literature on the verbal morphology of some of the languages of South Sulawesi Indonesia, in particular Bugis, Makassar, or Toraja?"\
d) "22100200"\"Can anyone provide information, anecdotal or otherwise, about the quality of the programs at those schools?"\
d) "22100200"\"Can anyone recomend any good introductory books or articles on the topic of Free Energy Perturbation Theory?"\
e) "22100200"\"Can anyone suggest software (preferably cheap or free!) which can extract and produce this kind of output?"\
f) "22100200"\"Can anyone suggest some material--opther than creating it myself."\
g) "22100200"\"Can anyone who has a computer lab recommend a good pronunciation program-ELLIS looked good but the sound came out blurry- that students could use on their own?"\
h) "22100200"\"Can somebody help me?"\
i) "22100200"\"Can you help?"\

Typically, when we demand information or action directly, we do not know whether the hearer-based conditions are true or not. This need not stop us making our demand -- we can ask anyway, and be rebuffed (if the hearer either cannot or will not comply). However, the indirect strategy is possible. Before risking a rebuff (humans are sensitive creatures), we can attempt to elicit the state of the sincerity conditions. (O'Donnell 11 Jun. 1996)

Thus the questions, "Can you pass the salt?" and "Can you send me your full name and street address?" would imply the following sequences:
 

O'Donnell (11 Jun. 1996) suggests a slight variation on this and proposes that a form of condition checking goes on in this type of situation. "A dialogue planning system set with the goal of discovering some fact might choose to verify the sincerity conditions first. The system could set the discovery of the values of the sincerity conditions as a pre-goal to the greater goal. Thus we would get exchanges such as:

4.13
C: Can you tell me my balance?
O: Yes.
C: Do you want to tell me?
O: Yes.
C: O.K. then, tell me!
O: It's seven dollars twenty.
C: Thank you (O'Donnell)"

Halliday describes the Commodity, information, as something which can exist only in verbal form (1994: 69). Where the Internet and e-mail are concerned, all responses are, by nature, verbal. However, all of the examples (example 10, above) solicit some sort of action, such as: "help", "explain", "point me", "provide", "recommend" and "suggest". Are we to understand that the examples are not the same as the question, "Can you pass the salt?" If so, what makes them different? If the answer is in the process, i.e. , lies within Transitivity, then how can they be given different, exclusive, classifications in the Mood system? If, "Can you pass the salt?" (demanding + goods&services) and "Can you send me your full name and street address?" (demanding + information) are the same in all Mood related aspects, then how can we differentiate between Commodities in the Mood system? It may be that Commodities are not part of the Mood system, after all. Perhaps they belong to the system of Transitivity, where specific types of processes and participants signal what kind of Commodity a modulated polar interrogative is exchanging. Therefore, for the purpose of Mood analysis, the nature of the Commodity being exchanged (goods&services OR information), seems to be of no consequence.

Further classificatory work within the Mood system should perhaps depart from Hasan's solution to the problem:

[...] it useful to think of the primary lexicogrammatical options of the MOOD system as construing "interactive modalities" rather than the "packaged entities" called statement, question, command .. whatever. I suggest there are in English three primary interactive modalities, known in my network as the semantic options: [assertive] realised as declarative; [consultative] realised as interrogative, and [exhortative] realised as imperative. This semantic network will differentiate not simply the favourite Searle-based pair "pass me the salt" v "can you pass me the salt" but such messages as: I'd like the salt, would you mind passing me the salt, could you pass me the salt, pass me the salt will you, pass me the salt won't you... not to mention such things as: how about a speck of salt, lets have that salt etc. When I say differentiate, what I mean is that it specifies the semantic structure of such messages in terms of the set of semantic options (their underlying selection expression) and relates them systematically to the lexicogrammatical systemic options as realisations. As the description advances in delicacy, it ideally specifies features that become criterial in differentiating -- if you are keen on classifying -- different categories of messages, which are at once alike and different to varying degrees from each other. (Hasan 12 Jun. 1996)

Analysis #3: message type - 122 (giving + information + responding)

Message #001 Subject: The Pearl

Somebody asked (sorry, but the delete finger is fast) why _The Pearl_ was banned in the Orient. I really don't know. In fact, I am not sure that it was. What I do remember is that in the Catholic Seminary I was in high school and early college, it was claimed that _The Pearl_ was banned and therefore it was acceptable. (Probably why I never read it.)

This whole conversation is giving me flashbacks to high school and college. The power of the teacher to influence what you think is fantastic. Flashing to comments former English students have given me, the EFL/ESL teacher is in a very powerful position. One aspect is the material that is espoused as being good. Another is the style of life that we show. (And I have no idea where this discussion might take us.)

The message is divided into two paragraphs which perform distinct functions on their own. The first introduces the topic and indicates that the message is a continuation of an active "thread", "The Pearl" by Steinbeck, which was the subject of several exchanges centering around the use of short novels in the ESL / EFL classroom. The second introduces commentary that the writer thinks is relevant and interesting in the current context. While this is not always so clear in other messages, this text seems to divide neatly along THEME and RHEME (Halliday, 1994) boundaries, the first paragraph being the THEME or "given" and the second being the RHEME or "new" information.

As one might expect, the first paragraph is made up almost entirely of responding FCs, since, by their nature, they are concerned with prior information or THEME, while the second contains only FCs which are initiating types.

What follows is an example of the analysis process. Each FC is numbered and accompanied by its eight-digit tag, a short verbal evaluation of functional elements, and a brief discussion.

The message opens with a restatement of the question and an apology for erasing the original message without noting who it was from. From a functional point of view, it is a rather straightforward announcement of the topic which immediately places the message within the thread about "The Pearl".

1) Somebody asked (sorry, but the delete finger is fast) why _The Pearl_ was banned in the Orient.
[12100010] Proposition giving + information + initiating + polarized declarative

(11)
BW>Dear Jack,
BW>Please send me a copy of your article about...

2) I really don't know.
[12200012] Proposition giving + information + responding + polarized declarative + confronting

3) In fact, I am not sure that it was.
[12200022] Proposition giving + information + responding + polarized declarative + confronting

4) What I do remember is that in the Catholic Seminary I was in high school and early college, it was claimed that _The Pearl_ was banned and therefore it was acceptable.
[12100010] Proposition giving + information + initiating + polarized declarative

5) (Probably why I never read it.)
[12100020] Proposition giving + information + initiating + marked-modal declarative

6) This whole conversation is giving me flashbacks to high school and college.
[12100010] Proposition giving + information + initiating + polarized declarative

7) The power of the teacher to influence what you think is fantastic.
[12100010] Proposition giving + information + initiating + polarized declarative

In FC7, the speaker (implied SUBJECT = "I" + FINITE = present + PREDICATOR = "think") is masked by the use of a CLAUSE as SUBJECT. This implies that the evaluation ("fantastic") is universal and, as such, is a fact. However, assertions about values which may not be shared by all the members of a group must be treated as opinions. As was mentioned earlier, this flexibility of usage was the cause for changing the names of the systemic choices from fact OR opinion to polarized OR marked-modal.

8) Flashing to comments former English students have given me, the EFL/ESL teacher is in a very powerful position.
[12100010] Proposition giving + information + initiating + polarized declarative

9) One aspect is the material that is espoused as being good.
[12100010] Proposition giving + information + initiating + polarized declarative

10) Another is the style of life that we show.
[12100010] Proposition giving + information + initiating + polarized declarative

11) (And I have no idea where this discussion might take us.)
[12100010] Proposition giving + information + initiating + polarized declarative

Messages of this type (122) were by far the most common and contained an average of 10.8 FCs each. The example analyzed here has 11 FCs and, as such, is representative of the type. On average, nearly 80% of the FCs in this type of message were of the type: giving + information. In this particular case, the percentage is above the average (100%).

This message is typical of the majority of texts of the type: giving + information + responding) encountered in the corpus. It is a staple item of e-mail communication. Even though there may be many, and varied responses, for any request, the form those responses take is generally the same, regardless of their length.

Analysis #4: message type - 122 (giving + information + responding)

M#09 Subject: Re: Spelling rules
In-Reply-To: Your message of Fri, 28 Apr 1995 00:11:49 -0400 TESL-L
>
> ( -----------Original message-----------)
>
What can " 'Alright' is not a correct word in English (yet)" [NAME]) mean? Doesn't it depend on which 'authority' you turn to? Longmans Dictionary of Contemporary English says: "USAGE Alright is very common now, but some people think all right is better English." The BBC English Dictionary says: (admittedly under the entry for ALL RIGHT) ..'Also spelt alright. 'My ancient, misnamed: Shorter Oxford English Dictionary (2515 pages) reads: "ALRIGHT. Frequent spelling of all right 1893." Doesn't this example show how unimportant such matters are in terms of learning and teaching priorities?

1) What can " 'Alright' is not a correct word in English (yet)" ([NAME]) mean?
[22200102] Proposition demanding + information + responding + querying + confronting

The function used to perform this affirmation is demanding + information. Since no answer is expected or admitted, the question takes on the character of a statement of belief. FC1 calls into question a polarized declarative posted by another list participant. The use of a categorical statement always incurs the risk of rebuttal. Hodge and Kress (1988) equate this with the degree of "affinity" a writer has with the "system of classification" in discussion (123). They go on to say: A high degree of affinity indicates the expression of solidarity between participants. A low degree of affinity indicates that power difference is at issue. Either power (difference) or solidarity may be expressed via a modality of high affinity with the mimetic system. (Hodge and Kress, 123)

In cases of confrontation, the issue is "power (difference)" and in cases of support, the issue is "solidarity". Where marked modality is used in giving information, it indicates that the writer is not particularly committed to a proposition.

In effect, the question indicates low affinity on the writer's part by denouncing the declarative as non-sensical and derides the writer as ignorant or, at least unsophisticated. The same holds true for FC2 which introduces the idea that authority is relative, in this case and questions the authority of the other writer.

2) Doesn't it depend on which 'authority' you turn to?
[22200200] Proposition demanding + information + initiating + questioning

3) Longmans Dictionary of Contemporary English says: "USAGE Alright is very common now, but some people think all right is better English."
[12100010] Proposition giving + information + initiating + polarized declarative

4) The BBC English Dictionary says: (admittedly under the entry for ALL RIGHT) ..'Also spelt alright.'
[12100010] Proposition giving + information + initiating + polarized declarative

5) My ancient, misnamed: Shorter Oxford English Dictionary (2515 pages) reads: "ALRIGHT. Frequent spelling of all right 1893."
12100010] Proposition giving + information + initiating + polarized declarative

6) Doesn't this example show how unimportant such matters are in terms of learning and teaching priorities?
[22100200] Proposition demanding + information + initiating + questioning

Analysis #5: message type - 121 (giving + information + initiating)

M#052 LINGUIST Subject: Langues et Grammaire Conf. Practical Information
Here is some practical information about the conference. The Conference will be held at the University of Paris 8 at Saint Denis. Access: From Paris Metro line 13, direction SAINT DENIS BASILIQUE station SAINT DENIS BASILIQUE, then Bus No 255 stop Universite Paris-8. Accomodation : two Hotels near the University have set aside a block of rooms each: Hotel Campanile: Tel. +33+1+48.20.29.88 Fax +33+1+48.20.11.04 single room 372 FrF including Breakfast Double room 200 per person/night including Breakfast. Hotel Fimotel : Tel : +33+1+48.09.48.10 Fax : +33+1+48.09.85.14 Single room 330 FrF Double room (Twin beds) : 380 FrF To make a reservation : contact directly the hotels BEFORE JUNE 10. Registration Fees : 150FrF (+50 FRF For the proceedings, (optional)) Students 100 FrF (+50 FrF for the Pro ceedings, (optional)). For more information contact us at sdl@univ-paris8.fr. Information will also be sent out by regular mail. HOPE TO SEE YOU IN JUNE

The text opens with two polarized declaratives: the first identifies the nature of the message contents and the second offers an important bit of information, where the conference will be held.

1) Here is some practical information about the conference.
[12100010] proposition giving + information + initiating + polarized declarative

2) The Conference will be held at the University of Paris 8 at Saint Denis.
[12100010] proposition giving + information + initiating + polarized declarative

3) Access: From Paris Metro line 13, direction SAINT DENIS BASILIQUE station SAINT DENIS BASILIQUE, then Bus No 255 stop Universite Paris-8.
[12100010] giving + information + initiating + polarized declarative

4) Accommodation : two Hotels near the University have set aside a block of rooms each: Hotel Campanile: Tel. +33+1+48.20.29.88 Fax +33+1+48.20.11.04 single room 372 FrF including Breakfast Double room 200 per person/night including Breakfast. Hotel Fimotel : Tel : +33+1+48.09.48.10 Fax : +33+1+48.09.85.14 Single room 330 FrF Double room (Twin beds) : 380 FrF To make a reservation : contact directly the hotels BEFORE JUNE 10.
[12100010] giving + information + initiating + polarized declarative

5) Registration Fees : 150FrF (+50 FRF For the proceedings, (optional)) Students 100 FrF (+50 FrF for the Pro eedings, (optional)).
[12100010] giving + information + initiating + polarized declarative

6) For more information contact us at sdl@univ-paris8.fr.
[21102000] demanding + goods&services + initiating + imperative: exclusive (you)

7) Information will also be sent out by regular mail.
[12100010] giving + information + initiating + polarized declarative

8) HOPE TO SEE YOU IN JUNE
[21103010] demanding + goods&services + initiating + declarative + polarized

4.3 Analysis of the more common FC types

One result of grouping and classifying data was the possibility to examine the language used in each FC type. After tagging and alphabetizing, the FCs were then searched to find groups which share common characteristics. What follows is a brief description of the more commonly used types, which indicate the preferred ways of expression in the respective discussion lists.

The analysis is centered on the MOOD element and the rest, the RESIDUE, is ignored. The motive for this is that it is basically the former that contains the constituents which give important information on the relationship between the writer and reader. Politeness strategies (Brown and Levinson 1987) and indicators of indirectness are also usually contained there. The MOOD element is also the key for developing an indexed language base which groups FCs by their language use.

As mentioned earlier, the MOOD element has two parts: the SUBJECT and the FINITE. The SUBJECT element was used to organize and categorize the FCs. Table 4.6 shows the SUBJECT and FINITE content which was focused on during the analysis. Three categories were decided on for analyzing the SUBJECT element: FCs which take the SUBJECT "I", FCs which take the SUBJECT "you" and FCs which take a NOMINAL GROUP as SUBJECT. Eggins describes a NOMINAL GROUP (NG) as "the part of the clause that contains nouns and the words that accompany nouns" (Halliday 1994: 60). A NOMINAL GROUP COMPLEX (NGC), then, is a combination of two, or more, NOMINAL GROUPS (Halliday 1994: 97). Furthermore, each NG can contain "embedded material" such as "prepositional phrases" or "clauses" (Halliday 1994: 97). For the purpose of this analysis, use of the term NG is restricted to those groups which do not have "I" or "you" as their HEAD.

Also, consideration was given to the position of the SUBJECT within the FC. While in most instances the SUBJECT element appeared at the beginning of the FC, there were a number of cases where it was preceded by some form of ADJUNCT.

In keeping with the tagging system, the FC groups will be presented by Speech Role, giving OR demanding, and Commodity, goods&services OR information.

4.3.1 Ways of informing

Of the 569 FCs which are type: giving + information, 133 have "I" as the SUBJECT in the MOOD element, while more than 364 have a NG as the SUBJECT. Table 4.6 shows the most frequently used ways of informing that were found in the corpus.

Under the heading "SUBJECT = 'I' ", there are 5 forms which account for nearly half of the FCs in that group. Of these the most recurrent forms were those which followed the SUBJECT with verb tense. Example 12 shows some of the more common ones.

(12)
SUBJECT ("I") + tense
"12100010"\"I begin one of these survey classes with Chaucer and proceed to Shakespeare, Donne, and many other poets which are even difficult for native speakers."\
"12100010"\"I know it from my own work and from my own calculations."\
"12100010"\"I know of a nice job opportunity for a person or persons with linguistic expertise in Vietnamese, Laotian, and/or Cambodian (Khmer, I guess?)."\
"12100010"\"I need the name of the journal or book in which it appeared, date of publication, city of publication (if book), volume & number (if journal), etc., etc.--all the usual bibliography requirements."\

(13)
SUBJECT (NG) + "is" + ATTRIBUTE
"12100010"\"Enrollment is limited to 20."\
"12100010"\"FUN101 is a free, non-credit course which will take place online 6/5-7/28, and is designed to help non-native speakers of English living outside of an English-speaking country improve their writing in English."\
"12100020"\"GEOPOL is very fast and you, as I have, could easily compute higher dot densities by computing the centers of the tesserae and extrapolating to the sphere's surface"\
"12200012"\"The following statement by Joe Landman is either incomplete or wrong:"\
"12100010"\"The language is so similar to many of our students' interlanguage - no copula, no third person 's', etc."\

4.3.2 Ways of requesting

As mentioned above, both Halliday and Eggins offer what seem to be clearly defined "typical" choices for each Speech Function. The corpus, however, presented a different picture. In the analysis which follows, the difference between Eggins' "typical" choices for demanding, imperative for demanding + goods&services and interrogative for demanding + information (Eggins 1994: 216), and the preferences found in the corpus will be the focus.

In the corpus, 54% of the FCs of type: demanding + goods&services were declaratives, not imperatives. Furthermore, of the FCs which are type: demanding + information, 47% were polar-interrogatives, 40% were declaratives and only 13% were wh/ interrogatives. The underlining in the previous sentences serves to note that those categories don't exist in the original system diagram. Nevertheless, declaratives do make up a significant percentage of the requests in the corpus.

Another finding is that, of the total of all the interrogatives used in information requests, 75% were type: polar. This is significant because the use of polar-interrogatives, in this corpus, is not explicit; the literal answer to the question ("yes" or "no") is not the information being requested (see discussion p. 66-70). Thus, they are used in a manner which infers the desired response.

(14)
Querying and Questioning
Querying
a) "22100100"\"What was it like? Who did you work for? What opportunities are there around Poland? Etc."\
b) "22100100"\"What do others think?"\
c) "22100100"\"What ""functional"" forms do you teach?"\

Questioning
d) "22100200"\"Is there an intensive course I can take via distant learning mode,ie. correspondence or short intensive program?"\
e) "22100200"\"By the way, does anyone know of a publisher who might be interested?"\
f) "22100200"\"Do you ""balance"" your writing program?"\

(12)
Declaring
a) "22100310"\"suggestions on how to prepare for this adventure will also be appreciated inasmuch as living in japan and teaching english will be novel experiences for me."\
b) "22100310"\"Right now I will take any general information on the subject, but great if you have specifics on issues dealing with parent control/parachute kid situation."\
c) "22100320"\"Any info on newspaper articles, magazine articles, TV programs, school programs, districts to contact, where to get hard data (if available), your experience with parachute kids.... will be greatly appreciated."\
d) "22100320"\"I would appreciate any input you all can give to me personally or to the network if the program is worth further use."\
e) "22100320"\"I'd appreciate your response to my private address as well as to the list."\
f) "22100310"\"I am interested in receiving suggestions of titles, preferably full length works, for intermediate to advanced ESL college freshmen."\

The most used method for requesting some action (goods&services) on the part of the reader is a polite form of imperative. But, there are examples of declaratives used to express the desire that some action be performed. In general, they use the same structure that is found in declarative requests for information.

Usually, writers preceded demands with "please" (example 15a,d and e). It is the most common treatment of commands in the corpus. Another type of politeness which shows up is the use of adverbs to cover the bluntness of a direct imperative (example 15c). An additional factor in example 15c is the adverbial phrase of purpose which precedes the main clause. It serves two purposes. First, it qualifies the command by adding the conditional meaning: "Do this only if you are interested in our Web site." and, second, it reduces the bluntness of the command by dislocating it further back in the clause complex. Example 15b shares these characteristics, although to a lesser extent, since the phrase of purpose follows the main clause.

When imperatives were used without politeness, they were generally devoted to giving instructions and were frequently part of announcements or calls for papers. As part of such, they offer little threat to face (example 15b).

(15)
Commanding
a) "21102000"\"So please remove me from your list."\
b) "21102000"\"Send me your snail mail address for the guide to contributors."\
c) "21102000"\"To access Fluency Through Fables, simply point your favorite web browser to: http://www.interport.net/~comenius/fable.html"\
d) "21102000"\"Please send any comments/suggestions/criticisms you might have!"\
e) "21102000"\"Valerie - please provide your e-mail address to those of us who would like your handout."\
f) "21102000"\"If anyone does send me information, please include the source (and your name) so I can accurately document this."\

Declaratives are the very soul of indirectness. They can be so indirect that it can be plausibly denied that a request has even been made. Indirect declarative requests take two principal forms in the corpus. One uses "I" or "we" as the SUBJECT element and a modal FINITE. The writer, or responsible body, expresses desire or interest that a certain act be performed or condition fulfilled (examples-16a, g and h, below). These were often preceded by a conditional clause beginning with "if" (examples-16g-i). The other uses a NG as SUBJECT in a polarized attributive context and has no politeness marker. The impersonal quality, resultant from the use of the passive voice, gives these FCs an abstract, matter-of-fact tone (examples 16b and c). They are the most indirect type of request found in the corpus.

On the surface, these FCs are type: giving + information, but, as in the case of polarized questions, the reader can detect another function beneath. As Searle (1969) implied, the reader will try to find the relevance of the statement. The question, in the case, is, "Why am I being given this information?" It is natural for the reader to try fitting the information into the immediate context of situation. The obvious answer is that the writer wants something in the form of information or goods & services.

This one situation where that the line between the two Commodities becomes blurred. In examples 16b-e, the ambiguity becomes apparent. The possibility of the FCs meaning both Commodities at the same time is concluded later, in section 4.4.

(16)
Declaring
a) "21103020"\"We would be interested in hearing from potential contributors interested in relationships between (mainly) literary theory and current and emerging technologies."\
b) "21103010"\"Abstracts are invited for 20 minute talks in all areas of Spanish Linguistics (History/Dialectology, Phonology, Syntax/Semantics, Applied, Sociolinguistics, etc.)"\
c) "21103010"\"Submissions are encouraged by e-mail."\
d) "21103010"\"Abstracts are to be submitted in both a short and long version."\
e) "21103010"\"The short abstract is to be prepared for photocopy reproduction in the meeting handbook."\
f) "21103020"\"Reservations must be made at least 24 hours in advance by calling (913) 842-2432."\
g) "21103020"\"If anyone can pass on that information to me I would be grateful."\
h) "21103000"\"If anyone has information or handouts that I could include in a manual or a workshop, I would appreciate a copy."\
i) "21103000"\"If not, maybe someone can at least tell me who to ask!"\

Even when the manner is more direct, as in example 14e, an impersonal distance is established by the absence of the SUBJECT "you". It is literally directed at no one in particular. For Brown and Levinson this is a form of Negative politeness where the FTA is performed, but with redress to the reader's "want to be unimpinged upon (1987: 131)."

Recapitulating, the largest group of the messages in the corpus performed the function of demanding + information. The next largest group performed the function of giving + information. Messages whose function was information exchange (giving OR demanding) accounted for 96% of the corpus. Almost all of the messages, regardless of function, contained FCs of type giving + information. Message length varied from 1 to 40+ FCs with an average of 8 FCs per message. Requests (demanding) tended to be slightly shorter than responses (giving).

The most frequent FC type in the corpus is giving + information which, in the majority of cases, has a Nominal Group as Subject. The FC types for demanding make up the next largest group. Since the first-level choice for Commodity (goods&services OR information) becomes ambiguous in many instances, it is disregarded here. In most cases, imperatives use "please", and indirect requests which use the declarative mood contain marked modality.

4.4 Discussion

Difficulties encountered during the analysis of the corpus raised questions about the organization of the system and some of its fundamental constituents. In Halliday (1994), each Speech Role + Commodity combination (Figure 2.1, p.14) maps to one of the four primary speech functions (offer, command, statement, question) (Table 2.1, p. 15). But, as it turns out, that is only true when dealing with direct speech acts. Indirect speech acts, which choose non-typical clause mood, break the one-to-one relationship between speech functions and Speech Role + Commodity combinations (Table 4.5, p64). To accommodate this change, more second-level options were introduced into the system. These additions reflect politeness strategies in Brown and Levinson (1987). Of the four new options, three are "Off record" choices. They are: choice 3 in the fourth digit, declarative mood used for giving + goods&services, choice 3 in the fifth digit, declarative mood used for demanding + goods&services, choice 3 in the sixth digit, declarative mood used for demanding + information. The remaining new option falls into the category of "conventional indirectness" as described in Brown and Levinson (1987: 132-44). This choice specifically covers the use of modal interrogatives to make requests (choice 4 in the fifth digit). The bold face type in Figure 4.5 indicates these additions.
 
 

Although Halliday (1994) and Eggins (1994) mention the fact that non-typical mood choices and indirect speech acts occur, they don't go into detail. The system in Eggins (1994) which serves as the model here, does not accommodate indirect speech acts. The corpus, however, demonstrates clearly that, in the context examined, indirect speech acts are used frequently, and must be explicitly integrated into the system, if it is to be used for more effective classificatory work.

The relationship between speaker and listener, or TENOR (Halliday 1994) of a text, is revealed, in part, through the levels of politeness and indirectness employed. When these levels are high, it is a sign that speakers are trying to diminish the imposition of their requests or threats to the addressee's face (Brown and Levinson 1987). Thus, choice of clause mood (declarative, interrogative, imperative) for a speech act and, consequently, of typical or non-typical forms, is determined by the relative status of the participants and the level of politeness they feel compelled to use.

As was discussed earlier, indirect requests tend to cause some confusion about whether they belong to the category demanding + information or demanding + goods & services. In practice, there is no grammatical difference between indirect requests for the two Commodities. The inclusion of new second-level options (Figure 4.5, above ), while helping to classify indirect requests, destroys the binary nature of Eggins' (1994) original Figure. Furthermore, the fact of two different Speech Role + Commodity combinations sharing common speech functions, runs contrary to the system's organization.

In Eggins (1994: 216), the available choices for demanding + goods&services are in the imperative , while the available choices for demanding + information are in the indicative. These choices, however, only apply to direct speech acts. When the same requests are made indirectly, the choices for demanding + goods & services undergo a shift to the indicative. This causes all indirect requests to be formed from the same choices.

Analysis revealed two frequent types of indirect requests which lie at the heart of the issue: one which uses modal interrogatives and another which uses modal declaratives (examples 16a-f and 17a-f, below).

(16)
Indirect requests using modal interrogative
a)"\"004.09"\"21103000"\"Can you help?
b)"\"006.05"\"21103000"\"Can anyone provide information, anecdotal or otherwise, about the quality of the programs at those schools?
c)"\"019.01"\"21103000"\"Can anyone who has a computer lab recommend a good pronunciation program-ELLIS looked good but the sound came out blurry- that [...]?"
d)"\"067.02"\"22100200"\"Can anyone explain why fluent English speakers feel uncomfortable if the order of adjectives in either of these examples is [...]?"
e)"\"087.03"\"22100200"\"Does any one can help me ?" [Can anyone help me?]
f)"\"096.01"\"22100200"\"Can anyone recomend any good introductory books or articles on the topic of Free Energy Perturbation Theory?"

(17)
Indirect requests using modal declarative
a) "\"079.07"\"22100320"\"I would also be interested in knowing about any other hypertext linguistics papers that may be out there.
b) "\"093.02"\"22100320"\"I would be greatly thankful if you could send me any information you probably will receive about this problem.
c) "\"068.07"\"22100320"\"I would be very grateful if you could give some information about any other languages you may know or speak which may present the[...]."
d) "\"015.03"\"21103000"\"I would also be interested in hearing from anyone thinking about going to Poland, or Eastern Europe, to teach.
e) "\"019.05"\"22100320"\"I would appreciate any input you all can give to me personally or to the network if the program is worth further use.
f) "\"028.07"\"21103000"\"I would appreciate information about resources available on ESL Tutor Training.

1. Indirect requests may imply both Commodities at the same time.
2. Indirect requests may make the addressee and/or the Commodity indefinite.

Interestingly enough, there is never any doubt about the writers' goals in examples 16 and 17; just as hearers usually understand what speakers are requesting through indirect speech acts. The problem that arises here is systemic, consisting of the need to accommodate indirect speech acts within the system while maintaining the ability to assign a unique grammar structure to each Speech Function. As the system stands, there is a conflict which eliminates that possibility. The solution that was applied (new second-level options) is not satisfactory. Though it accommodates some indirect speech acts, it doesn't differentiate their grammar.

Any solution will require basic systemic modifications. One possible answer is to create a first-level choice for Directness (direct OR indirect) and make Commodity a second-level choice for the Directness. This solution groups the choices for indirect requests together in one, binary choice (modal interrogative OR modal declarative) and direct requests in another (goods&services OR information), which lead to their own, separate, discrete choices, as they exist in the current system.

Another more drastic solution would be to altogether eliminate Commodity as a choice. This move might be more difficult to justify, but could use the line of reasoning that Commodity is not actually part of the Mood system; that in reality, it belongs elsewhere, possibly the grammar of processes, Transitivity. This, however, remains to be seen. These solutions cannot be pursued here and will remain for future research to sort out.

5 Conclusion

The tagging system proved to be both flexible and durable. The design of the tags allows them to be augmented whenever necessary. Their eight-digit format is compact, machine readable and allows considerable power in searching. Although the format was worked out early in the project, when it was believed that all of the choices would be binary, it was easily adapted to the changing classification system.

Originally, Standard "C"⁸ language comment markers were used to delimit the tags. Future implementations of the tags will be SGML8 (Standard Generalized Markup Language) compliant and use the ubiquitous SGML brackets ( "<", ">"). A full DTD⁹ (Document Type Definition) for the tagging system is also planned.

The tagged corpus offers easier access to information on functional choices and their frequency. A researcher or student could take the corpus and find out how list users on the Internet use language and how they prefer to be addressed. The corpus, to paraphrase Eliot, tells us which are the words "neither diffident nor ostentatious."

Two of the several possibilities for application suggested by the data will be considered here. They are: a writing course in English for Internet list participants and a software application for the automated functional tagging of texts, based on the classification and tagging systems developed here.

For Internet users who are not native speakers of English, the reality of easy international communication can bring a heightened awareness of their linguistic needs and shortcomings. It is for researchers and educators to meet those needs and provide appropriate educational opportunities. The prospective beneficiaries of the suggested language description and a consequent ESP based language course would be academics and professionals who are interested in participating in e-mail Discussion Lists. Some of these characteristics and needs are outlined below:

1. All list participants can be considered to be competent computer users.
2. All list users who participate in international lists need English language skills.
3. A significant number of list participants in Brazil do not participate actively. This may be a result of insecurity about writing in English or a lack of English skills.
4. It is also assumed that there is a significant number of Brazilian academics and professionals who are computer users and who would participate in lists if they had an appropriate level of English language competence.

The classification and tagging systems also lend themselves to software application. The idea of automated software tagging of texts is attractive for a number of reasons, two of which are of particular interest here. First, developing the software would, necessarily, drive the development and extension of the system. The software would serve as a test bed for the theory behind the classification system.

The basic idea would be to develop software which could parse and compare clause complexes to a language base of previously tagged FCs. Upon finding partial or complete matches, the program would then display the possible classifications along with its best guess. At this point, at least in the early stages, there would be user interaction. It is believed that, the larger the language base, the more accurate the program's guesses. The results would be probabilistic, relying on the program's knowledge (the language base serving as experience and memory). Language bases could be developed for other discourse communities and, eventually, comparisons could be made between groups.

The second reason is that, even during developmental stages, the software would alleviate some of the drudgery of tagging large amounts of text by hand and, consequently, reduce the time needed to produce language descriptions which would be useful for other research or course design. The software application would not be an all or nothing proposition. From its first implementation, it would produce meaningful results. Improvements made thereafter would extend an already functioning system.

What would Dionysius of Halicarnassus think of all this? It is tempting to believe that, despite the difference in technologies, he would recognize a familiar spirit in the work that linguists and educators do today, the effort to further the power of human expression and to have, as Eliot says,
 

 Notes

1. George Kittredge (Harvard U.) and Frank Farley (Wesleyan U.) were turn-of-the-century educators and grammarians. Their compact text (1913) is representative of what may be called traditional, main-stream, prescriptive English grammar. They leave no doubt that the book's purpose is normative when they state that the rules of grammar derive from "good usage, - that is, from the customs or habits followed by educated speakers and writers (1913: xv)." They are cited here in contrast to the concept of modern descriptive linguistic practice.
2. Halliday has specific categories for the exchange of commodities. "When language is used to exchange information, the caluse takes on the form of a PROPOSITION. [...] the semantic function of a clause in the exchange of goods & services is a PROPOSAL (1994: 70-1). " Thus, offers and commands are PROPOSALS while statements and questions are PROPOSITIONS.
3. "Occam's razor is a logical principle attributed to WILLIAM OF OCCAM, although it was used by some scholastic philosophers prior to him. The principle states that a person should not increase, beyond what is necessary, the number of entities required to explain anything, or that the person should not make more assumptions than the minimum needed. This principle is often called the principle of parsimony. Since the Middle Ages it has played an important role in eliminating fictitious or unnecessary elements from explanations. In the development of logic, logicians such as Bertrand RUSSELL removed traditional metaphysical concepts by applying Occam's razor. Questions have been raised, however, as to whether a person can determine without any doubt that given entities or assumptions are not needed in an explanation. Unless this determination can be made, it is impossible to tell with complete certainty when the principle can be applied"(Popkin).
4. "disclaim    (dis klam)  v.t.1. to deny or repudiate interest in or connection with; disavow; disown: disclaiming all participation. 2. Law. to renounce a claim or right to. 3. to reject the claims or authority of.  v.i.4. Law. to renounce or repudiate a legal claim or right. 5. Obs. to disavow interest. [140050; late ME < AF disclaimer, desclamer. See DIS-1, CLAIM]" (Random House Unabridged Electronic Dictionary).
5. "disclaimer    (dis klamŸr)  n.1. the act of disclaiming; the renouncing, repudiating, or denying of a claim; disavowal. 2. a person who disclaims. 3. a statement, document, or assertion that disclaims responsibility, affiliation, etc.; disavowal; denial. [140050; late ME < AF: to DISCLAIM]" (Random House Unabridged Electronic Dictionary).
6. "Flaming" is  the name given to emotional and, frequently, offensive responses  to messages which contain,  real or imagined insults.
7. Actually, it is unclear in the system diagram exactly what is to be done in the case of giving + information, since the two terms do not connect at any point.
8. "C" is powerful, general purpose structured programming language developed at Bell Laboratories in the 1970's.
9. "A DTD is a formal description of the structure of a particular class of documents (Tittel et. al.,. 1995 , 29)."

Works Cited