5. Kenneth Pike¹

5.1 Kenneth Lee Pike's weighty volume (762 pages) Language in Relation to a Unified Theory of the Structure of Human Behaviour (hereafter LB) documents his ‘ambitious’ ‘attempt’ to ‘revise the conceptual framework of language study’ and to foster ‘extensive deep-seated changes in language theory’ (LB 5f). Though published as a whole in 1967, its chapters had been composed (and some published) between 1945 and 1964. Pike made no ‘full revision’ to enhance ‘consistency’, but left the parts largely intact, even ‘sections’ that seem to ‘come out of another age, so fast have battle grounds shifted’ (LB 85n, 424n, 389n, 6, 8). ‘The reader is warned in footnotes’ that ‘points of view in the text have been modified’ or expressly ‘withdrawn’ in Pike's later thinking, and ‘changes’ have been made ‘in terms’, or new ‘views’ ‘adopted, but not integrated into the early chapters’ (LB 10, 424, 232nf). Sections with newer references were inserted next to older ones. Though at odds with the book's title, the disunified quality allows us to follow a gradual evolution spurred by steady ‘confrontation with a wide variety of natural-language data’ (cf. LB 9).

5.2 Pike's ‘total work arose from a struggle to describe empirical data (especially the Mixtec and Mazatec languages of Mexico’) in the absence of ‘a satisfactory basis’ in ‘the current literature’ (LB 5, 34). With a team of students and colleagues, including his wife Evelyn and his sister Eunice, he developed ‘principles’ for ‘the analysis of scores of languages’, chiefly under the auspices of the Summer Institute of Linguistics, which by 1964 had studied ‘more than 350’ ‘languages’ (LB 9) (cf. 5.89; 13.56). Pike himself ‘often’ ‘took the first steps in the analysis of various languages of Asia, Africa, Australia, New Guinea, Europe, and North America’ (LB 29f). Through such uses, the ‘approach’ began to ‘meet many requirements’ on ‘theory and method’ and to ‘provide the theoretical basis’ for notions previously ‘postulated on empirical grounds’ (LB 555n, 398) (cf. 13.43).

5.3 Pike's ‘tagmemic’ approach differed from mainstream American linguistics in many ways (5.6, 30, 35, 54ff, 61f), but most of all in its sheer elaboration and complexity. The organization of language was to be treated in: (a) ‘variable depths of focus’ determining which data or aspects merited attention (5.16); (b) a dyad of ‘approaches’ (etic, emic) to units seen either outside or inside a system (5.22); (c) a triad of ‘views’ (particle, wave, field) on the interrelatedness of units (discrete, continuous, arrayed) (5.31f); (d) a matching triad of ‘modes’ (feature, manifestation, distribution) (5.33); (d) a triad of ‘hierarchies’ (phonological, lexical, grammatical) (5.36f, 39f); (e) a structure of indefinitely many ‘levels’ (morpheme, word, phrase, etc.), arranged chiefly according to unit size (5.34f); (f) a miscellany of ‘styles’ related to social and geographical dialects, social roles, individual personalities, emotions, or voice quality (5.82); and so on. Although Pike gives sporadic examples from many languages, he nowhere fully analyses a discourse in terms of all or even most of these constructs. Their justification rests mainly on theoretical arguments that are sometimes intricate and provisional, as can be expected for so complicated an approach.

5.4 This profusion is partly offset by the absence of familiar schemes and dichotomies. The ‘parts of speech’ are not reconstructed (5.73). Mainly to facilitate data-gathering, the division into ‘langue’ and ‘parole’ is expressly discarded, and language is not separated from non-language, nor verbal from nonverbal (5.7f, 25, 32, 48). The observer is included in the observation, and the analyst in the analysis (5.9, 11, 16, 22f, 36, 44, 71). Form and meaning are handled not in opposition but as two sides of a ‘composite’ (5.48, 64, 76). And above all, language is ‘unified’ with ‘human behaviour’, as the book's title portends.

5.5 Pike's ‘tagmemic’ approach seeks an ‘oscillation between theory and method rather than a one-way priority’ on either side (LB 509) (cf. 13.42ff). On the method side, the ‘principles’ are intended ‘as sign-posts’ for ‘structures to be discovered’ and as ‘exploratory tools for further work’ (LB 518; cf. LB 70). A ‘multiple-stage and multiple-level approach’ should ‘reflect’ ‘the practical working procedure of every practising descriptive linguist, even of those who most vigorously attempt to eliminate’ or ‘reduce to a minimum’ ‘cultural references’ in ‘linguistic analysis’ (LB 215, 59). Pike doesn't ‘object to an “as-if” procedure’ of ‘temporarily or deliberately ignoring data’, but only to ‘forgetting’ that one's ‘description’ ‘is limited in validity by the initial selection’ and ‘insisting’ that the ‘selection’ is ‘the only’ ‘scientific one’ (LB 59) (cf. 5.35, 57). In ‘practical fieldwork’, even the ‘methodologically helpful’ tactic of ‘working with “cleaned-up text”‘ and ‘sentences’ ‘separated’ or ‘dictated’ ‘by the informant’ can be ‘fatal’ for ‘theory’ (LB 571; cf. 5.12f).

5.6 A ‘theory may be viewed rather broadly as a statement purporting to describe, or to explain, or to help one to understand a phenomenon’, not merely to ‘present a claim of truth, or assert relationships between phenomena, or predict the occurrence of phenomena’ (LB 68) (cf. 12.84-98). This goal requires ‘a unified theory’, ‘set of terms’, and ‘methodology’ for ‘analysing’ ‘any kind of complex human activity’ ‘without sharp theoretical or methodological discontinuities’ (LB 26). In contrast to ‘American writers’ who ‘place priority on dichotomous constructions’, Pike expects ‘theoretical advances’ from an ‘emphasis upon unity’ (LB 358; cf. 5.35f, 38, 54, 63).

5.7 ‘As more and more materials in speech begin to appear structured, the traditional view that “language” as a structure differs from “speech” as activity is threatened’ (LB 536). So Pike ‘abandons the distinction between “langue” and “parole” proposed by de Saussure’ (cf. 2.20; 6.33, 46; 7.12; 8.30; 9.5; 12.12, 26, 47, 55, 67; 13.36). He extols ‘the strength’ of ‘a theory of linguistics’ ‘which has a few theoretical constructs applying to the whole system of human behaviour’ and which can ‘portray adequately the universe as a system in which units can interact’ (LB 547; cf. 5.38f, 89). He braves ‘the dangers of leaving one's own discipline’ in search of ‘analogies between linguistic structure and the structure of society’ (5.84), and advocates ‘unity’ among ‘linguists, archaeologists, ethnologists’, ‘anthropologists, sociologists, and students of personality’ (LB 6, 641) (13.35).² Though one might ‘come from the opposite direction’, e.g. from ‘ethnological theory’ into ‘linguistics’, ‘recent formal studies in the linguistic area’ offer ‘a base which at the moment is easier to build on’ (LB 6f) (cf. 2.7; 6.10, 51; 8.6, 12, 33; 11.5; 13.21).

5.8 ‘For best results, linguists’ should ‘raise their focus’ to ‘a high level of abstraction’ and ‘generalization’ by ‘treating language data within a single hierarchical structure along with nonverbal activity’ (LB 111; cf. LB 26, 6, 120; 5.37, 60 78f; 8.43; 9.42, 49; 11.7, 9, 86). We should ‘give for the total event, as a unit, a unified description’ which ‘would simultaneously analyse and describe non-linguistic behaviour as well as the smallest and most intricate elements of linguistic structure’ (LB 26). Motives for this demand include: ‘language and non-language behaviour are fused in single events’; ‘verbal and nonverbal elements’ may be ‘interchangeable’ or ‘substitute structurally for one another’; the respective ‘structures’ are ‘partly alike’; and ‘language behaviour’ ‘obtains its structuring only in reference to’ the ‘larger behavioural field’ (LB 26, 30, 32, 35, 68; cf. LB 26, 134f, 120). Ultimately, a full ‘description’ might ‘allow an outsider’ ‘to act as would a native member of the culture’ (LB 121n).

5.9 Pike ‘insists that an observer component enters into all ability to discover, understand’, ‘report on, or act as a member of a community’, and forms ‘part of any equation where perception is involved’ (LB 58, 659). The main tactic of the ‘analyst’ is therefore to ‘observe regularities of sequence in events’ (LB 156). ‘The study of observables’ transcends any ‘solipsistic’ confinement to ‘the study of one's own speech’ and fosters a ‘belief in the possibility of empathy with other persons’ (LB 289). ‘The scholar's observation and understanding of his own behaviour’ can then enhance the ‘fruitfulness’, ‘elegance, and simplicity’ of the ‘approach’ (LB 663). Admittedly, ‘one cannot guarantee uniformity of judgment in natural language’ when ‘observer intuitive components must enter analytical procedure’; though not ‘unique’, ‘solutions’ may be at least ‘alternative, simple, and mechanically convertible’ into each other (LB 259n, 289) (cf. 8.35; 11.90; 12.55; 13.3).

5.10 Since ‘language’ cannot be ‘analysed or described without reference to its function in eliciting responses’, we should consult ‘the normal expected response of the community’ (LB 40, 27). Pike ‘insists’ that ‘explicit observable reactions’ constitute ‘data’ and ‘objective evidence’ (LB 352, 63, 67, i.r.; cf. 5.15, 70; 8.25, 41, 91; 11.16). Citing the concern of ‘psycholinguistics’ for ‘“publicly observable indices of subjective events”‘ (Carroll 1953:72), he discounts demurrals that ‘“the linguistic processes of the ‘mind’ as such are quite simply unobservable”‘ (Twaddell 1935:9), and that ‘“the native speaker's feeling”‘ ‘“is inaccessible to investigation by the techniques of linguistic science”‘ (Bloch & Trager 1942:40) (LB 351f, 66f) (cf. 4.8f; 7.9ff; 8.24; 12.38).

5.11 Problems do persist, because, as Pike concedes, ‘observable’ ‘clues’ are ‘a usable’ but not a ‘precise criterion’ (LB 81) (cf. 4.8; 5.11, 13, 16, 18, 27f, 30, 43). ‘Responses’ may be not ‘immediate’, but ‘delayed for a much longer time’, or may not be ‘uniform’, or may be either ‘conscious or unconscious’ (LB 43, 66, 83). A ‘relativity of phenomena’ can arise when ‘equally true’ ‘observations come from different standpoints’ and ‘observers’ (LB 659). ‘Observers’ may ‘differ’ in ‘ability’, ‘training’, ‘hearing’, ‘memory span’, ‘attention’, or ‘interest’ (LB 46f). One's own ‘cultural background’ may foster a ‘bias’ toward ‘familiar’ ‘events’ or make one ‘notice’ ‘alien’ things ‘native participants do not’ (LB 45f) (cf. 3.50; 4.4, 72; 5.24, 78; 8.14; 13.1).³

5.12 Pike's own methods stemmed from using a ‘monolingual approach’: working ‘without written or translated documents and without an interpreter’, and relying heavily on ‘gesture’ (LB 29f, 34f, 40, 61, 225, 601). There, ‘the interdependence of language and non-language behaviour’ is ‘striking’ and must be reflected in ‘the analysis’ and ‘description’ (LB 29). ‘Working through’ the ‘interlacing problems’ of ‘data’ helps ‘make concrete’ the ‘principles of linguistic structure’ (LB 215). The ‘student’ who ‘views data from many viewpoints’ can ‘treat the problem as a whole’, and ‘dive into the structure’ (cf. 5.23, 25, 89).

5.13 Even using a monolingual approach, it is hard to ‘avoid encroaching on the innate linguistic naivity of the informant’ (as ‘recommended by Bloomfield’ [1942]) (LB 62) (cf. 4.6; 13.47). ‘Verbal’ ‘data’ in ‘normal cultural settings’ may become ‘abnormal’ ‘due to the intrusion of the linguist’ (LB 68). ‘Too rigorous a direct attempt to observe persons’ or ‘test’ them may ‘change their reactions’ by making them ‘self-conscious’ and ‘forcing’ ‘items’ into ‘awareness’ -- a ‘tool for a psychoanalyst’ perhaps, but a ‘detriment’ for a ‘linguist’ (LB 191, 238, 159, 657; Zipf 1935:12). When asked ‘for the meaning of words’ by a ‘language analyst’, ‘informants’ may be ‘unable to express’ the ‘meanings’, or ‘misinformed’, or even ‘deliberately deceptive’ for the sake of ‘psychological comfort’ (LB 90, 156) (cf. 13.49). Their ‘observational and analytic ability may be poor’; or their ‘variation of purposes’ may render their ‘reports confused or conflicting’ (LB 156).⁴

5.14 The modest tactic of asking informants ‘whether elements are the “same” or “different”‘ (cf. 4.26, 31; 5.61, 65) already entails ‘difficulties’ because ‘the reaction’ is ‘an exceedingly complex process’, not a ‘simple’ ‘“yes” or “no”‘ (LB 61f, 223f). Here too, ‘the ‘questioning’ adds a new ‘context’ and may ‘change the structure we were investigating’ or ‘destroy the naivity’ of the ‘informants’ by making them ‘quasi-analysts’ (LB 159, 224, i.r.). ‘An elaborate gestural situation’ or ‘specific linguistic training’ may ‘help the informant identify the level on which the linguist is working’, but may also ‘destroy the apparent simplicity of the procedure’ (LB 62). Do the ‘units’ ‘differ by their distribution’ in ‘larger units’, ‘by their structural function within the total system’, or by their ‘purpose or meaning’ (LB 160)?

5.15 Various solutions are aired for these dilemmas with informants. Pike mistrusts the ‘elaborate distributional substitute for this procedure’ proposed by Harris and others (LB 613), because of the unacknowledged steps involved (cf. 5.61). A better tactic is to ‘check’ one's ‘conclusions’ ‘against fresh data unbiased by such questioning’, e.g., to ‘gather comments when’ ‘the actor’ ‘is “off guard”‘, or to examine ‘a body of textual materials’ obtained from ‘recorded conversation’ among ‘other speakers’ (LB 159, 656). Even ‘more objective’ is to ‘study the kinds of slots in which activity occurs’ and to look for relations between an ‘activity’ (as a ‘cause’) and a ‘response or sequel’ (as an ‘effect’) (LB 90f, 663; cf. 4.8; 5.10, 47, 49ff).

5.16 Observation vitally depends on ‘focus’, i.e., what a person attends to. ‘Observer status affects the focal hierarchy’, and any ‘change in focus is necessarily accompanied by a sharp change in observer attitude or participant type’, especially when acting as ‘analyst’ (LB 107, 111). ‘Individual differences’ imply an ‘indeterminacy of focus’ (LB 80). Depending on their ‘attention’, ‘interest’, or ‘concentration’, ‘participants’ may raise focus to the ‘whole’ or ‘lower’ it to ‘a shorter sequence’ (LB 110, 79f). ‘We can never be certain’ what is ‘essential’, ‘since any particular person may be acting analytically rather than participating’, ‘and this difference of purpose’ affects the recognition of ‘units’ (LB 80). Whereas an ‘ordinary participant’ does not ‘normally go’ beyond certain ‘lower limits’, ‘an analyst’ might do so, e.g., as ‘a linguist’ ‘quoting’ ‘items’, or as an ‘observer’ ‘noticing’ only some ‘grammatical infelicity or noncurrent pronunciation’ (LB 80, 155; cf. LB 129; cf. 5.36, 46). ‘Teaching situations’ also ‘change the focus’ and ‘thresholds of attention’ by ‘focusing attention on details which later must drop below the threshold’ (LB 292f, 154f)

5.17 Pike applies the term ‘hypostasis’ when, ‘in order to give it separate analytic attention’, ‘any unit of activity is abstracted’ from the ‘purposive activity sequence expected by regular participants’ (LB 107) (cf. 5.36, 76; 6.16; 8.71). We can ‘try to pick a predominant focus unit’ inferred from ‘cultural evidence’, or ‘start at some arbitrary but convenient’ ‘threshold’ (LB 106, 153, 293). ‘In a systematic hypostasis’, ‘scientists’ may ‘arbitrarily set upper and lower limits’ and ‘concentrate on a smaller body of data’ -- whence ‘the difference between disciplines’ (LB 111) (cf. 9.3-9). But ‘normal’ and ‘hypostatic verbal systems may be’ only ‘partially congruent’ (LB 107, 155).

5.18 Accordingly, a ‘complete theory’ ‘must include the theorizing of the theorist’ and distinguish between ‘analytical and nonanalytical’ ‘activity’ (LB 222) (cf. 5.13f, 16, 20, 36, 46; 13.36). ‘The analysis of words or sentences outside of normal behavioural contexts’ ‘itself constitutes an activity’ meriting ‘analysis’ ‘in structural terms’ by ‘students of human behaviour’ (LB 134; cf. 5.54). ‘In some situations’, ‘the activity of bringing an item into focus’ can be ‘partly observed’ as a ‘physical act’ (e.g. ‘turning’ one's ‘head or eyes’), but not when ‘inward concentration’ or the ‘use of memory’ is involved (LB 112f). So ‘the objective study of concrete verbal utterances’ of ‘participants’ must be accompanied by ‘tests to determine the accuracy of one's powers of discriminating’ (cf. 13.36). After ‘the intuitive steps’ whereby ‘a system is arrived at’ through ‘guess and check’ ‘procedures’, ‘description’ should not ‘present’ ‘only’ ‘the formal part’ of ‘the data’; instead, we should ‘attempt to understand and systematize this heuristic’ of ‘intuitive’ ‘steps’ as a ‘possibility of knowledge’ (LB 225n, 317n) (cf. 7.9ff). This project would be ‘profitable’ provided ‘its results and procedures are not allowed to vitiate the results of an analysis of normal participant activity’ (LB 134).

5.19 One key problem of focus and analysis is to decide ‘“the proper size”‘ of the ‘“unit”‘ and to ‘differentiate’ ‘units of size’ (LB 96, 42; cf. Zipf 1935:12) (cf. 13.29). In a ‘unified theory’, ‘large and small units’ should be ‘mutually defining’: ‘minimum’ ‘units’ ‘can be defined only as relative points in the larger units and systems’; and ‘one must start with some knowledge’ ‘of large units before studying smaller’ ones (LB 150, 120; cf. LB 72, 142; 11.32; 13.57). The lower limit can be set by consulting ‘meaning or purpose’, ‘cultural’ ‘relevance’, and ‘observable native reactions’, rather than relying on a ‘structural regress from the point of view of physiologist or biophysicist’ (LB 130, 83, 304, 306, 409; cf. 2.31, 68; 3.9; 4.18, 32; 5.15, 28, 42; 6.7, 54). For the ‘top limit’ on ‘maximum’ units, Pike can ‘find no evidence’ for ‘setting up theoretical limits’ (LB 130; cf. 5.51).

5.20 An equally ‘difficult’ ‘theoretical problem’ with units is ‘the balance’ between ‘giving priority to relationships’ or to the ‘items’ they relate, e.g., to ‘oppositions’ or their ‘poles’ (LB 179, 358). A ‘theory will look very different’ depending on which receives ‘emphasis’ (LB 358). Pike's basic tendency is to ‘place attention on units’: ‘elements viewed’ as ‘whole entities set apart’ (LB 9) (5.32). In his opinion, ‘we perceive a structure being made of units’ as ‘parts of a system’, just as ‘“the speaker acts as if he were using units which start and stop”‘ (LB 271f; Longacre 1964a:14; cf. 5.26). ‘“Purely relational units”‘, in contrast, might ‘“have nothing to relate”‘ (LB 357; Vachek 1936:38). These arguments oppose the ‘glossematic’ claim that ‘to the scientific view, the world’ ‘consists’ ‘only of functions’ (LB 271, 282) (cf. 6.25, 28; 12.25; Uldall 1957). Yet Pike doesn't mean to ‘imply that units exist apart from their occurrence in any relationship’; ‘the present theory’ merely ‘interprets’ ‘relationships’ as ‘conceptualized hypostatic constructs’ of ‘the analyst’ (LB 282).

5.21 Pike's ‘interest’ thus lies ‘in a system of units’, not ‘of oppositions’ -- a point of conflict between ‘American’ and ‘European’ work, respectively (LB 345, 358) (cf. 2.57f, 70, vs. 4.45). He questions the idea of ‘“binary oppositions”‘ being the only ‘“distinctive”‘ kind, or ‘the “most advantageous way of coding any verbal behaviour’, or the ‘“child's first logical opposition”‘ (LB 359; Jakobson 1949, 1962; Jakobson & Lotz 1949; Jakobson & Halle 1956). He foresees ‘confusion’ when ‘binary oppositions as wholes’ get ‘treated as distinctive features’, or when ‘absence’ as ‘one pole of an opposition’ ‘is treated as essentially present’ (LB 348, 358; cf. Note 12). ‘Binary emphasis’ also makes ‘immediate constituent analysis’, which seems ‘intuitively’ ‘valid’,⁵ ‘extremely difficult to handle empirically’ (LB 477) (cf.4.60ff, 5.50, 62; 7.36f; 9.33). In Pike's ‘theory’, ‘immediate constituents’ are not ‘the point of initial attack’ as in ‘current linguistics’, but ‘the end product of analysis’; and ‘binary’ ‘end products’ are a just ‘special instance’ among all kinds of ‘series’ (LB 477, 444, 244; cf. Pike 1958). ‘Starting from unity’ is better than the ‘traditional’ ‘starting’ by ‘looking for “cuts” in the string of materials’ (LB 478; cf. 5.6).

5.22 To reform and refine linguistic analysis, Pike found it ‘convenient -- though partially arbitrary -- to describe behaviour from two different standpoints’, whose ‘results’ ‘shade into one another’ (LB 37). ‘The etic viewpoint’ is ‘an essential initial approach’ that ‘studies behaviour as from outside of a particular system’; it ‘treats all cultures or languages, or a selected group of them, at one time’.⁶ ‘The emic viewpoint’ ‘studies behaviour as from inside the system’, sees ‘every unit as ‘functioning within a larger structural unit or setting in a hierarchy’, and ‘treats’ ‘only one language at one time’ (LB 37f). ‘Etic units’ can be ‘created by the analyst’ or can come from ‘broad samplings, surveys’ or ‘training courses’, and may thus be ‘available in advance’ of the ‘analysis’ (LB 37f, 55). ‘Emic units’ ‘must be determined during the analysis of the language; they must be discovered, not predicted’.

5.23 Of course, we have to ‘assume a philosophy of science’ ‘granting that in the universe some structures occur other than in the mind of the analyst’ (LB 38).⁷ ‘Structure really exists in language’ ‘“as much as any scientific structure really obtains in the data which it describes”‘ (Harris 1954:149) (LB 56). ‘This viewpoint’ ‘does not rule out alternate descriptions’.⁸ If ‘“the constructs were in the metalanguage”‘ only, it would not be sensible to ‘“look for behavioural correlates or psychological reality”‘ (LB 72; Saporta 1958:328) (cf. 5.23; 11.42, 10³²; 13.57). Moreover, if ‘the linguist’ ‘denies structure’, ‘his own statements, descriptions, or rules’ must be ‘without publicly available structure or ordering; linguistic statement comprises a subvariety of language utterance, and hence can have no structure if language has no structure’ (LB 38) (cf. 13.48).

5.24 ‘An etic system may be set up by criteria or a “logical plan” whose relevance is external to the system’ or comes from ‘partial information’ (LB 38). ‘Emic’ ‘criteria’ must be ‘relevant to the internal functioning of the system’ and hence ‘require a knowledge of the total system’. So ‘etic data’ are ‘tentative’ and ‘preliminary’, whereas ‘emic data’ are ‘refined’ and ‘final’ (LB 38f). Still, ‘etic and emic’ may ‘often’ be ‘the same data from two points of view’ (LB 41). ‘Many’ ‘etic units turn out to be emic’; or, when ‘emic units’ are ‘compared’ from one language to another, they ‘change into etic’ by being ‘viewed as generalized instances of abstract stereotypes, rather than as living parts’ (LB 41f, 75).⁹ Moreover, the ‘ultimate’ ‘replacement’ of an ‘etic description’ by a ‘totally emic’ one, though foreseen ‘in principle’, ‘probably never’ occurs ‘in practice’ (LB 39). Still, ‘emic procedures’ help to ‘eliminate’ ‘etic’ ‘distortion’ or a ‘margin of error’ in the ‘preliminary recording and analysis’ due to a bias toward or against ‘Indo-European languages’ like ‘Latin’, or to ‘over-recording’ ‘more elements than can be relevant’ for ‘the whole system’ (LB 182, 142; cf. LB 72, 213, 141, 173; 3.5, 50; 4.4, 72; 5.11, 60, 78, 92; 8.24).

5.25 Some units are ‘etically’ ‘similar but emically different’ (LB 43, 47, 105).¹⁰ Far more are just the opposite, because ‘etically, each repetition’ of ‘any unit’ is ‘distinct’ in respect to ‘absolute physical differences’ (LB 44) (cf. 5.19, 28f; 7.91; 13.45). ‘Delicate measuring instruments’ show it's ‘impossible to repeat any movement’ ‘exactly’, and ‘every movement’ ‘differs etically according to the sequences’ wherein ‘it occurs’ (LB 316, 44ff, 164). ‘Variations’ may also happen ‘below the threshold of perception’ (LB 45, 87f). So we could have ‘an infinite number of etic differences’, while ‘the emic unit is a composite of all’ (LB 87) (3.3; 4.13; 8.42). ‘The investigator’ should thus assume every ‘form of purposive activity’ to be ‘a variant of an emic unit’ (or a ‘part’ or a ‘sequence of units’); what seems ‘random’ is ‘not structureless’ but is ‘emically unanalysed’, or shows ‘a greater range of variation’ than other forms, or ‘does not forward the purpose of the activity’ ‘in focus’ (LB 518, 115). In this spirit, ‘the linguist’ should try to ‘use any observable data’ to ‘discover the emic units of a language’, including ‘extralinguistic’ ‘actions’ ‘eliciting’ or ‘resulting in speech’ (LB 68).

5.26 The special ‘value’ of ‘the etic approach’ is to ‘give to the beginning student’ an overview of ‘the kinds of behaviour occurring around the world’ and to enable a ‘faster handling of the data of an unfamiliar language’ (LB 40, 182). Pike envisions an ‘etic “lens”‘ making ‘tacit reference to a perspective oriented to all comparable events’ ‘of all peoples of all parts of the earth’; but a more modest hope is that ‘sufficient uniformity throughout the world’ will enable ‘the analyst’ ‘to make early guesses’ (LB 41, 176) (cf. 5.84; 9.18; 13.38). The special ‘value’ of ‘emic study’, on the other hand, is to show ‘a language or culture’ ‘as a working whole’ and to ‘help one understand the individual actors in such a life drama -- their attitudes, motives, interests, responses, conflicts, and personality’ (LB 40f). We might strive toward a ‘predictive science of behaviour’; ‘even statistical’ ‘studies’ require that ‘homogeneity in behaviour’ be ‘emically defined’.

5.27 A particuarly compelling motive for the ‘etic’ viewpoint is Pike's loyalty to physicalism (cf. 4.8; 6.26; 7.15; 8¹⁴). In fact, ‘physical’ can appear instead of (or with) ‘etic’ as the counterpart to ‘emic’ (e.g. LB 43f, 87, 89, 99, 105, 120, 151, 164, 168, 677). By using an ‘etic physical description’, ‘emic structural units’ can be ‘presented’ ‘not only as algebraic points’ in ‘a structural system, but also as elements physically described’ (LB 39, 120).¹¹ Indeed, ‘we never’ ‘completely “abstract” a behavioural emic unit’ ‘away from the actual physical action’ (LB 89, 187, 645). This loyalty explains the preoccupation with ‘the physical setting of society’ and ‘language’, witness such insistent locutions as ‘physical dwellings’, ‘physical clothing’, ‘physical bodies’, etc. (LB 121, 128, 169, 645, 658; cf. 5.82). Language entities are also said to have a ‘physical order’, ‘position’, or ‘place in the uttereme’ (LB 209, 281, 246, 253, 511, 251, 457, 196) (cf. 13.33). Pike admits ‘all “facts”, all “things” reach’ human beings ‘only through perceptual psychological filters’ for ‘the physical data he observes’ (LB 645), but such data seem to make him feel most comfortable (cf. 5.84).¹²

5.28 If ‘each overt or covert type of physical movement must ultimately enter into the analysis of activity’ (LB 112f), we might end up considering events down through ‘muscles’ to ‘molecules’ and ‘atoms’ (cf. LB 90, 106, 111, 118, 130, 292f, 306, 365n, 393n, 516, 590, 660, 662f) (cf. 13.45). Pike recognizes an ‘insoluble’ ‘difficulty’ for ‘science’ here if the search for ‘minimum units’ enters such ‘an infinite regress’ (LB 303fn). It would be a ‘reductio ad absurdum’ to ‘treat the movement’ of a ‘molecule’ in a ‘nerve fibre’ in a ‘muscle’ of ‘the vocal organs’ as a unit (LB 130). And ‘the study of the movement of muscles’ is ‘more complex than’ ‘the anthropologist's’ ‘techniques of observation can handle’ (LB 304). Anyway, ‘describing’ something as ‘a concatenation of molecules’, as a ‘physicist or chemist’ might, still entails ‘the emic structuring of a person -- the scientist’ and his whole ‘cultural history’ upon which his ‘“understanding”‘ depends (cf. 4.13); so ‘the observer is not bypassed by going to the microscopic level’ (LB 660). We should be content to ‘treat subperceptual variants’ only ‘in special studies’ of ‘a small amount of data’, e.g., a ‘microscopic and physiological analysis’ of ‘movements even down to submolecular size’ (LB 88, 292). But when ‘analysing behaviour as the actors themselves react to it’, such minuscule actions ‘appear noncontrastive’ (LB 292f, 297).

5.29 ‘The emic analytic process’ also has its ‘problems’ (LB 95). It ‘must deal simultaneously with emic units as discrete parts’ both ‘of a system’ and ‘of sequences’ within ‘complex events’. ‘Models of the emic unit’ as ‘a mere sum’ of ‘separate parts’ ‘must be rejected’ (LB 513) (cf. 2.29). In line with his devotion to observed behaviour, Pike says ‘the basic problem’ for ‘linguistics’ and for ‘epistemology and theory of perception’ is to find ‘a theory and procedure for analytically breaking up the physical etic continuum into a sequence of discrete emic units’ (LB 94).¹³ ‘Sharp points of change’ inside a ‘continuum’ or ‘fusion’ may be ‘impossible’ to ‘set up in theory or practice’ (LB 95; cf. 5.47, 77). ‘No absolute physical criterion’ or ‘simple measurement’ can apply when ‘physical variants overlap’ (LB 94). Moreover, ‘a complete description of the manifestation modes of an emic unit’ could include’ an ‘infinite number’ of ‘actual and potential variants’ (LB 88) (cf. 5.25).

5.30 Such factors render Pike uneasy about the American trend whereby ‘the approach through item and process was largely replaced’ with one through ‘items and their arrangements’ (LB 502, 556) (cf. 7.75). ‘In this respect, linguistics’ ‘was moving counter to the general stream of the philosophy of science’, which was ‘“thinking more in terms of process”‘ than ‘“of things”‘ (LB 557; Sinclair 1951:80). Carried too far, either ‘approach’ ‘leads to serious problems’ and ‘distortions’ (LB 550, 547). The ‘item-and-arrangement approach’ entails ‘arbitrariness’ in ‘requiring sharp-cut segmentation’, even if this tactic has ‘advantages’ for ‘dealing’ with ‘actual utterances, not constructs’, and is ‘effective’ for ‘listing morphemes separately, as in a dictionary’ (LB 551, 553, 558; cf. Lounsbury 1953:12, 15). ‘The resulting discrete localizations of meaning’ ‘may be highly unrealistic’ (LB 558; cf. 5.65; 6.47f; 13.51). And ‘the approach may be forced to list laboriously every stem’ ‘as if no regularity could be observed’ (LB 551). ‘The item-and-process approach’, on the other hand, entails the ‘distortion’ of ‘setting up’ ‘“norms”‘ as ‘convenient theoretical starting points’ which ‘are at times arbitrarily chosen’ or even ‘imply’ an ‘irrelevant normative judgment’ (LB 553) (cf. 13.27). The ‘approach’ also ‘implies’ that ‘forms’ ‘“first” come together in sequence and “then” are modified’; this ‘pseudo-history’ overlooks the prospect that the ‘unmodified forms’, though ‘present in the description’, ‘may never have occurred in sequence in the observed data’ (LB 552; cf. 5.87; 7.48, 51; 13.54).¹⁴

5.31 To incorporate item, process, and arrangement, Pike finds it ‘highly attractive’ to combine ‘three separate technologies’ or ‘theoretical concept sets’: a ‘static view’ of ‘a sequence of units’ as ‘discrete segments’ or ‘particles’; a ‘dynamic view’ of ‘a sequence of units’ ‘treated as waves flowing into one another’; and a ‘functional’ view of a ‘field’ of ‘complex but unanalysable units’ ‘with unpredictable unitary characteristics’ (LB 468, 511n, 545f, 563, 553). ‘The particle view’ can use ‘psycholinguistic data on segmentation’, ‘the wave view’ can use ‘physical data on continuous articulatory movements or sound waves’, and ‘the field view’ can use ‘a theory of classes and systems of phonemes, morphemes, and tagmemes’ (LB 513).

5.32 In the 1954-60 sections of the book, Pike ‘gave special attention to particles’ and ‘less’ to ‘waves’; later he turned to ‘fields’, e.g., in his ‘matrix studies’ (LB 512n).¹⁵ He extols the ‘field’ view for ‘helping explain’ how ‘some high-level units’ ‘are semantically relevant as a whole’ (e.g. ‘compounds’ and ‘phrasal idioms’) (LB 554) (cf. 2.61; 4.60; 5.54). The ‘field’ view also stresses the role of ‘reacting to a pattern’ within ‘the theory of learning’: ‘the child’ can ‘learn complex expressions as wholes’ before being able to ‘manipulate substitutable items’ (LB 554; cf. LB 547) (cf. 9.11). ‘In practical language learning it is often simpler’ ‘to memorize a few “exceptions” than to learn complex analytical descriptions’ not vital to the ‘recurrent pattern in the language’. However, the field view will need some new ‘techniques’ in ‘theory and method’ to handle ‘non-segmentable’ ‘junctions’, ‘mixtures’, ‘diverse’ ‘inventories’, ‘disparate’ ‘fusions’, ‘concentrations of energy’, and so on (LB 555). And we must not ‘adopt the field approach’ exclusively, lest we ‘no longer find’ ‘language’ ‘in our data’. As we will see, most of the book is in fact dominated by particle views.

5.33 Pike matches his triad of views with a triad of ‘modes’, defined as ‘distinct’ ‘simultaneous structurings’ of ‘activity’ (LB 86, 93, 513).¹⁶ The three ‘modes are not parts nor pieces of the whole: they each comprise the entire substance’ or ‘physical data’ (LB 93, 86). The ‘modes’ are also not ‘merely’ points of view’, but ‘reside in the behavioural data’ (LB 514). The ‘feature mode’ shows units ‘statically’ as ‘discrete particles or segments of activity’; ‘the manifestation mode’ shows them ‘dynamically’ in ‘continuous waves’ (as ‘simple vs. complex’, ‘free’ vs. ‘conditioned’, or ‘fused vs. clearly segmented’); and ‘the distribution mode’ shows them ‘functionally’ in a ‘total field’ (LB 511, 463f). The three modes cover ‘(a) contrast or identification’, (b) ‘complementation or free variation’, and (c) ‘distribution or ‘class membership’, respectively (LB 85f, 426, 510f).

5.34 Pike also distinguishes ‘levels’,¹⁷ each one ‘representing some phase of structuring in the material examined’ (LB 480). As units on respective ‘levels’ he mentions (in descending order or size) ‘conversation’, ‘topic’, ‘monologue’, ‘utterance-response’, ‘utterance’, ‘sentence’, ‘clause’, ‘phrase’, ‘word’, ‘morpheme’, ‘stem’, and ‘phoneme’ (LB 441ff, 437f, 517, 362). These ‘levels’ ‘are quasi-absolutes’ ‘in that etic criteria’ can ‘differentiate them’ (LB 437).¹⁸ Yet ‘the levels’ are ‘still somewhat relative’ in ‘specific details of the available criteria’, and in the ‘numbers which are structurally relevant to any one language’. The ‘crucial’ ‘requirements’ are that ‘unit types on one level’ must ‘control the occurrence and relative (fixed or free) order of included constituents, and be structurally organized’ ‘sharply in contrast’ to the ‘next higher or lower’ level, despite some ‘indeterminacy’ of levels’ reflecting ‘built-in indeterminacies in the system’ (LB 482; cf. 5.37, 39, 45, 53; 13.27, 57).

5.35 So despite all ‘contrastive’ ‘criteria’, ‘levels’ cannot be studied in the ‘complete separation’ some American linguists demanded in the belief that ‘rigid, water-tight compartments or levels are aesthetically satisfying and provide the only valid scientific conclusions’ (LB 443, 59f; 66; cf. Hockett 1942, 1955; Moulton 1947; Trager & Smith 1951) (cf. 3.60; 7.46; 13.27). They excoriated the practice of ‘level mixing’ as a ‘sin’, a ‘dragon’ to be ‘slain’, or a ‘ghost’ to be ‘exorcised’ (LB 410, 362; cf. Martin 1956; Joos 1957). They condemned ‘“the Pike heresy -- introducing morphological considerations into phonemic analysis”‘ (Voegelin 1949:78) (LB 362) (cf. 5.45; 8.68). ‘Grammar’ might be used at most ‘heuristically’ in ‘search’, but not in ‘presentation’. For Pike, however, ‘to insist on a rigid separation of levels’ is ‘to fail to report the empirical data’ in its ‘integration’ (LB 591, 406). ‘Compartmentalization’ fosters a ‘fragmentation’ he cannot judge ‘elegant’ or ‘rigorous in scientific description’ (LB 406; but see LB 555). Moreover, ‘a rigid separation of levels’ relegates ‘meaning’ to a level ‘beyond the sentence’, ‘gives priority’ to ‘phoneme’ over ‘morpheme’, and ‘leaves no room for the tagmeme’ (LB 586; cf. 5.50f).

5.36 To re-integrate the levels, Pike undertakes to portray ‘human activity’ within a ‘pyramided hierarchy’ (LB 194, 226, 245, 409, 479, 586). Here, the ‘traditional’ scheme with ‘phonemes combining to make morphemes, morphemes to make words, words to make sentences, etc.’ is ‘rejected’ in favour of ‘three hierarchies with partial overlap’: ‘phonological’, ‘lexical’, and ‘grammatical’ (LB 409; cf. LB 586; cf. 4.50; 5.45; 7.56; 13.27).¹⁹ Thus, several levels of ‘hierarchical structure’ can ‘occur pyramided’ ‘within a unit’ (LB 434, 109).²⁰ Our perspective can vary according to (a) ‘height’, i.e., ‘the hierarchical element cut out for attention from a sequence’, or the ‘place’ of ‘a unit’ ‘in a hierarchy’; (b) ‘depth’, i.e., ‘simultaneous attention on both the high and low’ ‘units of one or more hierarchies’; and (c) ‘breadth’, i.e., ‘the composite range’ of ‘hierarchies’ under ‘attention’ (LB 177). But size still seems to be the chief factor, e.g., when ‘high’ is opposed to ‘small’, or ‘depths’ are labelled ‘large’ or ‘small’ (just the terms often used for ‘segments’ or ‘parts’) (LB 109ff; 75, 79, 83, 79, 135). ‘Limits to the lowest level of focus are set by the purpose of the participant or observer’ (5.16f), but with some ‘indeterminacy’; for example, ‘focus’ on ‘a single word or syllable’ is ‘best treated as’ a ‘hypostasis in which the hearer becomes analyst’ before ‘“shifting gears” back into normal communication’ (LB 111) (5.17, 20). ‘The upper limit fluctuates greatly’ with ‘permanent or temporary purpose and interest’ and is ‘less rigid culturally’ ‘than the lower limit’ (LB 111; cf. 5.19, 46, 59).

5.37 A total event (like a ‘football game) is composed’ of an ‘enormously complex network’ of ‘interwoven hierarchies of activity’ (LB 117f; cf. 5.78ff). ‘The classes of units’ in each ‘included hierarchy constitute a simultaneous componential system’, ‘the most important and obvious’ being the ‘verbal’ and the nonverbal’; beyond that, ‘linguists differ’ about ‘how large’ or ‘how small a part of language is best called a “system”‘, ‘according to their area of attention’ (LB 132; 584f) (cf. 13.43). ‘The hierarchies’ are ‘relatively or partially independent’ and ‘interpenetrate’ with a ‘margin’ of ‘indeterminacy’ (LB 132) (cf. 5.39, 45, 53). Pike contrasts this approach to one based on ‘a simple linear sequence’, which ‘conceals the hierarchical structuring of the data’ by ‘squeezing data’ and ‘mashing a hierarchy’ ‘into a linear sequence’ (LB 406, 589). Chomsky, Halle, and Lukoff (1956:79), for instance, propose to ‘“simplify linguistic theory by restricting it to the consideration of linear systems”‘ (LB 417f; cf. 9.109).

5.38 Pike judges his ‘behaviouremic theory elegant and fruitful’ because it ‘describes’ the ‘enormous complexity of interlocking systems, levels, and units’ in terms of ‘a few simple components’, rather than seeking ‘simplicity’ ‘by a rigid separation of levels’ (LB 586f; cf. 5.3, 6, 35; 13.45). All things in a ‘system’ ‘are mutually defining’; indeed, ‘all the terms in any system, in our view’, are ‘in part circular’ (LB 555, 440) (cf. 8.40; 9.29; 13.48). ‘Neither unit types, nor subsystems, nor levels of structure’ can ‘be discovered or adequately described’ without regard for others ‘which are diverse from those momentarily under’ ‘attention’ (LB 555f). ‘High-level material must be anticipated’ sometimes when ‘presenting’ ‘low-level’ and vice versa in order ‘to make the empirical situation clear’ (LB 577; cf. 5.19, 34; cf. 13.43).

5.39 If ‘no language system can occur’ without ‘interlocking’, ‘tagmemic theory’ must ‘stand or fall by demonstrating’ how ‘hierarchies interlock’ (LB 565f) (cf. 9.34). The ‘details’ of this ‘interlocking’ ‘must be discovered through empirical research for any particular language’ (LB 581). The ‘interlocking’ may be ‘lateral’ among ‘units’ on one level (e.g., ‘fusion of units at their borders’), or else ‘vertical’ among ‘units’ on ‘low’ and ‘high levels’ (e.g., ‘inclusion of smaller units within larger units’) (LB 565, 582; cf. LB 474). ‘Structure’ itself is ‘in part a function of the interlocking’ of ‘levels’; and ‘fusion’, ‘double functions, and indeterminacies’ enhance the ‘integration’ and ‘dynamics of a system’ (LB 566, 582; cf. 5.34, 36f, 58f; 13.59).

5.40 ‘Hierarchies’ are mutually ‘relevant’ if at some ‘points or regions’ their ‘units’ are ‘co-terminous or co-nuclear’, but not all (or we would have ‘merely a single structure’) (LB 566, 581). Although ‘parallels’ may be ‘aesthetically satisfying’, the ‘nonconformity’ of ‘borders’ was what originally ‘started’ Pike ‘on the quest for a new’ ‘theory’ to supplant ‘theory’ which uses ‘artificially selected data’ to ‘eliminate empirical areas where multiple hierarchies must be postulated’ (LB 570ff; cf. 5.48, 89; 9.46, 55, 75, 109; 13.50). The vital question is how ‘grammatical’, ‘phonological’, and ‘lexical hierarchies’ exert ‘mutual control’ on ‘patterns’ and ‘borders’ (LB 573). For example, ‘the units of the lexical’ and ‘phonological hierarchies often have borders in common’, though not ‘every word’ is a ‘phonological’ ‘unit’ (LB 567) (cf. 5.52). Or, the ‘presence’ of a ‘morpheme’ can ‘signal the presence’ of a ‘lexical unit’, a ‘lexical class’, and a ‘grammatical unit’ (LB 576) (cf. 4.53; 5.46, 53). Still, ‘the lexical and grammatical hierarchies’ are ‘distinct’ in several ways: ‘in two utterances’, ‘the same ‘lexical unit’ may have ‘different grammatical functions’; two ‘sentences’ can be ‘lexically different’ but ‘grammatically the same’, whence the distinction between ‘sentence and sentence type’; ‘lexical units’ can be ‘expressed’ in ‘phonemes’ but ‘grammatical’ ‘structures’ only in ‘formulas’; and so on (LB 577f).

5.41 When one unit or ‘sequence’ ‘manifests’ ‘two or more levels simultaneously, Pike ‘postulates portmanteau levels’ (LB 440, 452, 483, 548n) (cf. 5.52ff, 59, 5¹⁶, 5²⁷). On the middle levels, ‘portmanteau’ relations are found among ‘word’, ‘phrase’, ‘clause’, and ‘sentence’ (LB 441f, 455, 459) (cf. 2.55; 3.26; 4.61; 5.51, 54; 8.55; 11.40, 79; 12.75, 93). On the low levels, ‘portmanteaus’ may combine the ‘phone’, ‘syllable’, and the ‘stem’ (LB 317, 548, 330, 443).²¹ On the high levels, the ‘sentence’ may be ‘portmanteau’ with a ‘monologue’ or even a ‘total discourse’, and the ‘utterance-response’ may be ‘portmanteau’ with a ‘conversation’ (LB 442, 466) (cf. 5.59). To prevent pormanteaus from unduly clouding his hierarchy, Pike stipulates that a level counts as ‘higher in the hierarchical scale’ when ‘some’ of its ‘units’ ‘are larger than the longest units in the next lower’ level (LB 364, 404f) (cf. 13.29). This manoeuvre allows him to retain size and length as organizing criteria even when they don't happen to differ, and thus to make his hierarchy quantitative as well as qualitative.

5.42 The phonemic level is the lowest, and the most basic to Pike's approach. The ‘phoneme’ is ‘probably the one unit which can be demonstrated to exist both linguistically and psychologically’ (LB 352; cf. Saporta in Osgood & Sebeok [eds.] 1954:62). Its ‘threshold criteria’ relate to ‘articulatory movements’ (LB 432, 78, 25) (cf. 2.70f; 3.14, 18, 21; 4.29, 34; 8.66, 70; 13.26). So it is a ‘threshold unit’ of ‘behaviour’, not of ‘acoustics’; its ‘locus’ is not ‘the sound wave’, but the ‘actor and his actions’ (LB 306, 309). ‘The essential physical substance’ of its ‘manifestation’ is ‘the physiological movement of the body parts’ during its ‘production’ (LB 306) (13.26). These ‘movements’ are not ‘absolutes or constants’, but ‘relative’ to the ‘identificational-contrastive features of the phoneme’.

5.43 If we ‘focus’ ‘on the speaker’, we face a ‘difficulty’: ‘how can a listener “hear” a phoneme?’ (LB 309f) (cf. 2.71). Maybe the hearer ‘reacts’ ‘by empathy’ ‘to the physiological movements’, such that ‘spoken and heard phonemes’ function within ‘congruent systems’ sharing ‘neurological movements’ (LB 310). Even ‘the phonemes in a thought sequence’ could be ‘neurologically’ ‘congruent’ (or entail ‘suppressed articulatory movements’), though this activity’ can be ‘observed’ ‘only through gross patterns of electric activity’, and ‘the person thinking’ has ‘no proprioceptive sense’ of it (LB 311; cf. 3.10; 4.9).

5.44 At all events, Pike devotes his most massive efforts to describing ‘the phonological hierarchy’ in terms of movement. He meticulously labours upward through ‘the phoneme’, ‘the hyperphoneme’, ‘the syllable’, ‘the rhythm group’, ‘the stress group’, ‘the pause group’, and ‘the breath group’, frequently referring to movements of the ‘chest’ and ‘abdomen’ (LB 290-423) (cf. 8.22). He invokes ‘the chest pulse’ to ‘differentiate the syllable from the phoneme’; ‘the abdominal pulse’ to ‘differentiate’ ‘the rhythm group from the syllable’; and so on (LB 432).²² The degree of detail and the physical grounding encourage Pike to see in ‘phonological movement’ the ‘clearest’ model to ‘be generalized to other linguistic levels’ (LB 547) (cf. 13.27). His ‘tagmemics’ ‘looks forward to a universal etics of grammatical types’ ‘analogous’ to ‘phonological types’ (LB 470n). He started out by coining his main terms ‘etic and emic’ (5.19) ‘from phonetic and phonemic’ and seeking ‘“cultural equivalents of phonemes”‘ (Kroeber & Kluckhohn 1952:124), such as ‘the behavioureme’ (LB 37, 33, 121) (cf. 5.60).²³

5.45 The morphemic level is also treated in detail, though less so than the phonemic. Here, ‘a language’ is ‘constituted’ by the ‘set of all systems of morphemes which are congruent and/or simultaneous’, plus ‘all other verbal elements congruent with them’ (LB 178; cf. 4.46). A ‘system of morphemes’ is a ‘class of distribution classes’; a ‘large system’ may ‘include’ ‘smaller’ ones, e.g., for ‘segmental’ and ‘intonational morphemes’.²⁴ ‘The internal structure of the morpheme’ has ‘a sequence of actions’ ‘resulting in some typical sequence of sounds’ (LB 175). ‘Morphemes’ can thus be ‘characterized’ ‘by a transcription of a series of single phonemes, linearly’; the fact that ‘phonemic features’ can also have ‘morphemic status’ indicates a ‘fuzzy’ or ‘indeterminate threshold’, as with ‘many classificatory boundaries in the theory’ (LB 305) (cf. 4.50; 5.36; 13.27, 59).

5.46 ‘Ordinarily, a morpheme’ ‘is below the threshold of participant awareness and shorter than a unit’ of ‘focus’ (LB 175; cf. 4.54; 5.13, 48; 9¹⁵; 13.49). It goes unnoticed unless it ‘happens to manifest a complete uttereme under conscious attention’ (e.g. ‘“Boy!”’) ‘or is under the immediate scrutiny of the analyst’ (e.g., in a ‘dictionary listing’). Consequently, the methods for discovering morphemes can be complex. Pike advises us to start with an etic ‘procedure for identifying morphs’ by ‘finding the least common denominator’ in ‘two utterances’ ‘partly alike in form and meaning’, or an ‘element’ whereby ‘the two differ’ (LB 179) (cf. 4.52). ‘Once a morph is identified’, we should decide if it is a ‘free or conditioned variant’ of a ‘morpheme’, or ‘part’ of a ‘complex variant’, or ‘a sequence’ of ‘variants’, or ‘a fused composite’ -- or just ‘an error’ to be ‘eliminated’ or ‘corrected by later comparison with other occurrences’ (LB 182, 179).²⁵ ‘No technique’ ‘can lower to zero the margin of error’ in a ‘guess’, but ‘errors’ can ‘in principle’ ‘be corrected’ by ‘more adequate data’ (LB 222).

5.47 An ‘important’ ‘characteristic’ of both ‘morph’ and ‘morpheme’ is the potential to be ‘active (or live, productive)’, ‘entering regular analogies’; but ‘serious practical difficulties’ can arise in telling which ones qualify (LB 169ff, 190) (cf. 2.52f; 7.76; 8.58). Pike suggests a battery of ‘tests for activeness’: if the morph or morpheme is ‘one of a large number’ that can ‘occur’ in ‘a slot’ or ‘larger unit’; if it ‘can occur in a wide variety of different kinds of functional slots’; if it is used ‘in new combinations’ created ‘in recent times’; if its ‘meaning’ is ‘easy to determine’ and ‘contributes in a regular fashion to the total meaning’ of the ‘sequence’; if a ‘native speaker, upon questioning, can describe’, ‘discuss’, or ‘define’ it; if an ‘investigator’ can ‘easily segment’ it ‘out of a continuum’; and so on (LB 170f). These tests too leave some ‘indeterminacy’, so that between ‘active’ and ‘inactive (or passive, dead’) we may find ‘semi-active’ ones, or several ‘degrees’ within ‘a progressive gradation’ (LB 170ff, 174, 191). ‘Indeterminacy’ also enters when morphemes ‘fuse’ or their ‘boundaries blur’ (LB 177; cf. 5.39, 45, 53, 77, 87; 9.29; 13.59).

5.48 ‘The morpheme’ is a ‘form-meaning composite’ (LB 163) (cf. 5.64, 76). Although the ‘relationship’ is not ‘one-to-one’ (‘biunique’), the ‘formal component’ is not ‘separable’ from ‘the meaning’, and ‘neither may be abstracted as a unit for normal participants’ in ‘nonhypostatic behaviour’ (LB 162f; 187, 189) (cf. 13.54). To be sure, when a ‘morpheme’ is ‘below the threshold of focus of nonanalytic members of a culture, they will be unable to report, on questioning, any meaning’ (LB 157f). Only ‘lexical’ or ‘dictionary meaning’ is ‘sometimes above the awareness threshold of the untrained native speaker’ (LB 160). Besides, ‘the meaning of a morpheme can be so greatly weakened in certain contexts’ (e.g. ‘“terrible'“ vs. ‘“terribly good”’) that its ‘variants’ ‘have little meaning in common’. No meaning at all ‘can be detected’ for ‘an empty’ ‘morpheme’, which is ‘identified only as a residue’ after analysing out the ‘meaningful elements’ (e.g. ‘“does”‘ in ‘“does he go to the school?”‘ (LB 160f, 199). Still, the ‘majority of those sequences of sounds’ that ‘appear meaningless’ ‘early’ on can later prove to be ‘meaningful’ ones, or else ‘mistakenly segmented parts of a larger morpheme’ (LB 161). ‘Communicative’ functions’ ‘could not be served by a language’ with too many ‘meaningless elements’; and ‘few’ ‘are in fact found’.

5.49 Although the ‘morpheme’ makes a fairly tidy ‘correlate of the phoneme’, Pike's method was crucially shaped by his conviction that ‘grammar’ might contain ‘some other correlate’ (LB 5, 9). ‘Just ‘as the phoneme was reflected in practical orthographic work for millenia before being “found” by the scientists’, so some ‘unknown unit of grammar’ might have remained hidden from ‘current theory of linguistic structure’ (LB 5, 287). Here also, his reasoning was steered by the idea that the ‘inclusive analytical principle’ is to ‘treat items in slots on a higher level’ (LB 477) (5.47). ‘The most basic relationship is between a unit and its slot-occurrence’, rather than a ‘binary’ link ‘between units of equal rank or of the same level’ (LB 282; cf. 5.21; 13.29). At ‘each level of focus’, ‘classes of segments occur and are determined by the slots they fill’ (LB 83). In return, ‘each kind of slot’ has ‘a class of appropriate segments’, though ‘appropriateness’ can be ‘indeterminate’ (LB 83f). ‘This slot-class potential’ for ‘occurring’ engenders ‘positive and negative predictions’ about what might happen or be said, even if ‘slots’ in ‘lower-level focus’ ‘need not be filled in every manifestation’ within the ‘activity’ (LB 86, 88). ‘The probability’ or ‘prediction’ applies to ‘structural components’, not to ‘particular’ ‘words’ (LB 605) (cf. 7.90f; 9.93; 11.16, 56).

5.50 To designate this key ‘correlation’ between a ‘slot’ and a ‘class’, Pike selected the ‘tagmeme’, a term of Bloomfield's now given a new meaning²⁶ and declared ‘more basic than’ ‘immediate constituents’ (LB 194f, 490, 282). The tagmeme embodies the theses that the ‘function in the slot of a higher structure’ ‘is always immediately relevant to the nature of a unit’; and that ‘a unit cannot be defined’ as ‘discoverable or describable in itself’, but only within ‘a larger unit of behaviour’ (LB 451, 195; cf. 5.8, 14f, 19, 22, 39, 47). Because ‘the total’ of a ‘class’ ‘cannot all occur at one time as an event’, ‘the tagmeme seems’ ‘much less concrete than the morpheme’, yet is claimed to be ‘an objective emic unit’ in ‘normal participant behaviour, not a mere conceptual construction of the linguist’ (LB 203). Predictably, Pike rests this claim on the ‘physical basis’ in a ‘manifestation’ (cf. 5.24, 28, 31, 39).

5.51 More encompassing than the tagmeme is the ‘hypertagmeme’ for the ‘high-level’ ‘members of the grammatical hierarchy’, such as the ‘conversation’, ‘monologue, utterance’, and so on (LB 432, 517) (cf. 5.59). Pike is uncertain whether ‘hypertagmemes’ and their ‘levels’ should be ‘absolute’ or ‘relativistic’ (LB 443, 446f), i.e., whether they could be ‘listed’ independently of ‘focus’, using a fairly ‘mechanical’ ‘methodology’ (LB 443, 445ff). A ‘relativistic’ outlook enables ‘a unit to simultaneously represent a low level and a higher one’ (e.g. ‘word’ and ‘clause level’); and we can ‘abbreviate complex formulas into simple high-level forms in a less arbitrary’ and more ‘consistent’ way (LB 443f). Yet ‘disadvantages’ ‘result’ as well (LB 445). The ‘tagmeme’ comes to depend on ‘the temporary focus of attention’, their ‘number’ ‘fluctuates’, and ‘levels’ ‘proliferate’ (e.g. ‘within the word stem’). The ‘minimum’ ‘tagmeme’ is not ‘as directly seen’ as are ‘phoneme’ and ‘morpheme’. ‘Threshold criteria’ are needed ‘for determining when one has passed from one level to another’, and ‘the total coherence of the theory’ is affected (LB 445f). ‘In some sense’ then, ‘terms need to be made absolutistic’. To resolve the problem, however, ‘a major change is required in the theoretical framework’ (LB 450).²⁷

5.52 A closely related quandary is whether the ‘hypertagmeme’ should be ‘obligatorily complex’, i.e., ‘composed of two or more’ units or sequences (LB 432f). At first, this standard seemed ‘useful’ because it was ‘simple’, ‘sharp-cut’, and ‘easy to state’, ‘understand’, and ‘apply’; and it allowed a ‘simple discovery procedure’, ‘hierarchical results’, a ‘quick recognition’ of ‘minimum formulas’, and ‘immediate work’ ‘without determining’ ‘the point of word boundaries’ (LB 435). Moreover, ‘data’ could be ‘structured’ ‘in an elegant, concise fashion’ ‘easily transferable to pedagogical treatment for language learning’. Yet because a ‘hypertagmeme’ may sometimes have ‘a single emic slot’, the standard raised problems in ‘field studies’, until it ‘collapsed’ and was ‘withdrawn’ (LB 447, 433ff, 439). ‘Just as a morpheme’ ‘could be simultaneously a word and a phoneme’ (5.50, 58f), a ‘tagmeme could also be a hypertagmeme’ -- another ‘portmanteau’ (LB 449; cf. 5.40). ‘The theory’ becomes ‘more coherent’ when ‘one point of view’ includes ‘reference to minimal units, to portmanteau levels, to division subclasses, and to hypermorpheme classes’ (LB 452).

5.53 Moving upward from morphemes brings us to ‘the word level'²⁸, for which ‘general etic criteria are available’, but may not always ‘apply’ nor ‘lead to the same results’ (LB 437f) (cf. 2.18; 3.31; 4.54; 6.23; 13.29). To count as ‘words’, units must be ‘isolatable’, ‘interruptible’, ‘versatile of occurrence’, and ‘rigid in order’ of ‘parts’ (cf. 3.62, 4.53, 60; Nida 1949). Further factors are ‘special relationships’, ‘junctures’, or ‘sequences’ within the unit’, as well as ‘phonological markers’ like ‘pause’ and ‘rhythm’, and so on. Yet Pike admits that the ‘word level’ may not be ‘structurally relevant and useful’ for every ‘language’, e.g., not in ‘the Mayan family’ (LB 481f). And the ‘indeterminacy between the levels’ is acute when ‘border-line instances occur between word and bound form, and between word and phrase’ (LB 438) (cf. 13.28). High ‘frequency’ can convert a ‘sequence of morphemes’ into a ‘fixed’ ‘idiomatic unit’ (LB 605). Or, a ‘single word’ may ‘constitute an entire’ ‘phrase’, ‘clause’, or even a ‘sentence’ (more cases of ‘portmanteau’) (LB 439f, 483; cf. 2.55; 3.34; 4.67; 5.51; 6.45; 8.56f; 13.28).

5.54 The next higher level is that of the ‘phrase’: ‘a unit’ ‘filling an emic slot in a clause or sentence structure’ and ‘composed’ of ‘two words’ or ‘one word which is optionally expandable in that same slot’ (LB 439) (cf. 13.54). Here again, Pike departs from ‘Bloomfield’ and ‘the American scene’, for whom ‘the phrase’ is ‘a free form’ made of ‘two or more lesser free forms’ (LB 486) (cf. 4.42, 65f). Pike wants to allow for ‘a single word’ being ‘portmanteau’ with ‘a phrase’, and rejects ‘obligatory complexity’ (cf. 2.55; 3.26, 34f; 4.60; 5.51; 8.56; 9.75; 11.40; 12.75). Indeed, he warns that the `dichotomy between morphology and syntax’ should not be made ‘too early’ or ‘rigid’; it is not ‘sharp’ for some languages (e.g. Mixtec, Chinese) and creates problems with ‘phrase-words’ (e.g. ‘“the king of England's hat”’) or ‘stereotyped phrases’ (e.g. ‘rack and ruin’) (LB 580, 479, 481, 162; cf. 2.61; 4.60; 5.32; 13.28). Still, in ‘English’ at least, the ‘phrase has a much greater expansion potential than the word’ and more ‘freedom’ to ‘vary’ ‘the order’ of its parts; and ‘is more likely to be interruptible by parenthetical forms or phonological junctures’ (LB 440f).²⁹

5.55 Next comes the ‘level’ of ‘the clause’, which had previously been ‘undefined’, ‘due to the great influence of Bloomfield’, for whom it was ‘never an integral part of his description’ (LB 486) (hence no mention of it in Ch. 4). Following Longacre (1964a), Pike accords ‘the clause level, though definable separately for each language’, ‘a place in the grammatical hierarchy between phrase and sentence, as the syllable is between phoneme and rhythm group’ (LB 441). ‘Clause’ is an ‘especially useful’ ‘term’ for ‘subject and predicate’,³⁰ since the ‘typical’ ‘overall structural meaning’ is ‘predication’, ‘equation’, ‘query, or command’ (LB 425, 441) (cf. 3.36; 4.69; 8.55; 12.78f).

5.56 Next comes the level of the sentence, described as ‘a minimum utterance’, ‘isolatable in its own right’ (LB 442). ‘Some but not all are clauses’, while others are ‘nonclause phrases or words’ (cf. 4.67; 9.82). And whereas a ‘clause’ can add ‘tagmemes of time, manner location’, etc., ‘the sentence’ can add ‘further clauses in coordinate, subordinate, and paratactic relations’. In ‘its broader setting’, the sentence entails ‘the deep problem of identity of unit against ground’ and ‘remains’ ‘immune from attack’ only if ‘it is taken, in a regularized form, as an axiomatic starting point’ (LB 8n) (cf. 13.54). That ‘the sentence is “the unit of language, not the word”‘ was asserted by Sweet, Cassirer, Humboldt, and Firth, though Sweet (1913:5) too hedged by terming the word ‘“an ultimate or indecomposable sentence'“ (LB 482, 146) (cf. 8.56; 13.54). In America, ‘linguistics in the past’ had ‘made its most striking progress by dealing with units no larger than the sentence’ (LB 145). Yet ultimately, Bloomfield's definition of ‘the sentence’ as ‘an independent linguistic form’ (4.67) imposed a ‘limitation which has prevented, in this country, the development of linguistics’ (LB 146, 484). ‘Large language units’ were ‘left to students of literature’ (more ‘written’ than ‘spoken’), ‘metrics’, ‘public address’, and ‘speeches’ (LB 146). Among the ‘few linguists’ to address such units, Jakobson and his co-workers ‘studied verse patterns’, while Harris proposed a ‘“discourse analysis”‘ which Pike finds ‘atomistic’, based on ‘assumed, not procedurally identified sentences in juxtaposition’ rather than in ‘integration’ (cf. Jakobson 1960; Harris 1952) (cf. 11.2).

5.57 For dealing with ‘the total language event in a total cultural setting’, ‘the sentence is a totally inadequate starting’ or ‘ending point’ (LB 484, 147, 484). ‘Sentences must not be studied outside of total concrete behavioural contexts’; ‘conclusions’ about ‘isolated hypostatic data’ (in the sense of 5.17) may not be ‘valid for the description of units of normal contextual speech’ (LB 155) (cf. 11.3; 13.55). ‘The abstracting out of sentences for study’ is ‘legitimate and useful’, but must be recognized to be an ‘“as-if” procedure’ and ‘a deliberate distortion’ for ‘handling data’ (LB 484; cf. 5.5; 13.39). ‘Many important characteristics of sentence structure can be adequately handled’ ‘only in reference to discourse structure’, e.g., to tell whether or not a structure is ‘independent’ or ‘complete’ (LB 485f, 148).

5.58 In Pike's classification ‘full’ ‘sentence types’ include ‘sentence-word’, ‘question’, ‘actor-action’, ‘instrument-action’, ‘equational’, ‘narrative’, ‘emphatic’, ‘surprise’, and ‘disappointment’; ‘minor sentence types’ include ‘interjections’, ‘completive’, ‘exclamatory’, and ‘aphoristic’ (LB 139; compare BL 171-77). By taking discourse into account, Pike can identify types by quite diverse criteria, ranging across ‘form’, ‘elements’, ‘constituents’, ‘order’, and ‘pitch’, plus ‘meanings as determined by their occurrence in the cultural setting’ (LB 139). Still, the examples from Menomini given by both Pike (LB 139) and Bloomfield (BL 175f) suggest that a sentence type (e.g. ‘disappointment sentence’, BL 176) can be set up whenever some language marks it formally (cf. 4.68; 13.54).

5.59 The even ‘higher levels’ -- ‘monologue’ for ‘the connected discourse of a single speaker’ (e.g. ‘lecture, soliloquy’), ‘utterance-response’ for a ‘change between two speakers’, and whole ‘conversation’ -- are barely described aside from the relative size, the number of speakers, and the flow (‘merging’, ‘diverging’, ‘overlapping’, ‘interrupted’, etc.) (LB 442, 125). Perhaps Pike supposes that portmanteau relations and level-independent concepts like ‘tagmeme’ and ‘hypertagmeme’ provide a channel for transposing up to these higher levels the results for the lower levels he explores in much greater detail. For instance, he suggests that the ‘topic’ as a ‘unit in between the utterance-response unit and extended conversation’ might be ‘treated somewhat like the meaning of a morpheme’ (LB 442, 136) (but cf. 11.63-69).³¹

5.60 Pike devotes more concern to the ‘behavioureme’: the ‘emic unit of top-focus behaviour’ ‘related to its cultural setting in such a way that cultural documentation may be found for its beginning, ending, and purposive elements’ ‘within the verbal or nonverbal behaviour of the domestic participants or observers’ (LB 121; cf. LB 128ff, 140, 153f). ‘The size of behaviouremes’ and their ‘closure’ help indicate ‘when one's analysis is complete’ rather than ‘arbitrarily’ ‘ended’ (LB 129f; cf. 5.51). ‘An acteme’ is the ‘minimum ‘segment or component of human activity’ in a ‘behavioureme’ (a ‘verbal acteme being a phoneme’ and a ‘nonverbal’ one a ‘kineme’) (LB 291).³² ‘A verbal behavioureme is an uttereme’ -- a ‘unit which receives participant focus in nonhypostatic situations’ -- ‘large’ ‘types’ being ‘hyperutteremes’ and ‘small’ ones ‘minimum utteremes’ such as the ‘single sentence’ (LB 157, 121, 133). ‘In the analysis of language’, ‘uttertics’ would be a ‘classification’ of ‘utterance types around the world’ (LB 133, 135; cf. 5.23).

5.61 It can been seen that Pike provides no special ‘level’ for ‘meaning’ nor a separate hierarchy for ‘semantics’. At first, we might be reminded of the deliberate exclusion proposed by other American linguists (cf. Morris 1946; Chomsky 1957; Lamb 1962) (cf. LB 148f, 279, 474f, 497, 500, 617, 620; 4.15, 26). Lacking an ‘algorithm for the discovery of semantic components’, some ‘analysts’ adopted a ‘formalistic approach’ using strictly ‘distributional criteria’ and ‘eliminating’ even ‘structural meaning’ both from ‘definitions’ and from ‘procedures of analysis’ (e.g. Harris 1951) (LB 620, 277). ‘“Meaning”‘ got set aside as a ‘“metalinguistic”‘ aspect of ‘“the material”‘, or postponed ‘“until the linguistic system has been completely described”‘ -- ‘“phonology”‘, ‘“morphology”‘, and ‘“syntax”‘ (LB 61, 278; Smith 1952:59; Trager & Smith 1951:68). At most, ‘meaning’ was to be ‘the linguist's and the layman's shortcut to a distributional’ result, a source of ‘quick clues’, or an aid for deciding ‘whether elements are the same or different’ (LB 60f; 180; Harris 1951; Fries 1952; cf. 5.14, 65; 4.26, 31). ‘“Grammatical meanings”‘ might be admitted if they are ‘“definite and sharp, essential features of every utterance”‘ (Fries 1962:99) (LB 279).

5.62 Pike also started out in 1948 seeking ‘formulas’ for the ‘immediate constituents of any utterance’ ‘without reference to meaning’, but ‘later’ gave up and ‘abandoned an algorithm for analysis’ (LB 286fn) (cf. 5.86; 9.110; 11.14; 13.50). ‘Bloomfield overstated his case’ in arguing that ‘since meanings cannot be known exactly, they cannot be utilized’ (LB 148; cf. 4.14ff, 26, 51, 68). So Pike's pique is now added to the ‘protest’ against ‘“the stultifying exclusion”‘ or ‘“delay”‘ of meaning (e.g. Pulgram 1961; Haas 1960) (LB 188). In his opinion, ‘the use of no meaning’ ‘implies that the linguist’ ‘is not interested in language as it functions as a communicating device, and cannot analyse the communicative process’ (12.32) and its ‘content’; and one ‘attempts to reject the implications of one's own procedures’ (LB 60ff). ‘Semantic components’ are essential for the ‘presentation’ of ‘language as a communicative system’; we need to know what a ‘structure’ ‘means, not merely that it is well-formed’; and to ‘generate sentences which are meaningful and usable by the speaker’ (LB 225n, 280). To exclude ‘meaning or purpose’ is ‘to abandon the most useful structural threshold between the reciting of a poem and the minutiae of atomic structure’ (LB 304; cf. 5.28). Linguists can only ‘bypass the mention of meaning’, not ‘the use of meaning’ (LB 61).

5.63 On the other hand, Pike ‘rejects with Fries and Harris an analysis by meaning alone’ just as much as one by ‘form alone’ (LB 278, 181). He also ‘rejects the dualism of Hjelmslev’ with a ‘functional dichotomy of expression and content’ (6.25ff) for ‘leading to a theory in which “signs”, “symbols”, and “semiotics”‘ are ‘too widely divorced’ from ‘human behaviour’, and for implying ‘emes of meaning’ as ‘abstracted relationships’ with no ‘physical manifestation’ (LB 187) (cf. 6.50-56). He joins Firth in ‘rejecting the theory of signs of de Saussure and Hjelmslev’, who imply that ‘a “sign”‘ ‘“is the bearer of a meaning”‘ (LB 63; cf 6.23; 8.20). Pike predictably likens ‘form’ and ‘meaning’ to ‘physical’ and ‘functional’ ‘characteristics’ (LB 55).

5.64 Thus, meaning is for Pike not a level or a hierarchy, but an omnipresent aspect of all levels and hierarchies. ‘Every step in linguistic analysis’ must deal with ‘a form-meaning composite’, in which ‘form and meaning do not have to be in a one-to-one relationship’ (LB 278, 63n, 55, 63, 149, 162, 472, 516; cf. 5.48, 76). ‘Meaning has its locus not in the individual bits and pieces’, but ‘within the language structure as a whole’ or within ‘verbal behaviour’ in a ‘frame’, i.e., in an ‘identified context’ (LB 609, 134). For example, ‘each morpheme ultimately obtains its meaning only in relation’ to others in ‘the total system’ and in ‘particular sequences’ (LB 605) (cf. 5.67). So by ‘defining the morpheme only in relation’ to a ‘total structure’, we can resolve ‘difficulties’ when morphemes seem ‘lexically meaningless’ or ‘lack’ an ‘unchanging core of meaning’ (LB 184, 186, 598f; cf. 5.48; Bazell 1949; Bolinger 1950; Hockett 1947; Nida 1948, 1951). Or, we can handle ‘semantic variants’ by asking how they are ‘conditioned by the universe of discourse’, ‘the style’, the ‘physical matrix’, or ‘the neighbouring morpheme’ (LB 599; cf. 5.84f).

5.65 ‘The sharp-cut segmentation of meanings’ ‘is therefore in principle impossible’ (LB 609; cf. 5.27; 13.59). ‘The meaning of one unit in part constitutes’ and ‘is constituted of the meaning of a neighbouring unit’. ‘Meaning’ is a ‘contrastive component of the entire complex’ and ‘occurs only as a function of a total behavioural event in a total social matrix’ (LB 148f, 609).³³ So we must foreground ‘the social components of language meaning’ by focusing on ‘the activity of the communicating individuals’, both ‘overt’ (‘verbalization’, ‘physical activity’) and ‘covert’ (‘intention’, ‘understanding’) (LB 598). ‘Perhaps the answer will lie’ in ‘finding a statistically’ measurable ‘set of common contexts’, or in ‘testing for native reaction’ (e.g. ‘“same or different” tests’), or in consulting ‘the common’ ‘cultural effect’ of ‘physical events’ or ‘behaviour’ (LB 600; cf. 5.10, 26, 70).

5.66 Pike suggests that ‘meaning in verbal behaviour’ has as its ‘analogue’ ‘cause and effect in physical matters’ (LB 663; cf. 4.8, 80; 6.62; 7.33; 8.41; 13.11).³⁴ ‘Units of physical motion’ ‘underlie all the physically manifested units of purpose and meaning’ (LB 290) (cf. 5.27). This outlook leads him to imagine a metaphoric ‘orbit’ of ‘meaning’ (LB 603). ‘The central meaning’ applies when ‘words occur’ in ‘descriptions’ ‘close to the physical situation which they name directly’, and gets ‘more difficult’ to state when no ‘reference’ is made to ‘physical objects’, ‘actions, or qualities’ (the case with ‘small distribution classes, such as “if”, “an”, “who”, “the”, “to”’) (LB 602f) (cf. 4.24). Or, the ‘central meaning’ may have ‘greater frequency’ among ‘the community’ than ‘marginal meanings’, except for ‘special universes of discourse’ (LB 601). Or, ‘the central meaning’ may relate to ‘the physical context in which the words were first learned by a child’ (LB 603f, 600f).³⁵ ‘As the speaker grows older’, ‘the central meaning’ may become ‘relative to the universe of discourse’ (LB 601). Pike even envisions ‘a hierarchy of universes of discourse with progressive degrees of centrality’ (LB 602).

5.67 Pike's scheme has ‘no specific number of distributional orbits, or degree of remoteness from the central’ (LB 604). ‘The outer’ ‘orbits carry the greater communication energy’ for ‘hearer impact’, e.g. in ‘poetry’, ‘puns’, and ‘slang’. Major examples of the outer ‘dependent or derived meanings’ are the ‘idiomatic’ meaning not ‘predictable’ from ‘the meanings of its parts’, and the ‘metaphorical meaning’ (LB 601ff) (cf. 9.97ff). ‘Nonsense’ ‘results if one attempts to carry back the “meaning”‘ of a ‘metaphor’ ‘“to the primary physical context”‘ (Urban 1939:639f) (LB 632). ‘Metaphor’, and ‘poetry’ in particular,³⁶ are domains for ‘going beyond verbal responses to physical stimuli’ and ‘discussing nonphysical problems’ (LB 615). But a ‘metaphoric meaning’, which starts out being ‘less frequent for the community as a whole’, can ‘spiral down into the central orbit’, ‘gradually becoming the only linguistic item’ to ‘label’ an ‘object or situation’ (LB 604).

5.68 This orbit scheme befits a physicalist approach. Also fitting is the ‘behaviourist’ notion of ‘“nonlinguistic reality serving as a guide”‘ (LB 635) (cf. 13.24). But Pike approves Malinowski's (1935:64f) warning against the ‘dangerous assumption that language mirrors reality’; and Cassirer's (1946:9) tenet that the ‘“mutual limitation and supplementation”‘ among ‘symbolic forms’ is a more ‘“basic philosophical question”‘ than ‘“their relation to an absolute reality”‘ (LB 625f) (cf. 4.71; 6.12; 11.10). For Cassirer (1946:8), these ‘forms’ are ‘not imitations, but organs of reality’ whereby ‘“anything real becomes an object for intellectual apprehension”‘ e.g., ‘codable colours’ are ‘recognized’ ‘more often’ (cf. 4.22; 6.54; 7.31, 71); ‘the grasp of reality’ must be ‘mediated’ by ‘concepts’, even ‘mass and force’. For Pike, to know ‘the ultimate truth, the ultimate structure of physical reality’ would require ‘the emic perception of God’, whereas that of ‘individual men’ is a ‘component of the total reality available to His observation’ (LB 659). Since ‘His views’ ‘are not available to us’, they don't figure in ‘our discussion’.

5.69 Of course, Pike is less willing to identify language with thoughts and concepts than was his teacher Sapir (3.10f) (to whose ‘memory’ the book is ‘dedicated’, LB 3). ‘“Units of thought and speech do not coincide”‘, ‘“showing unity but not identity”; ‘thought is “there simultaneously, but in speech it has to be developed successively”‘ (Vygotsky 1939) (LB 640) (cf. 11.15). Also, ‘a person may attain a concept’ ‘without verbalizing it’, or may have the ‘ability to think’ despite ‘incapacities for speech’ (e.g. in ‘expressive aphasia’) (LB 634, 544, 639; cf. Goldstein 1948; Miller 1951). Even so, Pike concurs with Cassirer that ‘“the chaos of immediate impressions takes on order and clarity’ only by means of ‘“linguistic thought and expression”‘; and he ‘closes the volume’ by quoting George Herbert Mead that ‘the perception of objects as enduring is not possible without language’ (LB 639, 678) (cf. 2.17, 27, 32; 3.10ff, 17, 32, 35, 40; 6.2, 6f, 26, 30f; 7.3, 44; 8.24f; 12.17f, 22, 60f).

5.70 From a more behaviourist standpoint, ‘concepts’ are said to ‘“tie together”‘ ‘“sensory experiences”‘, organize ‘“data”‘, and provide ‘“hypotheses about the way one should react to one's environment”‘ (LB 633, 635; cf. Vinacke 1951:1ff; Postman 1951:267f).³⁷ ‘Meaning’ and ‘purpose’ form ‘the bridge by which’ ‘a physical behaviour pattern enters into the structuring of society or of the individual’, and allow ‘similar or identical events’ to ‘function differently’ (LB 660, 42). We can ‘determine the presence and nature of meaning and purpose’ from the ‘elicitation of response’ -- a domain whose ‘data’ are open to ‘objective study’ and ‘evaluation’ (LB 156, 158; cf. 4.14, 27; 5.10, 15; 7.10; 8.23; 11.92; Fries 1954:65). ‘Purpose’ is strategically placed here alongside ‘meaning’ because it is ‘frequently obvious’ and ‘easily’ ‘detected’ in an ‘observable larger sequence of events’ as a ‘regular association of units in activity’ toward ‘a goal’ and as ‘a response’ or ‘impact on the hearer’ caused by an ‘utterance’ (LB 42, 116, 598). We can ‘assume that human activity is purposive if it regularly elicits either (a) a positive response activity’ ‘or (b) a deliberately verbalized or nonverbalized resistance’ or ‘rejection’ (LB 156). ‘To study behaviour as it actually functions’, we must ‘assume that the analyst can detect’ such data (despite ‘a margin of indeterminacy and error’) by ‘studying the formal components of the physical activity’ (LB 158, 156). Even ‘epistemological attitudes, belief systems’ (‘conscious or unconscious, vague or organized’), and ‘character’ can be discovered by ‘examining overt behaviour’ (LB 533).

5.71 Since ‘participants in human behaviour’ ‘affirm an awareness of meaning or purpose’, the ‘analyst’ can also combine ‘technical’ and ‘lay analysis’ by consulting ‘the popular hypostatic reaction to meaning’, e.g., when participants ‘discuss’ ‘meaning and purpose’ (LB 156f). Or, we can study the ‘deliberate conscious choice’ among ‘alternative purposes’ or ‘meanings’ when ‘“hunting for the right word”‘ (LB 197).³⁸ There, ‘attention is likely to be focused’ on ‘higher levels’ and ‘purposes’, in contrast to the ‘intricate unconscious selection’ of ‘internal elements’. When the latter do become ‘conscious’ during ‘disputes’, ‘challenges’, or ‘new situations’, we have ‘a different kind of behavioureme’ from a ‘routine’ one (cf. 5.11, 46). Pike raises but skirts ‘the metaphysical problem’ of ‘free will’ versus ‘biological and cultural determinism’ by remarking that ‘people talk’ and ‘act’ ‘as if they thought they had free will, purpose, and choice’.

5.72 Instead of addressing ‘the general nature of a theory of meaning’, Pike only shows how ‘a hierarchical view of language illuminates a few special problems’ in the ‘general study of meanings’ (LB 598). One problem is ‘class meaning’, which may be ‘detectable within many members of the class’ and ‘separable from the structural meaning’ of ‘the slot’; or may be ‘in part observable’ through ‘the statistical probability’ of any ‘member’ having a certain ‘lexical meaning’ or ‘semantic component’ (LB 253; 198, 227, i.a.). A ‘strong class meaning’ may even ‘clash’ with, ‘modify’, ‘eliminate, or reverse the meaning’ expected in a context (LB 199, 226). But ‘class meaning’ may also be ‘vague’ or ‘derived from only some members of the class’ (LB 606). ‘Function words’ form ‘a marginal group of lexical items in which tagmemic, morphemic’, and ‘class meaning become fused or indeterminate’ -- evidence for Sweet (1900:74) that ‘“there is no absolute” “demarcation”‘ between ‘“the grammar and the dictionary”‘ (LB 274). So ‘making class meaning a basic starting criterion for determining the classes is fatal to any structural analysis’ (LB 201).

5.73 Glaring mistakes of this kind were made in ‘the study of parts of speech’, e.g., the ‘incomplete’ and ‘inaccurate’ ‘definition’ of the ‘noun’ as ‘“the name of a person, place, or thing”‘ (LB 488, 181) (cf. 2.65; 3.23; 4.55; 6.49; 9¹³; 13.7). But ‘“it is easier”‘ ‘“to muckrake the parts of speech than to replace them with word-categories valid for every language”‘ (LB 285; Sheffield 1912:20). For Henry Sweet (1913:16, 18), ‘“the real difficulty of determining the meaning of the parts of speech lies in the fact, which logicians and grammarians obstinately ignore, that they often have no meaning at all”‘ -- ‘so that the “definition of a part of speech must be a purely formal one”‘ (LB 273) (cf. 13.54). For Fries (1952:73), ‘“a part of speech is a functioning pattern’ and ‘“cannot be defined by means of a simple statement; there is no single characteristic that all examples of one part of speech must have in the utterances of English”‘ (LB 272).³⁹

5.74 Pike hopes his own ‘tagmemic model approximates our traditional feeling as to the relevance of certain components of language structure’, but he doesn't try to resolve the ‘difficulties in determining membership in parts of speech’ (LB 444, 488; cf. Paul 1889:403ff; Jespersen 1940-49:2). To ‘set up’ ‘a part of speech’ or to ‘preserve’ ‘the groupings of traditional grammar’ would ‘require criteria’ of considerable ‘complexity’: ‘identity of stem across the paradigm, comparability of kinds of relations between stem and affix, mutual exclusivity’ among ‘stems’, and so on (LB 489). Pike's own examples of ‘class meaning’ mix grammatical aspects (e.g., ‘person, tense, aspect’) with conceptual ones (e.g., ‘action’, ‘quality’, ‘substance’) (LB 198f, 253; cf. LB 180, 284f).

5.75 Another problem is ‘segmental meanings’, which are ‘components of structurally segmented verbal material’ (LB 611). They can ‘frequently’ be ‘abstracted by speakers’, ‘“put into words”‘, and ‘discussed overtly’ with ‘some assurance of accuracy’, however ‘naive’ from the standpoint of ‘scientists’ (cf. 4.22). ‘Linguists or dictionary makers’ also set up ‘segmental meanings’ by giving them ‘explicit verbalized attention’ (LB 612). ‘Subsegmental meanings’ in contrast, tend to be ‘vague’, supplied by ‘a “hunch or “feeling”‘; Pike likens them to ‘subsegmental phonological features’ in that both are ‘covert backgrounds’ (LB 611, 615).

5.76 Yet another problem is the ‘sememes’, defined as ‘minimum structural units of meaning’ by Bloomfield (LB 187, 620f; cf. 4.45, 50, 68; 6.47; Nida 1951; Joos 1958; Lamb 1962) (LB 187). For Pike, these are ‘units of functional or analytical conceptualized hypostasis’ and cannot be ‘on an entirely different plane of behaviour’ from ‘morphemes’, which are ‘form-meaning composites’ (LB 187, 162) (cf. 5.48). Yet ‘the entire semantic system is in a fluid state’: ‘human communication’ requires the ‘possibility of change in meanings’ (LB 623).⁴⁰ ‘Language’ ‘functions’ by ‘extending the meanings of words to a variety of contexts which are only vaguely related’ (LB 600) (cf. 4.16). A ‘theory’ of ‘semantic markers’ (e.g. Katz & Fodor 1963) is unrealistic in ‘omitting’ a speaker's ‘knowledge “about the nature of their world and about momentary situational, motivational and linguistic contexts”‘ (LB 628; Osgood 1963:738) (cf. 7.67, 77; 13.59).

5.77 A final problem is ‘hypermeaning’, appearing when ‘participants’ identify’ ‘two or more utterances’ ‘as having the same meaning’ ‘even though the morpheme sequences’ are ‘different’ (LB 612f). This direct appeal to ‘informant reaction’ allows for ‘hypermeanings’ differing among various ‘native speakers’, e.g. ‘scientists’ and ‘poets’, or showing ‘indeterminacy at the borders’ (LB 613f; cf. 5.39, 45, 47, 53, 87). ‘Hypermeanings can ‘enter into a hierarchy of meanings’ formed by ‘inclusion’ and ‘subtype’, and ‘come into play’ in ‘translation’ (LB 615f; cf. Nida 1955). They ‘may be considered as conceptual quanta’ that aid ‘useful and rapid reaction to one's environment’, but ‘may lead to stereotyping’ (LB 614).

5.78 To provide major demonstrations of his ‘unified theory of verbal and nonverbal behaviour’ Pike ‘discusses’ ‘large units of activity’ (LB 72; cf. LB 142), such as a football game, a church service, and a family breakfast. Having been present himself as participant and observer, he could collate his recorded data with his own impressions. We thus are not shown how an alien fieldworker could extract form, meaning, and purpose by observing such events. Being inside the culture, Pike attends the ‘homecoming’ game at ‘dear old Michigan’ (LB 100) without using observation to decide why people gather to witness a violent struggle (seldom played with the foot) for an odd-shaped ball that hardly even rolls or bounces.⁴¹ He can tell the ‘fans’ are animated by ‘nostalgia’ and ‘sentimentality’, the University by ‘public relations’, and the teams by ‘the purposes of playing the game according to the written rules’ (in ‘contrast’ to those of ‘baseball, tennis, hockey, or warfare’), gaining ‘the Conference lead’, and qualifying for ‘the Rose Bowl’ (LB 100f, 649). His insider knowledge also filters out ‘irrelevant’ or ‘unofficial, related activities’ and focuses on those governed by ‘official’ ‘rules and unwritten customs’ or announced in a document calling itself ‘“Official Program”‘ (LB 81, 99, 104, 649, 100).

5.79 Nor does Pike use the other techniques he recommends for discovery and fieldwork, such as consulting informants, making phonetic transcriptions, deliberately eliciting grammatical forms, or communicating by gesture as a substitute for language. So we do not get to see him at his best, analysing, say, a complete Mazatec ‘planting or harvest’ ceremony (cf. LB 27). In addition, he fails to take his own advice by ‘including the theorist’ in the ‘theory’ (5.18). This failure is serious if ‘each person to some extent constitutes a separate sub-culture’; ‘a common experience’, whether ‘verbal or nonverbal’, ‘never occurs for any two people’ (LB 51). He does not explicate his own sense of the occasions as typical ceremonies of the white middle class of the 1950s, where meals like ‘breakfast’ have an ‘official opening’ (‘saying grace’) and ‘closing’, where children must arrive at the same time (after being wakened by a ‘symphony on the phonograph’) and ask to be ‘excused’, and ‘have been taught to take turns’ ‘to talk’ or to ask for ‘the floor’ (LB 122-25, 193). A deep irony of some such ceremonies (like the church service) is that children's behaviour, other than sitting silent and motionless (i.e., non-behaviour), is seen as ‘misbehaviour’ (LB 84).⁴²

5.80 Another problem with the demonstrations is the staggering explosion of data implied, even though Pike wants to show not so much ‘the details of any complex behavioural pattern’ as the ‘structure’ of ‘wheels within wheels’ (LB 78f). For the church service, he considers time (‘a few minutes later’, ‘simultaneously’, ‘immediately’; ‘day’, ‘week’, ‘year’), space (‘auditorium’, ‘pew’, ‘aisle’), and (of course) the ‘continuum of physical activity’, ‘divided into significant major chunks’ ‘during which the purpose’ is ‘vigorously forwarded’, but also including ‘noise’ and ‘non-emic’ or ‘nondirected activity’ (LB 73f, 77f). The ‘behaviour sequence’ contains ‘pulsations of activity’; ‘segments end’ or ‘begin’ with every ‘appreciable change of activity’ (‘stand up’, ‘sit down’, ‘look up’, ‘sing’), ‘actor’ (‘organist’, ‘song leader’, ‘usher’), or ‘motion’ of a ‘body’ ‘part’ (‘arms’, ‘eyes’, ‘legs’, ‘lips’, ‘tongue’, ‘vocal chords’) (LB 75f). For the football game, Pike distinguishes (a) ‘human activity’; (b) ‘products of human behaviour’ which are either ‘relevant’ (‘stadium, field, goalposts, whistle, horn’) or are not (‘coats, hats, cigarettes’);⁴³ and (c) ‘behaviour of non-human elements which are not products’ (‘dogs’, ‘sun’, ‘winds’, ‘molecules’) (LB 118). The breakfast includes not merely ‘conversation’ with ‘slots for give and take, i.e. utterance and response’, either ‘integrated’ with ‘eating’ (‘refusal of bananas’) or ‘unrelated’ to it (‘husky warning’), but also the ‘physical setting’ (‘house’, ‘dining room’ ‘table’, ‘pots and pans’) with its ‘contrasts’ (‘between bowl and plate, or dining room and kitchen’) and ‘variants’ (‘heat of the stove’, ‘cloudiness of the sky’) (LB 125, 128). ‘“Adults eating a bowl of cereal” is an emic motif’ with its ‘purpose-meaning of sustenance’ and its ‘physical components’ (‘filling the spoon’, ‘swallowing’) (LB 151). ‘A toast-eating motif’ gets special focus for ‘a young child’ learning the ‘emic motif of buttering toast’ (LB 151, 292f) (an emetic buttereme?).

5.81 Pike concedes that ‘if all components of the spectacle were to be treated on a par, the data would be unmanageable and a description hopelessly unwieldy and unintelligible’ (LB 114f) (cf. 13.45). He hopes to ‘avoid this chaotic result’ by postulating ‘sub-assemblies of component hierarchies’ of ‘structurally related’ ‘items’ within the ‘activity’ (cf. 5.17, 36-40). But the data would still be enormous if we are resolved to describe a ‘total system, ‘event’, or ‘structure’ (5.8, 14, 24, 37, 57, 64f).

5.82 Or, a ‘congruent system’ of ‘permitted’ ‘types’ might be set up under the term ‘style’ (LB 132, 208, 599; cf. LB 463-67) (cf. 3.69; 4.40; 5.3, 64; 6.52, 54; 8.23, 70, 77, 83; 9.102f; 11.57, 86). Pike's behavioural outlook would focus on ‘the style of speaker at that particular moment’ (LB 152). The ‘style’ may be ‘careful’ or ‘uncareful’, ‘slow’ or ‘rapid’, ‘normal’ or ‘lively’, ‘trite’ or ‘literary’ (LB 316f, 412, 320, 343, 462, 550, 236, 605). ‘Informal’ or ‘colloquial’ contrasts with ‘formal style’, where ‘formal’ has the everyday sense of ‘rigid’ or'ceremonial’, e.g. in ‘public address’ (LB 72, 89, 155, 169, 208, 316, 343, 427).⁴⁴ ‘Style’ can also vary with ‘voice quality’, e.g., ‘high-pitched’, ‘harsh’ or ‘emphatic’, or used in ‘whisper, song, shout’, or ‘chanting’ (LB 311, 397, 400, 343, 378, 395, 582). ‘Dialects’ may also differ ‘somewhat as styles’ do; any one ‘style or dialect’ may ‘contain coexistent systems or fragments of systems’ (LB 582, 643).

5.83 ‘Style’ acts as a ‘systemic conditioning’ for ‘variants’ that ‘help to signal its presence’ and to make a ‘text’ ‘a uniform document’ (LB 168f, 208). Variants include ‘nouns’, ‘verbs’, ‘slots’, ‘lexical sets’, ‘morphemes’, ‘tagmemes’, ‘pauses’, and ‘vocabulary’ (LB 208, 235f, 169, 238, 466f). Pike recognizes the perplexing ‘implication that the system of morphemes must be determined separately for each’ ‘style’, even though we might later ‘find’ ‘many or most morphemes in common’ (LB 169). He reassures us that ‘styles share most of their units’ despite ‘different physical manifestations’ (LB 132). And ‘variants’ of ‘phonemes’ ‘are structurally the same in a topological sense’: ‘the two total patterns are identifiable point by point’ (LB 312).

5.84 Pike closes his book with an ambitious sketch of ‘analogies of society with language’ (LB 642).⁴⁵ Both sides have a ‘structure comprising a set of relationships within a network’; this ‘structure’ ‘can be detected only by observing individuals in action’, yet is ‘relatively stable, outlasting the lives’ of ‘members’ or ‘speakers’ (LB 643). ‘A society constitutes a system of individuals’ ‘just as a language has a network of sounds’, ‘syllables’, ‘words, and sentences’ (LB 644). ‘Groups of individuals’ resemble ‘phonemes, syllables, and stress groups’ in having ‘identifying criteria’ that (a) are ‘physical,⁴⁶ essential, universal or nearly-universal in all societies’, (b) ‘cut across the entire population’, and (c) are ‘functioning units in the society’ (LB 646). ‘The formal features of society’ ‘identify and contrast it with other societies’ and include ‘language as part of a communication system’ integrated with ‘habitual shared patterns of behaviour essential for coherence in self-awareness, for maintenance of life’, and for ‘carrying out tasks’ related to ‘economic, normative, political, or educational’ ‘goals’ (LB 644).

5.85 If ‘anthropologists’ find ‘this approach’ ‘merely formal’ and ‘lacking in insight’ into ‘deeper problems’, ‘the same difficulties may be seen in linguistic analysis’, whose devotion to ‘phonemes and morphemes’ overshadows ‘the ultimate goal of the study of communication’ (LB 642). ‘Nevertheless’, ‘formal studies have been very stimulating to understanding the “mechanics” of language activity’. But those studies have succeeded only through drastic limitations on the data; they would explode if applied to a whole ‘hierarchy’ of ‘personal activities and history’ and to each person's ‘total personal or social participation’, including ‘movements’, ‘facial expressions’, ‘utterances’, ‘restless activity’, and so on (LB 646, 115ff) (13.45). ‘The individual is as least as complex in his internal structure as the language he speaks’ (LB 655). ‘Something within his structure’ may be ‘responsible for the structuring of his language’, though Pike is ‘not prepared for’ ‘seeking’ ‘analogous materials in detail’.

5.86 And after all, the wide scope of ‘tagmemics’ is precisely what sets it apart from formal studies. Pike has little use for ‘formal algebraic systems’ or ‘mathematical networks of abstracted relationships’, such as the ‘algorithmic view of the theory of grammar’, ‘automata’, and ‘machines'“‘ (LB 501, 645, 69; Bar Hillel & Shamir 1960:156) (cf. 2.70, 82; 3.72f; 6.8; 7.40, 7¹⁸; 9.110; 11.14; 13.17, 50). A ‘theory influenced by logic sets up axiomatic affirmations’, ‘presents a mechanical device’ like ‘a known mathematics’ ‘to predict some’ ‘phenomena of the real world, and tests’ ‘this prediction’ against ‘a few observed data’ (LB 68f). But a ‘theoretical system’ can have fully ‘axiomatic form’ only when it remains ‘uninterpreted’ and is hence not ‘applied to facts of nature’ -- ‘a “mere calculus”‘ ‘floating in the air’ and ‘constructed’ ‘from the top’ rather than ‘anchored at the solid ground of the observable facts’ (Hempel 1952:33, 53f; Carnap 1955:210, 207) (cf. 6.56, 64). Also, ‘mathematical notation’ makes no ‘reference to meaning or purpose’ (LB 663).

5.87 Distinctly non-formalist also is Pike's thematic recognition of ‘indeterminacy’ in both theory and data (e.g. LB 64, 159f, 192, 230, 232f, 237, 251, 356, 546, 552f, 594; cf. 5.16, 34, 36f, 39, 45, 47, 49, 53, 70, 72, 77; cf. Bazell 1952, 1953). For ‘fidelity of description’, ‘the indeterminacy in the theory’ should ‘reflect’ that ‘in the activity of the community’; an ‘arbitrary attempt’ at a ‘“clear cut” theory’ might ‘conceal’ ‘the facts’ and ‘do violence to the structure’ (LB 129, 159) (13.27). ‘Indeterminacy’ might ‘never be resolved’ or even ‘increase as one attempts to resolve’ it ‘more vigorously’ (LB 222, 158; cf. LB 248, 356; 13.52). Pike's preoccupation with units and segments (a particle view) is thus attenuated by his attention to the ‘indeterminacy’ of their ‘borders’ or ‘boundaries’ (LB 45, 79, 95, 113, 116, 132, 180 342, 381f, 468, 552f, 585, 645f) (a wave view) (13.59).⁴⁷ ‘Well-described units’ need not be ‘well-defined units’ (LB 121n).

5.88 As I remarked at the outset, Pike's book shows us work in progress, a continuing effort to develop a complex theory based on a limited set of notions for a wide domain. To satisfy the scientific climate of the times, he tries to remain loyal to a conception of objective observation whose last recourse is always the physical domain (5.18, 25, 27ff 31, 33, 42, 44, 50, 63f, 65-68, 70, 80, 83f). The complexity of language had to be reconstructed by multiplying levels or hierarchies and justifying them by the (at least partially) different borders of the respective units (5.36ff 41). Pike tends to proceed as if he felt reality of units and constructs were somehow in the material itself. Yet he acknowledges that the analyst is more likely than the ordinary native speaker to be aware of them (5.11, 13f, 16, 18, 22, 46, 48, 71; cf. 13.49). If language ‘causes’ a ‘hearer’ or a ‘community’ to react (5.15, 66), linguistic analysis is a special reaction, and Pike gets himself into his sights when he confronts technical factors (e.g. discovering the ‘tagmeme’) but not raw data (e.g. attending a church service). He does not, therefore, seriously question whether his ‘unified theory’ can genuinely be achieved by reconstructing the organization of discourse in terms of slots and fillers. That view may appear plausible to the analyst after the fact, when suitable end-results and protocols are made available, but hardly plausible to the participants during the event, when the discourse is in progress.

5.89 Still, Pike's method fostered the incontestable accomplishments of describing hundreds of previously little-studied languages and widening the scope of linguistics (5.2). He strove not ‘solely to catalogue units, or provide expanded paraphrases’, but to provide a ‘description’ that is ‘useful for productive’ ‘purposes in the community setting’, such as ‘learning to speak’ or ‘read’ (LB 121n, 493, 388; cf. LB 43, 51f, 65, 68, 352) (cf. 13.61).⁴⁸ Moreover, ‘the human observer’ he presents ‘resists being dissected into logical parts’ or ‘forced into a single logical-coherent Procrustian view or set of one-dimensional rules’, and ‘demands the right of multiple perspective’ as he ‘reacts to criss-crossing, intersecting vectors of experience, mental tools, values, and psychological presets’ (LB 10). By insisting that ‘the structure of language shares’ many ‘characteristics with the structure of society’, Pike ‘hopes to demonstrate’ ‘more pervasive’ ‘structural traits’ of ‘man’ ‘than have previously been suspected’ (LB 641). ‘Behaviouremic theory’ may ‘bring into coherent, organized relationship many facts’ ‘which before this were isolated, ignored, buried in footnotes, or treated in an offhand manner’; or may ‘lead to an observation of new relationships in old trouble spots of theory’ and ‘stimulate the creation of new hypotheses’ (LB 519f). At the end of the quest may lie ‘a theory, a set of terms, and an analytical procedure’ to make ‘intelligible’ ‘all human overt and covert activity’, ‘all psychological processes’, all ‘responses to sensations, all of thinking and feeling’ (LB 32).

NOTES ON PIKE

¹ The key for Pike citations is LB: Language in Relation to a Unified Theory of the Structure of Human behaviour (1967 [1945-64]). The title maybe intended to echo the Encyclopedia of Unified Science, of which Pike cites the volumes by Carnap, Hempel, and Morris (cf. 4.7).

² Actually, disciplines like ‘psychology’, ‘psychiatry’, and ‘personality theory’ (with its familiar triad of ‘“id, ego, and superego”’) appear only sporadically; ‘anthropology’ is more prominent, due to its interest in a ‘universal cultural pattern’ (Wissler 1935), and ‘standardized’ ‘behaviour’ (Nadel 1951) (LB 537, 674f, 144f; cf. LB 71, 183, 444, 641).

³ Pike ‘remembers the surprise’ of learning that ‘the two “p” sounds in “paper” were not the same’, and he needed ‘two years before he heard the difference easily’ (LB 45f).

⁴ According to Opler (1948:116), ‘culture is what the investigation of its carriers by the anthropologist proves it to be, not what informants think it is’; their ‘rationalization, idealism, self-righteousness, and hope’ are ‘part of the culture’, but ‘not a definition’ of it (LB 157).

⁵ Yngve (1960:445) proffers the ‘immediate constituent framework’ as ‘a model for sentence production’ with the ‘rules’ ‘unordered’ ‘in the memory’ (LB 479) (cf. 7.54, 11.33, 10¹⁸).

⁶ ‘The etic approach’ ‘might well be called comparative’ if that term did not suggest an approach for ‘reconstructing parent forms’ (LB 37; cf. 2.5, 10, 53, 63; 3.19f; 4.1. 73; 12.90f). Yet the etic method is mostly applied in Pike's book to one culture at a time (cf. 5.78ff). And for linguists like Hjelmslev and Chomsky, a systematics for ‘all languages’ would be emic (cf. 6.11, 35; 7.20).

⁷ Quine's (1953:14) claim -- ‘“logicism holds that classes are discovered, while intuitionism holds that they are invented”‘ (LB 58) -- seems backwards to me. Surely logic invents classes, and intuition is a domain where real classes of things are rarely questioned. Quine hedges by siding logic with ‘realism’ and yet making it ‘epistemologically’ rather than ‘physically fundamental’.

⁸ ‘The emic system’, however, counsels against ‘accepting too readily’ alternative ‘analyses which appear to be equally valid but contradictory’ (LB 56). The ‘etic criteria’ were revised, at the ‘suggestion of Longacre’ (1964a), to eliminate ‘alternates’ being ‘theoretically correct even when one’ ‘was empirically undesirable’ (LB 471).

⁹ Or, one could expect the emic to be ‘generalized’ and ‘abstract’, and the etic to include ‘living parts of an actual sequence of behaviour events’ (LB 41).

¹⁰ Examples include ‘homophones’, ‘mimicry’, ‘irony’, and ‘a lie’ -- outward sameness, but different intentions (LB 43, 132f). Pike describes a ‘lie’ differing from ‘a parallel normal sentence’ by ‘breaking the normal connection between’ ‘sentences and reality’ (LB 602; cf. 5.66, 76f) -- but surely many hypothetical or counterfactual statements do so. Elsewhere, ‘deliberate deception’ is more appropriately described as a case of a ‘hidden purpose’ or an ‘inappropriate response’ (LB 42f, 105, 115).

¹¹ An ‘etic approach’ might ‘ignore meaning or purpose’, apply ‘absolute’ ‘criteria’, and rely on ‘instrumental measurements’ (LB 44, 38, 46). Or, it might use only ‘differential purpose and meaning’, but not ‘the full systems’ (LB 44; cf. 4.26, 31; 5.14). These provisions suggest that some American linguists were only working etically (cf. 5.61).