From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

1. Text-immanent vs. Contextual and Theses vs. No-theses Approaches

Methodological issues in the interpretation of Wittgenstein’s Philosophical Investigations (PI) include dichotomies between text-immanent and contextual and theses and no-theses approaches. Baker and Hacker’s Analytical commentary refers continuously to Wittgenstein’s Nachlass and thus involves a contextual approach, while von Savigny’s Kommentar für Leser consciously adopts the method of strict text-immanent interpretation. While von Savigny’s argument for his text-immanent approach is not impeccable, prevalent defences of contextual approaches are equally problematic. The dichotomy between theses- and no-theses-approaches concerns PI’s relation to theses. While one group of scholars, though on different levels and in varying degrees, sees the PI as a work that advances philosophical theses and also theories, another sees it as a work that rather dissolves philosophical theses and theories through a number of “therapies”, which are not systematically interrelated. Representatives of the first group would also claim that the PI propose philosophical therapies; however, they regard them as connected in a system, and even their most minimalist interpretation understands the PI as developing and defending theses, e.g. about grammar and linguistic rules. In recent years, this position was attacked by the late Gordon Baker, who says that in the PI “[n]o fact (even one about ‘our grammar’) is stated, no thesis advanced” (Baker 2004:45).¹

Interestingly, Baker used earlier to hold the opposite position and argued that Wittgenstein’s collaboration with Waismann on a systematic presentation of his philosophy shows that Wittgenstein, incl. his PI, was positive about systematic philosophising, and that therefore “it cannot be mistaken in principle to try to fit Wittgenstein’s remarks together into a philosophical system” (Baker 1979:245). The early Baker's reasoning seems to rest on certain false methodological assumptions. First, the justification for “fitting” an author’s work “into a philosophical system” is not dependent on whether this undertaking is in agreement with the author’s intention. Second, the fact that Wittgenstein may have been positive about systematic philosophising at a certain point does not entail that this is also valid for the Wittgenstein of the PI. Now, where the later Baker argues contextually for the contrary therapeutic reading, this is, for the same methodological reasons, equally problematic. Consequently, notwithstanding the great significance, which the later Baker's work has for understanding the PI's therapeutic nature, his justifications are nevertheless often contextually fallacious and therefore as problematic as were his earlier for the opposite position. Contextual fallacies deceive both defenders and opponents of the no-theses approach, when they argue their respective and opposed approaches with reference to earlier or later statements or activities by Wittgenstein. This is partly due to the fact that both defenders and opponents of the theses-approach often adopt an uncritical attitude to the issue of the special place the PI occupies, or at least, might occupy, in the context of Wittgenstein’s work, extending views valid for certain parts to the whole “later Wittgenstein”, including the PI, or views valid for the PI, to the whole "later Wittgenstein".

Looking at PI interpretation in general, the four approaches – text-immanent, contextual, “theses”, and “no-theses” approach – combine to four principal readings: (i) a text-immanent “theses” reading; (ii) a contextual “theses” reading; (iii) a text-immanent therapeutic reading; (iv) a contextual therapeutic reading. Thus, the adoption of a text-immanent vs. a contextual approach seems to be neutral to the choice of a “theses” vs. a “therapeutic” approach. While both von Savigny and Cavell read the PI text-immanently, the former reads it in the “theses” way, while the latter reads it in the therapeutic way. While both the early and the later Baker read the PI contextually, the former reads it in the “theses” way, while the latter reads it therapeutically. While both von Savigny and the early Baker read the PI in the “theses” way, the former reads it text-immanently, the latter contextually. Thus, it seems that to read the PI therapeutically or as advancing theses depends as much on text-external factors, such as one’s conception of and expectations to philosophy, as on the PI text itself.

2. Elements of an Argument for a Therapeutic Reading of the PI

Each of the mentioned four types of principal readings of the PI entails a different set of answers to a group of questions, including whether it is sufficient or necessary to rely on the text alone, or whether it is necessary to include contextual elements in order to pick the correct interpretation; whether it is necessary for philosophy to advance theses and theories; whether text-immanent approaches yield “therapeutic” or theses-readings, and, whether contextual approaches yield “therapeutic” or theses-readings. Arguments against strict text-immanentism or strict contextualism will be able to point out that the questions and answers of both partly rest on (in fact, shared) fallacious metaphysical assumptions about what a text is and what text-understanding is. Arguments against theses-readings will be able to appeal to the fact that the truth of the position that it is necessary for philosophy and philosophical interpretation to advance theses and theories is not established, but rather debated, and that text-immanent theses-readings of the PI are yielded by this position's premises rather than by a text-immanent reading as such. (It should, however, not need mentioning that it is not this particular position which is under scrutiny here, but the view, that the PI encourage this position.)

In the following, the outline of an argument is presented, which defends the view that a therapeutic reading of the PI is proper, and which proceeds text-immanently in the first place. The argument consists of two steps, where first evidence is put forward for the view that a text-immanent reading yields a therapeutic reading, and where it subsequently is established that also contextual data support the therapeutic reading. What this therapeutic reading amounts to in detail and in what sense it informs the discussion of prominent philosophical issues, cannot, however, be elaborated here further.

A crucial issue in the argument concerns the role of certain obstacles, difficulties - both of structural nature and at specific points in the text - which one meets in the PI (I would like to call them “skandaloi”, Greek for “stumbling blocks”). A close text-immanent reading of the PI cannot ignore frequent ambiguities in phrasing and, more generally, a prevalent lack of precise meaning and a dominant presence of incoherence and also inconsistencies in the text. These obstacles are largely addressed by von Savigny in his commentary and yield the three questions which he considers crucial from a methodological perspective: "Wer spricht?", "Worum geht es?", "Wo gehört das hin?". Interpretational responses to these obstacles generally adopt one of the following approaches: (A) The obstacles result from Wittgenstein’s “incapability” to produce a linear and systematic text; this is thought to be supported by the PI Preface, where the author describes his problems in bringing his thoughts in a linear way together into a "whole" (e.g. Hilmy 1987). It is argued that Wittgenstein himself, however, wished to achieve a systematic presentation of his theses, and consequently, it is in the interest of Wittgenstein that we supplement what Wittgenstein himself was unable to achieve by providing the text of the PI with an explication of its inherent (but stylistically contaminated) argumentative and linear structure, thus rendering the thesis content visible and systematic. (B) An alternative approach, however, is to regard the obstacles not as accidental mistakes and deficiencies, which Wittgenstein himself would have wanted to avoid, but to try to go along with the text as it stands, without attempting to unify its content into a linear argument – an approach exemplified early on by Cavell 1962. (B) can be applied and carried out on a purely text-immanent basis. However, defenders of A often invoke contextual confirmation, and therefore, anticipating contextual support, it can be pointed out already here that there actually is strong counterevidence against (A) stemming from extra-textual sources. Thus, insofar as (A) rests on the argument that Wittgenstein was unable to produce a linear and systematic text and therefore produced the PI in their peculiar form, (A) and the theses-reading rest on false premises. Clearly, Wittgenstein has shown in a number of cases that he was indeed quite able to present and develop theses in a systematic, linear and argumentatively well-connected manner.

The therapeutic reading can, first of all, claim that it is more adequate to the text to proceed on its premises and to accept its obstacles-dimension, without excluding in advance the possibility of the obstacles having an important function, than to alter it in favour of a smoothed coherent reading which is required by the theses-approach. There are two other features, which are easily underplayed by “theses” readings, but taken seriously by therapeutic readers who are willing to see the form of the PI as PI internal; these are PI's strong dialogical structure and their metaphorical language. In addition, the defender of the therapeutic reading will surely also be able to better utilize specific points made in certain individual remarks (e.g. PI §18), or in larger portions of text (such as the passages collectively referred to, justly or unjustly - see Savigny 1991, as the “philosophy chapter”) and, last but not least, the paratexts, which include the motto, the title and the preface.

Reference will be made below to contextual data that support the view that the PI's special form, including their non-linear and non-systematic character, are not the result of a personal shortcoming, but rather the vehicle for the content and method which the PI sought to carry. However, if the argument is to be established primarily on a text-immanent basis, then for this purpose, only text-immanent data can be admitted. But in order to validate the claim that a text-immanent reading yields a therapeutic reading we nevertheless need contextual corroboration of a special type. For it is possible, that our PI reading is based on a text that is not authentic and authoritative. Indeed, it has sometimes been questioned whether the paratexts, which are an important element of the argument for the therapeutic reading, are part of the PI that Wittgenstein intended, or whether they were put there by the editors without authorisation. Contextual data clearly show that the PI paratexts are authentic parts of the PI, and, more generally that the PI edition of 1953 and its follow-ups are in fact fairly trustworthy editions. Consequently, if it is true that the PI elements described here, incl. the paratexts, support a therapeutic reading, then such a reading cannot be weakened or falsified by the claim that it is based on a PI edition which is misrepresentative of its source.

Before we can deal with the contextual support, in order to corroborate further the text-immanent therapeutic reading, we must first distinguish two kinds thereof. One can for example refer to other contexts and invoke passages, which support the view that Wittgenstein and the PI have a therapeutic program. Or, sections of the PI can be interpreted in the light of their earlier versions, as for example thus: “The ambiguity of section x is clarified if we interpret x in the light of remark y, which is a precursor of x.” This kind of methodology is often used in Wittgenstein scholarship; however, it can be fallacious, as pointed out above. The contextual support, which is in our focus here, is of the following kind: We should refer to the specific textual and stylistic developments that occurred during the composition of the PI rather than to precursors, ancestors and relatives of the PI remarks. Consequently, this support is of a kind that, although concerned with the context of the PI’s origin, is much more internal than the first can be, since it is intimately connected with the shaping of the PI and its remarks. Therefore, in this argument one will not so much use the context to interpret the meaning of certain remarks in the light of earlier remarks or meanings, but rather study differences and changes in order to see what new elements were introduced in the making of the PI, and what old elements were left out.

Here the relevant contextual data can only be sketched (see further Pichler 2004). In short, they include the following elements: (a) In late 1936, Wittgenstein dismissed the form of the Brown Book as suitable for his envisaged second (third) book and consciously adopted for the PI a different form, that of the album. The development from Ms115, second part, to Ms142 is the primary witness for this shift from the book-form to the album-form. (b) For the PI, in late 1936 content, method and form were merged to such an extent that they became inseparable. The development that led to this merging and the reasons for it are described by Wittgenstein himself in a series of preface drafts. (c) The genesis of individual remarks and parts of the PI bears witness to their conscious redaction towards a text that becomes unfavourable to a theses-reading and encourages a therapeutic reading. That fits well with the often-described experience of PI remarks being much "clearer" in their earlier versions.²