1. Phonological Theories
Session 2
Phonological Theories
From the Phoneme to Distinctive Features
Bibliography (works cited in Wk. 2)
Anderson, S.R. (1985) Phonology in the Twentieth Century. Theories of Rules and Theories of Representations. Chicago, Chicago University Press.
Fischer-Jörgensen, E. (1975) Trends in Phonological Theory. Copenhagen, Akademisk Forlag.
Saussure, F. de (19484) Cours de linguistique générale. Lausanne
Trubetzkoy, N.S. (1939) Grundzüge der Phonologie. TCLP VII

2. Origin of the phoneme concept
Ancient forerunners of modern descriptive linguistics (Paņini, Patañjali (India), the Greeks & „Anon“ (Iceland, 12th C.)) clearly recognised the systematic nature between distinctive sound properties and the identity of words in their languages.
de Saussure ( ) used ‚phonème‘, first as a term for speech sounds, later as a purely functional entity.
Baudouin de Courtenay ( ) and Kruszewski ( ) used the term phoneme for linguistic units underlying sound alternations between related forms.
Without using the term phoneme, many 19th century phoneticians focussed on sound differences with a distinctive function in their language descriptions.
Since we are discussing „phoneme“ as a a technical term in linguistics, this recognition of early scholarly insights into the nature and function of sound contrasts merely highlights the fundamental (God‘s truth?) nature of the sound-meaning relationship. All theoretical variants of the phoneme concept are based on this axiom of sound-based communication.
But there are fascinating parallels between the ancient works and some structuralist concepts of the 20th century: Patañjali (150 BC) defined the ‚sphoţa‘, which is the invariable entity of a structure. Its realisation is the ‚dhvani‘. This dichotomy shows clear parallels to Saussure‘s distinction ‚langue‘ – ‚parole‘. The smallest entity in language is the ‚varna sphota‘
The Greeks also had an ultimate sound unit in speech – the ‚stoicheion‘ (literally: „element“) – from which other, larger units are formed. (Fischer-Jörgensen, p. 4-5)
Saussure‘s definition of the ‚le signe‘ as the basic language unit comprising – indivisibly – the content or meaning (‚signifié‘) and the expression or phonetic form (‚signifiant‘) is arguably the first formulation of the old semiotic signans-signatum relation as a linguistic sound-meaning relationship.
Regarding the phoneme, he states: „Ce qui les caractérise, ce n‘est pas, comme on pourrait le croire, leur qualité propre et positive, mais simplement qu‘ils ne se confondent pas entre eux. Les phonèmes sont avant tout des entités oppositives, relatives et négatives“ (de Saussure, p. 164).
I.e., it is not their phonic substance that characterise them positively, but the properties that distinguished them among one another.

3. Baudouin de Courtenay and Kruszewski, (who worked together at the University of Kazán from 1878 onwards) foreshadow one of the important principles of Generative Phonology in that they link the sound structure (which they call „phonetics“) directly to the morphological structure. Their „phoneme“ might be more justly called a „morphophoneme“ although they also used the term both for purely phonetically determined alternations (e.g., bid – bids (/z/) vs. bit – bits (/s/)) and for morphologically determined alternations (e.g. wife (/f/) – wives (/v/))
In 1895, Baudouin de Courtenay published his best known work: „Versuch einer Theorie phonetischer Alternationen“ where the original morphophono-logical definition of phoneme was abandoned, and the definition for which he is better known was formulated. It was now defined as the psychological equivalent of a speech sound: „Eine einheitliche der phonetischen Welt angehörende Vorstellung, welche mittels psychischer Verschmelzung der durch die [unterschiedliche] Aussprache eines und desselben Lautes erhaltenen Eindrücke in der Seele entsteht.“ (quoted from Fischer-Jörgensen, p. 9).
This psychological definition was well known to de Saussure (they met several times and wrote to each other), and although he never defined the phoneme as a psychological unit there are several statements in his Cours de Linguistique Générale to the effect that la langue has psychological reality (despite the contradiction to his definition of langue as „la partie sociale du langage, extérieure à l‘individu, qui à lui seul ne peut ni la créer ni la modifier; elle n‘existe qu‘en vertu d‘une sorte de contrat passé entre les membres d la communauté“). (quoted from Fischer-Jörgensen, p. 13)
As early as 1877 (Sweet), „Classical“ phoneticians (and the dialectologist Winteler 1876) made a very clear distinction between sounds that change the meaning of words and those that are „not used for linguistic purposes“ (Noreen, 1903) or are only „als Lautschattierungen zu betrachten“ (Forchhammer 1924). Passy (1888), Jespersen (1897) are also documented as having made the explicit distinction.
Fischer-Jörgensen stresses, however, that awareness of the fact is not the same as developing a new theory of language description. So, although later protests to Prague phonologists that the ideas being propounded were not new may be justified, the Prague phonologists presented them as part of an overall theoretical framework.

4. The phoneme develops
The Prague School (1926 ff.) was the first group to formulate an explicit phonological theory (in The Hague 1928)
Sprachgebilde/Sprechakt reflected the strong influence of de Saussure.
Likewise the principle of phonological opposition („a difference of sound in a given language that may serve to distinguish intellectual meaning“).
A phonological unit manifests an opposition, and the phoneme is the minimal phonological unit.
Since the phoneme consists of only the phonologically relevant properties, a (realised) speech sound cannot be a phoneme.
Roman Jakobson presented the theses in 1928 (signed by Jakobson, Karcevskij and Trubetzkoy) for the aims of a phonological theory:
To set up phonological systems. 2) To account for significant differences.
3) To find correlations (e.g. p/b, t/d, k/g.
To formulate general laws concerning the structure of phonological systems.
To account for historical change in terms of a teleological development of the system.
A standard phonological terminology was presented at an international conference in Prague in 1930 and affiliated circles formed throughout Europe.
The Saussurean dichotomy was taken further: the Sprachgebilde/Sprechakt distinction also covered form/substance and separated phonology from phonetics.
(Jakobson objected to the exclusion of phonetics from the realm of Sprachgebilde, arguing that phonemes have to be realised in speech).
Martinet (and later Jakobson) adopted the definition of phonology as „functional phonetics“.
The Saussurean idea is also taken over, that meaning is differentiated by one sound NOT being another sound (a system of oppositions rather than a system of sounds with certain properties).
The phoneme defnition developed from „a phonological unit … that cannot be decomposed into smaller units“ to „ … that cannot be decomposed into smaller successive units“ (because Vachek objected that bad vs. pad is an opposition of voicing vs lack of voicing, which is a smaller „unit“).
In addition, it was defined as „die Gesamtheit der phonologisch relevanten Eigenschaften eines Lautgebildes“ (Trubetzkoy p. 35). This is the basis of Jakobson‘s definition as „a bundle of distinctive features“.

5. Types of opposition
Originally (1929) only correlative, e.g. p/b; t/d or i/i: o/o: (i.e., presence vs. absence). All others are disjunctive.
1936/1939 opposition classification was elaborated to cover:
Their relation to the overall system
- bilateral or multilateral
- isolated or proportional
The relation between the members of the opposition
- privative, gradual or equipollent
Their distinctive validity
- constant or suspendable
With the concept of „opposition“ so central to the phonological theory, it is not surprising that the Prague phonologists tried to systematise it.
The two correlative examples show the presence and absence of voicing in one case and the presence and absence of vowel length in the other.
1a. In relation to the overall system, bilateral oppositions are equivalent to the correlations in the original classification. Multilateral oppositions, logically, have more than two members; e.g. p/t/k.
Discussion point: Is Korean ph/p/b a multilateral opposition?
1b. An isolated opposition has no other cases with the same feature opposition; e.g. r/l in many languages. If an opposition is not privative it is proportional; e.g. p/t/k , b/d/g and m/n/ (they have the same multilateral place opposition)
2. Regarding to the relation between the members of an opposition, an opposition is privative if one member lacks a feature that the other has. The member with the feature is „naturally marked“ (merkmaltragend); the one lacking the feature is „naturally unmarked“ (merkmallos).
A gradual opposition is one in which the same property is present to a differing degree; e.g. i/e or u/o.
Equipollent oppositions have no degree or absence of a property in one or the other member; The two members are considered logically equal; e.g. p/t or f/x.
3. With regard to distinctive validity, a constant opposition has no position where the opposition does not apply; if it is suspendable in any position, the opposition is neutralised in that position.

6. Neutralisation
• Context-determined vs. structure-determined neutralisation:
Context: voiced-voiceless consonants preceding stops or fricatives in Russian.
Structure: voiced-voiceless in in syllable-final position in German.
Only minimal oppositions (1 feature) can be involved in neutralisation.
In neutralisation, only common features are relevant. The neutralised sound is the archiphoneme
Except when context-determined the form of the archiphoneme corresponds to the unmarked member of the opposition
When different forms of the neutralised opposition are found in different positions, the position where the greater number of phonemes are distinguished has the unmarked member.
Neutralisation is particularly important to the Prague School (and to the Danish school of Glossematics – Hjelmslev), a consequence of the central concept of „opposition“.
Discussion point:
1. What other cases of structural neutralisation can you identify a) in German, b) in any other language you know?
2. Does equal voicing in consonant clusters containing a fricative and a stop in English constitute neutralisation of voicing? If so, what sort of neutralisation is it? (examples: lvd - lAft, reIvd - rAft, leIzd - lAst, pUzd - pUst, pts - pdz, kAts - kAdz, lks - lgz, lUps - lUbz
Trubetzkoy went further in restricting the type of opposition that is open to neutralisation. Not only should the opposition be minimal, it must also be bilateral.
1. Can an opposition be minimal and NOT bilateral?
Can you find any reasons to disagree with Trubetzkoy that only bilateral oppositions can be involved in neutralisation?
3 The two conditions given for determining the unmarked member are problematical. Discuss in relation to /s/ and /z/ in German.

7. American Descriptive Linguistics
Theoretical developments in USA were less coordinated (less centralised) than in Europe .
Several different standpoints were represented by different linguists or groups: Sapir; Pike & Nida.
„Descriptive“ linguistics strove for clearly defined methods. No unobservable facts could be considered..
Procedures needed to be so explicit that they were completely replicable.
Typical reply to a (palpably true) statement: „I don‘t care if it is true. How do you justify having found it?“ (Anderson p. 184)
- Bloomfield is generally considered the father of American Structuralism and was recognised as the main originator of the ideas by his successors (after 1933 (Language) he did not play the central role that might have been expected).
- „Bloomfieldians“ like Bernard Bloch, Charles Hockett, George Trager, Henry Lee Smith, Archibald Hill, Martin Joos, Rulon Wells all subscribed to the Bloomfield position but were in many ways more extreme than Bloomfield had been: a) in their exclusion of mentalism and any recourse to meaning, b) in the separation of linguistic levels, c) in the advocation of unprejudiced description and inductive procedures, i.e., the avoidance of any a priori assumptions.
- This contrasts strongly with the Prague assumption of (universal) distinctive features that underlie the phonological sound oppositions for all the languages of the world. Deductive procedures were applied = seeking evidence to support the assumptions.
In the USA, scholars like Sapir were more „mentalistic“, though they accepted the methodological rigour of the „Bloomfieldians“. Kenneth Pike and Eugene Nida were often healthily critical of their exaggerated methodological approach, though they to were positively inclined in principle.
The „Bloomfieldians“ were, on the contrary, often very negative and derisive towards those who did not follow them completely.
- It should be stressed that Bloomfield and his (sometimes more extreme) followers were very much in line with the current trends of philosophy of science, and its reflex in psychology and theories of learning: positivist, mechanist, operationalist, behaviorist.

8. Bloomfield‘s Phoneme
„The smallest units which make a difference in meaning“ „A minimum unit of distinctive sound feature“ (p. 77). I.e. an externally defined, non-mentalistic unit. Phonology is „the study of significant speech sounds“ (p. 78)
He identifies „primary“ (segmental sounds) and „secondary“ (stress and tone) phonemes according to their function in language (primary: syllable forming; secondary: structuring larger units).
Phonemes are defined by their participation in structural sets.
(syllabic, open-syllable, closed syllable, non-syllabic, initial, medial, final, initial cluster, final cluster, etc.)
Contrary to the focus of the „school“ for which he is considered responsible, Bloomfield is very vague about the procedures for identifying phonemes: „…trust our everyday knowledge to tell us whether speech-forms are ‚the same‘ or ‚different‘“(Language, p. 77); „a little practice will enable the observer to recognize a phoneme even when it appears in different parts of words ..“ (p.79)
He also reveals thoughts which are reminiscent of Prague: He distinguishes distinctive and non-distinctive features and states, „these distinctive features occur in lumps and bundles, each of which we call a phoneme“ , and later: „The phonemes of a language are not sounds, but merely features of sounds which the speakers have been trained to produce and recognize in the current of actual speech sound.“ (p. 77).
But he is interested in the phoneme, not its component properties. The part they play in language should guide their classification – therefore the division into primary and secondary. Primary: vowels (always syllabic); consonants (may be syllabic). Intersecting structural sets are defined according to distributional regularities (initially, medial, final in words; phonotactic regularities). Any one phoneme is a member of particular structural sets, (mostly) different from any other phoneme.
Discussion point: a) What vocalic and consonantal structural sets can you define for German (or your L1)?
38 sets were defined for English: E.g., final consonants; pre-final consonants in consonant clusters; second pre-final consonants; post-final consonants. b) How would you classify the consonants in Engl. „texts“; „fifths“

9. Underlying Forms
Bloomfield recognised the need for underlying forms to simplify the description of morphophonemic alternations.
Only later (1939) did he call for a separate discipline called morphophonemics whose basic units were morphophonemes.
He chose the forms and used ordered rules to achieve the simplest possible description.
He even set up „artificial“ underlying forms to achieve a simpler description.
Post-Bloomfieldians were strictly insistent on the separation of levels (morphophonemics from phonology) and did not accept ordered rules.
It was not until 1939 that Bloomfield strictly separated morphophonemics from phonology. In Language he deals with them as part of morphology, and morphemes – the minimal linguistic unit – consist of phonemes. - It is interesting to note that in 1933 he already had underlying forms and ordered rules for deriving the surface forms – features which are usually attributed to developments linked with generative phonology. E.g., /Iz/ = underlying form for Engl. plural. /I/ is deleted unless the preceding sound is a sibilant. /z/ becomes /s/ if the preceding sound is voiceless. He regarded the ordered rules as a fiction – resulting from the descriptive method – and saw the simplification of the description as the goal influencing the choices. - In his discussion of alternants he allows zero (e.g. sheep (sing.) – sheep (plur.) and substitution alternants (goose – geese). He avoids any discussion of the logical problems arising from the view that morphemes consist of phonemes: a) with two different morphemic structures having the same phonemic structure. and b) two different (but co-extensive) phoneme strings sharing the same morpheme. The post-Bloomfieldians didn‘t accept such looseness (see post-Bloomfieldian developments below)
Another aspect of Bloomfield‘s description of sound variants, which attracted a lot of discussion (not to say criticism – particularly by Trager) of the phoneme concept, was his treatment of reduced vowels (schwa) when unstressed. He saw them as variants of the corresponding full vowels. (/kntrkt/ – /kntrkt/)But the impossibility of knowing from the surface form which of the full vowels the schwa belonged to (unless there were morphological alternations) made it the object of criticism.

10. Post-Bloomfieldian Phonemes 2
Bernard Bloch & George Trager saw the phoneme as a class of sounds (physical definition, cf. Bloomfield) .
„A phoneme is a class of phonetically similar sounds, contrasting and mutually exclusive with all similar classes in the language.“
Zellig Harris, on the hand, saw the phoneme as a „purely logical symbol“ (cf. Twaddell half a generation earlier).
Part of the problem underlying these fundamental disagree-ments is the amount of variation to be catered for by the description (idiolect, dialect, pan-dialectal language).
Non-uniqueness of the phonetic-phonemic relationship; the non-determinability of the phoneme from the phonetic properties and the non-prediction of the phonetic properties from the phoneme (lack of bi-uniqueness) was a problem.
The „class of sounds“ having the same feature in common is what Bloch considers important (so the view is similar to Daniel Jones‘ „family of sounds“). Physical variants with shared properties.

11. Morphemes and Phonemes
Hockett addressed the unclear relationship between morphemes and phonemes. It is clearly illogical to say:
On the one hand, Morphemes consist of phonemes
On the other hand, Morphemes have alternants (morphs) … … and morphs have differing phonemic structure!
Following Hjelmslev, Hockett distinguishes content units (morphemes) and expression units (phonemes). He also makes a distinction between representation and composition.
Morphemes are represented by morphs.
Morphs are composed of phonemes.
The indirect relation between morphemes and phonemes is one of „programming“ (i.e. encoding).
Hockett brought a logical structure into the relationships by defining different categories of relationship between the same and different levels of description (Manual of Phonology 1955, p. 15, quoted in Fischer Jörgensen, p. 111)
He took terms from Hjelmsev as a point of departure. Hjelmslev was, of course, expressing the formal analytic separation of the two indivisible parts of le signe linguistique (signifié and signifiant) identified by de Saussure.
Though this redefinition was perfectly logical, Hockett himself rejected it as a local rather than an overall solution.
He set up two strata, phonology and grammar. The phonological stratum has a hierarchy of units: distinctive feature, phoneme, syllable, phonological word (he calls them the microsegment), macrosegment (approximately a intonation unit) and an utterance. The hierarchy in the grammatical stratum comprises the morpheme, word, phrase, clause, sentence.
He then applies the above principle of „programming“: The grammatical stratum is „programmed into“ or „mapped into“ the phonological stratum. The morphs and morphophonemes are then said to be artefacts, units of analytic convenience, not elements of the language.
Morphs are sliced out of the phonological stratum as matches for units in grammar; morphophonemes are identified in the grammar stratum as matches for phonemes in the phonology.
Hockett no longer accepts morphophonemics as a valid descriptive level.

12. US-Structuralism vs. Prague Phonology
Prague dichotomy (Phonology vs. Phonetics) vs. US hierarchy (von Phonetics to Phonology.
Prague allowed meaning to be considered, US (theoretically) excluded meaning from consideration (though not Bloomfield himself, and the others not in practice!)
Prague focussed on paradigmatic oppositions (and employed commutation tests), US focussed on syntagmatic structures (combinatory possibilities).
Prague considered the phoneme to be analysable as a bundle of distinctive features, US regarded the phoneme as the smallest unit of analysis and refrained from decomposition (except Hockett & Harris).
Prague does not „phonemicize“ prosodic phenomena, US has a system of stress, intonational and junctural phonemes.

13. Status of the Distinctive Feature
Distinctive property of a phoneme or distinctively used dimension?
Distinctive feature as the defining property of a natural class of sounds?
Are distinctive features permanent or variable properties of a sound(class) depending on the opposition?
Are feature oppositions always binary or can they be unary or multilateral?
How many different distinctive features are there?
How should the distinctive features be defined?
We have seen that the idea of a phoneme as comprising a number of defining properties, i.e. the decomposability of the phoneme had been present right from the start of modern phonology (see previous sessions). But the theoretical status of the feature as a) defining the sound element or b) defining the opposition has always been open to discussion. There are, of course, clear echoes of the phoneme as a physically defined unit vs. an abstract, negatively defined member of an oppositional relationship (see previous slides).
Both as sound-defining and as sound-distinguishing properties distinctive features define groups or classes of sounds that share the same properties (cf. Gussenhove p.64-5).
As sound-defining properties, feature values can be linked permanently to particular sound (classes); e.g. [voiceless] and [plosive] identify the /p t k/ class in German. But as a sound-distinguishing property, [voiceless] only keeps /p t k/ apart from /b d g/ in the syllable onset, not in the coda. I.e. coda stops have less distinctive features.(cf. Fischer-Jörgensen 147-8, cp.Heike 1961 for German)
Phonetic description led the Prague group to bilateral (e.g. voicing) and multilateral (place of articulation) oppositions. Jakobson later became the champion of binarity, but the idea has always been subject to discussion.
In contrast to American Structuralism, Prague Phonology postulated universal principles behind the sound structuring of language. The distinctive features should reflect these universal principles not differences in phonetic detail – therefore the overall aim is for a minimal set. But here, as so often, the degree to which the features are to be defined by phonetic substance has an influence….… should they be defined articulatorily, acoustically or auditorily (does it matter?)

14. Discussion point
What is your standpoint regarding the restriction to binary feature oppositions? Are there advantages in strictly binary features ...
a) ... as a formal framework for classifying the sound inventory of a language?
or is there any validity in the assumption of binary features ...
b) ... as an explanatory framework of the way the human speech-perception and/or production mechanism works?

15. Feature Systems 1
The formal development of distinctive feature theory is due primarily to Roman Jakobson.
a) DFs are the minimal linguistic units (not just classificatory dimensions).
b) Only binary oppositions are accepted.
c) Descriptions should be based on a minimum number of DFs.
d) These are selected from a limited set of universal DFs.
e) The phonetic description of the DFs is important.
f) The DF values for the sounds of a language are arranged as a matrix with +, – and 0 (not relevant) values.
The roots of Distinctive Feature (DF) theory lie clearly in Prague Phonology (where Jakobson contributed from the start), but Jakobson developed the theory in a way that clearly went further than classical Prague theory. These points of development (and divergence) are summarized in the 6 points (cf. Fischer-Jörgensen, p ).
Upgrading the DFs to a linguistic unit went beyond Trubetzkoy‘s position, who retained the phoneme as the smallest linguistic units. The principle of sequential organisation of units is also abandoned with this (cf. de Saussure; compare also morphological analysis, which does not require sequentiality).
Binarity, a minimal number of DFs (cp. economy, parsimony, simplicity) and universality are much discussed general principles of phonological (and general linguistic) theory.
Jakobson‘s belief in the importance of the phonetic basis of phonology (in terms of its role in language acquisition and language change) is also behind his use of the term „functional phonetics“ for phonology (cp. also Martinet). Most important for him – as the closest point to communication in the sense of meaning transfer from speaker to hearer – was auditory description.
The modern matrix representation of sound inventories goes back to Jakobson.

16. Inherent Features Sonority:
vocalic/non-vocalic: glottal source; free vocal tract; formants;
conson/non-cons: low F1, low intensity; obstruction in v. tract.
nasal/oral: nasal formant, low intensity; oral + nasal resonator
compact/diffuse: narrow, central frequency energy; horn-shape resonator
abrupt/contin: no energy above voice-bar; burst or fast transition
strident/mellow: high intensity in high frequency, supplementary obstruction.
checked/unchecked: higher energy discharge in shorter time; stoppage of pulmonic participation
voiced/voiceless: periodic low-frequency excitation
Originally, vocalic vs. consonantal, then split into two “±” features to cater for the four major phoneme groups: vowels, consonants, liquids & “glides”:
vowels: +voc, -cons; consonants: -voc, +cons; liquids: +voc, +cons; glides: -voc, -cons
Liquids (r and l) are acceptable as +voc, +cons, but grouping [w j h ?] as “glides” with –voc and -cons is often criticized as arbitrary (cf. Fischer-Jörgensen, p.156-6)
nasal-oral is an unproblematical feature opposition.
The acoustic definition of compact-diffuse is problematical because of the application to vowels and consonants. The articulatory description applies more satisfactorily to both categories (open vowels and velar consonants). Velar consonants have a narrow band of energy, but not always central; [a A] have central bands of energy but not narrow.
The abrupt-continuous opposition is satisfactory in its separation of stops and fricatives, but it is also applied to [r] vs [l], which is more problematical. Nasals have some (though rather damped) resonances above the voice-bar, but they have rapid transitions due to the closure-opening movements.
Some applications of strident-mellow are problematical. Bilabial vs. labiodental, sibilants vs. non-sibilants; affricates vs. stops appear acceptable, but velar vs. uvular has no real acoustic justification and [v]-[] is not articulatorily justified.
Checked-unchecked as a definition of ejectives, implosives and clicks (vs. pulmonic consonants) is difficult to characterise in common acoustic terms.
Voiced-unvoiced is both acoustically and productionwise unproblematical.

17. Inherent Features Protensity: Tonality:
tense/lax: longer duration of steady state; greater deviation of vocal tract from neutral configuration.
Tonality:
grave/acute: predominance of energy in lower part of spectrum; peripheral artic. /less compartmentalized oral resonator.
flat/non-flat: lowering (and weakening) of higher frequency energy; narrowing at front or back of resonator
sharp/non-sharp: raising and strengthening of higher frequency energy; dilation of back resonator with palatal stricture.
Protensity: Like some other feature oppositions (e.g. compact-diffuse) the tense-lax feature has difficulties in the phonetic definition because it is applied both to vowels and consonants, i.e. to tense vs. lax vowels (e.g. in German) and to fortis-lenis consonants and aspirated-unaspirated plosives.
Tonality: grave-acute has less problems than compact-diffuse through being applied to both vowels and consonants.The peripheral places of articulation (lips, velum) give bigger (less-divided) oral resonators than dental and palatal.
+flat is meant to capture the effects of lip-rounding, pharyngealisation or retroflexion. The term itself is auditorily motivated (from music); the acoustic description is satis-factory, but the articulatory description does not cover retroflexion.
+sharp is intended to capture palatalisation (a raising of the front of the tongue). The acoustic characterisation covers the phenomenon well, but the articulatory description dwells on the part played by the pharynx, whereas it is the front articulation that contributes most to the raised higher frequency energy.

18. Problems with (Jakobson’s) features
• The use of [+flat] to cover 3 different articulatory modifications presupposes that they don‘t co-occur in any one language.
• Applying the same features to vowels and consonants stretches the plausibility of the phonetic basis.
• The same feature can be manifested very differently in different positions.
• Allophonic variants may have opposing feature specifications.
Economy in the descriptive frame was an important criterion for Jakobson. In principle, his observation that rounded, pharyngealized and retroflex consonants do not occur together in a language is (almost) correct. However, there is evidence that some less well known (e.g. Caucasian) languages do have two modifications. Anderson (p. 124 ff.) recognises that – in absolute terms – the assumption is incorrect (and [± flat] would need to be supplemented. But he also argues for the importance of Jakobson‘s economy principle, and for the auditory basis on which Jakobson grounds his collapse of three different articulatory patterns. Clearly, the auditory (see acoustic) effects are similar, so the distinctiveness of the articulations is not sufficient to make them good feature differences. The fact that only an occasional language does use them in opposition to one another is a sign of this. He assumes that accompanying realsiational differences probably strengthen the auditory differences between them
A similar criticism is possible of the strict economy approach – using features for both vowels and consonants.(e.g. compact-diffuse, tense-lax).
The criticism that different manifestations occur in different contexts was countered with the reminder that oppositions are relational statements. As long as the differenc is valid under the same contextual conditions, the criticism is invalid (this is easily demonstrated in the case of vowel length, for pitch contrasts, even for vowel differences). Fischer-Jörgensen (p ) presents Danish examples of syllable-initial and -final allophones of /t/ ([t d]) and/d/ ([d D])which would be considered different phonemes if the two contexts were analysed separately.

19. Acoustic properties: Flat (retroflex) / Plain

20. Acoustic properties: Flat (pharyngealized) / Plain

21. Acoustic properties: Checked / Plain

22. flat
plain
grave  acute
grave  acute

23. Acoustic properties: strident / mellow

24. Feature Matrix for English (Jakobson Fant & Halle p. 43)
Only 9 of the 12 features are needed. No [sharp], [±checked], [±voiced]

25. Feature Matrix for German (Halle 1954, nach Fischer-Jörgensen, 168)
The same 9 of the 12 features are needed as for English, but …..

26. Feature tree for Swedish consonants (Fant 1961, nach F-J, 172)
8 features; no [±strident] or [±tense] but [±voiced] (but …..)

27. Exercise (written)
Prepare notes on the „Discussion Point“ (slide 14) in preparation for discussion in Übung (hand in notes
with other answers)
Compare the distinctive-feature matrices for English and
German (slides 24 & 25). Do the features cover all the sound distinctions in each language? What differences are there in in the status and treatment of features in the two tables?
Try to construct trees for English and German that compare
with the tree presented for Swedish (slide 26).
Please hand in sheets by Friday or answers to by Sunday (cc. wbarry)