English Phonology: Study Guide - 6 2.4 Phonemes as Categories [eɪ] vs. [e ɪ] as the same vowel (i.e., nondistinctive, noncontrastive) A group of physically different but mentally the same sounds forms a phoneme, an abstract phonological category. The phonetic ( physical) variants are called its allophones. The variation depends on the phonological and phonetic context/environment the segment appears. (Diagram on p. 23) 2.5 More Instances of Allophonic Variation /n/ dentalization in English: [n] vs. [n ] (data and distribution, p. 24) [n ] before [θ] [n] elsewhere /l/ velarization, (partial) devoicing, etc. in English: [ɫ] vs. [l l] vs. [ɫ ] vs. [l] (p. 25) [ɫ] word-finally [l l] after a voiceless consonant [ɫ ] before [θ] [l] elsewhere In full phonetic detail, thousands of sounds Phonologically, only a small subset of such differences is used to distinguish words. The noncontrastive differences are allophonic, and determined by rule. 2.6 Phonemic Transcription Phonemic transcription: Representation of words as sequences of phonemes Phonetic transcription: includes allophonic detail Orthographic transcription (ordinary spelling) Phonemic transcription, e.g., as given in dictionaries, needs phonological rules to get detailed allophonic information. Phonemic transcription is simpler and carries the same information, because allophonic details are predictable in a given language. 2.7. Phonological Rules A sample phonological rule (in prose): /eɪ/ Shortening (p. 26) Characteristics Language-specific; not universal; must be learned by language users (native or nonnative) Productive: readily applied to new words Well-formedness intuitions: rule violations such as [se ɪv] and [seɪf] (by phoneticians and speech synthesis) are sensed intuitively. Untaught; unconscious knowledge (not a normative/prescriptive rule) Must be studied through data gathering, experiment, and construction of theories.
English Phonology: Study Guide - 7 2.8 Formalizing Phonological Rules Need to be expressed precisely; formal notation is used. 2.8.1 Expressing environments A B / C D A becomes B in the environment of..., in this case, between C and D. C and D can be null: if C is null (if there is no C, D is read before D). /n/ [n ] / [θ]: An alveolar stop becomes dental before a voiceless interdental fricative. A, B, C and D can be a group of segments. Then list relevant properties to indentify the group. Partial devoicing of /l/ in English [l l]: after a voiceless consonant +consonantal voiced C and D can be a word boundary: Word-initially (OR at the beginning of a word): / [ wd (OR / # ) /l/ [ɫ] / ] wd (OR / #) A lateral approximant becomes velarized word-finally (OR at the end of a word). 2.8.2 Underlying representations and derivations Underlying OR base representation/form undergoes a series of phonological rules. Phonological rules to apply to /l/ in English (p. 29) /l/ Devoicing /l/ Dentalization /l/ Velarization The result: surface OR phonetic representation/form Derivation: the whole process of turning an underlying form to its surface form Sample derivation of /l/ to [ɫ], [l l] and [ɫ ] (p. 30) Typically, the elsewhere form is (identical to) the underlying form (but see fn 5). 2.9 Phonemes in Other Languages Different languages have different phonemes Different languages have different ways of organizing allophones into phonemes. English [ɾ] and [t] are not distinctive; [ɾ] is an allophone of /t/ (pp. 31-32). Derived by the Tapping rule. /t/, /d/ and /ð/ are separate phonemes. Spanish /ɾ/ and /t/ are contrastive and hence separate phonemes (middle of p. 32). [ð] is an allophone of /d/ (derive by spirantization, p. 33).
English Phonology: Study Guide - 8 Two languages may have exactly the same (phonetic) sounds; The phonological organizations can be radically different, if they select different phonetic distinctions. 2.10 Phonemicization (a.k.a. discovery procedure ) 2.10.1 Minimal pairs 2.10.2 Near-minimal pairs Examples for [ð] vs. [ʒ] on p. 36 2.10.3 Using local environments to establish complementary distribution Method of compiling local environments E.g., Maasai obstruents: [p, t, k, b, d, ɡ, β, ð, ɣ] Group the individual environments for each segment: p. 39 Generalize the distribution (p. 39 bottom) The elsewhere form has the widest and heterogeneous distribution (here /k/). Phonological rules (for /k/): /k/ Spirantization; Postnasal Voicing (/k/ voicing after /ŋ/) Sample derivations (p. 40 middle) Big question: Why do the same rules recur in many different languages? 2.10.4 Natural classes A complete set of sounds in a language that shares the same value for a feature or set of features 1) undergoes the same process... (A in the rule schema A B / C D) 2) results from the process... (B) 3) behaves as (part of) the environment... (C, D) In Maasai: [k, ɡ, ɣ] [p, b, β] [t, d, ð]; [p, t, k] [b, d, ɡ] [β, ð, ɣ]; [m, n, ŋ]; etc. bilabial dental velar vl stops p t k... (A) plain vd stops b d ɡ... (B) vd fricatives β ð ɣ... (B) vd nasals m n ŋ... (C) vowels... (D) Sample rules with natural classes: Spirantization in Maasai Postnasal Voicing in Maasai Vowel Shortening in English: [+vowel] [ long] / [+stop, voice] Spirantization in Spanish: [+stop, +voice] [ stop] / [+vowel] [+vowel] Some natural classes recur in different phonological processes in different languages Natural classes share one or a few phonetic features Assignment 3: Exercises 1 and 2 (pp. 43-45)
English Phonology: Study Guide - 9 Chapter 3. More on Phonemes 3.1 Phonemic Analysis and Writing In phonology and linguistics, spoken language is primary; written language is derivative. A practical application of phonology (esp., phonemic analysis) is the design of alphabets. Each letter is for a phoneme, not a phonetic/allophonic sound. 3.2 The Psychological Reality of the Phoneme Do speakers actually produce and perceive language in terms of phonemes? 3.2.1 Audibility of fine distinctions Bengali speakers: /t/ vs. /t / easily detected in (near) minimal pairs English speakers: not able to hear the difference (without training) Speakers of a certain dialect of English hear the difference between the vowels in caught and cot, while speakers of other dialects do not. Acquired quickly: Ten-month-olds hear only those phonetic differences present in their language. Humans change from unbiased perceivers to phonological perceivers. 3.2.2 The notion of same sound Allophones of the same phoneme considered/heard/felt as the same sound [ε] vs. [ε ] in English and in French [t] vs. [t ] in English and Bengali 3.2.3 Foreign accents and transfer Transfer: mispronunciations in a second language due to (the phonology) of the first, i.e., native language Illegal phoneme: [θ] rendered as [s ] or [t ] in Korean Illegal sequence of phonemes: strike [straɪk] as [si-t i-raik i-] in Korean Illegal distribution of allophones: silicon [sɪlɪkən] as [ɕillik h on] (s ɕ / i) L1 phonological rules applied in L2 German name Gödel [ˈɡødəl] as [ˈɡɚdɫ ~ ˈɡɚdəl] in English French thé tea [t e] as [t h eɪ] in English Learners of foreign languages are prisoners of their own phonologies (with the exception of about.2 % of the entire population). 3.3 The Criterion of Phonetic Similarity [h] and [ŋ] in complementary distribution in English But considered separate phonemes; phonetically too dissimilar But how about [ɾ] and [ʔ] for /t/ in English? And how about [ɾ] and [l] for ㄹ in Korean?
English Phonology: Study Guide - 10 3.4 Other Problems in Phonemicization 3.4.1 Contour segments and the segment/sequence problem Contour segments: change of a phonetic quality within a single complex segment Affricates, diphthongs, and prenasalized stops Clear cases Polish: [tʃɨ] trzy three vs. [t ʃɨ] czy whether English: [bɔ ɪŋ] boing vs. [sɔɪŋ] awing Many cases are unclear. 3.4.2 Unclear status of borrowed sounds [v] centuries ago in English: a phoneme or an allophone of /f/? [ɸ] in Japanese (p. 58) [x] and [t s] in English 3.5 Free Variation Saying the same thing, i.e., meaning, in different ways, i.e., different forms Phonological doublet: [ˈεnvəˌloʊp] or [ˈɑnvəˌloʊp] envelope Two forms listed in the lexicon for one meaning Free variation: a single phoneme has two or more phonetic forms. /æ/: [æ ] ~ [ε ə ] before /n/ in English Optional application of the rule /æ/ Diphthongization (p. 59) Branching derivations (p. 60) Tapping in English /t/ [ɾ] is optional, showing /t/ ~ [ɾ] variation in data, outer, water etc. Not all rules are optional. The same rule can be optional in one dialect and obligatory, i.e., not optional, in another 3.5.1 Optional rules and speech style Application of an optional rule is generally determined by the speech style. 3.5.2 Variation and elicitation 3.6 Contextually Limited Contrasts and Phonotactics E.g., only voiceless stops are possible word-finally in Toba Batak: *[sukkub] 3.6.1 Analyzing phonotactics and contextually limited contrast Two ways/approaches of characterizing contextually limited contrasts Rule: Final Devoicing (p. 63) Phonotactics: Constraint against Final Voiced Stops (p. 64) 3.6.2 Disallowing a segment in certain positions Toba Batak disallows more than one consonant word-finally Rule: Consonant Cluster Simplification (p. 64) Constraint: Ban on Final Clusters (p. 65) Assignment: No.2 (p. 66-67)