12. Peter Hartmann¹ 

12.1 Though rarely available in English, the work of Peter Hartmann has been highly influential in the reestablishment of linguistics in Germany after World War II. To present his ideas, I shall translate from the German, contending with his dense, complicated style and his meticulous, specialized use of terms.² His book Theory of Language Science (Theorie der Sprachwissenchaft)³ (1963a) offers ‘not an introduction to the science of language nor to the history, representatives, methods, and results of linguistics, but a fairly abstract theoretical argument, intended to be useful for theory, i.e., for a scientific treatment and estimation of the concern for language’ (TS 3) (cf. 6.1).

12.2 Hartmann's resolve to cover language study at large does not serve the ambition -- detectable in the surveys of Saussure, Bloomfield, Pike, Chomsky, Firth, and Halliday -- of validating the scientific credentials of one particular approach over others. Instead, he is the most balanced, syncretistic, neutral, and thorough of all the theorists in my survey. I thus saved him for the last because he organizes and sums up the issues in the widest way. Throughout his career, he advocated ‘a truly encompassing theory of linguistics’ to ‘take account of all forms of concern for language’, with ‘scientific linguistics’ ‘situated among them and distinguished from them’ (TS 5, 13). Though his overview is ‘just a specimen’ of that ‘wide-ranging’ ‘project’ and indicates ‘only some basic trends’, he is convinced that ‘higher theoretical insight can best be achieved if the framework is chosen to be as wide as possible’, probing ‘the various statements about language in all cultures’ (TS 5, 50, 6) (6.20). ‘The goal in seeking a theory of linguistics is to subject to theoretical scientific contemplation the procedures which engender and control all concern for language and hence all statements about this object’ (TS 7, 168).

12.3 Because ‘representations of the object phenomenon’ ‘can differ according to the direction of research’, we should use ‘more than just linguistics to characterize the work of the language researcher’ (TS 167, 136) (cf. Kainz 1960; Ullmann 1957) (13.35). ‘The linguist must recognize that his manner of research pursues just one mode among others, and only in combination with them constitutes the totality of knowledge about language’ (TS 5f). We might hope not merely to ‘characterize and understand linguistics’, but to ‘further it as a group’ of ‘research methods’ whose ‘scientific quality’ depends on ‘discovering the specialness of its object’ and on ‘“being explicit about the primary data on which our statements are based; these data are human utterances”‘, which ‘“in every instance”‘ ‘“must be interpreted by a linguist, with the unavoidable interposition of a subjective element, and they must be related to the context and situation from which they sprang”‘ (TS 6, 160f) (Reid 1960:23) (1.9; 13.24).

12.4 Being ‘scientific’ is depicted as ‘demanding insights’, ‘producing knowledge’, pursuing a ‘concern for formality’, and ‘attempting to interrelate different modes of characterization within a determining context’ (TS 3f, 159). ‘The word science [Wissenschaft] designates the activity that results in knowledge [Wissen], the knowledge produced, and the totality of such knowledge’ (TS 13f). To decide ‘what kind of knowledge science seeks’, we should follow John Dewey (1910) rather than Plato by stipulating that ‘science must have a genuine purpose’, and that ‘the knowledge must be operationally usable’ for ‘analysing, constructing, concluding, etc.’ (TS 15f) (13.58). Also, the knowledge must be ‘intersubjective, i.e., usable’ for ‘everyone who proceeds according to it’ (TS 16, 19, 59, 157). So the ‘cognitive’ ‘side’ should be ‘democratic’ in respect to the ‘practical side': that the ‘practice’ be ‘accessible to everybody’ (TS 16). Equally broad is ‘the main requirement’ proposed for ‘science': ‘to achieve all that can be known about a field; to strive for a totality of possible determinations; to notice as many facts as possible; to vary the standpoint’ and ‘intention’ ‘in hopes of a later’ ‘epistemological’ ‘unification’; and ‘to work through the manifestations in as many directions as possible’ (TS 129f). ‘The history of language research shows that progress comes from shifting and enriching the problematics’ (13.1).

12.5 Hartmann's ‘deliberations’ belong to ‘second-order science’ or ‘metascience’, ‘the science of language science’ (TS 3, 168). As signalled by ‘the prefix “meta-”‘, this domain has ‘primary sciences as objects’, including their ‘elements, forms, structures, and methods’ (TS 3).⁴ Within this ‘“higher” order of generality and abstractness’, where ‘naivete’ is ‘reduced’, there ‘arises a new level of criteria for forms’ of ‘elements, procedures, and relations’ (TS 3f, i.r.). ‘What might otherwise occur “unconsciously” or come about “on its own”‘ is made ‘the object of treatment’. ‘A second-order science’ can attain ‘higher-level formal and controlling insights’ from ‘a scientific reflection and deliberation on a particular science’ -- here, ‘a theory of what people do when they work in linguistics’ (TS 168, 4f) (cf. 12.84; 13.36).

12.6 ‘Such a theory can assist the whole circle of research directions’ involving the ‘functioning of language’ (TS 168). The ‘linguist assumes the role of an important member of research communities’ when ‘research moves from registering data’ over to ‘understanding facts according to transfactual rules or laws’ (TS 110) (cf. 12.93, 95f; 13.21). ‘Since we don't know how far the influence of language extends’ or ‘how much is “only” language’, it is ‘advisable for research and knowledge to be of the greatest possible breadth’ (TS 137f). ‘Bound to data and mainly empirical, yet primarily interested in forms and structures’, ‘linguistics is the neighbour of three general theories dealing with signs, sense, and being’ (13.22). But it is ‘self-sufficient thanks to the data placed at its disposition, and probes the structures in statements made about all possible objects by means of signs that have or rather acquire⁵ sense’ (i.r.). ‘To have scientific force’, ‘linguistics needs a theory as free as possible’ of ‘extralinguistic theorems’ and not subordinated to a ‘philosophical or psychological system’ (TS 52; cf. TS 135) (cf. 13.10f, 14, 16f).⁶

12.7 ‘The difficulty for a specifically linguistic theory lies in two facts’ (TS 51). ‘First, a plausible science of language gains its intersubjectivity and general validity by ‘de-theorizing’ its object’, i.e, by ‘dispensing with all categories which are not appropriate to the object or which add something to it’. ‘Second, every researcher tends to follow a latent position or “philosophy” that can best be neutralized by moving into the totality of all possible positions’. ‘But it would be practicable for only a few people’ to ‘study all possible standpoints with the intent of transcending and setting them aside’, i.e., to ‘confront a set of systems (theories) of transcendentals only to totally dissolve their transcendent status’ (TS 51f) (cf. 6.5f).

12.8 ‘Still, this would be a path of inductive theory, in accord with the usual maxims of inductive science’ ‘attending to factual features of phenomena’ (TS 52, 120) (cf. 4.7, 67, 76; 5²⁸; 6.16f; 7.6f; 12.16, 95f; 13.44). ‘In addition, a deductive way is permissible for conducting research free of theory, e.g., by immediately recognizing every theory or theoretical entity’ and ‘allowing only statements that expressly “say nothing” beyond the actual phenomenon’ (TS 52) (6.17, 33). ‘Terminology’ would be kept ‘free of metaphors’, and ‘every formulation would stipulate how it is meant, e.g., as description, interpretation, explanation’, ‘evaluation’, ‘introspection’, ‘motivation’, ‘experience’, etc. (cf. 12.98)

12.9 A further obstacle lies in ‘language being such a familiar, important, and central phenomenon for humans that it is hard to even notice all its basic features and their significance’ (TS 21f) (cf. 2.8; 3.1; 4.2; 6.6; 13.1). ‘It is a peculiarity of humans to interpret, i.e., to use something to indicate something’, and to ‘appropriate the world by interpreting it’. ‘Humans are highly hermeneutic, dependent on and predisposed toward signs’; and ‘human freedom'⁷ includes ‘using and understanding signs freely within certain limits’ (TS 161; cf. TS 62). ‘Most human achievements are decisively influenced by the use of language’ as ‘an interpersonal medium of communication, a practice of using and presenting signs’ (TS 262, 20) (cf. 13.22). ‘The elements, forms, structures, and practices discoverable in language’ are ‘open to the most general membership, subsumption, or identity’ and thus ‘have a normative exemplary value for semiotic procedures in general’ (TS 161, 262) (cf. 2.8; 6.53; 13.21).

12.10 ‘Sign theory’ is accordingly situated at ‘the beginning of all language study’ (TS 70) (2.8, 30; 6.50-56). ‘The sign is a typically human product': it ‘arises from an intent to mean, indicate, etc.’, and the ‘communal inventing and using of means for variation and “free” combination are found only among humans’ (cf. Hockett 1958: 570ff) (TS 173f) (cf. 3.15; 4.28; 7.35; 8.27; 12.10; 13.12). ‘The sign vehicle [Zeichentrager]⁸ is the perceptible manifestation’ ‘bound’ to ‘the intelligible manifestation’ in a ‘functive relation’ (TS 174) (cf. 12.54; 2.25; 6.25f). Though ‘vocal language’ ‘is the largest and most important area of sign use’, ‘the broader sense includes every kind of sign’ in ‘communication’, e.g. ‘gesture’ and ‘mimic’ (TS 171, 177f) (cf. 2.8; 3.10; 4.25; 5.12, 14; 6.50-55; 8.22; 9.11; 11.86; 11²⁹). ‘The narrower sense’ includes ‘only “the linguistic sign”‘ as ‘a special kind’ ‘not found elsewhere’, arising from ‘its own constellation of means and intent’ (TS 176ff). ‘Perhaps influenced by the term “speech”‘, ‘language science’ has, ‘until recently’, ‘usually studied’ only the narrower domain of signs (TS 178, 171).

12.11 Against Saussure, who treated the sign in dualities like ‘sound’ and ‘idea’, or later ‘signifier’ and ‘signified’ (2.25), Hartmann argues that ‘“four entities must be distinguished: sound and phoneme, abstract meaning and intended referent”‘ (TS 176n) (Lohmann & Brocker 1942:2). ‘“Saussure did not draw this consequence”‘ of his ‘“emphasis on the two abstract, inner formal components”‘, in part because he ‘“neglected the referential”‘. Moreover, ‘if a sign can be whatever is not the final purpose of an utterance, then the contents of language elements can be signs’ (TS 172). Thus, if ‘the signified can be a further sign for another, we can postulate various layers of sign quality: direct and indirect’ (TS 172f). ‘Generally, however, signs with more than two layers’, as in complex ‘metaphors’, are left ‘outside linguistics’ and assigned to ‘stylistics and literary studies’ (cf. 6.54).⁹

12.12 A prominent tactic of Hartmann's is to draw frequent parallels between science and language (cf. 13.48). They both ‘depend on symbols’ to render ‘intersubjective knowledge accessible in a communicative form’, and on ‘procedures’ for ‘determining, designating, discussing, and making statements about objects or situations’ (TS 17f). ‘Science is a language in the sense of being communicable (e.g. in terminology), existing in an intersubjectively valid system of knowledge “above the things” it classifies, and operating with systemic-functional units attained by analysis and usable for basic discourse about something’ (TS 146; cf. TS 16, 21n, 159). Therefore we might view ‘language science’ as ‘research attaining intersubjectively usable insights into structure-creating’ ‘forms, relations, and functions within a phenomenon that itself consists in utilizing intersubjectively valid relations for the purpose of communication’ by ‘everyone, at all times and places’ (TS 165, 16) (13.48). ‘Language science’ ‘shares the borderline status of language, compelled to run on multiple tracks, combine standpoints and results, and adopt a general outlook’, whereby it becomes ‘superior to sciences whose object domain has only one sort of manifestation’ (TS 168, 39). More specific parallels compare the ‘relation’ between ‘formality’ and ‘primary science’ to the ‘relation’ between ‘grammar’ and ‘actual language’; and portray a ‘theory of language science’ as the ‘langue’ of the linguistic ‘parole’, i.e., as ‘a system of rules for making conclusive or classifying statements about language as the object of research’ (TS 4, 135). We'll see more parallels later (12.40, 44, 46, 48, 69, 87, 93, 98).

12.13 Yet Hartmann notes divergences between science and language too. ‘The scientific treatment of language is unlike language in being concerned with the constitution of its object’ (TS 146). In ‘language’, we use a ‘relatively unconscious kind’ of ‘knowledge’ and ‘communicative value’ , which is ‘more vague and variable’ and ‘not constructed according to scientific principles’, but ‘fairly empty’ until being ‘further interpreted and utilized’ (TS 21, 255, 17) (cf. 2.35, 2¹⁶; 12.27, 42, 45, 53, 57f; 4.22; 13.49). ‘Science is a language vis-a-vis its object in the sense of characterizing and foregrounding from a distance, and has a language as expression of its activity; its terminology is the language of a special’ ‘determining context’, whereas the ‘determining context’ of ‘ordinary language’ is ‘not based on some real states or distinctions of things’, but ‘enables communication and discourse about any correlates whatsoever, including ones unknown as facts’ (TS 159) (cf. 12.60f, 63). In this manner, ‘language science applies a highly rigorous, intersubjectively valid procedure to a less rigorous intersubjective procedure which is however never without rules’ (TS 21, 165). Hence, ‘language science represents the scientific analysis of a non-scientific analytic practice’ (cf. 3.14; 6.58; 12.64, 86f). ‘Language could be aphoristically described as a pre-scientific type of intersubjective knowledge’, or as ‘a pre-science of everything humans encounter’ (TS 17, 21) (cf. 13.24).

12.14 ‘Science thus needs more than just knowledge of names and terms’ and ‘approaches language in a non-linguistic, practical way’ (asking ‘what happens when..., what to do such that..., why it happens that...’), which ‘in other domains enables technical construction’ or ‘prediction’ -- a possible ‘parallel to natural science’ (TS 146f) (cf. 2.13; 4.8, 18; 7.11; 9.112; 12.13, 48, 80, 99; 13.11, 58). At one point, Hartmann indeed suggests that ‘by virtue of its multi-level object, language science is able to register forms of mental procedure such that they can be largely treated with the exactness of natural science’ (TS 39), but he doesn't elaborate (cf. 2.13; 4.8, 18; 7.11; 9.112; 12.99; 13.11)

12.15 ‘Throughout history, all directions in linguistics have been attempts and results of various representations and transpositions into other’ ‘determining contexts’ ‘that were supposed to explain the constitution or so-called essence of language’ (TS 159) (cf. 6.22). ‘Linguists’ can ‘suddenly find themselves in the midst of broad and ancient research which can be useful to know’, since the same ‘issues’ still form the ‘background’ for ‘linguistics’ (TS 51) (cf. 1.5; 11.1; 13.64). We can ‘join with earlier and often more general theories and measure ourselves against them’, though they were ‘naturally not all equally scientific by today's standards’ (TS 51, 5). So Hartmann acknowledges all kinds of prior sources, both ‘pre-scientific or scientific’, and both ‘extralinguistic or linguistic’ (TS 113f) (cf. 12.94).¹⁰

12.16 One major source is ‘the ancient concern for language in philosophy’, dealing with ‘knowledge in general, correct determination’, ‘valid conclusion’, and so on (TS 8) (13.11). Here, ‘language’ is treated as ‘a means or indicator’, a domain whose ‘commonalities’ reflect the ‘“deeper” causes’ presumed to be the ‘basis for interpreting language and its manifestations’, e.g., ‘logical forms’ (TS 122, 115) (13.17f). ‘The statements’ of ‘philosophy’ ‘should properly be attained via inductive linguistics, but may be usable for more general statements about language’, e.g. for ‘founding a general grammar’ or a ‘linguistic semantics’ (TS 164, 137). We find many ‘statements about constitutive motives [or grounds, Grunde] for language in general’ and ‘speculations about its essence’ (TS 32). ‘How sense is constituted’ is ‘the main question’ in ‘language philosophy’ (witness the notion of ‘“senseless”‘ for ‘disallowed, inappropriate, or misleading combinations of words’), but is an ‘antecedent’ issue merely ‘presupposed’ for the ‘data’ in ‘linguistics’ (TS 137; cf. TS 185) (cf. 8⁷, 8¹⁴). Also, linguistics is not expected to search for the ‘essence of language’ (TS 116, 145f, 156 165f, 198) (8.19; 9.1), though Hartmann doesn't exempt the question from other types of inquiry (TS 122, 128, 143, 147, 149, 167) (12.15, 38f).

12.17 ‘The attempt to understand the fact of meaning’ led philosophers to contemplate ‘the origins of language’, variously seen as ‘a product of human “intellect”‘ (Leibniz, Herder, Humboldt), ‘divine spirit’ (Hamann), or ‘natural development’ (Darwin) (TS 122, 122n, 114f). The plan to explain language from its ‘mutual relation to the intellect [Geist]’, a ‘phenomenon’ on ‘another level’, gave ‘impetus’ to ‘the Romanticist search for knowledge about primordial times’, following a ‘less historical than philosophical orientation’ (TS 122f). Herder (1772) envisioned a ‘spontaneously uttered’ ‘signal’ which ‘the intellect binds to a judgment of an object and thereby makes into a sign for the object’ (TS 123n). Humboldt (1836) believed ‘“language was indispensable”‘ for the ‘“subjective activity”‘ of ‘“synthesis”‘ whereby a ‘“representation becomes objective”‘ and a ‘“concept is formed”‘ (TS 123f). The ‘“sound”‘ of language ‘“makes the internal and evanescent activity”‘ become ‘“external and perceptible to the senses”‘ (TS 124n). So in Humboldt's opinion, ‘“language is the formative organ of thought”‘ and enables it to ‘“objectify impressions”‘ and ‘“attain clarity”‘, thanks to the ‘“sharpness and unity of the sound”‘ (cf. 2.32; 3.3, 10ff; 6.2, 31).

12.18 Humboldt separated ‘technical forms’, as in ‘grammar’, from the ‘so-called “inner forms”‘ of a ‘semantic and combinatory nature’ (TS 124f) (cf. 7³). The ‘inner forms’ ‘originate in the “intellect of a people”‘ in order to ‘establish meaningful connections to the world’. Hence, scholars used ‘language forms’ ‘to infer the “world view of the people”‘, ‘concentrating on semantics’ or ‘assigning syntactic facts a semantic value’ -- claiming for instance that ‘Amerindians depict a situation as exactly as possible with polysynthetic expressions; that Chinese don't recognize simple connectives and demand more powers of decision’; and so on. Such speculations influenced Sapir's belief in the ‘diagnostic value’ of ‘linguistic forms’ for ‘the psychology of thought’ (SL iii) (TS 125n) (3.10).

12.19 Leo Weisgerber (1929) in turn based his conception of ‘native language’ or ‘mother tongue [Muttersprache]’ on Humboldt's idea of ‘language as “energeia” (activity)’ interacting with ‘the intellect’, and on Cassirer's (1923-29) ‘theory of the artificial sign required to fixate impressions’ (compared to Saussure's ‘arbitrary’, cf. 2.28) (TS 126). ‘Inner form’ was claimed to ‘reveal itself’ in ‘recreating the world as the property of the intellect’. ‘Language “radiates its effects on the life of the community”‘ and is ‘hence a “force”‘ ‘in cultural events and historical life’, just as ‘real as the reality of things'¹¹ (another idea of Herder's) (TS 126f, 115). ‘The world divides into the outer world of “being” [Sein], the inner world of human cogitation, and the intermediate world’ (‘cosmos of content and meeting-ground of forces’) wherein ‘a mental image [Bild] arises from natural conditions and the work of the intellect’ (TS 126f, n). ‘“The whole human being” is always contained in the sense of signs': ‘the sign value unites naming (denotation) with sensation (connotation), reason with feeling, thing with human being’ (cf. 8.20). If ‘content is not just meanings, but “transformative” forms of the effects of language’, then ‘word formation’ and ‘sentence patterns’ can be ‘investigated for their content values’, and ‘certain cultural forms’ can be seen as ‘traces of the power of language’. This ‘peculiarly German trend’, though ominously implying ‘one's native language is “better” than a foreign language which allows only “foreign thinking”‘, ‘was welcomed by pedagogy, which is, to an amazing degree, oriented toward tradition and habit, and decisively hostile to structuralism’ (TS 127f, 78n) (cf. 4.5f, 85).

12.20 Another major source is ‘traditional grammars’, ‘the oldest and best-known form of description’, circulated ‘since antiquity’ (TS 63) (cf. 2.5f; 3.36; 4.4f; 6.5; 7.4; 8.5, 7f; 13.7). ‘The concept of “grammar” was at first very broad’ and ‘led to general contemplations of language': ‘in various philosophic schools’, such as ‘the stoics’; ‘in rhetoric as the doctrine of language formation’, and thus as ‘the foundation of later normative grammar’; and finally ‘in philology’, which ‘turned to “outer forms”‘ and ‘coined the first neutral terms like “medium”‘. ‘The Greek inheritance’ with its ‘multiple roots’ can be traced in ‘Latin grammar’ and in ‘its effects on the Middle Ages’ (TS 63f). In contrast, ‘Sanskrit grammar’ ‘consulted only the actual functioning of the language, though from a decidedly normative standpoint, as was also the case among theoreticians of written Arabic: no conscious use of language form without conscious grammar’. ‘Sanskrit grammar alone’ ‘coined artificial and fully unambiguous terms’ ‘for language practice, i.e., for the combinability of elements’, in contrast to the Western drive to ‘reinterpret terms’ ‘in a logical or ontological perspective’ (e.g. Harris 1751) (TS 64f; cf. SB 131f) (cf. 2.5; 4.4, 40; 13.17f).

12.21 Like our other theorists, Hartmann notes the drawbacks in earlier grammars (cf. 2.5f; 4.4f; 8.5, 7f; 13.4). They are ‘often charged with lacking purity’, ‘usefulness’, and ‘unity’ (TS 63f, 235). ‘Constitution’, i.e., ‘operational’ ‘production’, ‘was considered a secondary, merely accidental aspect’ reserved for ‘specialized research’ (TS 223, 210). ‘Correctness’ was attributed to ‘grammatical form’ rather than, say, to ‘units of designation’, because ‘a feel for formal characteristics and differences arose quite early’ among ‘grammarians’ of ‘Greek, Sanskrit, Assyrian, Japanese, and so on’, and all these ‘model languages demand formal correctness’ (TS 210f, 218). ‘The Indo-European language type’ with its ‘formal requirements’, although not the only ‘godfather’, was ‘naturally influential’ (TS 210n). Many ‘grammars’ and ‘descriptions’ were ‘oriented toward Latin’, which was not too ‘disadvantageous because Indo-European languages belong to a shared type’ (TS 63ff, 70) (but cf. 2.5; 3.50; 5.24; 6.5; 8.5; 9.25).

12.22 Hartmann dates the modern period of ‘language science’ proper from the late 18th century. He sorts the ‘themes’ in early ‘language science’, which failed to adopt a ‘metascientific’ viewpoint, into five groups in roughly chronological order (cf. 2.5): ‘(1) language in its peculiarity as a changing and hence “living”, “organic” manifestation’ (Friedrich Schlegel, Bopp, etc.); ‘(2) language as a typical phenomenon of human beings’ and the ‘precondition and expression of thought’ (Humboldt); (3) ‘the lawfulness in observable changes of language’, whereby ‘individual manifestations (language forms)’ are ‘partial, temporally conditioned states, related to others and formed by the same tendencies’; (4) ‘the general linguistics’ ‘proposed by Saussure as research devoted directly to language as such’, and still ‘fruitful today’; and ‘(5) subsequent schools staying within the methods of linguistics (phonology, structuralism)’ and ‘based on theoretical deliberations about language’ (TS 7f). Scholars in (3) postulated ‘sound laws’ and ‘applied them to etymology’ while neglecting the ‘language system’ in regard to ‘word formation and syntax': ‘sound analyses reach far back, while syntax was postulated roughly as it is found in historically documented times’ (TS 79). Whereas those scholars held ‘statements’ to be incontrovertible as long as they refer’ to ‘historically attested language forms’, scholars in (4) (notably Saussure) argued ‘how little’ such statements were ‘really laws’ and ‘how little such research takes account of language itself’ (TS 7f) (cf. 2.13f, 38).

12.23 These ‘themes’ demarcate a shifting epistemology, with ‘older forms of language science’ being ‘transcendent’ (‘language as a means of knowledge, belief, correct expression’, etc.), and ‘recent’ ones ‘immanent’ (‘language as a phenomenon of its own -- what its nature is and how it works’) (TS 54, 9) (cf. 6.6, 8, 10, 20f). ‘In contrast’ to previous work, ‘the modern linguistics of form directly explored the functional mechanism’ (TS 32, 139). Saussure's ‘new’ ‘synchronic and descriptive’ ‘outlook’ treated ‘forms’ as ‘indexes of constitution and relation, as well as signals of achieved possibilities’ (TS 139n, 218).

12.24 Of course, ‘formality had been emphasized much earlier’ (e.g., by Humboldt), and ‘the naive linguistics of form is as old as the study of grammar itself’ (TS 32) (cf. 12.41, 88). Indeed, ‘no study could dispense with formality’, though the latter may not be as ‘thematic’ or ‘intentional’ as it was in ‘the relational linguistics of Saussure, the oppositional linguistics of phonology, or the dependency linguistics of structuralism’. ‘Methodological deliberations’ were ‘relatively difficult and late’, ‘not until the start of the 20th century’ (TS 55). ‘Recent linguistics’ became more ‘rigorous’ and ‘theoretical’, ‘based on insight, not accidental knowledge’, and undertook to ‘describe languages already described in other ways’ (TS 65, 105, 139, 165, 235, 70). ‘Progress in linguistics’ consisted in ‘discovering new objects’, ‘unknown languages’, and ‘new forms of description’ that ‘presupposed a certain theory inquiring into the validity of descriptive methods’ (TS 65).

12.25 ‘The schools of recent linguistics’ in the ‘descriptive direction’ agree that ‘difference is the contribution of any relevant characteristic’, and seek ‘relations among sounds, sound functions (phonology), forms, form functions (morphology), words, word functions, and so on’ (TS 67f) (cf. TS 201ff, 216; 2.26; 12.50, 82; 13.26). This direction ‘clearly contrasts with’ ‘comparative’, ‘unifying’, and ‘totalizing’ ones (TS 68). ‘Recent’ work ‘bypasses the definition of totality and yet delivers the structures that compose the “totality”‘ (cf. 12.95, 98). According to the ‘underlying epistemological theory’ (‘discussed by Hjelmslev’), an entity is ‘seen only in confrontation and relation with others’ ‘with which it tends to co-occur’ (cf. 5.20; 6.44f). ‘The language forms being investigated exist only to indicate relations’. Here Hartmann hails ‘one genuine theoretical and methodological change’ that was ‘not dictated empirically’ (TS 68f).

12.26 Despite their ‘divergences and controversies’, ‘the schools of Geneva, Prague, Copenhagen, and so on’ were ‘close in goals and methods’ (TS 69). Their ‘quarrels’ were chiefly about ‘the relevance of emphases and the purity of description, i.e., the categories being applied’. In ‘the Saussurian “system”‘, ‘the basic structure’ emerged from ‘the tasks, possibilities, and necessities of elements’ (cf. 2.27). ‘“Langue” was circumscribed’ with ‘the question: what can occur as factual language (“parole”) and yet enable and control it as a system?’ (cf. 2.20; 13.36). ‘In order not to transcend the phenomenon’ (12.23), ‘descriptive linguistics did not move beyond technical functions’. ‘The epistemological position that the single entity is registered through the multiplicity¹² surrounding it, was joined by the methodological position that a phenomenon may be described only from within its own domain’. An extremely ‘strict’ consequence was to ‘consider meaning’ an ‘extralinguistic’ ‘effect of expression signs’ and to ‘leave it aside’ (cf. 4.15, 26; 5.61f; 7.56f). Yet this ‘position’ will not serve for ‘discussing the total phenomenon of language’, whose ‘peculiar property’ is its ‘relation to the aspect of sense’ (Hjelmslev's ‘“content plane”‘, 6.24, 29, 42) (TS 69f) (cf. 12.65).

12.27 ‘The special nature’ of ‘linguistic treatment lies in demonstrating ‘issues’ or ‘structures’ that ‘need not be grasped by non-scientific acceptance [Hinnehmen] nor become conscious during the naive production of language’ (TS 162, 72n) (cf. 12.13, 42, 45, 53, 57f; 13.49). ‘The major practice of language research consists in isolating, characterizing, and determining language manifestations according to issues that are structurally latent in the handling of language, yet relevant for constitution': the ‘constant forms within the functioning’ (TS 167, 163f) (cf. 12.31f; 13.54). Since ‘linguistic formalization is an observational procedure’ that ‘emphasizes forms where none would be noticed by a different method’, ‘the observer regularly discovers more forms than the language user’ (the ‘participant’ in Pike's scheme, cf. 5.16f) (TS 132; cf. TS 227; SB 133) (13.49). ‘In the observer system’, ‘even the occurrence of manifestations is described in abstraction from realization’, including whatever ‘other aspects are involved for participants’ (TS 131). ‘In general practice, in contrast, forms (structures) are obeyed, but not as conscious formalizing’ (TS 132). So ‘scientific treatment’ ‘makes an issue out of functional forms that are normally not an issue, and foregrounds them’ with the aid of ‘specialized’ ‘concepts and terms’ (TS 162f) (13.38).

12.28 ‘A methodological problem arises: the systematic, “automatic” methods of formalizing linguistic objects can lend an “unnaturally” high importance to formal items in expressions’, until ‘the analyst’ ‘readily believes that formality is the “essential” and that form is the vehicle of all discourse’ (TS 131) (13.49). Yet a ‘formal level of representation’ can be ‘language only in a derivative, expanded sense’, containing merely ‘forms of formality’; the ‘terms cannot be freely applied elsewhere’ (TS 158). Still, ‘it is understandable and justifiable’ that the ‘scientific’ ‘interest’ of ‘formal linguistics’ ‘isolated and objectified’ ‘forms’ and postulated ‘new properties’ resting on ‘basic features abstracted away from real material’ (TS 208, 263, i.r.) (cf. 13.54). ‘Though the features emerge only during a certain kind of analysis, they can clarify the ways and forms whereby language functions’ (TS 263).

12.29 ‘The forms of language’ have three ‘indexical values’ for ‘the formation of expression’, namely to ‘indicate': (1) ‘a mode of proceeding’ in ‘constitution’; (2) ‘a basic precondition’ in ‘relation’, and (3) ‘potentiality’, ‘regularity’, and ‘freedom’ in ‘possibility’ (TS 209). ‘Forms in languages’ have been ‘conceived as manifestations that can change in respect to others they co-occur with’ -- whence ‘the conception of form as transformation [Abwandlung]’, as ‘accident within the domain of non-accident’ (TS 208) (2.13, 6.16). Thus, ‘discussions’ of ‘substance and accident, content and form, inner and outer’ reflect ‘two kinds of classification’ that are both ‘interdependent and yet opposed': ‘for constants and for variables’ (TS 208f, 211). Hartmann proposes a somewhat Hjelmslevian four-part scheme of ‘relative instantiations': (1) ‘between constant and another constant (e.g. between word-stems)’; (2) ‘between constant and variable (e.g. between word stem and suffixes)’; (3) ‘between variable and constant (e.g. between endings and the cases, persons’, etc. they signal)’; and (4) ‘between variable and another variable (e.g. in paradigms)’ (TS 212f) (cf. 6.25, 29, 33f). The first is ‘represented in the lexicon’, and the other three in ‘word-formation and grammar’.

12.30 For Hartmann, ‘the content-form distinction is largely given in language’ and is thus ‘legitimate’ (TS 212) (but cf. 6.24, 47, 50; 9.38). ‘Befitting the demand for the greatest possible variety of standpoints’ (12.2), ‘the linguist’ must ‘deal with form and content’ in ‘every manifestation of language’, even where ‘only one side’ appears to ‘dominate’; we ‘always need to determine how far form “carries” content and what kind’ (TS 131). Nevertheless, ‘chiefly form is encountered in language science proper, firstly because it alone emerges from comparable entities that hold manifestations together; secondly because scientific insight is directed to what remains the same within all factual change’ and can thus be ‘reliably communicated and intersubjectively used’; and thirdly ‘because analysis’ ‘does not collect undivided manifestations, but systematizes them’. In this framework, ‘innerlinguistic forms are facts’; and ‘formal inventories’ are ‘more constant than the non-forms they must enter to occur within a phenomenon’ (TS 131, 163f). (cf. 13.54)

12.31 Yet Hartmann acknowledges the wide variety of ‘claims about how language unites the semantic content with the non-semantic form’ (TS 130). He conjectures that ‘form classes are more stable over time’ ‘than meanings’ (12.66), and that ‘the semantic domain’ is complicated by ‘connotative differences in content’ and by the ‘non-factual, but projected sameness’ in ‘metaphor and analogy’ (TS 205, 119, 28). He depicts ‘forms’ to be ‘relatively constant characteristics’ whose ‘principled invariance’ ‘guarantees intersubjective usability’ and ‘a relative permanence of languages as communicative systems used by communities’ (TS 132, 25). Against the backdrop of ‘invariance needed’ for ‘utterances to be understood by several partners’, ‘variance’ appears when ‘different kinds of things are combined’ (TS 25). Evidently, both form and content can be seen as either constant or variable, depending on how we collate them (cf. 6.24; 11¹⁸).

12.32 A related problem is where to draw the line around forms. They ‘can be “asemantic” only as long as they are kept distinct from the motivation of discourse’, and even then they can have an ‘instrumental semantic value’ (TS 132) (cf. 5.62). ‘When we evaluate a formal, asemantic or combinatory symbolic mechanism, a semanticizing usually occurs, because language is generally used that way’ (TS 244). And ‘motivation’ is not so easily ‘left aside by science’ or ‘linguistics’ (TS 221, 137). ‘A structural description may, for certain limited purposes, work with just one aspect’ such as ‘formal inventories’, as ‘Hjelmslev tried to show’ (TS 130). But ‘linguistics as a whole cannot ignore the semantic-content side’, ‘because the use of language in real events can't be separated from motivations’ which, on ‘the level of sense’, may be ‘more significant than structural issues’ -- one reason why ‘structuralism failed to attain pure formality’ (TS 130, 221).

12.33 Nonetheless, scholars have hoped for a ‘handbook of linguistic statements, formulated as formally as possible’ and ‘prescribed for valid work in linguistics’ (TS 136) (cf. 5.2, 12ff; 7.7; 8³). ‘Linguistics’ might be the model for ‘formalization in all sciences’ if it attained a ‘highly accurate division between form and non-form’ (TS 136f) (cf. 2.8; 5.7; 6.9f; 8.16; 12.6, 12; 13.21, 49). ‘Examples include: constancy and relevance of grammar versus innumerable possibilities of expression’; ‘constancy of logical syllogisms versus variable content of the premises’; ‘formality in transcendental philosophy’ (compare ‘Kant's a priori “before all perception”’); ‘formal displacements in the metaphoric’ (compare ‘recent English philosophy since Wittgenstein’); and ‘numerical and relational modes in mathematics’ (TS 136n). And the claim in recent physics that ‘the electron is nothing but its properties’ can be compared to Hjelmslev's vision of the ‘language element’ as ‘nothing but the intersection of bundles of dependences’ (TS 137n; cf. TS 62n, 251, 256; SB 131; 6.45, 62).

12.34 One way to promote formality might be a ‘mathematicized treatment of language’ (TS 52) (cf. 4.21; 13.15). ‘Mathematics’ is the mode that most ‘appears to be free from standpoints and universally applicable, and to have a terminology and determinacy that falsify the least’ (TS 160). ‘Mathematics’ ‘works exclusively with relations that can be calculated, i.e., treated, described, and reconstructed via commensurate operations within the framework of the multiplicity of quantities’. ‘Almost every manifestation’ ‘can be subjected to quantitative manipulation’ in terms of ‘form or documentation’, including ‘language’ and ‘many products of the human use of symbols’ (TS 160f; cf. TS 155).

12.35 On the other hand, ‘mathematics applies only partially to language’, chiefly to ‘what is constant’ and ‘documented’ in ‘a fixed form or an unambiguous concept’, and thus not to ‘matters of a continually varying nature, such as assignment, naming, value’, ‘hermeneutic interpretation’, and ‘claims about identity’ (TS 167, 160, 163, 262, 264).¹³ ‘In respect to’ the ‘freedom’ of sign use (12.9, 29, 58), ‘quantification is largely “powerless” and irrelevant’ (TS 161). Also, ‘statistical questions about how many, when, where, can seldom be answered exactly, even though natural limits obtain in principle for the structural givens of the reality of language’ (TS 160). Even such ‘quantities’ as ‘sets of similar elements’ ‘are not decisive’ (TS 161). ‘Instead, language consists’ (1) of ‘mixed sets from several levels at once, e.g., the semantic and the operational’; and (2) of ‘domains, e.g., of validity, appropriateness, and membership within a subjectively variable radius, such that people (a) can talk imprecisely, without addressing the essence (or the “Ding an sich”), (b) can talk in detachment from and outside the phenomenon, without considering science and its characteristics, and (c) can characterize things in idiosyncratic and subjective ways, with individualized expressions’ (TS 161f, 166) (cf. 12.60f; 13.41). ‘Semantic classes have a hermeneutic value, range, and applicability’ that can ‘always be widened or narrowed’ (TS 205) (12.57). For all these reasons, ‘mathematics cannot be the ideal method’.

12.36 Moreover, Hartmann does not favour the solution of ‘tying language to logic’ (TS 29) (cf. 13.17). He does conjecture that ‘the units whereby language transcends the merely physical result from a logical capacity’. But he reminds us of the ‘contrast between linguistic’ and ‘logical form’ (cf. Flew [ed.] 1951-53) (TS 138n). ‘Logical forms need not coincide with forms of sense or meaning in language’, e.g. when ‘uniform expressions’ in language ‘mask logical’ ‘differences’ (TS 261, 138n). ‘Logic’ is limited to inquiring ‘how far an expression has sense or meaning when investigated “only from a grammatical standpoint”‘ (as in ‘the Vienna Circle’, Carnap, Wittgenstein, Neurath, etc.) or to using ‘syllogisms’ to ‘draw correct conclusions without knowing what one is talking about’; and so on (TS 137n, 145) (cf. 6.56, 64).

12.37 Language science faces the further ‘difficulty’ of ‘finding a method’ that can ‘unite the factual with the necessary’ (TS 121). When we renounce ‘seeing language manifestations as effects of logical grounds’, we must seek ‘general and necessary traits in the language facts themselves’. These traits can be ‘taken as structural a prioris or as the most general structural laws’ ‘“in”‘ ‘factual manifestations’, and can be ‘designated with concepts like sociality, differentiation, reference, determination, identity, assignment, arrangement, elementarity’, ‘combination, complication, complementation’, ‘(in)variance, opposition’, ‘abstraction, applicability’, ‘translatability’, ‘redundance’, and so on (TS 121, 22). ‘Ideally’, these would ‘emerge from the commonalities of specific facts in all languages’ (TS 121). Though ‘science can use everything in the representation of facts that contributes to comprehending the occurrence of the manifestations in question’, ‘a general method’ does not ‘ask about functional forms in realizations of language’, ‘but about typical ways of functioning’ in ‘the phenomenon linguists must address’ (TS 166, 164).

12.38 In ‘method and terminology we should distinguish between registering language as a phenomenon and scientifically treating’ ‘language as an object’ (TS 139) (cf. 13.58). The ‘phenomenon’ appears in ‘its suchness’, whereas the ‘object’ is ‘treated’ in its ‘otherness’ (TS 140). The ‘pure phenomenon’ is ‘not linked to others, either proximate or remote’; it is ‘characterized with its own means’ (TS 142). ‘Understanding is attained if we can relate our terms to certain states of affairs within the phenomenon’. This direction was pursued in ‘language philosophy’ (e.g. by Heidegger 1959) using ‘subjective methods’ (TS 142, 144). ‘Statements’ were ‘usually metaphoric on purpose’, e.g., ‘“language speaks”‘ (TS 142f). ‘Words of the language’ were ‘interrogated for their deeper, ambivalent sense’, wherein language should ‘declare itself’. Hoping thus that ‘the phenomenon would reveal itself’, philosophers took ‘the basic stance that the original experience should not be disturbed by dividing or combining’ it, or by ‘forcing language into predecided representations’ (TS 142f, n) (cf. 6.4; 12.87). We must now consider whether such ‘strivings for a transscientifically true essence’, such a ‘passive attitude waiting for the self-revelation of things’, are to be ‘dismissed’ as ‘mere thought-experiment’ or ‘Romantic enthusiasm’, or are a ‘result of the same justified striving for pure experience that is also at the base of all science’ (TS 143f). Although the ‘scientific linguist’ wants to ‘control or prevent’ ‘the danger’ of ‘hearing one's own thoughts’, ‘science’ can ‘investigate this approach’ in order to explore ‘what happens when so-called language-internal characteristics’ are ‘taken seriously’ (TS 144, 142) (cf. 12.5, 84; 13.36). This strategy could meet the ‘criticism’ of ‘scientific attitude’ ‘for limiting the experiential aspect’ (TS 144).

12.39 The reproach that ‘science covers up the phenomenon with classifications in order to “put it aside” has no force because the classifications never need to cover the whole phenomenon, but only to bring to awareness a partial domain or characteristic’ (TS 144). ‘Language science need hardly be concerned to state what language ‘actually’, ‘really’ is “in its essence”‘, but can still ‘take note of such statements to judge the plausibility of its own statements pursuing a more limited intention’ (TS 144f) (cf. 8.19; 9.1; 12.15f). The ‘image’ of ‘language’ ‘in linguistics’ is after all ‘incomplete’ because it ‘must allow the right set or combination of features’ to ‘emerge from the progress of research and insight’ ‘by means of successive correction through new findings’ (13.63).

12.40 In contrast to ‘“language as phenomenon”‘, ‘“language as object” is treated by being interpreted, analysed, characterized’, and ‘factored’ (TS 147).¹⁴ It is ‘reasonable’ to ‘treat language ‘“only as an object”‘ if ‘the cited characteristics allow a truly scientific exploration of language’ and a ‘foregrounding of formal features’ (TS 149, 162). ‘Further links’ can be established’ by means of ‘comprehension’, ‘attribution’, ‘confrontation with other knowledge’, and ‘identification’ of ‘the whole and the parts’ (TS 147-50). Again, we do ‘not encompass the whole phenomenon, but the manifestations do receive formal names; and the foregrounded structures do belong to the phenomenon’ (TS 162). Hartmann sees another parallel, namely to the way ‘language’ ‘assigns symbols and arranges by means of an arsenal of classes (vocabulary)’ (TS 149) (cf. 12.12).

12.41 We can also distinguish between ‘structural’ and ‘non-structural’ approaches, where ‘structure’ is ‘a term for a collection’ of ‘forms in a limited phenomenal domain’, or for ‘a similarity of types’; ‘no grouping is without structure’ (TS 141, 49, 154, 45).¹⁵ ‘Higher-level formality’ requires a ‘shift from available achieved facts over to the structures that can be grasped in them’ (TS 4). The ‘non-structural’ mode (the ‘main one in earlier language science’) ‘registers’, ‘accepts’, ‘interprets’ in ‘general terms’ (e.g. ‘expression, activity’), and ‘analyses’ into ‘units of chiefly semantic kinds’ (e.g. ‘meanings, motifs’) (TS 141). The ‘structural’ mode (the main one in modern ‘linguistics’) ‘analyses formally’ and ‘interprets’ according to a ‘formal typology’. In between, ‘traditional grammar’ has its ‘formality’ ‘distributed across several levels, each with its own structure’; these can now be subjected to ‘a rigorous structural treatment’ to ‘decide whether or not structure and grammar correspond’ and thus how far ‘the grammatical analysis was structurally adequate’ (TS 235, 237) (cf. 2.6; 6.49; 7.4, 75; 8.38; 12.88, 90; 13.7).

12.42 ‘The structure’ of ‘sets of signs’ in an ‘expression’ can be attributed to ‘obeying rules of formation’ that ‘constitute a higher-order system’ (TS 33, 97, 222). ‘Insights into structure’ are ‘attained by a survey of discoverable forms and functions in such a system’ (TS 97). The ‘rule-governed’ ‘practice’ of ‘language’ appears in the ‘occurrence of an element with something else’ and in ‘the regularity’ of ‘constituting sign sequences’ (TS 225). ‘“Rules”‘ is ‘the typical designation’ for such ‘observable interdependence’ and hence a tool for ‘description’ (TS 225, 148). Just as ‘people speak without noticing grammar’, they also ‘follow rules without knowing them’ (TS 148) (cf. 12.13, 27; 13.49). So ‘rules’ may function with ‘zero consciousness’ and no ‘actual motivation’, in contrast to the ‘semantic consciousness’ about what is to be ‘communicated’ and ‘achieved’ (TS 34, 223). Still, Hartmann concurs with Wittgenstein that ‘rules of language are ways to use language’ (TS 222f) (cf. 8.47; 13.36).

12.43 ‘A language expression is multiply structured’, and its ‘usefulness for communicative purposes rests on simultaneous utilization of all the structures encountered in it’ (TS 234f). Hartmann lists ‘four main structures': in the ‘arrangement’ ‘within a spoken chain’; in ‘the dependence between elements’ ‘uniting in groups’; in ‘the semantics of the element’ whose ‘components’ ‘are already signs’; and in ‘the semantics of the expression’ ‘being a sign, a combined totality formed, used, and understood’ as such (9.34, 37; 12.72). ‘The circle of possible objects’ is ‘open wherever new objects can emerge from a new formation of structures’, or wherever we can ‘discover objects with a new structure’ (TS 18). Hartmann suggests that ‘the smaller the structural sectors to which we are limited, the easier it is to find commonalities’; ‘the more we move toward the total extent of the text, the harder’ this becomes (TS 236) (but cf. 13.39). This ‘proportion’ holds for research in both ‘morphology’ and ‘syntax’ (TS 236f). In contrast, ‘the full consideration of semantic coherence leads toward powerfully individualized stylistic objects, whose comparability diminishes’ as ‘the extent of the evidence’ increases (but see 13.39). Here too, content seems less constant than form (cf. 12.29ff).

12.44 Structures appear both in the research and in its objects (cf. 12.12). ‘Language science proves to be structural research about an object that itself works by means of continual structuring’ (TS 45) (cf. 12.12). ‘Structure’ is ‘the possibility of possibilities’, ‘the image of possible constitutions’ (TS 223f). ‘Structuration is the epistemological product’ of ‘scientific factoring’, whereas ‘grammaticality’ and ‘grammar’ are the ‘product’ of a ‘historical structuration according to naively discovered relations’ (TS 237, 233). In contrast, ‘the structure of an expression, in each case, is a fact of “parole”‘, in that it ‘can be done wrong without causing damage’; ‘right consciousness’ is unaffected by ‘occasional distortions of the system’ (TS 49) (cf. 7.12).

12.45 ‘The relative firmness of structure’ lies in being ‘realized relatively automatically as semantic sentence components’ (TS 49) (cf. 11.11). ‘Structures can remain unconscious’ and be ‘followed’ through mere ‘imitation’ (TS 44n; cf. TS 26; 12.13, 27, 42, 53, 57f). Indeed, ‘treatment’ is more likely to be ‘objective’ and ‘free of interpretation’ when ‘the object is normally situated below the threshold of consciousness’ (SB 134). ‘Structural research’ can show ‘necessities’ without involving ‘the motivation for people using elements’ (TS 45n; cf. TS 132, 158, 221; 12.32, 42, 76, 79; 2.28). We ‘think of structures only when discrepancies are consciously noticeable’; otherwise, ‘we do things the way we know how, and do not ask how we can know’ (cf. 9.6).

12.46 ‘Language science’ therefore pursues ‘research intending to gain knowledge about structures inherent to language, independent of time’, ‘place, and space’ (TS 20, 28) (cf. 13.43). Correspondingly, ‘as a manifestation of signal values’, ‘language’ itself ‘proceeds by suspending (neutralizing) characteristic differences of the designated’ ‘physical’ ‘world’ (TS 29). It ‘suspends space in the typical unity of the name’. It ‘suspends time’ in the ‘conceptuality of class formation’. And it ‘suspends motion’ in the ‘“rigid” form of the idea or the concept’.

12.47 Time is highly relevant here because, due to the ‘relation’ between ‘vehicle and sense’ (or ‘“signifiant” and “signifié”’), ‘the sign’ can be depicted as ‘a duality bound in simultaneity’ (TS 175) (cf. 5¹⁶). ‘Simultaneity’ also ‘applies to the fact that elements’ in both ‘systems’, ‘the virtual (“langue”) and the actual (“parole”), are given’ at once (TS 91) (13.39). The idea of ‘simultaneously available possibilities in the virtual system’ (TS 91, 95, 96) is familiar from ‘Saussure's “etat”‘ (state), postulated as ‘a research category’ for ‘systemic linguistics’ in order to ‘detach itself from all aspects of non-simultaneity, like historical development’ (TS 91) (cf. 2.40). But the idea applies to the actual system in more complex and less familiar ways. In addition to the ‘spatial, temporal relations among sign vehicles in a sequence’, Hartmann postulates ‘simultaneous’, ‘non-spatial, non-temporal relations among sense units’ (TS 214, 183, 91f) (cf. 9.34, 48, 103; 13.33). ‘The speaker’ ‘converts’ the mode wherein ‘sense elements are with and for each other’ into a mode wherein the ‘vehicles are one after another’; ‘the hearer’ goes the other way, working ‘serially’ in ‘small steps, often in twos’, up to ‘the simultaneously conscious determination’ (TS 93, 176; cf. TS 27, 43) (cf. 7.83; 11.81, 85; 13.57). When ‘thoughts are formulated’, ‘the total intention comes first (like a bridge), and then comes the local filling and shaping’; when ‘thoughts are understood’, the ‘reverse’ occurs (TS 92n). In the latter case, ‘the sequence is given and needs little notice, except when unclear or erroneous; the hearer “assumes” that the chosen form will work for communicating the thoughts’ (TS 92fn). Still, this ‘simultaneity of sense relations is only a functional mechanical background for the relations being formed’, and a perspective for ‘insight’ into ‘relations outside time and space’ (TS 93).

12.48 In a comparable vein, ‘language science’ is required to ‘seek immateriality in structures of relations’ (TS 20, 19, 45) (cf. 2.17, 20).¹⁶ Only ‘by accounting for immaterial modes of constitution can science attain intersubjective validity’ (TS 19). Hartmann sees yet ‘another parallel to language': ‘language science’ is ‘research leading to intersubjectively usable results and knowledge according to structure-creating immaterialities -- forms, relations, structures, functions -- with respect to a phenomenon that also consists of using intersubjectively valid structurations for the purpose of communication for everybody, everywhere, always, and about everything’ (TS 20) (cf. 5.23; 12.12). ‘The term “immateriality” designates a purely analytic contrast’ between the ‘occurrence’ and the things that ‘do the occurring’, or between ‘structuration’ and what gets ‘structured’ (TS 47f).¹⁷ ‘Immateriality obtains in arrangements, relations, forms’; it is ‘the structure of the form and is thus formal to the second degree’ (TS 198, 48). ‘Immateriality applies to various facets’, such as: ‘determination’, ‘distinctiveness’, ‘rule application’, ‘interchangeability’, ‘combination’, and ‘grammaticity’ (TS 33f, 42).

12.49 If ‘immaterial relations are revealed only in structures that determine and control the material’, we might ask ‘how far the material and the immaterial’ are ‘symmetrical’ (TS 19, 265). Though Hartmann doesn't give an answer (the question comes at the very end of TS), he ponders ‘the curious fact that humans express their intellect in language by producing structured material’; ‘why does the structuredness of matter, extending from the inanimate to life forms, even constitute the medium that seems to be the least “given in nature”?’ (TS 264; cf. TS 262). On the other hand, ‘all sciences are “intellectual”‘ in that they ‘deal with immaterial relations’; ‘differences between physical and nonphysical sciences, natural and humanistic sciences, are due to the structures they address and the methods they use’ (TS 19f) (cf. 12.14, 99). In any case, the ‘parallel to the binding forces of matter is not pervasive, and is useful only for the material side of language’ (TS 264) (cf. 4.8; 6.62; 7.16, 36; 12.59; 13.43).

12.50 To say that ‘immateriality is revealed in distinctiveness’ is to ‘address the fact’ that ‘elements must differ’ and to recognize ‘not the processual, but the real structure of language’ (TS 42) (cf. 2.26; 13.27). ‘Distinctiveness’ is ‘an elementary structure in the most rigorous sense’; ‘elements’ ‘must be distinctive’ in order to be subject to ‘disposition and combination’. So ‘differentiation is the basic precondition for language’, ‘communication’, and ‘analysis’, and therefore ‘a legitimate domain of research for linguists’ (TS 181f, 42) (cf. 2.26; 12.25; 13.26). ‘The sign vehicle’ consists either of ‘one unit differing from units not uttered’ or of ‘several differing units’ into which it ‘can be divided’ (TS 182). The ‘a priori’ ‘unit’ of ‘language always occurs where others are either present’ or ‘merely possible’ -- ‘potential existence has the same effect as real’ (TS 182f). Or, ‘the unit’ may ‘emerge from continuation or from non-continuation’ (e.g. a ‘pause’). Or, ‘the same’ unit can occur ‘twice but with a different function each time’, as in a ‘tautology’ (TS 42).

12.51 The upshot is that ‘all languages’ have ‘elements -- words, word-parts, sentences’ -- which ‘are normally present in a shared multiplicity and arranged in a speech sequence divided by differences’ (TS 24f) (cf. 2.58; 13.28). ‘To the degree’ that ‘an element’ ‘is identical to itself, it is also different and special’ ‘in its environment’ (TS 26). ‘Language relations are rendered a bit opaque in that difference can appear either in the modification of one type (a constant) or in the opposition between types’. ‘The combined expression consists of otherness as well as of elements’, the more so if ‘the element is what is other than the others’ (TS 26, 181).

12.52 ‘The conception of system’ is pertinent ‘both for the functioning in (or of) structures and for the application or construction of procedures’ (TS 83). ‘Recognizing functions in structures and using the procedures for structures go together like theory and practice’ (TS 83n). We should thus expand the familiar conception of ‘system’ as ‘a union characterized by the co-occurrence of single entities whose basic function is to form this union’ (TS 82f). ‘The system allows a more dynamic interpretation’ than ‘the static, set-theoretical view of whole’ and ‘parts’. Still, ‘language research might use set theory for certain problems, such as quantities, relations between quantities, and the constitution and description of sets’ (TS 83n) (cf. 12.34).

12.53 ‘Function’ ‘designates the dependency of elements that jointly organize a system’ (TS 84) (cf. 3.16, 22, 24, 33; 4.47, 49; 7.63; 8.49f, 58, 61; 9.3, 11, 18f, 28, 33; 12.42, 50). ‘“Function” can be seen as the complement of “whole”, “system”, or “set”‘, and is ‘closely related to value’ (TS 85). ‘An element acquires (not has) a value by virtue of the function in which it is used’ (TS 86). ‘The properties of an element are determined by which functions it receives, while the structure of the system rests on which functions an element can receive’ (TS 89). Hence, ‘“function” designates the “performance” of form’ (TS 88). ‘The structurally same form’ can ‘have different consequences’ ‘in different data, e.g., the same forms in declensions of both adjectives and nouns or predicate nominatives’ (TS 140f). We might ‘ascribe a “functional effect” to the element’, though it would be ‘intransitive’ and ‘automatic’; we should ‘be wary of metaphoric “activities of language”‘ or ‘“responsibilities of elements”‘, which after all can occur only ‘in consequence of a schema’ (TS 86, 89) (cf. 12.57). The ‘schema’ has a ‘predecided, yet open set of possibilities’, and thus ‘forms the background for functions’ involving ‘unconventional uses of elements, e.g., for emphasis and affect’ (TS 87).

12.54 ‘Linguistics must consider various functions on the levels of sign vehicle’, ‘sense elements, vocabulary, and sentence’ (TS 85). To distinguish ‘the functions’ of ‘levels’, Hartmann proposes the ‘terms’ ‘“opposition” on the sound level and “determination” on the sense level’ (TS 85, 90) (cf. 13.59). ‘In the material of signs, only such functions occur as are compatible with the physical aspect of language’, e.g., ‘collocation in a rhythmically bound linear realization’ (cf. 8.66; 12.49, 80). ‘The functions of sense elements are essentially different, despite some traces of the sequential character of language, e.g., mental steps’ (cf. 6.47f; 12.47, 76; 13.30).

12.55 When ‘elements from a virtual system’ are ‘activated’, they form a ‘factual or actual system’ -- a transition like that between ‘“langue” and “parole”‘ (TS 87) (cf. 12.47; 13.39). ‘In principle, the virtual system is open, whereas the actualized system is closed, finite, smaller, and formed for the sake of determination’ (TS 87f).¹⁸ ‘In the virtual system’ we have ‘not static fixed unions of elements’ in ‘functions’ but ‘sets of variable groupings and hence a variability of functions’ (TS 99). ‘Static description’ serves to reveal ‘structure’, whereas ‘operational production’ is the basis of ‘constitution'

(TS 223). Also, ‘the scientific description of a language manifestation, e.g., a discourse or a text’, ‘puts it’ into a ‘static’ ‘form’ by ‘detaching it from its original or direct status’ (TS 157) (cf. 2.36; 3.54; 5.31, 33; 6.33; 8.30; 9.95; 13.31).

12.56 The ‘autonomy of the system’ as a ‘set of functions’ ‘possible in a domain’ enables ‘creative individuality': ‘the individual person uses, builds up, and changes the system’ while ‘forming actual partial systems’ (TS 87, 89) (cf. 7.44; 12.58; 13.41). No such process could occur if ‘the system’ were ‘firmly bound to self-sufficient elements’ (TS 90). Yet ‘the autonomy is relative’ in that ‘certain elements’ ‘demand other elements be situated in certain ways and thus predetermine actualizations’, e.g., the case of the ‘object’ of a verb or preposition (TS 90, 99n).¹⁹ So ‘possibilities are conceivable’ that cannot be ‘realized’ or ‘considered valid’ ‘on closer examination’ (cf. 6.11): ‘combination’ is expected to ‘follow some system’ based on ‘compatibility’ (TS 99). Still, if ‘everything is combinable except where a definite system of requirements (fixed usual bindings) applies’, then a ‘virtual system’ comprises chiefly ‘domain-building functions’; ‘complexes can be formed in which virtual elements and possible bonds converge’. Such a ‘system has the total domain of possible combination, where possible functions’ can form ‘commensurate’ or ‘fuzzy sets’ [unklare Mengen]’ (TS 100, i.r.).

12.57 ‘Functions’ can be classed into ‘domains’ by their capacity ‘for forming names (associations in name sets, e.g., synonyms, word-fields)’ and ‘for forming statements of an additive or alternating kind (schemas)’ (TS 100). ‘Name domains create functions that are difficult to determine’ because ‘words can be reinterpreted to widen or reduce a domain’ (cf. 12.35); ‘so we can postulate variable structures within changeable sets’. Because ‘a continual drifting among concepts alternates with the actualizing of factual givens’, ‘we live among continual possibilities of combination': ‘not among words and sentences but among meanings that can but need not be actualized via words’ (TS 101). Thus, ‘functions appear coupled with a continual activity of deciding’ -- ‘seldom a conscious one’, because ‘language is mainly used through imitation’ (TS 100) (cf. 12.45; 13.49). ‘Due to the speed’ and ‘familiarity of discourse, a process is experienced only for longer expressions produced consciously, e.g., discerning discussions or public speeches’ (cf. 11.83). Hence, we cannot ‘expect the effects of functions’ to be ‘felt by any speaker’, and we must take care not to ‘ascribe to functions such meaning-creating tasks as they could only have if someone really used them as signs’ (TS 100nf). ‘Discursive consciousness can be viewed as a ceaselessly running mechanism for identifying and representing [Vorstellung]’; only in such cases as ‘dreams’ and ‘illnesses’ does it ‘overstep the limits of the combinable’ (TS 101, 98n) (12.73).

12.58 ‘Classification’ is a similar ‘interesting and important aspect’ ‘permeating the whole phenomenon of language’, and must be addressed in any ‘structural description’ or ‘scientific treatment of language’ (TS 201, 207) (cf. 2.26, 59; 3.40; 12.12, 29, 39, 64, 86). ‘Nobody intends merely to classify when speaking, but we cannot speak except via classification’ (TS 196). ‘Meaning has class properties in that (1) a word can be used’ to ‘select an object, designate it as special, and make it the representative of a class (a group of similar objects)’; and (2) ‘meaning’ is ‘at least potentially an instance of grouping similar things’ (TS 35). ‘The forming of classes in all discourse remains unconscious and latent’, ‘but the fact is plain’ wherever ‘manifestations (or their parts and features’) reveal ‘regularity, similarity, and limitation’ (TS 201; cf. TS 259). The ‘similarity’ of ‘functions’ within a ‘class’ complements the ‘difference of manifestations’ within the ‘perceptible side of a language’, as seen for instance in ‘sounds’, ‘syllables’, ‘suffixes’, and ‘sentence formations’ (TS 201f, 204, 206).²⁰ ‘The individual can discover possible classifications just as functions can vary in forming domains’. ‘Decisions and trials include language creativity, such as the poet's’, and the freedom to ‘expand the lexicon’ and ‘form metaphors’ (TS 102) (cf. 13.41). Paradoxically, ‘the most abstract mechanism allows the individual personality the surest chance to perform’ (cf. 12.55).

12.59 Moreover, ‘a new production of “more fitting” classifications and names is always in progress’ in response to changes in ‘knowledge’ of ‘the forms of reality’, e.g., those in ‘physics, energy, intention, logic, “thought”‘, and so on (TS 252, 257). ‘The more a designation refers to the real’ and ‘the more a change in knowledge about the latter is expected, the more likely unfitness is to appear’ (TS 258) (cf. 9.67). But though ‘further exploration of the real’ ‘can show earlier designations to be unfit’, we cannot ‘bypass language’ ‘except by not speaking of the matter at all’; and ‘improved knowledge does not change the fundamental character of language’ (TS 258, 256). ‘Inadequacy appears’ when we need to ‘speak of’ a ‘new object of discourse’ ‘for which a language has no “fitting” expression’, such as the ‘new discoveries in the natural sciences’ expounded by Heisenberg (1960) (i.e., uncertainty relations and quantum phenomena in physics) (TS 62) (cf. 12.33).

12.60 We needn't be distraught if ‘natural language is not adequate for reality in many ways’, especially not for ‘the reality accessible to natural science’ (cf. MacDonald 1951) (TS 151, 150n) (cf. 3.23; 4.22; 5.68; 9.14f, 44, 112; 11.83; 12. 63, 76, 11³⁰; 13.11). ‘Language’ may for example ‘simplify, or encourage errors through metaphors and analogies’ (TS 151). Yet ‘communication requires no “exact correspondence” of an expression to what is meant’ (cf. 4.15). ‘Language does not have the task of containing or reflecting “the essence of things”; if tied entirely to language, thought’ would have ‘no capacity to recognize any such essence, because language is not a repetition of correlates, but their designation’ (TS 151; cf. TS 57, 145, 149, 151, 166, 198, 255) (cf. 3.10ff). At most, ‘language contributes to determinability’, which is what ‘leads to reality by making everything into something determinate’ (TS 252).²¹ But ‘existence does occur before a mode of determining it has been found’, and ‘other forms of understanding, insight, and reaction elude designation and classification’. ‘Statements about reality are made and understood by means of experience beyond language’; ‘`humans live only partly in a world of language’ and ‘have the option of more direct experience’ (TS 257; cf. TS 36) (cf. 13.24).

12.61 ‘Debates about language and thought’ might be clarified by distinguishing among ‘types of knowing’ (TS 146n) (cf. 4.9; 5.10; 11.20-29). Though ‘the access to language depends in practice on factual knowledge about things’, ‘knowing what a discourse is about’ ‘need not include knowing how the things under discussion are constituted’ (TS 166) (cf. 12.35). The ‘real nature’ of a thing does not ‘appear in language’ except insofar as ‘what is known about it’ is ‘mentally added and understood’ ‘via the act of naming’ (TS 253f). ‘Speaking inexactly is the rule in spontaneous discourse and is sufficient; imperfection is only detected when we need to determine something more exactly by means of a formulation alone’ (TS 151). ‘Reality is falsified only if the form of naming is the only source of information about the “essence” of what is named’; or when ‘inappropriate values’ and ‘false associations’ intrude (TS 57).

12.62 We should therefore keep in mind that ‘the sign basis of language’ is ‘merely a specialized and limited consequence’ and a ‘communicable expression’ of ‘the overall unity of understanding, knowledge, and volition’ (TS 254). ‘The assumption that different language communities think differently’ ‘is “correct”‘ only ‘if thinking is taken to be the constitution of insight with the aid of sense forms’ (TS 188) (cf. 3.11f; 9²; 12.19). ‘Senses are aids to the understanding’ and ‘communication’ of ‘knowledge, wants, and insights’; ‘people don't actually understand sense, but by means of sense’. ‘Language is not the vehicle of understanding or knowledge’ ‘but their mediator, and the arranger of the ideal elements’ that ‘assist comprehension’ (TS 256).

12.63 Since ‘reality’ -- when it is ‘spoken about, designated, objectified, or meant’ -- ‘is a collective term’ for ‘facts’ of many ‘very different kinds’, and ‘language is related to reality’ ‘in multiple ways’, ‘we can formulate the question’ more simply: ‘how is language’ related to ‘types of correlates spoken about?’ (TS 251f) (cf. 13.24). The term ‘“correlate”‘ covers whatever ‘people intend or understand’ an ‘utterance’ to ‘indicate’ (TS 23f). ‘All manifestations of language of any extent are assigned to correlates': ‘the isolated language element’ in a ‘simple’ way, and ‘multiple expression or groups of elements’ in a ‘complex’ way (cf. 12.78f). Hartmann distinguishes three types: the ‘external correlate’, a ‘real thing’; the ‘internal correlate, a meaning’; and the ‘ideal²² correlate’, a ‘classification’ (TS 257ff). ‘Yet the meaning relation is the same for both ideal and real correlates’; ‘the difference lies in the form of existence’. ‘Sign-units are in principle independent of real correlates, to which they correspond by virtue of a referential [Bezug] intention’, yet which they must ‘take into consideration’ (TS 253). ‘The language type decides which forms’ are assumed by ‘signs for ideal’ or real ‘correlates’ (TS 259) (cf. 3.32). Often, those for the ideal ‘correlates are of a special kind’, (‘non-lexical, non-inflected, etc.’), as in ‘polysynthetic languages’ like ‘Eskimo’ (TS 258, n) (cf. 3.53; 4.64; 12.18).

12.64 To fully explore the role of meaning, Hartmann ‘contemplates a comprehensive research project that might place all concern for language under the main heading of semantics’, including not merely ‘language science’ but ‘scholastics’, ‘philosophy’, ‘psychology’, and ‘logic’ (TS 50, 130) (cf. 6.50, 53). Again, he notes that ‘the semantic side’ of language is ‘not prefigured in “nature”, though related to correlates to the degree that classifications correspond to givens outside language’ (TS 36) (cf. 12.58f; 13.24). ‘Once present, analytic signs exist by virtue of referring to correlates despite the principled arbitrariness of labelling (e.g., in semantic fields)’.

12.65 Equally prominent is Hartmann's insistence that ‘since no sign could exist without sense’, and ‘the potential for sense is the purpose of the sign’, ‘linguistics’ must be ‘concerned with sense’, at least ‘as far as sense appears in the form of sign-value’ (TS 185). The old claim that ‘sense’ is not ‘a legitimate object’ for ‘linguistics’ (12.26) was just a protest against ‘research concerning itself too early or exclusively with sense’ (TS 185, 186n). ‘That signs in language have sense appears’ in their ‘constitution’, ‘use’, and ‘achievement’ (TS 186). But since ‘linguistic utterances reflect mixed sets’ (12.35), ‘we can expect different kinds of sense’, as indicated by ‘numericals, pronouns, verb forms, and so on’ (TS 187).

12.66 Again like Wittgenstein, Hartmann stresses that ‘sense values result from usability’ (TS 189) (cf. 12.42). ‘Sense’ is what ‘the individual forms in an utterance’ ‘take on’ when they ‘mediate comprehension’. ‘The meaning of a word is what appears when a unit of comprehension is decomposed, and what contributes to building such units’ (TS 190). ‘Sense types are formed’ when they ‘appear in various positions in an utterance’. Besides, ‘translation reveals that sense can be independent from the forms of elements’; and history reveals that a ‘sense type need not change’ when ‘the word forms it subsumes are changed’ (TS 187; cf. TS 194f) (cf. 3.58; 8.48; 12.31). However, the idea of ‘fixed’ ‘independent sense-units’ and ‘meanings’ (e.g. 4.50) is favoured by the ‘treatment of word-meaning’ in ‘the special-purpose language of linguists’, and by their ‘standard practice’ of trying to ‘make sense “visible” by interchanging sign vehicles’ and ‘noticing what's different’ (TS 190, 193, 196) (4.52; 5.46).²³

12.67 Invoking once again the ‘“langue” and “parole”‘ division, Hartmann proposes to ‘distinguish the meaning a sign has as a component of a language system’, i.e., the ‘usual’ and ‘potential sense registered in the lexicon’, from ‘the meaning a sign has as a component of an actual language manifestation’, i.e. the ‘factual sense’ (TS 191, 188n) (cf. 12.55f; 13.39). Each kind should be ‘a research domain’ of its own, though ‘potential senses’ always presupppose ‘actual manifestations’, even if the latter are merely the ‘syntax and sentences’ within the ‘explanations of words glossed in a lexicon’ (TS 191, 88n).²⁴

12.68 To specify the issues further, Hartmann proposes to ‘divide’ the ‘communicative’ or ‘informational value’ ‘into the nominal value and the understood value’, ‘the nominal’ being ‘the literal meaning’ and thus ‘the decisive structural part in linguistic utterances’ (TS 254f) (cf. 4.24; 5.66; 7.61; 9.97). ‘Speaker and hearer’ (or ‘encoder’ and ‘decoder’) can ‘increase the value far beyond the nominal’, and ‘what one wants to achieve with an utterance can be very different from what one must say to make it understood’. The ‘nominal content of an utterance’ is also given a more decisive role than ‘purpose’, ‘emotion’, and the ‘extralinguistic reactions’ elicited by ‘intonation, irony, allusion’, and the like (TS 172f).

12.69 As we can see, Hartmann goes further than our other theorists in foregrounding and detailing the semantics of language. He is strongly disposed to inquire how far ‘grammar’, in ‘transmitting information’, ‘may run parallel to the structure of sense’, since ‘categories’ and ‘word types’ ‘differ in accord with the purposes of determination’ (TS 240, 232) (cf. 6.24, 26; 9.13; 13.29, 59). Whether ‘grammatical’ and ‘semantic combination’ ‘are equated depends on whether one finds grammar in the sequences typical of languages or in the sense formation that controls these sequences’; and ‘on whether one sees the purpose of the linguistic utterance in its designation or in the form in which it characterizes’ (TS 228). ‘The structure of a sentence’ or ‘message’ can be seen as a ‘second-order sign for its mode of disposition’ and ‘explication’ (TS 49).

12.70 When ‘treating grammatical phenomena’, ‘linguistic analysis’ may inquire ‘how far descriptive methods can be independent of sense’, or, conversely, how far ‘technical formality is controlled by sense’ (TS 227). We may elect to find ‘grammar’ in ‘arrangement’ (‘the language combinations’ that ‘generate expressions’), or else ‘in assignment’ (‘the combination of material and sense within the sign’) -- two aspects’ ‘not adequately distinguished’ ‘in the usual grammars’ (TS 226, 200) (cf. 12.76). ‘Grammaticity'²⁵ would be ‘the fact that all elements in a given bond stand in determining relations’; ‘all statements in grammar’ are ‘interpretations of types of arrangement’ to which ‘a firm sense can be assigned, e.g., subject and predicate, noun and verb’, etc. (TS 90f). ‘Grammar’ can designate either ‘the operational linkage’ of ‘language elements or the scientific registration’ of this linkage (TS 225) (13.45).

12.71 ‘Grammar occurs in language’ as ‘the analytic demonstration of the constitution of sequences': its ‘analytic method corresponds to’ ‘the combinatory formation of expressions and sets in spoken language’ (TS 225f) (cf. 6.59; 12.14, 87). Its ‘paradigms’ are ‘schemas of forms assigned to several elements when placed in an appropriate position’ (TS 118). Yet ‘grammar’ is a ‘knowledge of possibilities’ and ‘deals with’ ‘facts that are not identical’ with ‘events or productions’ in ‘language realizations’ (TS 226). So ‘grammar is a special sector of the theories of formulation, information, and communication’, and is ‘antecedent to the decision of whether something forms a sentence (syntax) or a word (morphology)’ (cf. 2.55; 3.26, 34f; 4.60; 5.51, 53f; 6.45; 8.56; 9.75, 9¹⁵; 11.40, 79f; 12.75, 77; 13.28).

12.72 In the same spirit, ‘grammar’ can be ‘found in the different positions within a series of language elements that must differ from each other’, and is thus ‘an indicator of difference’, a ‘metalinguistic’ ‘symbolism’ of ‘role, position, and form’ (TS 229, 231) (cf. 2.57f; 12.43). This stipulation fits the notions that ‘linguistic sequences consist of positions standing in opposition’; that ‘grammatical gestalts’ are composed of ‘non-identical mutual determiners’; and that although ‘grammar’ is ‘based on recurrence’, ‘multiple repetition of one kind of element’ is ‘excluded’ (TS 44, 233, 229f). But unlike most structuralists, Hartmann asks: ‘how far does one think about oppositions when forming expressions?’ (TS 45) (cf. 2.57). Is ‘“opposition”‘ only ‘a concept from structural analysis’ (TS 44n)?

12.73 ‘To underlie the practice’ of ‘spoken language’, ‘grammar’ must ‘presuppose some ability’ among ‘speakers and communities’ (TS 228) (7.12, 14; cf. 13.49). ‘Grammar is an aid to understanding semantic statements and must be articulated because every partner expects the aid’; yet ‘the intended meaning is still recognizable within limits’ when grammar is ‘inadequate or incorrect’ (TS 148) (cf. 12.44). Normally, though, ‘there is nothing ungrammatical in the human practice of language’, except in ‘what does not yet fully belong to the language’ (speech of ‘children or foreigners’) or no longer belongs (speech of the ‘mentally ill’) (TS 230) (cf. 12.57).

12.74 ‘Since the earliest times, grammar has been an interpretation’ of ‘language-internal’ ‘connected forms described in a functional, deterministic, or set-theoretical manner’ (TS 90). Now, Hartmann is still searching for ways to circumscribe the notion of ‘grammar’, as attested by his largest volume, ‘Theory of Grammar’ (Theorie der Grammatik) (1963b), which ‘probes the general motives for structural elements like opposition, formality, relation, etc.’ ‘well beyond the frame of usual linguistics’ (TS 32f). For example, he abstractly circumscribes ‘grammar’ as ‘the assembly of usual indicators of combinatory possibilities’; or the ‘system’ ‘containing whatever keeps appearing as the same in the formation of sentences’, thus ‘uniting commonality, sameness, form, and recurrence’ (TS 218, 118). ‘Grammar seems ‘difficult’ when it ‘demands ‘numerous possible bonds’ (e.g., in ‘Amerindian languages’), or ‘varied indicators for the same form’ (e.g., in ‘declensions of Indo-European languages’) (TS 218f) (cf. 12.18).

12.75 ‘Grammar’ is closely allied to ‘the special province of syntax’, i.e., ‘sign formation’ and ‘arrangement’; ‘“syntax” is the application of “grammar”‘, the two being on ‘different levels’ with ‘grammar’ ‘one level deeper in consciousness and relevance’ (TS 245, 213, 36; SB 136f). ‘Appearing together’ is ‘characteristic of all original, natural, spontaneous language’ and ‘counts’ as a ‘group with a determining intention’ (TS 216, 240); ‘things can be actually unconnected only through deliberate isolation, as in science’ (TS 216, 2.40) (cf. 3.25). ‘The necessarily syntactical character of all discourse’ ‘emerges wherever syntagmation, determined chiefly by semantics, appears in language formations’ (TS 245). ‘Several symbols of different value are needed’ unless ‘one of the determining elements is clear or understandable by itself’ so that ‘an expression’ can consist of ‘just one word’ (TS 239; cf. TS 59).

12.76 Since ‘the fundamental practice in assignment and referral to correlates’ lies in ‘the formative element (word, affix), the formulation (sentence, expression), and in the formulative sequence (argument), the basis of language manifestations is assignment in arrangement’ (TS 24) (cf. 12.70). ‘Both semantic assignment’ and ‘syntactic arrangement’ are ‘arbitrary’; but ‘the syntactic aspect’ ‘is far more independent from any forms of reality’, and its ‘motivation’ is ‘internal to language’, namely, to ‘symbolize the mode of combination’ (TS 36f). So ‘a second-order arbitrariness appears in sentence formation’, both in ‘the application’ and the ‘disposition of symbols’ (but cf. 2.29). The ‘sign character of sentence formation’ therefore entails ‘two levels of signage’, such that ‘one-dimensional syntactic bonding joins multi-level units’ (TS 37, 60). ‘The syntactic aspect of words “converts” certain object phenomena into contexts’ by ‘varying the determining role of these words’ (TS 61f). ‘The contexts of words and syntax’ in ‘language in no way disturb a fact by following their own structure’ (TS 61). The ‘linear’, ‘sequential character’ of ‘arranged parts’ cannot be ‘parallel’ to the ‘non-linear structure of content’ (TS 39f) (cf. 12.53).²⁶ ‘Syntagmation disposes by comprehensively positing relations': ‘in the semantics of words, a delimiting opposition by means of neighbouring elements’; ‘in the semantics of syntax, however, a constraining disposition (determination) among neighbouring elements’ (TS 37).

12.77 Due to syntax, ‘one finds a fixed definite structure’ in ‘actually uttered sentences’; ‘a sentence has a given structure’ ‘“always” and “everywhere” to the degree that it occurs by virtue of a sentence pattern’ (TS 97) (cf. 3.38; 4.68f; 5.40, 58; 7.51, 90f). ‘The obligatory variation in the sentence vis-a-vis the potential variation in the word demands that a sentence consist of more than one word’ (TS 26) (but cf. 4.67; 5.58; 6.45; 12.75). So ‘as far as the linguist is concerned, syntax is to be approached operationally, as the production of sentences in combinatorial groupings’; and ‘descriptively, as the discovery and analysis of such groupings’ (TS 245) (cf. 13.45). ‘How far word formation or morphology is included is left to the practices of each individual language, e.g., when sentence formation is expressed in word formation’ (cf. 12.71).

12.78 ‘The sentence’ can also be viewed as ‘the manifestation of a communicative and judgmental form’ (TS 61) (3.35; 4.68; 5.57f; 8.56). Whereas ‘in the lexicon’ the ‘assignment to correlates is found’ ‘between word and meaning, signal and concept, idea and thing, and in grammar’ it is found ‘between form and functional meaning, in the sentence’ it is found ‘between sentential sign and determining situation’ (TS 24). The ‘constellation of names in a sentence corresponds to a communicable object that exists’ ‘primarily as a mode of determination’; ‘determination is the structural object of sentence expressions’ (TS 246, i.r.).²⁷ The ‘predication’ is a ‘complete determination’, whereas the ‘attribution’ is a ‘partial’ one (TS 238). ‘If we rank predications by the quantity of determining parts’, we find that ‘the more gets included in an expression, the less a known correlate needs to be given; the less gets included, the more must “go without saying”‘ (TS 239). We must also acknowledge ‘redundance': ‘the multiple grounding of sense in the sign material’, in that ‘an utterance usually has more elements than are necessary for understanding’ (TS 26).²⁸

12.79 ‘The distribution of determining values among elements of the same ranks, such as subject and predicate, is not that the first is “known” and the second is “new”‘, but that ‘the second supplies a role or function of the first’ (TS 241) (but cf. 9.49). ‘Verbs’ are ‘predicate words’ ‘bracketed with subjects’ and have a ‘structural, not ontological class meaning’ of ‘expressing a functional position’ or ‘role’ for ‘a “subject” or “agent”‘, such as ‘activity, state’, ‘motion, action, and the like’ (TS 242f) (cf. 9.46). This can be ‘joined by typical verbal adjunct formations’ like ‘voice’, ‘tense’, ‘aspect, and mood’. ‘But these formatives are ‘properly’ -- i.e, structurally and without recourse to ‘content’ -- language-internal reflexes of class formation’; ‘only secondarily’ are they ‘designations for forms of the real contexts’ that are ‘primary for motivation and content’ (cf. 12.32). Still, ‘factual, mainly anthropomorphic, activities have crept into definitions of the class meaning’ of verbs.

12.80 Again unlike our other theorists, Hartmann does not devote much attention to language sounds, precisely because they present the fewest theoretical difficulties -- no doubt the very reason why ‘the phonological scheme dominated’ and ‘shaped recent linguistics’ and why ‘language science sometimes stopped in the domain of sound structures’ (TS 44) (cf. 13.27). He makes the standard referral of the ‘physically perceptible’ ‘material’ of ‘language’ back to ‘a differential set of sound variances’ (TS 40) (cf. 2.70f; 3.14, 18, 21; 4.29; 5.42; 8.70; 13.26). Here too, ‘language functions by exploiting commonalities’ rather than ‘unique and individual facts': in this case the ‘“purely phonetic” vocal sound’ which ‘belongs in the domain of so-called natural science and is treated by ‘phonetics’ (TS 118, 40). ‘Important is not the produced succession, but the structure in the sequence according to which mere sounds become phonemes’, i.e., ‘sound-classes’ for ‘the smallest basic components’ (TS 40f, 46) (2.69; 4.29f, 33; 5.42f; 6.43; 8³⁵; 13.26). ‘By entering relations’, ‘sounds’ serve ‘certain functions made possible’ via ‘otherness, opposition, arrangement, combination, distinction, complication, system, etc.’ (TS 46, 41). So ‘phonology works with systems of possibilities’, and ‘language is also in its sound basis a varying phenomenon composed of invariants’ (TS 136n, 44) (cf. 12.29).

12.81 However, ‘the sound of the sign vehicle depends on the organs producing it’, and ‘its presence and duration’ depend on ‘an irreversible linear sequence’, ‘a spoken chain of mutually differing vocal tones’, the ‘differences appearing in sound type, tempo, rhythm, grouping, part-wholes, etc.’ (TS 183, 41f) (cf. 2.57f; 12.51). ‘The mechanism of difference’ works not merely with ‘opposition’ but with ‘alternation’, e.g. between ‘opening and closure’ in ‘vowel and consonant’, respectively (TS 44, 183f; cf. 4.34f; 8.68).

12.82 Hartmann portrays the extrapolation from sounds to other levels as a projection based on ‘whole’ versus ‘parts': ‘sounds as repeatable vocal realizations’; ‘syllables as repeatable sound groups’; ‘roots, stems’, and ‘affixes’ in words; and words in ‘sentences’ (TS 105f) (cf. 2.62; 3.27; 4.50; 6.45; 9.30, 9¹⁶; 12.58). To use the same ‘terms’ in phonology and ‘syntax’, we must ‘assume that: (1) certain basic forms of signs, such as opposition, permeate the whole of language, so that words in a sentence can be seen like phonemes in a word; and (2) a separation between the formation of sound, word, and sentence is not justified: the differences lie in content’ and ‘can be arranged in steps or levels’ (TS 41). Although ‘even in the “smallest” manifestation of language we find a surprising complexity of connections in several levels’, this ‘complexing “neutralizes” the divergence of levels by combining them all’ into ‘new wholes’ (TS 264) (cf. 13.57). Therefore, ‘it does not matter if seemingly incompatible things are combined, e.g., inner and outer, content and form’.

12.83 Significantly, Hartmann sees the levels as a ‘visual representation’ and ‘imagistic interpretation’, not as an ‘explanation’ (TS 41). They are devised because ‘the first real manifestations we find in all languages’ are ‘sounds’ we ‘actually hear’, and ‘the sign vehicles’ ‘have no visible manifestation’ (TS 39).²⁹ It should follow that ‘all terms and descriptions suited for the visible must be non-essential and metaphoric’ (TS 40) (cf. 13.33). However, ‘writing is a special derivative stage of visible mediation and preservation of language, materialized in relatively constant forms’ (TS 39). Writing allows ‘vocal sign vehicles’ to be ‘transposed’ into an ‘independent’, more ‘permanent’ ‘medium’, and ‘represented’ in ‘varying degrees of precision’ (TS 178). ‘Pictorial writing’ and ‘ideograms’ are the ‘closest to content’, whereas ‘syllabic or phonetic writing’ are ‘closer to the sign vehicle’; but ‘direct’, ‘reproducible’ ‘conservation of sound’, as on ‘tape recording’, is ‘the only fully satisfactory mode for investigating the sign vehicle in phonetics and phonology’. Historically, of course, ‘a major part of the work in linguistics had to rely on language conserved in the “imprecise” secondary symbolism of writing’ (TS 178f) (cf. 2.23; 4.43f; 6.50; 8.72ff; 9.42f; 13.33). ‘This limitation was not particularly noticed’ as long as there was no other ‘form of the data’ and the main concerns were ‘semantics (content, philology)’ and ‘the categories of grammar’. Even ‘the sound laws’ ‘could only be determined from the regular attestation’ of ‘correspondence and change’ in ‘written form’.

12.84 As we can appreciate by now, Hartmann's major ‘line of argument is directed not to the structures of language, but to the structure in the procedures of treating language’ (TS 4) (cf. 12.5). Its ‘type of structural forms’ yields ‘the best means for bringing a science into a system with others’ (TS 175). He accordingly proposes a scheme for the various ‘general linguistic methods’ (TS 53), which can be explored or combined in many ways (see Fig. 12.1 below).³⁰ The two main headings are ‘obtaining data and evaluating data’. The four modes of ‘obtaining data’ are ‘descriptive’, ‘analytic’, ‘comparative’, and ‘explanatory’; the two modes of ‘evaluating data’ are ‘generalizing’ and ‘interpreting’. Alongside his two main headings he places the ‘metalinguistic study of methods, the theory of language science’ and cites Hjelmslev as ‘the main representative’, though his own work certainly goes here as well. I shall briefly survey each of his six ‘modes’, beginning with those for ‘obtaining data’.

12.85 ‘Descriptive research’ is concerned with ‘documentation’ and ‘designation': it ‘assembles and systematizes discovered language data’ and ‘tells what belongs to what’ (TS 53, 59). ‘Pure description’ ‘means discovering a formulation’ which ‘describes something as such’ ‘without adding anything to what was found’ (TS 58). This is ‘the hardest problem of research’ and ‘the central question of science’, though ‘the idea of pure description is as old as philosophy itself’, and ‘countless descriptive statements have always been made’ (TS 62). ‘How can an object be represented so as to remain “itself” with nothing added?’ (TS 56). It would be possible if we have ‘a method whose categories’ ‘indicate only relations inherent to the phenomenon’. ‘Description’ can examine language in terms of its ‘inherent basic features’, ‘structure-giving immaterialities’, ‘analytic procedures’, and ‘the factual gestalt of manifestations’, in that order (TS 23).³¹

12.86 ‘Analytic research’ is ‘the chief method’ of all and provides ‘access to formality’ and ‘intersubjectivity’, ‘the central criteria of science’ (TS 15, 19) (cf. 12.4, 7, 12, 14, 33, 48, 97; 13.54). ‘Science’ does not ‘contemplate a unitary phenomenon’, but ‘analyses everywhere’ in order to ‘investigate characteristic partial domains’ (TS 17f). So ‘analytic treatment’ is ‘the proper domain of language science’, although ‘the analyst knows the manifestations being investigated are only a part of language’ and ‘do not have their own purpose’ (TS 167, 156). The resulting ‘representation’, though ‘not proper to the postulatable pure phenomenon’, ‘is relevant and useful’, because ‘the forms’ it ‘extracts and classifies’ are the ones that ‘result in the observed phenomenon’ (TS 167). ‘Linguistic analysis’ ‘divides into multiples in order to make statements about unities’; or ‘specifies an exponent in order to characterize a solidarity’ (TS 18, 37) (cf. 6.41).

12.87 In yet another of many parallels, ‘the functional analysis of systems as the mainstay of language science’ is said to investigate ‘the procedures that create language and are themselves analytic in both semantic and communicative ways’ (TS 103; cf. TS 18, 34-38, 203; 12.12). ‘Linguistic analysis is based on the actualized and creative semantic-grammatical analysis or factoring in language, which operates by decomposing contexts or situations’ in order to ‘communicate’ them (TS 104).³² ‘Systemic analysis is directed to the factual events of language and their preconditions’ and is ‘prior to all “applications”‘ (TS 103). ‘Yet analysis is just an auxiliary method’ and becomes ‘an object of research in its own right only when questions are posed about knowledge and its results’. ‘For instance, the analytic method in language research has been criticized on epistemological grounds by those who insist on the “unity of the phenomenon”‘ (cf. 12.38). In this ‘remnant of earlier philosophy’, the ‘justified emphasis on the unity of thought was unjustly carried over to the elements of phenomena’ (TS 104n). The traditional ‘opposition’ makes ‘the real (the thing’) ‘divisible’, and ‘the ideal (the concept)’ ‘indivisible’; ‘but in practice the real can be indivisible (e.g. when dividing a thing makes it into something else), and the ideal can be divisible’ (e.g., when ‘classifying’ ‘decomposes unity into partial insights’).

12.88 Since ‘analysis’ ‘occurs everywhere where scientific results are sought’, it is used in ‘comparative, historic, functional, and descriptive language research’ (TS 102). Being also ‘presupposed by every grammar’, ‘analyses have been made since earliest times’, and ‘recent’ ones are only ‘more rigorous and scientific’; we need to ‘inquire whether older analyses (grammars) are usable for systemic linguistics’ (TS 107, 105) (cf. 12.41, 90). Whereas ‘recent work consciously proceeds from the system’, ‘earlier works’ may ‘retain their value’ via ‘restatements’ that ‘trace back their results to elements and relations in a systemic union’ (TS 107).

12.89 According to ‘Hjelmslev's work’ (engaged in detail in TG), ‘every language utterance made of signs’ is ‘realized’ in a ‘complex whole whose parts’ are ‘evident’ (TS 105). Thus, ‘language elements can be called partial types’ whose ‘main role’ is to ‘be special and to stand in opposition, function or non-function’ ‘relative to each other’. By ‘dividing the partial types’ into ‘those having semantic sense and those making it possible to have sense’, ‘we can define the means of language in a precise and general way’ (TS 106). ‘“Words” are parts of an expression that “contain” a semantic class to the degree that they are elicitable signals for classes’ whose ‘range is not within language itself, but in what is talked about’ (cf. 13.24).³³ ‘Affixes’ are ‘parts of an expression which contribute to forming a semantic class’ yet which are ‘non-elicitable signals for classes’ whose ‘range is a form, e.g., a position’. ‘“Phonemes”, finally, are parts of semantic signs and are usually meaningless (indicate no semantic class)’, ‘but help the sign to be meaningful’; by themselves they ‘designate their distinctiveness’ (TS 106f) (13.26). In this regard, Hartmann suggests they could be ‘formal signs’ ‘with no extent’. But his remarks on ‘immateriality’ and abstraction from ‘time and ‘space’ (12.46ff) might suggest that all forms may have no extent as forms. If so, the descriptive methods defining units and levels in terms of the size of segmentable units would signal a partial relapse into substantialism (cf. TS 31) (cf. 2.16; 13.26).

12.90 ‘Comparative research’ was ‘for a long time’ equated with ‘language science’, ‘especially in Germany’, ‘the classic form’ being ‘Indo-European studies’ (TS 71, 73) (cf. 2.5, 10, 52, 63; 3.19f; 4.1, 73). For Hartmann, ‘comparing is a typical human³⁴ performance’ that ‘asks not about the object, but its sameness or non-sameness’, and is given most ‘clearly’ in ‘the structure of utterances making comparisons’; but ‘every formulation’ offers ‘structural testimony’ of a ‘comparative’ ‘concern for fitting things together’ (TS 71f, 81, 73). We can ‘infer an underlying comparison wherever multiplicity has become a fact, e.g., in all syntax as a complementarity’ of ‘comparing and differentiating’ (TS 72). So Hartmann sees ‘comparing’ as a part not merely of ‘comparative language research’ but of ‘description’, ‘explanation, (e.g. Indo-European studies), interpretation (e.g. Humboldt), analysis (e.g. etymology)’, and ‘generalization’ (e.g. ‘general linguistics’) (TS 73). The modern ‘structural view’ might make ‘older research’ ‘useful by delineating not the elements, but the structures for whose sake the elements are present’ (TS 74nf) (cf. 12.41, 88; 13.27).

12.91 After ‘about 1800’, ‘historic’ ‘comparison’ treated ‘similarities as evidence of original unity’, unless they could be ‘proven to originate in local creation, convergence, or assimilation’ (TS 74) (2.5; 4.73). ‘Sameness’ was thought ‘typical’, whereas ‘deviations were due to historical conditions’. ‘Languages of the same type were called “related”‘, ‘stemming from an original unified form’. The project of stating ‘what manifestation is the same, similar, or related to what other’ led scholars to postulate ‘a basic form rather than a basic language’, because ‘the typical always appears’ in a ‘form’, and ‘the basic language is not reconstructable as a fact’ (TS 74f). This ‘basic form allowed one to renounce reality and notice only formal’ ‘structures that are “repeated” in individual languages’. ‘The question, “how did the ancient Indo-Europeans speak?”, was deflected to the level of the language system': a ‘formal schema reconstructed’ from the ‘features’ ‘left after a process of abstracting, unifying, and rarefying’ (TS 75f). Yet ‘only an exact inventory’ of ‘credible evidence can decide’ ‘how likely it is that the older language form was in principle more abstract and less diverse’. ‘It would also be important to know what chances the basic form’ had to ‘spread’ from its ‘native place’; the ‘spread’ need not entail ‘real events like war, conquest, or migration’ (TS 77n; i.r.). Or, we could ‘investigate why one form was “stronger”‘, e.g., ‘absentmindedness, compulsion, economy, simplicity, cultural superiority’, or greater ‘abstractness or clarity’ (cf. Tauli 1958) (TS 77, 79n).

12.92 ‘Explanatory research’ seeks to ‘find reasons for discoverable manifestations’ and say whether ‘accidental conditions’ or ‘effective rules’ are involved’ (TS 80). ‘Explanation’ can be either ‘internal’ or ‘external to the phenomenon’, as exemplified by ‘systemic’ and ‘historical research’, respectively. Even the ‘systemic’ approach deploys ‘means of description on a different level than the forms to be characterized’, namely ‘the meta-phenomenal level that determines phenomena’, as when we state that ‘all higher life-forms function’ by means of ‘comparison and differentiation’ (TS 81, i.r.). ‘The goal of explanation is reached when states given by a system or structure are furnished as reasons’ (TS 82). In contrast, the historical approach can be ‘an etymological search for connections between word and thing, can emerge from a specialized interrogation foreign to language, and can deal only with primal words which are expressly “similar” to the thing or correlate, e.g., in onomatopoeia’ (TS 151) (cf. 8⁴³).

12.93 The research methods for ‘obtaining data’, which we just reviewed, are presupposed by those for ‘evaluating data’, to which we now turn (TS 108). Hartmann admits that the ‘border’ between these two is ‘fluid’ and ‘soon crossed’, but he sees ‘a clear difference in the treatment of facts': only in ‘evaluation’ are ‘the facts’ or ‘data’ ‘understood as a representation, realization, or sign of something’ ‘trans-factual and inter-phenomenal’, rather than as ‘a goal in themselves’ (TS 109ff) (cf. 12.6). ‘Viewing data as an indicator of states of affairs, relations, or laws’ is likely to be ‘favoured wherever the data are well-known or no longer interesting in themselves’. Still another parallel between science and language (12.12) is suggested: ‘the word is related to the sentence as the single fact is to the multiplicity of facts: the single word allows no interpretation, and can at most name a thing’; ‘the multiplicity in the word-context of a sentence’ ‘results in’ ‘an insight or interpretation that is more than the parts’ (TS 109n) (cf. 2.27; 5.64f, 67, 75ff; 7.82; 11.36, 10²⁰; 13.59).

12.94 ‘As befits the many-sidedness’ of ‘evaluation’, it ‘necessarily unites several levels’ and gains ‘more general’ ‘superior’ ‘knowledge’ (TS 111f). In its full form, such ‘research’ ‘ideally’ ‘demands a combination of sciences’ (Hartmann calls it ‘syntactic’), which may be ‘hardly attainable’ (TS 112f). And we should ‘ideally evaluate from all languages’ and seek out what is ‘general’ or ‘universal’ (TS 121, 118).³⁵ At least ‘the demand for multiple determination’ can be ‘fulfilled’ by ‘dealing with the whole phenomenon, not just parts of it, in its dependencies, oppositions, and relations to others’ (TS 112). Alongside the older ‘pre-scientific’ and ‘extralinguistic’ ‘evaluation’ in ‘philosophy, psychology, theology’, and ‘history’ of ‘language’, Hartmann focuses on ‘formal’ ‘evaluation’ and subdivides this domain into ‘generalizing’ and ‘interpreting’ (TS 113f, 116, 53).

12.95 ‘Generalizing research discovers shared basic types (as in typology), establishes factual similarities (as in Saussure's general linguistics)’, or ‘demonstrates generally necessary features (as in a priori grammar)’ (TS 53). ‘The goal is to pursue insights into language facts by seeking commonalities’ and ‘unity in all realizations’ (TS 116). In ‘seeking formal commonalities’ and ‘features’, ‘modern research’ remains ‘inductive’ ‘even though it entails abstraction into transphenomenal contexts’ (TS 115f, 118, 120). ‘Individual facts in their totality’ are ‘unique’, but may reveal ‘partial coincidences': what makes ‘facts’ ‘partially comp