¹ Martin Joos, Lang. 12.196-210 (1936).

² G. K. Zipf, The Psycho-Biology of Language (1935).

³ L. Bloomfield, Language 385.

⁴ D. C. Jones, A First Course in Statistics 18 (London 1927).

⁵ This same arrangement of the material will also answer the hasty judgment in Joos (204): ‘He thoroughly confuses the synchronic and diachronic aspects of language-description… ’ The procedure of viewing the diachronous as a series of synchronous states is one with which few exact scientists would care to quarrel.

⁶ This has nothing to do with secondary sets of corresponding forces which are frequently assumed as present in most if not all behavior (and may easily be assumed here). Similarly (with Joos fn. 12) the factors involved are rarely simple, but presumedly very complex. For convenience however one speaks of e.g. intelligence or emotion without an ever-present qualification to the effect that they are probably very elaborate complexes because this qualification is invariably understood by persons familiar with the phenomena. Joos’s mention of ‘ spurious correlation’ (ibid.) would hence have been more telling if he had himself demonstrated the presence of the ‘third or several other causes’ of which the entire correlation might be considered as effects. They may indeed be there but I cannot find them and hope that someone else will.

⁷ ‘ Cosmic import’ (Joos 199) seems to be his gratuitous paraphrase of my feelings (as he reconstructs them) at the time of the discovery. Other scholars have been more generous toward me, possibly because they have personally experienced similar occasions of discovery.

⁸ W. Empson, The Spectator 270 (London, Feb. 14, 1936). Empson states 20,000 as the limit.

⁹ Joos 199, is correct about the nature of the diagonal line if one takes into consideration the chart on Plate IV.

¹⁰ Correctly stated Joos’s analogy (202) should conclude ‘certain formative elements are comparatively rare’, a conclusion which does not warrant his implicit assumption that Ames occurs less than 51 times in 400,000 running words of French.

¹¹ G. K. Zipf, Relative Frequency as a Determinant of Phonetic Change, Harvard Stud. Class. Phil. 40.1 (1929).

¹² R. A. Fisher, Statistical Methods for Research Workers 9 (London 1934).

¹³ That a ‘ rigid procedure for determining phoneme-membership’ may have its own difficulties I call attention to N. Trubetzkoy, D’une theorie des oppositions phonologiques, Jour, de Psych. 33.5-18 (1936), to my mind a masterly treatise on defining the phoneme, where (6) he expresses the difficulty of determining even whether the German affricate, is, is to be viewed as a single phoneme or as a group of phonemes. Is all quantitative research on any aspect of the phoneme to wait until this point is settled?

¹⁴ And not so unverifiable either; cf. C. V. Hudgins and R. H. Stetson, Voicing of Consonants by Depression of Larynx, Arch. Neer. de. Phon. Exper. 11.1-28 (1935) in which we find experimental support of Psycho-Biology 67; I am grateful to Dr. Hudgins for pointing this out to me. Similarly all the work in Phonometrie, e.g. E. & K. Zwirner, Phonometrischer Beitrag zur Frage der NHD Lautmelodie, Vox, 21.45-70 (Dec. 1935). My procedure, instead of being as described by Joos 208, is far more simple and incidentally thoroughly familiar to scientists: one can measure phonemes conceivably in many ways, including magnitude of complexity and relative frequency. Though relative frequency can be readily established, magnitude of complexity can be determined reasonably surely for only a comparatively few sets of phonemes. By establishing a correlation between relative frequency and magnitude of complexity in cases where both are measurable, we can make inferences about the magnitude of complexity of cases where only the relative frequency is observable.

¹⁵ The quotation from Joos has a familiar ring, cf. my Relative Frequency 40, or Psycho-Biology 106.

¹⁶ Cf. G. Panconcelli-Calzia’s fundamental work on the voicing of stops reported in Vox, passim.

¹⁷ My Relative Frequency 51. If Joos’s ‘good warrant’ 209 is correct for Russian, Bulgarian, and Czechish, it will make my statistics on these languages even more marked in the direction I indicate. Bulgarian is alphabetic in my tabulation, Russian though phonetic in my original presentation (ibid. 44-5) is nevertheless probably alphabetic (cf. ibid. 45 fn.1).

¹⁸ Spanish b and v are pronounced the same under the same conditions but may not for that reason be interchanged on the basis of an ‘orthographic equality’ as stated by Joos (209). Cf. T. Navarro, T. & A. M. Espinosa, A Primer of Spanish Pronunciation, 36f. (1926) : ‘The letters b and v are pronounced alike in Spanish, although orthographieally they are not interchangeable. … ’ Joos’s ‘v = 5’ states that they are interchangeable.

¹⁹ E. L. Thorndike, Jour. Ed. Psych. 27. 391 (1936).

²⁰ For a thorough and careful discussion of the methodological implications of these fundamental differences cf. J. F. Brown, Psychology and the Social Order 3-103 (1936).

²¹ E.g. C. L. Stone, Jour. Abnorm, and Soc. Psych. 30. 546 (1936): ‘In fact, Zipf has demonstrated convincingly that the social psychologist must add one more prerequisite for his work, dynamic philology.’

²² Somewhat different from Joos 196:‘the treatment almost uniformly evidences a belief that the author has attained valid formulations.’