How programming language influences interaction and culture
By Ivar Tormod Berg Ørstavik
This article illustrates how morphology in programming languages influences interaction among programmers. To facilitate this illustration three theoretical approaches (Dialogism1, Cassirer and Phenomenology) are presented and contrasted against each other. Based on this theoretical exploration and analytical toolsfrom dialogic theory an empirical study is performed. This study experiments with and exhibits how different type systems in object-oriented programming encourage and discourage different programming practices and power structures.
1.1 An applied linguistic approach to theories of language2
For an applied linguist theories of language are combined with empirical studies. These syntheses often focus on solving/describing language problems/phenomena from the ‘real’ world and place the theoretical foundation in the background. Such an approach fits well with the name ‘applied linguistics’. But, it is wrong to conclude that such a synthesis sums up applied linguistics. Applied linguists often place theories in the foreground and use the empirical foundation as a background against which to investigate and evaluate these theories. Hence, applied linguistics is not just about empirical knowledge and solving problems; applied linguistics also strives to enhance our theoretical understanding of language by constantly flipping empirical phenomena and theory as foreground and background.
Abstract discussion and comparison is one way to shed light on both similarities and differences between theories and the theories in themselves. But, evaluating theories through abstract or historical analysis alone is vulnerable to the personal, subjective taste and beliefs of researchers. Therefore, different theories on language should not only be described against each other, but also against some empirical background. By pulling in a third party, such as empirical data, a less biased mediator is involved in discussions on the ‘righteousness’, ‘truthfulness’, ‘usefulness’ or other value of different theories.
In this article I will try to use just this approach to better understand the nuances between three related theoretical views on language: Dialogism, Cassirer and phenomenology.3 This theoretical analysis will then be juxtaposed with an extensive empirical analysis of morphology and power in programming languages using a dialogic perspective. Based on this empirical analysis and the preceding theoretical discussion, the article will argue that and seemingly trivial attributes of Dialogism might be more important than often suspected and serve to demarcate Dialogism from both Cassirer and Phenomenology.
2.1 A hidden, historical legacy from Cassirer to Dialogism
‘Tragicomical’ is probably the best word one can use to describe the link between Cassirer and Bakhtin’s theories. Their common focus on laughter, comedy and carnival as philosophical subjects is a dominant, direct line between the two. In addition, the need for contemporary studies to uncover, or rather recover/salvage, such things as “Bakhtin’s verbatim translation of over half a page of Cassirer’s [writing]”(Poole 1998: 543) and other references of Cassirer’s line of thought in Bakhtin’s works, is a tragic example of the damning effect external and internal forces in men can have on research.
However, this tragicomic connection is not the only legacy left behind by Cassirer through Bakhtin. For me as an applied linguist the method Bakhtin adopts from Cassirer is the most pertinent: historical studies of texts and thought (ibid.:546-47: ‘historicism’). With Cassirer’s method Bakhtin turns to textual data spanning longer historical periods in order to understand the emergence of both language and thought and their symbiotic relationship. “To understand the historical dynamics in [literary-linguistic] systems and to overcome a simple (and in most cases shallow) mapping of existing styles and how they replace each other, to historically explain these changes, a review of the speech genres’ […] history is necessary, that more directly, flexibly, and more sharply reflects all changes that occur in society” (Bakhtin 1998: 7, Norwegian translation translated into English). Time is needed to understand languages evolving, and time is here understood in a loooong, historical sense, i.e. years and centuries, not hours and minutes. It is through this understanding of the emergent historical dimension of language, a.o., that one can understand a) how language come to form b) how men form c) their perceptions of ‘reality’ – ecologically. “[The language of techniques] first begins to open up and let go of its secrets when one, even here, goes from forma formata and back to forma formans, from that which is created, back to the principle of creation.” (Cassirer 2006: 91).
Cassirer and Bakhtin’s historical method has a second, important characteristic: the historical, textual data selected often exemplifies language revolutions and uses a broad conception of language (cf. Cassirer 2006, Cassirer’s variety of symbolic forms, and Bakhtin’s selection of literature). Language development, forms, and functions are investigated through illustrations of change and anomalies, not through descriptions of systematic evolution of normality in ‘natural languages’. This focus on abnormality differs from traditional linguistic studies of language development (cf. Chomsky 1965, Hopper 1987) or classical, philological studies as Humboldt (sited by Chomsky 1965). Their historical approach is revolutionary rather than stationary, focusing more on sporadic examples of ‘abnormality’ than on continuous descriptions of ‘normality’ or ‘unity’ in language.
This historical, revolutionary method goes hand in hand with the theoretical stance Bakhtin and Cassirer takes on intersubjectivity. Common in their understanding of language is a view of intersubjectivity rather than subjectivity as the driving force behind language evolution. However, their views on intersubjectivity differ: Cassirer takes a more abstract, general view of intersubjectivity while Bakhtin takes a very concrete, personified view; while Cassirer searches for a neokantian Basisphenomena, the ‘Objectivity of the symbolic form’(Cassirer quoted in Heidegger 1997: 205) grounding intersubjective processes4, Bakhtin’s descriptions seem more content with an open-ended, chaotic picture. But regardless of their differences, they both present an interactional, historical, intersubjective process driving language development, a process that is more erratic, incidental and revolutionary, than ordered, logical and steady.
In the Scandinavian dialogic tradition few studies use methods that enable a long historical perspective or revolutionary text focus. While the historical timeline and emergence in language is well described in theoretical models such as ‘diatope’(Evensen 2002) and ‘double dialogue’ (Linell, sited by ibid.), the associated methodological guidelines from Cassirer and Bakhtin are not as evident. When this theoretical and methodological synthesis is broken in empirical studies, discrepancies between theory and method might arise. First, macro-historical theoretical perspectives risk being set up against essentially micro-historical empirical data. Second, “only with the coming of ‘mnogoiazychie’, a pluri-lingual environment (polyglossia), can ‘consciousness be completely freed from the power of its own language and linguistic myths’” (Brandist 2004: 150 quoting Bakhtin 1981). Researchers need abnormality and differences between symbolic forms to create meta-perspectives on language and how it shapes their own language based perceptions. It can be argued that the contemporary Scandinavian dialogic tradition does not fully address these aspects in their choice of methods.
2.2 Dialogism inspired by the Soviet Union
However, Dialogism was not only inspired by Cassirer’s view on language development. “’Marxist understanding of the social-historical process’ [(Iakubinskii 1930-31)] constitute the main source of Bakhtin’s sociological and historical account of language in his essays of the 1930s.” (Brandist 2004:146). Partially isolated in the Soviet Union in the 1920s and 1930s Bakhtin was both inspired by and forced to relate his theories to Marxism and strongly socially-oriented views. Cassirer on the other hand, a neokantian German philosopher, discussed his theories with logically oriented philosophers heavily preoccupied with individual consciousness and subjectivity. Although both conclude that intersubjectivity drives language development, Bakhtin and Cassirer reach this conclusion from two almost diametrical theoretical arenas.
Common to Cassirer and Bakhtin’s theoretical and methodological approach on language and its development can therefore be understood to include their historical and change-oriented method and their conclusion on intersubjectivity. But the strong social and concrete starting point in Bakhtin’s theories seems political and different from Cassirer’s philosophical and abstract starting point. Bakhtin combined a socialist agenda with Cassirer’s historicism, “severed Iakubinskii’s argument from its historical moorings, and incorporated it into an account of literary history with different origins and historical coordinates” (ibid.: 150).
Another difference between the two approaches is Voloshinov, a contemporary Soviet scholar to Bakhtin, who “translated intersubjectivity into discursive forms: dialogic relations” (ibid.: 146). Giving an abstract concept of intersubjectivity concrete counterparts in various descriptions of ‘dialogic relations’, the dialogic approach thus enable researchers to more readily identify such phenomena in empirical data from text and discourse. Building on both historicism and a dialogic twist Bakhtin illustrates in several studies not only this directions value for theoretical understanding of language, but, perhaps even more so, the applicable value these ‘nuances’ made for empirical investigations.
Examples of terms and perspectives that modern dialogism cultivate from among others this socialist, Soviet origin are variations on ‘dialogue’, ‘sentrifugale and sentripetale forces of language’, ‘heteroglossia’ and ‘polyglossia’. These concepts and perspectives have their cousins in other theoretical traditions, such as for example Cassirer’s, but the socialist agenda underpinning these perspectives and the ‘dialogic’ twist they are given are historically and theoretically quite unique to Dialogism. In this way the social, dialogic starting point in this theoretical perspective differs from that of Cassirer.
2.3 A subjective presentation of Phenomenology and its time
“In the Logical Investigations [Husserl] conceives of phenomenology as descriptive psychology [and] analyzes the intentional structures of mental acts and how they are directed at both real and ideal objects” (Wikipedia 2007). Logic, psychology and analysis of intentions and mental actions constructed a platform for a philosophical tradition and a forum in which Cassirer participated, but that Bakhtin only observed. This difference is not only a historical, geo-political curiosity. Contrary to the socialist starting point of Bakhtin, the phenomenological platform started with the individual: “phenomenology studies conscious experience as experienced from the subjective or first person point of view” (Stanford 2007).
Both Dialogism and Phenomenology room both the subjective and the intersubjective. This article does not argue otherwise. However, historically the two traditions’ start off from opposite directions along this axis. This difference makes their approaches and preferences to subjectivity and inter-subjectivity differ. This article investigates the consequences of this diametrical methodological and theoretical starting point when applied to empirical phenomena. How do a) their different approach to subjectivity and inter-subjectivity shape b) how dialogism and phenomenology shape c) our perception of empirical phenomena such as language and its use? To illustrate how different approaches illuminate empirical phenomena differently, I will begin by illustrating how different starting points can color terms and perspectives.
‘Time’ is once more an appropriate example. Heidegger, a participant in the phenomenological discussion, “believes that we […] can only understand the phenomenon of time from our mortal or finite vantage point” (Alweiss 2002: 117). Taking a subjective, rather than inter-subjective approach, “time ‘is’ only for a being that lives with an awareness of its own mortality” (ibid.: 118). This perception of time was based on the necessity of using our death as individuals, rather than eternity, as an understandable point of departure against which to understand time.
‘Time’ in this particular phenomenological interpretation, is thus tightly linked with the ‘individual subjective’, both concepts informing the other. However, an individual, subjective experience of death (and thus time) is no more within our human grasp than eternity: “death comes from outside and transforms us into the outside” (Satre 1969, quoted by Alweiss 2002). This critique of time understood from individual death refers to the scope and boundaries of a subjective approach.
An inter-subjective approach to a similar timeline is however not open for the same criticism. Even though we may never experience the phenomenon of our own death, we can perceive the death of others: we can know and feel quite tangibly what another’s death means for me and for us (ibid.: 127). Individual death in general (as abstracted from that of others) and our own death (as a reflection of others) become meaningful within this approach. Having a reference for ‘death(s)’ thus gives a reference for a similar timeline -the only difference being the starting point on the subjective/inter-subjective axis.
Viewing time from an intersubjective stance highlights a different aspect of time: it’s plurality. Using a metaphor from computer science, our ‘environment’ can be understood as ‘multithreaded’, not only ‘single-threaded’. Several independent timelines and flow of control run concurrently, parallel, and across each others’ paths. In Dialogism this plurality of time is pivotal for its view of language as a chaotic sea of utterances and voices, each with it’s own time and direction, overlapping, interacting, and flowing in and out of each other.
A subjective, individual approach might also eventually in their argument support a multithreaded view of time and a view that such timelines cross, overlap and interact with each other. But, time’s multiplicity and heterogeneity would not be as glaringly apparent from an individual, subjective stance as it is from an intersubjective approach. The methodic starting point establishes a conceptual priority that consequently shapes our theoretical perceptions.
Therefore, understanding the priorities of different theories is especially important in empirical studies of for instance language. In such studies only the tip of the iceberg of theoretical conclusions makes it to the surface in the sea of empirical data, the rest being weighted down by the priority of other conclusions. But, crucially, this knowledge is not only important in order to understand empirical phenomena. Flipping the coin, the reliefs synthesized of theory and empirical data can also be read as a topographical map of theories priorities. Hence, both empirical phenomena and theories illustrate each other.
2.4 The semantics in dialogism – dialogic semantics
Focusing explicitly on the social character of language, dialogism has developed several analytical concepts to illuminate the intricate, reciprocate relationships between language, our thinking, interaction and culture. A central concept within these theories is the social dimension of utterances, texts and words (Bakhtin 1986). "[Language] is populated – overpopulated – with the intentions of others "(Bakhtin 1981: 294), and when we perceive and express reality with words in any language, we use the language of others in order to compose our own perspective on reality. Thus, in dialogic studies (Evensen 2002, Bostad 2005), our thoughts, actions and ontologies are composed as ‘polyphonies’ (Bakhtin 1984) or ‘refractions’ (Bakhtin 1981: 300, cf.276) of other's voices.
When viewing language and our perception of the world primarily as social references, 'semantics' is seen in a new light. Rather than categorizing semantic relationships between say a word and a physical world or a universal logic (figure left), the meaningful references or ‘semantic relationships’ in words become their relationships to other social actors and their texts, i.e. their 'inter-textual' role (figure right). The meaning ascribed to words are not viewed primarily in categories such as ‘information’ or ‘indices’ to domain-specific logics or lexica; the meaning-potential of words are viewed rather as social ques or voices of others.
Figures: The two simplistic models try to present the difference in focus and taxonomy of the two research approaches; the models do not present the approaches in themselves. Left side illustrates simplistic models of how a subjective, individual approach exemplifies index and logic oriented semantics. Right side illustrates how an instersubjective approach can exemplify social, inter-textual semantics. The arrows indicate the meaningful relationships that the different semantics focus on. Viewing language and words neither as individual property nor asocial realities, social forces become a primary reality in language. Social structures are seen not only side by side with language, but also as within language itself. Core aspects of language are viewed as sedimented patterns of interaction. Several separate linguistic traditions pick up on this point: Socio-constructionistic genre theory understands genres as ‘recurrent rhetorical actions’ (Miller 1984, Berkenkotter 1993). Other research on ‘grammaticalization’ tries to describe how “yesterday’s pragmatics is today’s syntax” (Givon 1979).
In order to understand the make-up of languages, it therefore becomes essential to understand the social interaction and relationships that languages are built up alongside. Interaction among language users, patterns of recurring rhetorical actions, practices, and other social dynamics all become dimensions potentially reflected within language. And since these languages are the ones we in turn use to form our individual utterances and perspectives on the world, the spiral is completed: social patterns of interaction shape language that shape individual actions and thoughts that in turn shape social interaction.
Turning to programming languages, a dialogic approach and social semantic taxonomy can seem irrelevant if programming is deemed solely utilitarian and asocial. Social dogma and stigma greatly encourages such a view on programming languages, and if a subjective approach and a conventional semantic orientation is taken towards this particular, empirical domain, it is more than likely that these views will only be enforced. However, though quite common, an asocial perception of programming languages and their use is not the only meaningful take on the subject. Programming can also be understood as highly social (cf. the open source movement). Software is an inter-text, and the words of programmers reflect not only logical functionality, but also echo the voices of other programmers. The more complex and socially entangled these inter-textual programs become, the more important it becomes to understand this social dimension of programming languages.
3. Empirical analysis
3.1 Scope of analysis
Morphology in object-oriented programming is a complex phenomenon, and the empirical scope of this study must therefore be specified and constrained.
First, the kind of words to be studied needs to be constrained. In object-oriented programming there are several ways to compose and use words. First and at the core is a set of ‘reserved words and operators’ defined and made significant by the compiler, interpreter and/or platform. These words constitute a lexical skeleton in all object-oriented languages. Second and fleshing out a body around this skeleton is a set of words that we will here describe as ‘types’. In this study we use a broad definition of types to mean all kinds of words that can be defined and referenced by any user alike. Third and within the reach of this body of words are other similar mechanisms to describe and refer to functionality. Relevant examples of such mechanisms are remote objects/methods (as in CORBA, RPC and RMI) and web services. In this study we will only focus on the second category: ‘types’. The reason for excluding the first category, the lexical core, is that these words are rarely, and if so very limitedly, defined by ordinary users. The reason for excluding the third category, similar mechanisms, is that these methods primarily use syntactical, not morphological forms to express references to other programmers’ texts.
Secondly, there are several nuances in the way types can be defined and referred: In Python types can be structured as both methods and classes, while in Java all types must be defined as classes; import and package statements can shorten type-names; object variables at different levels can function as ‘proforms’ or morphological placeholders; inner classes are stored differently than normal classes; etc. etc. However, all these nuances adhere to the same overall morphological form. We will therefore not address all such details, but only the general, overall structure of type references.
Third, this study approaches type morphology with a dialogic perspective and a social semantic taxonomy. Therefore, this study will primarily focus on the socially oriented actions and practices that directly surround the morphological forms, and secondarily on the symbolic morphological forms as such. This approach differs substantially from traditional linguistic descriptions such as BNF and traditional linguistics that derive their descriptions of morphology primarily from the symbolic forms alone.
With this scope, this article analyses two different morphological forms in object-oriented programming: ‘pc-based morphology’ in Java and Python and ‘web-based morphology’ in an experimental programming language theJ.
3.2 The path to the words of others: describing pc-based morphology
When writing a program in Java or Python, a programmer uses words that other programmers have defined as underlying resources in his own text. As mentioned earlier, these words can be operators and ‘reserved words’ defined by the programmers who made the compiler, interpreter and/or platform (figures: grey), or types defined by any other programmer (figures: italic). These words are fundamental and indispensable for all meaningful program texts.
Type names (such as org.jdom.Element in HelloJDOM and xml.dom.minidom in minidomPrinter) are always path-relative in Java and Python. Path-relative means that all type references depend on the context of a path delineating a certain group of texts, and type names must be interpreted within this inter-textual context. The act of wording type references is therefore not only to a) explicitly formulate a type name consistent with a morphological pattern. Implied in this act are also the acts of b) obtaining a resource (for instance downloading a .class- or .py-file) in which another programmer has defined the type and c) updating the path settings (i.e. CLASSPATH or PYTHONPATH) to associate words with resources. PC-based morphology can thus be seen as a compound act: a) write a word, b) obtain the word text, and c) update the path on your PC.
This individual, compound act of using other programmers’ words is not limited to a single step. A programmer can use another programmers’ word that in turn uses yet other programmers’ words, etc. etc. The morphological system thus creates relationships between programmers that span several individual words and actions.
Figure: Words are interpreted in an inter-textual context.
In principle, and contrary to popular belief, words such as types in a program text are therefore not specific to one particular interpretation: program words should not be mistaken for fixed indices to predetermined, universal functionality. The meaning of a word such as a type name must be determined within a path, and setting up different paths around the same program text thus allow for different interpretation of the same words. Program words rely on each other in order to function. Hence, program words can be viewed as inter-reliant voices that need each other in order to make sense and that echo across inter-textual webs of words. Only when situated in inter-textual environments enabling concrete interpretation of words do the texts become programs and start (to make sense).
Figures: Undetermined meaning potential in program words.
3.3 Social interaction and dynamics of pc-based morphology
The value and usefulness of using others’ words cannot be overstated. To be able to quickly and simply instantiate objects and invoke methods that other programmers have spent countless of hours developing is as enabling in programming as the concept of a round leather ball is in soccer. Also, to be able to delineate a space in which these words can be used with a specific path setting does as much for programmers as the concept of an enclosed pitch with two goals does for soccer players.
The reason, as might be guessed from the soccer analogy, is that this morphological act enables positive and competitive interaction among programmers. By rounding up programmable thoughts in types easily referable as words, a programmer can build his game with the input of other programmers working as his team mates. Switching to the social semantic taxonomy of dialogism, a programmer can construct his perspective on the world reusing ‘fractions’ of other programmers’ code. The previous examples HelloJDOM and minidomPrinter can thus be seen as textual, programmed compositions or ‘polyphonies’, where meaning is not just an individually formed idea, but a product or mosaic of social collaboration in a cultural sphere.
However, not all that glitters is gold. Several social, technical, ethical and practical concerns are associated with pc-based type morphology.
First, a well known ethical concern relates to ‘licensing terms’. When a programmer chooses to use another programmer’s word, he also presupposes that a text defining the word must be added to the path. A programmer doing act a) also implies doing b) and c). But, linked to these underlying texts are ethical, legal, commercial, political and cultural terms and licenses. Using another programmer’s word therefore presupposes accepting or breaking the terms of both the words directly used and all the words that these words iteratively use. “[Words are] populated with the intentions of others” (Bakhtin 1981), and in programming these intentions are long chains of both technical functionality and social, commercial and cultural terms and agendas.
As in soccer, the social game in programming is played to win. Using the words of others means in programming, as in for example academia, to accept the passes from certain players and choose sides. The programmer behind HelloJDOM and minidomPrinter has in effect chosen to side with the teams at ‘java.sun.com’, ‘jdom.org’ and ‘python.org’. These teams have their own agenda and terms associated with their words, and using these teams’ words involves to some extent accepting and pushing their agenda.
A second well known practical, technical and social concern relates to chained dependencies between words. Building on the minidomPrinter example, a new program text called printUseris set up to reference the minidomPrinter. This program text uses words to directly refer to a printDocumentfunction from minidomPrinter.
A Python module called printUser.
In order to interpret the words in printUser, we need to see it in an inter-textual context. In printUser printDocument is another programmed utterance. In order to interpret printUser, an inter-textual context must be set up in which printDocumentmeans some thing (i.e. text): b) obtain a file matching this description (the above minidomPrinter is a possible, but not compulsory alternative) and c) update the PYTHONPATH settings to include this file. If minidomPrinter described above is used, it references xml.dom.minidom. Therefore, if this interpretation of minidomPrinter is used, the inter-textual context of printUser will now also imply xml.dom.minidom. Thus a path with both a working definition of minidomPrinter and xml.dom.minidom must be set up. This contextual relationship from printUser, minidomPrinter to xml.dom.minidom constitutes a dependency chain.
Figure: Illustration of chain of dependencies around HelloJDOM.
As described in the above example a programmer using another programmer’s words will have to weigh the benefits of using the word against the cost of obtaining files and setting up a path needed to realize a certain interpretation of these words. If using a certain word demands obtaining dozens of files from different locations in separate steps, odds are that this word will not be used as often as other alternatives requiring less work. In addition to these practical concerns, programmers place a certain amount of trust in each other when they use each others’ words and types. Several technical and practical barriers stop programmers from knowing exactly what happens when the flow of control passes into an externally defined word in their programs. This lack of insight applies even more so when flow of control iteratively are passed to words within words. If the programmer does not trust other words with this control, he probably will not use them. Chains of dependencies thus also entail trust issues.
PC-based morphology enable the programmer to only perform act a) for only the first step, but in order to maintain control over the inter-textual context in which his program should run, PC-based morphology specifies that a programmer must perform acts b) and c) for all iterative steps in the dependency chain. A programmer may therefore get an infinite, illegal or broken path thrown into the bargain when using a simple word. His choice of words is therefore not only concerned with the words technical qualities, but also their ensuing practical tasks, ‘licensing terms’, his trust in them, and other social constraints.
3.4 Second and third party words in pc-based morphology
Because of these practical, social and ethical concerns, Java and Python platforms are distributed with a standard library of words and standard path settings. Distributing the language in this manner establishes in practice a ‘standard inter-textual context’. A programmer can anticipate that words from these libraries such as xml.dom.minidom and java.lang.Exception are always within the inter-textual contexts, while words such as org.jdom.Element and minidomPrinter are not. This segregates words within and outside the standard library into second and third party words respectively: when referring to second party words (a2), a programmer can presume that actions (b_ and c_) will not be needed; when referring to third party words (a3), a programmer can presume that actions to establish an inter-textual setting (b3n and c3n) are needed, and in all contexts were his programs/words are used.
These practical and social concerns play an important role in shaping program literature. A programmer using another programmer’s words orients not only towards the current program text and its relationship to prior program texts or types. The programmer of HelloJDOM implicitly makes the choice to perform acts (b3n and c3n) surrounding org.jdom.Element not only for himself and his own program, but also for any other user or programmer who wants to use HelloJDOM as an application or type. On the other side, the programmer of minidomPrinter chooses words such as xml.dom.minidom that with reasonable certainty can be assumed to be part of later users ‘standard inter-textual context’. Therefore, minidomPrinter does not place the extended burden on future users of this word that HelloJDOM does.
This future burden of using a program text is very important in programming. Thinking ahead programmers can foresee other program texts establishing links to their program texts. Because of this ‘responsibility’ to other texts potentially referring to ‘me’ in the future, ‘addressing future utterances’ is therefore as vivid in some program texts as programmer’s awareness of already existing program texts.
Because of these social, practical and ethical concerns related to path-relativity of program texts that 1) have been written, 2) are being written, and 3) might be written, a ‘centripetal force’(Bakhtin 1981: 270) in languages using pc-based morphology emerges. Because the use of second party words (a2) has so many clear advantages over technically similar third party alternative words (a3), both for present and future users, programmers flock towards the use of second party words. The social, practical and interactional structure of path-relativity combined with the distributors’ ability to set up a default inter-textual context thus gives a clear home-court advantage for the social groups uttering second party words over technically equal or perhaps even better alternatives from third parties.5
3.5 TheJ experiment and URL-types: Actions shaping morphology
Our informal experiment started with the development of a platform for programming multi agent systems in Java (Ørstavik 2005, 2006). One of the core issues we faced in this work was the need for different agent programs to interact with each other using the Java language as medium, while at the same time maintaining their ability to function autonomously in individual inter-textual contexts.
Due to the make up of the Java language morphology a problem quickly arose in our project. When one agent wants another agent to respond to a request, they need to involve words/types that are part of one agent’s inter-textual context, but not the other’s. There are many known mechanisms to work around this problem: type definitions can be passed directly between agents, or one agent can tell another agent to alter its path settings. However, such mechanisms provide alternatives to type morphology that make program texts more difficult and heavy-handed to read and write. Further, the problem of extra-contextual words/types in multi agent systems is in principle the same practical, social and technical problem concerned with PC-based morphology described earlier.
To tackle this problem, we therefore started looking more closely and creatively at type morphology. We found that the morphological forms quite easily could be changed so to enable us to reference extra-contextual types and tackle chained dependencies between such types. Based on the Java language we developed an experimental dialect called ‘theJ’ (Ørstavik 2006) that a.o. extended the PC-based morphology to also allow for types to be referred to as URLs (URL-types), an extension of the existing morphological forms.
A theJ class called thej.ExampleTwo.
Figure: To run this theJ program,
1) download http://the.hist.no/CoreClasses/agentcore.jar,
2) run java -jar …/agentcore.jar http://the.hist.no/ thej.ExampleTwo http://ntnu.no, and
3) hope that the system is up and running. The implementation of URL-types (Løkke 2005) extends both the front- and back-end of the Java type system: front-end type names can be written echoing complete URLs, for instance as http://www.jdom.org/jdom.jar/org.jdom.Element, and back-end the compiler and platform is extended so as to locate, download and utilize program files found using the URL-information in the URL-type name from the web.
The principal consequence of this change in morphology is primarily linked to the perception of inter-textual context and actions to control path (b and c). The path is no longer pc-oriented; the path is web-oriented echoing a protocol that describes how programmed words can be exchange both within and across different computers. The underlying compiler, interpreter and/or platform can thereby be told to relate to words on the web, using the web to automate the tasks b) and c), even in chained dependencies. Using a symbolic, morphological form to move actions into morphology as described is making “today’s pragmatics tomorrow’s syntax” (Givon 1979).
However, this move leads to a series of primarily social, security, and trust issues when the ‘pc-based’, centralized control over programmers’ inter-textual context so easily can be altered. Possible solutions to these security problems exist, but have not been implemented. This experiment is however not about producing a new programming language: the primary goal of this study is to learn more about the dialectics between human minds and languages.
By setting up a contrasting example this experiment illustrates both 1) how existing morphology tacitly may have guided our minds and perceptions and 2) how we can think of program words in new ways.6 This contrasting image creates a ‘hindsight’ view of current pc-based morphology, a perspective that is very difficult to achieve only having experienced existing programming language morphology. Exploring new possibilities highlight limits in existing technical framework and mindsets. Exploring alternative, exotic linguistic forms made me aware that I was continuously searching for second party words, even when I already knew better suited third party alternatives.
4.1 How theory has informed our empirical understanding
Applying this dialogic perspective to programming language should not be read as an attempt to capture a somehow ‘truer’ description of the world of programming. It is not. Using a theoretical perspective is a strategic choice taken in order to shed light on certain aspects or dimensions of phenomena that other theoretical approaches might miss. In this study the ‘new’ phenomena captured are how certain morphological forms can cluster or disperse certain patterns of interaction, and how this effect can be seen as crucial for how languages are shaped and made to use.
The ‘new’ phenomena captured can be described as 1) a hindsight view of current practices and 2) an insight into what might come. The first, our hindsight view has already been extensively described in the analysis chapter and will not be pursued further. However, a few words can still be said about the second.
The difference between pc-based morphology and web-based morphology is both technically and conceptually similar to the difference between pc-based and web-based file systems for ordinary text documents. This study shows how the centrifugal power unleashed with the advent of the Internet in the last decade has yet to be fully unleashed in programming languages. This study shows that incorporating URL into programming languages is simple to implement. This study also shows that the concept of URLs in program texts is simple to grasp for individuals. However, it remains to be seen if the potential centrifugal powers unleashed with such language use can be balanced properly in social interaction, especially with regards to issues of trust.
Having tested URL-types in action, what is most striking is the realization of the still untapped potential of ‘third party programming’ (i.e. the strength of the centrifugal power). The new morphological way of wording and seeing types encourages us to focus on the use of third party words, to widen our vocabulary and to extend our repertoire of voices. This has directed our research against creating tools and services to support this form of interaction among programmers, directing our attention to wikis for writing program text directly online, web-based compilers and search engines for programming texts. Such resources are necessary to fully support web-based morphology, but since this work has just begun, our perception of the world with web-based morphology is still misty.
4.3 How empirical study has informed theoretical understanding
The Marxist social agenda and theoretical view underlying the dialogic tradition of the 1920-1930’s Soviet Union and Bakhtin’s theory of language can clearly be seen as reappearing in this study. This was not the original intent of the author and not fully understood until very late in this study, as can be seen in an earlier version of this study (Ørstavik 2006 ECAP).
The social agenda finds a clear philosophical ally in the open source movement. Both the dialogic perspective and the open source movement advocate collective ownership and openness over corresponding individual rights. However, an interesting finding in this study is that ‘open source languages’ such as the liberal Python on Ubuntu employ the same centripetal forces to almost the same extent as commercially owned languages. This illustrates that power is not only accumulated for monetary ends.
The social agenda in the dialogic tradition, an agenda that neither Cassirer nor phenomenology theories use as their starting point, thus shapes my perception of programming languages and their morphology in a way that I do not believe a more subjective study would do. Assuming that a phenomenological or Casserian study would stand out similarly against similar empirical backgrounds would be to underestimate the consequences of the historical and theoretical social root in Dialogism. Thus, both theoretical differences and the findings of this particular empirical study suggest that Dialogism should be understood as more than a subset of phenomenology.
Secondly, the material underlying this study spans several years and spins around language forms at the border of established conventions. We believe that it is unlikely that a study employing a method within a single language and within established linguistic normality would illustrate the historical dynamics in this program-linguistic system in society as directly, flexibly, and sharply.
Alweiss L. (2002). Heidegger and ‘the concept of time’. History of the human sciences, vol.15, no.3, 117-132 (London: Sage Publications).
Bakhtin, M. M. (1981). Discourse in the novel. (In Emerson C. and Holquist M. (trans.): The Dialogic Imagination: Four Essays. Austin: Texas University Press.)
Bakhtin M. M. (1986). The problem of speech genres. (In McGee V. (trans.): Speech Genres and Other Late Essays. Austin, TX: University of Texas Press.)
Bakhtin M. M. (1984). Problems of Dostoevsky’s Poetics. Emerson C. (trans.). Minneapolis: University of Minnesota Press.
Bakhtin M. M. (1998). Spørsmålet om talegenrane. Slaattelid R.T. (trans.). Norway: Ariadne Forlag.
Berkenkotter C., Huckin T. N. (1993). Rethinking Genre from a Sociocognitive Perspective. Written Communication, 10, 475-509.
Bostad F. et. al. (Ed.) (2004). Bakhtinian Perspectives on Language and Culture: Meaning in Language, Art and New Media. (New York: Palgrave MacMillan)
Brandist, C. (2004) Mikhail Bakhtin and early Soviet sociolinguistics. In: Proceedings of the Eleventh International Bakhtin Conference. Universidade Federal do Paraná, Curitiba (Brazil), pp. 145-153. ISBN 859043981X. http://eprints.whiterose.ac.uk/archive/00001115/
Cassirer E. (2006). Form og Teknikk. (In Folkvord I. (trans.), Hoel A.S.: Form og Teknikk: Utvalgte tekster, 87-142 (Oslo: Cappelen Akademisk Forlag)
Chomsky N. (1965). Aspects of the Theory of Syntax. MIT Press. ISBN 0-262-53007-4.
Crowley T. (1990). That obscure object of desire: a science of language. (In Joseph J. E. & Taylor T. J. (eds.): Ideologies of Language, 27-50. London: Routledge.)
L. S. Evensen (2002). Convention from below: Negotiating interaction and culture in argumentative writing. Written Communication 19(3), 382-413.
Givon (1979). On grammar. (Sited by Faarlund J. T. (2002). Grammatikalisering av ordstilling. (In Moen I. et.al.(ed.) (2002). MONS 9 - Utvalgte artikler fra Det niende møtet om norsk språk i Oslo 2001. Oslo: Universitetet i Oslo)).
Gosling J., Bracha G., Joy B., Steele G. L. (2000). The Java Language Specification: Second edition. (Boston: Addison-Wesley)
Heidegger M. (1997). Appendix IV: Davos disputation between Ernst Cassirer and Martin Heidegger. (In Heidegger M., Taft R. (trans.): Kant and the problem of metaphysics, 5.ed., Bloomington: Indiana University Press)
Lindholm T., Yellin F. (1999). The Java(TM) Virtual Machine Specification (2nd Edition) (Paperback) (Boston: Addison-Wesley Longman Publishing Co.)
Linell P. (1998). Approaching dialogue: Talk, interaction and contexts in dialogical perspectives. (John Benjamins Pub Co.)
Løkke J. (2005). Implementing URL-types in Java. (Trondheim: HiST)
Miller C. (1984). Genre as Social Actions. (In Freedman A. and Medway P. (ed.) (1994). Genre and the new rhetoric. London: Taylor & Francis.)
Poole B. (1998). Bakhtin and Cassirer: The Philosophical Oirigins of Bakhtin’s Carnival Massianism. South Atlantic Quarterly 97:3/4 (Duke University Press).
Python (2006). The Python Programming Language. Retrieved June 20, 2006 from http://www.python.org/.
Saussure F. (1916). Cours de Linguistique Générale, Payout, France. (Eng. ed.: The Philosophical Library, 1959; McGraw-Hill, 1966.)
Ubuntu (2006). Ubuntu. Retrieved June 20, 2006 from http://www.ubuntu.com/.
Wikipedia (2007). Phenomenology. Wikipedia, The Free Encyclopedia. Retrieved March 1, 2007 from en.wikipedia.org/wiki/Phenomenology
Stanford (2007). Phenomenology. Stanford Encyclopedia of Philosophy. Retrieved March 1, 2007 from plato.stanford.edu/entries/phenomenology
Ørstavik I. T. B. (2005). Designing multi agent systems using Java as an Inter-Program language. (Paper presented at JavaZone 2005: http://www3.java.no/JavaZone/2005/presentasjoner/IvarOrstavik/html/Ivar_Orstavikjavazone2005.html)
Ørstavik I. T. B. (ed.) (2006). Site of theJ programming language. Retrieved June 1, 2006, from http://the.hist.no.
1 In this article Dialogism refers to a current, applied Scandinavian school of thought (Evensen 2002, Linnel 1998). This tradition takes much of its theoretical foundation from Bakhtin’s work.
2 In this article, ’language’ refers to ‘language, language use and communication’.
3 The article will begin with a description of a blurred connection between Cassirer and Bakhtin. This connection highlights the similarities between the Scandinavian dialogic tradition (represented by Bakhtin) and Cassirer, a philosopher linked to phenomenology. Second, this article will describe a sociolinguistic influence on Bakhtin that differs from Cassirer’s approach. This will illustrate a distance on a social axis between the two theoretical strands. Third, this article will try to place Cassirer and phenomenology along this same axis. This discussion will help illustrate subtle differences between the three theoretical viewpoints, a difference having more to do with research agenda and methodological starting point more than theoretical conclusions.
4 “Each of us speaks his own language, and it is unthinkable that the language of one of us is carried over into the language of the other. And yet, we understand ourselves through the medium of language. Hence, there is something like the language. And hence there is something like a unity which is higher than the infinitude of the various ways of speaking. Therein lies what for me is the decisive point. And it is for that reason that I start from the Objectivity of the symbolic form, because here the inconceivable has been done. Language is the clearest example. […] And I believe here that there is no other way from Dasein to Dasein than through this world of forms.” (Cassirer quoted in Heidegger 1997: 205).
5 On Ubuntu (2006) there is a platform tool to ease the practical task of obtaining files and updating path settings: apt. With tools such as aptitude, Ubuntu offers everyone the ability to semi-automate the process of establishing new inter-textual environments for both Java and Python programs via the Internet (b and c). However, these tools are also shipped with a standard setting, mainly due to concerns of trust. Therefore, even though apt makes several steps as to simplify the practical tasks of b) and c), the process is only partially automated and still specifies a ‘standard inter-textual context’. Ubuntu does therefore not alter the pc-based morphology in its programming languages, but mainly complements it with a web-based utility.
6 Since this experimental language has not yet accumulated a big body of interaction and practice, empirical data about interaction with URL-type is fairly limited. The data is constricted mainly to experiences of a couple of programmers related to the project and the subjective interpretations of our ‘old’ pc-based practices viewed in the light of a new morphological concept.