¹ T. Navarro-Tomás, Manual de pronunciación española ⁴ (Madrid, 1932), and, most important, Manual de entonación españda ² (New York, 1948). From the latter, other studies, none so extensive, have been largely derived, and have been similarly impressionistically phonetic rather than structural.

² See J. Donald Bowen, A comparison of the intonation patterns of English and Spanish, Hispania 34.30–5 (1956).

³ See Henry Lee Smith Jr., review of Intonation of conversational English (educated Southern British), by Wiktor Jassem, in Lg. 31.150–3 (1955); Linguistic science and the teaching of English (Cambridge, 1956) ; The communication situation, Department of State lecture outline (1954); also Uriel Weinreich, Stress and word structure in Yiddish, The field of Yiddish 15 (New York, 1954). The most complete treatment of vocalizations published so far will appear in Scientific linguistics and psychiatry, by Robert E. Pittenger and Henry Lee Smith Jr., in mimeograph but unpublished at this writing (and subject to title change), to appear in the journal Psychiatry.

⁴ For more discussion relevant to this point in English, see George L. Trager and Henry Lee Smith Jr., An outline of English structure (Norman, Okla., 1951; 2d printing, Washington, 1956) ; also Charles F. Hockett, A manual of phonology 45 ff. and 65 ff. (Baltimore, 1955).

⁵ Concerning dialects our information is based largely on firsthand observation, and necessarily incomplete. Prior to this writing, we have worked closely for three years with native speakers of Spanish from the following areas: central and northern Spain, urban Mexico, urban Guatemala, urban Costa Rica, urban Venezuela, upland Colombia, upland Ecuador, west-coast Peru, urban Chile. In addition, two of us have, within the past two years, visited every Spanish-speaking capital in the western hemisphere, though in each place for only a week or so. One of the secondary purposes of these visits was to gather dialect information. Siiva-Fuenzalida is a native of Santiago, Chile, and has done extensive field work throughout Chile. Among the informants who have aided us immeasurably in collecting data, and to whom we are deeply grateful, are Hugo Monterò, Guillermo Segreda, María Kothe, Rodrigo Gurdián, Victor Sánchez, Piedad Becerra, Oscar Salazar, Ernesto Williams, Hugo Pineda, Rosalinda Pineda, Bianca Russell, Enrique Forlong, Juanita Jenkins, Jesús Ríos, Yolanda Lastra, Dolores Oviedo, Mónica Da Rosa, Victoria Angulo, and Antonio Linares.

⁶ The phonemic interpretation of semivowels in Spanish, Lg. 31.236–40 (1955).

⁷ It is, of course, recognized that no one has, to our knowledge, actually described the conditioning on this higher level.

⁸ See Silva-FuenzaJida's analysis of the distribution of the allophones of /n/ in Estudio fonológico del español de Chile (Santiago, 1953).

⁹ The symbol [# ] means retardation of tempo with drop in pitch level, accompanied by gradual ceasing of phonation if followed by [...]. Neither phonemic nor phonetic pitch levels are indicated, since they are not pertinent to the examples, and only two levels of stress are indicated, strong by acute accent, weak unmarked, even though there are more than two levels (discussed below). Three dots mean a pause of some length (indeterminate : long enough that a second speaker knows that he may now speak without having ‘interrupted’ the first speaker). [=] means retardation of tempo with no change in pitch level, accompanied by abrupt ceasing of phonation if followed by [...].

¹⁰ Even in Chilean, where final [ŋ] is otherwise unattested.

¹¹ Progressive assimilation as here used means that the first of two phones becomes more like the second; regressive assimilation means that the second of two phones becomes more like the first.

¹² The phonemic interpretation of semivowels in Spanish. Since part of the utility of an analysis which includes plus juncture in Spanish depends on the validity of the interpretations in the earlier article, we insert here a further bit of evidence favoring that analysis which we had previously overlooked. There are contrasts which completely eliminate the argument that the nonsyllabicity of /i/ and /u/ is predictable next to other vowels; these contrasts are furthermore, for what it may be worth, indicated in the traditional orthography. Thus /piáda/ pīada ‘chirping’ vs. /pyále/ piale ‘cast of the lasso’; /piadór/ pïador ‘chirper’ vs. /pyanóla/ pianola ‘pianola’; /duéto/ dūeto ‘duet’ vs. /dwélo/ duelo ‘duel’; /duérno/ dūerno ‘double sheet’ vs. /dwéño/ dueño ‘owner’.

¹³ Sol Saporta of Indiana University, Frederick B. Agard of Cornell University, and Harold V. King of the Centro Cultural Costarricense-Norteamericano. As this article was being prepared in final draft, we learned that Saporta was publishing in this journal further discussion of the matter. He was kind enough to send us a copy of that discussion, and we have added some of his examples without modifying our analysis of the problem.

¹⁴ This is a point on which we do not wish to seem more vulnerable than we are. We recognize that this very matter will cause advocates of the alternative analysis (/aiūna/ for hay una vs. /ayúna/ for ayuna) to be more firmly convinced of the validity of their position than any other unresolved area of disagreement. We must point out that the language really is not very neat at this point, and that both solutions are about equally vulnerable. The vulnerability of our solution we have already acknowledged—it forces plus juncture into places where we would rather not have it, as in the nonfricative pronunciation of the palatal in ayuna. The vulnerability of the alternative solution is well stated in a letter from Harold V. King, dated 7 March 1956, which we quote with his permission: ‘There is one thing which I think will continue to bother me about this for a long time. As of now, I think forms like bien contain an occurrence of the phoneme /i/. And this allophone of /i/ contrasts with /y/. However, those speakers who have a non-fricative /y/ (like many Josefinos and most Bogotanos) sound to me as if they are making the same noise in mayor that they make in hielo. In other words, my difficulty is that some occurrences of /y/ sound practically the same as some occurrences of /i/. The main thing that makes them different is that among many other speakers a fricative y is much more frequent in mayor than in hielo.‘

¹⁵ It is assumed that strong stress can be identified even though its allophones have not been described. Note that the allophones of the pitch which occurs at this point are not described in terms of the stress; only the location of them.

¹⁶ Open juncture is assumed not to occur in the utterances here transcribed, though it may or may not in any one of them.

¹⁷ We would like to lay once, and we hope for all, the ghost of the descriptive approach which would say that weak stress is not a phoneme—it is only the absence of strong stress, so that there is only one stress phoneme in Spanish, which contrasts with the absence of stress. We hold that it is important to recognize an essential difference between segmental phonemes and suprasegmental phonemes : a segmental phoneme may contrast with its own absence: cat vs. at; but no utterance can be made without pitch or stress or, we believe, juncture phenomena being acoustically present. Since they are acoustically present, they either differentiate utterances or they do not. If they do not, the conditions that govern their occurrence and distribution must be stated. If they do, there cannot be fewer than two of each type that is significant, since they lack the possibility of being absent from an utterance altogether like a segmental phoneme.

¹⁸ George L. Trager, The phonemes of Castillian Spanish, Travaux du Cercle linguistique de Prague 8.217–22 (1939).

¹⁹ Unless, that is, a full primary stress is actually present. This is often the case, especially with enclitic pronouns: /dígamelo/ can be /digameló/, and usually is so in a formal style of speaking; in certain dialects, such as that of Montevideo or Santiago de Chile, this style seems to be generalized to completely colloquial usage.

²⁰ Parenthetically, our English-speaking students need constantly to be corrected for their tendency to put medial stress on the final syllable if it is at two or more removes from the primary, and English speakers tend to hear extra loudness here where we hold that none is present.

²¹ The phonemic interpretation of semivowels in Spanish.

²² Medial stresses on /en/ and /in-/; we thought that [oè] and [aì] occurred in our original data without shortening of the first vowel, corroborating the position of plus as determined by medial stress.

²³ Medial stress on /tra-/.

²⁴ Medial stress on /pa-/.

²⁵ Medial stress on /te-/.

²⁶ The phonemic interpretation of semivowels in Spanish 240.

²⁷ In our data up to this time we find no utterances with more than three phonemically strong stresses between the boundaries above stated. When we try to expand the utterances so constructed, a terminal juncture is invariably inserted to break the material into two phonemic phrases, each containing three strong stresses or fewer. We are, however, not yet sure that this is structurally mandatory; in fact, we doubt that it is.

²⁸ /dyó/ strongest even though on /¹/.

²⁹ /dí-/ strongest.

³⁰ /ká-/ strongest, again even though on /¹/.

³¹ /dón-/ strongest, though not last, because coincident with /³/.

³² /bá/ strongest, though on /¹/, because last phrase.

³³ /ké/ strongest in the first phrase, being coincident with the first /³/, but /-dád/ stronger still because last phrase.