SPE Features
The sound pattern of English is a major landmark in feature theory, a watershed in the history of phonology (Anderson 2021), and most in-depth works still cite it as their starting point. Often abbreviated SPE, the book was published by Chomsky & Halle in 1968, though it was in preparation years earlier Halle (1962:54). SPE posits that features have two different functions, a phonetic and a classificatory function. The phonetic function allows scales with a fixed number of values, and reflect “independently controllable aspects of the speech event or independent elements of perceptual representation” (Chomsky & Halle 1968). In other words, the phonetic function of features can be scalar values of whatever a speaker can articulate or perceive. In contrast, the classificatory function of a feature allows only two coefficients, + or -. Each feature is a “physical scale defined by two points, which are designated by antonymous adjectives” such as high-nonhigh, and which are specified as [±high]. Feature definitions are given primarily in articulatory terms. While the authors acknowledge features have acoustical and perceptual correlates, they do not explore those in depth due to considerations of space. Features are bundled into a rectangular bracketed matrix of features, typically in the order discussed here (an overview of which is in the Appendix at the end of this entry).
One of the earliest features employed is [segment], which distinguishes boundary types from segments, which in term can be fully specified, or partially specified as archi-segments.
SPE begins its segmental analysis with the major class features, a set of three features that provide the basis for the major divisions of vowels, consonants, obstruents, sonorants, liquids, and glides. Two have been carried over from Jakobsonian features, consonantal and vocalic, while sonorant was an innovation. Sonorant sounds are produced with a vocal tract configuration in which spontaneous voicing of the vocal folds is possible. In obstruents, this is not possible because constrictions more radical than those of glides like [w] and [j] were incapable of allowing spontaneous voice. In a similar fashion, vocalic sounds could not have constrictions greater than those found in the high vowels [i] and [u]. Consonantal sounds are produced with a radical obstruction in the midsagittal region of the oral tract.
Chomsky & Halle (1968:303) present the following table, which shows how the three major class features characterize major groups of sounds:
sonorant consonantal vocalic
voiced vowels + – +
voiceless vowels + – –
glides w, j + – –
glides h, ʔ + – –
liquids + + +
nasal consonants + + –
nonnasal consonants – + –
The next set, Cavity features, capture the activity in the main cavity of the vocal tract and are divided into seven groups: Primary strictures, features relating to the tongue body, lip rounding, tongue contact, covered, glottal constrictions and secondary apertures. Chomsky & Halle (1968:303-304) critique the Jakobsonian approach, but admit that a change in nomenclature makes their system seem more radically different than the Jakobsonian features. This is most evident in the primary stricture features [coronal] and [anterior], which displace [compact], [diffuse], and [grave]. Coronal sounds are produced with a raised tongue blade, capturing dental, alveolar, and palato-alveolar (i.e. postalveolar) sounds, to the exclusion of labial, velar, and uvular sounds. Anterior sounds are made with an obstruction in front of the palato-alveolar region, grouping labials, dentals, and alveolars and separating them from palato-alveolar, retroflex, and other posterior sounds, including pharyngeals. These main place distinctions were supplemented by tongue body features [high], [low], and [back]. The feature [high] was defined to characterize sounds with the tongue body raised above neutral position (roughly [ɛ], p. 300), while [low] is below that position, and [back] sounds are made by retracting the tongue body from neutral position. Sounds by definition cannot bear the specification [+high, +low]. These features interact with place features to create four posterior places of articulation:
palatals velars uvulars pharyngeals
[high] + + – –
[low] – – – +
[back] – + + +
(Chomsky & Halle 1968: 305)
The vowel features also worked to describe secondary articulations such as palatalized (high and nonback), velarized (high and back), and pharyngalized (low and back) sounds. And for vowels themselves, [+high] replaces Jakobsonian [diffuse], [+low] replaces [compact], and [+back] replaces [diffuse].
The feature [rounded] is applied to both rounded vowels like [u, o, y] as well as glides like [ɥ] (symbolized as [ẅ] in SPE) and labialized consonants.
To account for sounds like the difference between true dental (vs. alveolar) sounds, SPE posits the feature [distributed], defined as having “a constriction that extends for a considerable distance along the direction of the air flow” (SPE p. 312). The feature [covered], applied only to vowels, tentatively refers to sounds with a “pharynx in which the walls are narrowed and tensed and the larynx raised”. [+covered] sounds have since come to be described as [-ATR] (Advanced Tongue Root), after Stewart (1967).
The sixth Cavity feature describes glottal constriction, which does not obviously belong as a Cavity feature since it references activity within the larynx. The feature, in subsequent work dubbed [constricted glottis], covers tense glottalized stops, and laryngealized and creaky voice.
There are two secondary aperture features, [nasal] and [lateral]. The feature [nasal], as in Jakobson, is produced “with a lowered velum which allows air to escape through the nose” (SPE p. 316). The novel feature [lateral] is introduced and erroneously restricted to coronal sounds. It is produced “by lowering the mid section of the tongue” on one or both sides, and covers not only the “vocalic lateral (liquid)” [l], but lateral fricatives and affricates such as [ɬ, ɮ, t͡ɬ ].
The Manner of Articulation features include [continuant], release features, and supplementary movements. The feature [continuant] is much the same from Jakobsonian features, distinguishing oral and nasal stops and affricates from other sounds, as well as contrasting oral and nasalized vowels. The release features introduce [delayed release], used for affricates, while [instantaneous release] describes a normal plosive. Previously, Jakobson, Fant, and Halle (1963) used the feature [strident] to capture the delayed release of affricates. Chomsky & Halle (1968) sketch two release features for primary and secondary closures. The release feature of a primary constriction distinguishes affricates from plosives, also seen in lateral affricates, which have a lateral release different from the lateral liquid. A feature for release of secondary closure is proposed to account for clicks, which involve two or three simultaneous closures.
Two supplementary movement features, suction and pressure, are proposed for velaric and glottalic airstreams. The feature [suction] has two phonetic realizations: one for the downward movement of velar closure during production of a click sound, and the other for downward glottal movement of implosives. The opposite phenomenon, [pressure] could not be verified for velar pressure, but is posited to account for the compression that is characteristic of ejective sounds.
The final manner feature is [tense], for sounds involving “a deliberate, accurate, maximally distinct gesture that involves considerable effort” (SPE p, 324). Tense vowels display greater deviation from neutral position–they are more peripheral in the vowel space–than their lax counterparts. Tense consonants, associated with voiceless consonants, are [+tense], while voiced consonants are laxer, allowing vocal fold vibration.
The next category of segmental features are the source features: [heightened subglottal pressure], [voiced], and [strident]. Heightened subglottal pressure is a necessary but not sufficient condition for aspiration, which also requires no glottal constriction, so this feature was used primarily for aspirated obstruents. The feature [voiced], long recognized, is little changed, though it is discussed in the context of Lisker & Abramson’s (1964) study of voice onset time, with voicing lead, coincidence, moderate lag, and substantial lag, cross-classified with the four phonetic features of voicing, tenseness, glottal constriction, and subglottal pressure. A final source feature is [strident], as in Jakobsonian features, marked by “greater noisiness, generated by aerodynamic turbulence. SPE restricts [strident] to obstruent continuants and affricates, declaring plosives and sonorants nonstrident (Chomsky & Halle 1968:329).
Prosodic features are covered in three brief sentences in which the authors admit that their investigation has not progressed to a point of useful discussion. They did not cross-reference the one prosodic feature that they employed in earlier chapters–extensive use of the feature [stress]. Chomsky & Halle’s (1968) analysis of English stress employed a scalar value of four levels of stress, found in primary, secondary, compound, and unstressed syllables. The list of phonetic features found on page 300 of SPE anticipated eight prosodic features: stress, length, and six pitch (tone) features (high, low, elevated, rising, falling, and concave), but these are not explored.
Chapter Nine of SPE develops a theory of markedness such that only marked values of features need be specified on certain segments, e.g. [nasal] is marked for segments like [m n ɲ ŋ], but in unmarked for oral sounds. Full feature specifications are supplied by marking conventions and redundancy rules.
Diacritic features are proposed (Chomsky & Halle 1968: 373 and passim) to indicate that grammatical rules apply only to certain lexical items. For example, velar softening includes the change of k to s in words like criti[k] vs. criti[s]ize in words of Romance origin. Other diacritic features include, for Modern Russian, binary features [Slavic] and [Russian], or [H] for morphemes that undergo harmony in Igbo. SPE even discusses the possibility of diacritic features to mark exceptions to specific phonological rules, although this increases dramatically the power of the phonological component of the grammar.
Goyvaerts & Pullum (1975) contain a number of important essays reacting to SPE. Kenstowicz (2022:407-413) contains an insightful and concise overview of the features of SPE, concluding that ‘[i]n the fifty or so years since its publication no single work has come close to SPE’s combination of theoretical originality and descriptive insight’ (2022:415).
Unlike the state of Jakobsonian phonological features, which were developed and debated among the professional phonologists in the field of linguistics, the Chomskian Revolution of generative linguistics propelled the growth of the field and rapidly propagated the SPE feature system to generations of students, both graduate and undergraduate. Although Harms (1968) made some use of late Jakobsonian features within a generative framework, it was a transitional work. Within a few years after SPE, just in the 1970s, there appeared Schane (1973), Anderson (1974), Hyman (1975), Sommerstein (1977), Sloat et al. (1978), Kenstowicz & Kisseberth (1979) as major textbooks for varying levels, with a dozen or more appearing in the 1980s. These works clearly and simply defined phonological features, taught them to students, and refined the set of features through their use in problem sets both culled from the literature and brought to document and support the features for pedagogical purposes.
Chomsky & Halle (1968: 300) anticipated the subsequent development of feature theory, suggesting that “ultimately the features themselves will be seen to be organized in a hierarchical structure which may resemble the structure that we have imposed on them for purely expository reasons”. In fact, the development of feature geometry (Clements 1985, Sagey 1986, Clements & Hume 1995) did impose greater hierarchical structure on features. Major class features were incorporated into the Root node of a segment. Source features comprise the Laryngeal node. Cavity features are mostly within the Place node, and Prosodic features are found in suprasegmental levels of structure such as length in the skeletal tier, stress in the syllable and foot groupings of metrical phonology.
Appendix: List of Features described in The sound pattern of English (Chomsky & Halle 1968)
Segment
Major class features
Cavity features
Tongue-body features
Secondary apertures
Lateral
Manner of articulation features
Release features: instantaneous and delayed
Primary release
Secondary release
Supplementary movements
Velaric suction (clicks)
Implosion
Velaric pressure
Ejectives
Source features
Heightened subglottal pressure
Pitch
Elevated
Rising
Concave
Length
Diacritic Features
References
Anderson, Stephen R. 1974. The organization of phonology. New York: Academic Press.
Anderson, Stephen R. 2021. Phonology in the twentieth century: Second edition, revised and expanded. (History and Philosophy of the Language Sciences, 5). Berlin: Language Science Press.
Goyvaerts, Didier L.; and Geoffrey K. Pullum (eds.) 1975. Essays on the Sound Pattern of English. Ghent: Editions Story-Scientia.
Halle, Morris. 1962. Phonology in generative grammar. Word 18.54-72.
Harms, Robert T. 1968. Introduction to phonological theory. Englewood Cliffs, NJ: Prentice-Hall.
Hyman, Larry M. 1975. Phonology: Theory and analysis. Fort Worth, TX: Harcourt Brace Jovanovich.
Jakobson, Roman; C. Gunnar M. Fant; and Morris Halle. 1963. Preliminaries to speech analysis: The distinctive features and their correlates, 4th edn. Cambridge, MA: MIT Press.
Kenstowicz, Michael. 2022. The sound pattern of English and early generative phonology. The Oxford history of phonology, ed. by B. Elan Dresher and Harry van der Hulst, 396-415. Oxford: Oxford University Press.
Kenstowicz, Michael; and Charles Kisseberth. 1979. Generative phonology: Theory and description. New York: Academic Press.
King, Robert D. 1969. Historical linguistics and generative grammar. Englewood Cliffs, NJ: Prentice-Hall.
Lisker, L.; and A.S. Abramson. 1964. A cross-language study of voicing in initial stops: Acoustical measurement. Word20.384-422.
Schane, Sanford A. 1973. Generative phonology. Englewood Cliffs, NJ: Prentice-Hall.
Sloat, Clarence; Sharon Henderson Taylor; and James E. Hoard. 1978. Introduction to phonology. Englewood Cliffs, NJ: Prentice-Hall.
Sommerstein, Alan. 1977. Modern phonology. London: Edwin Arnold.
Stewart, J.M. 1967. Tongue root position in Akan vowel harmony. Phonetica 16.185-204.
Author
Paul D. Fallon