Unit 2 Speech Sounds PDF

Unit 2 Models of Speech Acquisition BBF (2021) Chapter 3 Theories must account for the following premises: “(a) the actual facts of children’s production & mismatches b/w a child’s output & the adult input forms (b) the generalities that span children’s sound systems, as well as a...

Unit 2 Models of Speech Acquisition BBF (2021) Chapter 3 Theories must account for the following premises: “(a) the actual facts of children’s production & mismatches b/w a child’s output & the adult input forms (b) the generalities that span children’s sound systems, as well as associated variability within and across developing systems (c) the changes that occur in children’s grammars over time An adequate model must be testable and falsifiable.” Barlow & Gierut (1999, p. 1482) Tradi8onal Model of Speech Acquisi8on: Behaviorist Models Behaviorist Model According to the model, language is a behavior that can be observed. B. F. Skinner was the most influential proponent of behaviorism Used principles of operant conditioning control behavior by changing the consequences following the behavior (Skinner, 1972; Thomas, 2000). In behaviorism, consequences can be positive or negative; be used as a form of reinforcement or punishment. Application to Typically Developing Children: Focus on observing environmental conditions (stimuli) that co-occur and predict overt verbal behaviors (responses) Behaviorists collected and documented normative behaviors of large groups of children (e.g., Age-of-acquisition data [Templin, 1957]) Behaviorist Model Criticism: Children develop speech and language quicker than can be explained by stimulus- reinforcement of each element alone Speech development too complex to be explained by reinforcement alone Application to SLP Practice Used extensively in application of children with speech sound disorders used in traditional motor-based articulation therapy (Van Riper & Irwin, 1958) within a stimulus-response paradigm E.g., child word on producing a sound or a word, and upon producing it correctly received praise, a sticker or a token (Winitz, 1969) Linguistic Models of Speech Acquisition 1) Generative Phonology 2) Natural Phonology 3) Nonlinear Phonology 4) Optimality Theory 5) Sonority Hypothesis Generative Phonology Proposed by Noam Chomsky to study the sound structure of languages Generative meant that speech sounds were produced by transforming the underlying representation (UR) into surface form using a language-specific rule i.e., theory applied to children’s speech acquisition with the intent of explaining the relationship between children’s productions and the adult forms (Grunwell, 1987) Premise: Child has perceptual mastery of adult phonemic system before he has productive control over it. Moved away from traditional phonemic analysis and introduced: 1) Phonological rules that code URs onto surface pronunciations 2) Phon.’l descriptions depend on info. from other linguistic levels (semantic & syntactic) Genera4ve Phonology Rules Example of Rule: “Vowels are realized as nasal in the context of … nasal consonants” (p.61) To the left of the arrow are segments that conform to the rule The arrow means “realized as” or “produced as” Diagonal line or slash means “in the context of” The right of arrow is the context(s) under which the rule for the segment works Generative Phonology Rules Phonemes are viewed in terms of their distinctive features (DFs). We use phonological rules to describe how phonemes combine & vary when used in speech. GP relates child production to adult production, Another example: /obstruent/ → [ø]/_____# Rule: An obstruent becomes deleted word finally Generative Phonology Criticism: It is questionable whether the child’s UR of sounds is same as the adults It is unclear whether we apply “rules” in our heads when we comprehend and produce speech Application to SLP Practice Not widely practiced This theory was helpful in the development of other theories (e.g., natural phonology) and was a stepping stone to phonologically based clinical analysis (i.e., phonological process analysis). Naturalness and Markedness in Phonology Naturalness - Some sounds, syllable shapes, phonological processes are more natural than others. The more natural elements occur more often than others; they are more widespread in language of the world; they are acquired earlier by children; they are less likely to be lost historically. Markedness – There are marked & unmarked options for all aspects of grammar. Distinctive features are used to determine markedness in phonology. Unmarked /t/ (- voice). More marked /g/ (+voice, +dorsal). Unmarked option is part of UG and does not have to be learned; Marked option is more unusual in a language & needs positive evidence from the input language. 11 Markedness and Naturalness Characteristics of Unmarked members (Lowe, 2002). requires less articulatory effort; is less acoustically complex; is less perceptually ambiguous; is acquired developmentally earlier; occurs more often in the languages of the world. Implicational Law In a language, the presence of a marked pair implies the presence of the unmarked pair. 12 Markeness & Naturalness continued Observations about Naturalness (Sloat, Taylor & Hoard, 1978). Voiced obstruents are more marked than voiceless; Obstruents are less marked than sonorants (i.e., liquids); Stops are more natural than fricatives; Fricatives are more natural than affricates; /n/ is the most natural nasal; /s/ and /h/ are the most natural fricatives; Low front vowels appear to the most natural vowels; CV syllable is the least marked followed by CVC. 13 Natural Phonology (Stampe, 1969, 1979; Ingram, 1979, 1989) Phonological development involves a set of innate processes (i.e., natural processes – e.g., Cluster ReducJon) that funcJon to simplify the adult target word. Children use these processes Jll they learn to suppress them. These processes are mo@vated by motor and mental constraints consistent with a young child’s psycho-physical limita@ons. A child begins to produce the appropriate sounds of his community through successive revisions of infan@le processes. Theory assumes the child’s phonological representa@ons are at least as deep as the adult. Surface structure differences are explained by the opera@on of natural processes. Natural Phonology Natural phonology rules are universal Speech acquisition is a progression from these innate speech patterns to the pronunciation systems of the target language. Critique: Assumption that phonological processes relate to mental operations that kids do in their heads. Best to be descriptive &use phonological patterns instead. Children manifest phonological patterns that are atypical/non-natural, not seen in typical development. Not all simplification patterns produced by children can be classified Negative progression, e.g., processes must be suppressed. Natural Phonology Application to SLP Practice Developmental data Assessment models developed Intervention procedures widely used COMMON PHONOLOGICAL PROCESSES/PATTERNS Found in English Speaking Children (Ingram, 1976; 1981) ØA phonological process/pattern is a systematic sound change that affects a class of sounds (e.g., fricatives or velars) or a sound sequence (e.g., /s/ + liquids; /f/ + liquids; /k/ + glides) (Edwards & Shriberg, 1983). ØPhonological processes/patterns provide a means for describing the error patterns manifested in young children's speech. ØProcesses/Patterns can be organized according to uSyllable Structure Processes, uSubstitution Processes and u Assimilation Processes SYLLABLE STRUCTURE PROCESSES: changes that simplify syllable structure of words. ØUnstressed syllable deletion (WSD): Common up until 3;6 for 3-syllable words & up to 4;6 for longer words. E.g., /ˈtɛləˌfon/ → [ˈtɛˌfon]. ØFinal Consonant deletion (FCD): resulting in an open syllable. Common up to 3;0 years. /kæt/ →[kæ]. ØInitial consonant deletion (ICD): (uncommon in English) /kæt/ → [æt]. ØIntervocalic singleton deletion (ISD): /bəˈlun/ → [bə.ˈun]. ØCluster reduction (CR). Tends to affect marked members of consonant sequences, (e.ɡ., /s/, liquids). Suppression ranges from 2;6 to 3;6 years. /stap/ → [tap] ØConsonant Sequence Reduction (CSR) /ˈbæθ.ˌtʌb/ → [ˈbæ.ˌtʌb] SYLLABLE STRUCTURE PROCESSES continued ØSyllable reduplication (R): Early process used by some children. Uncommon after 2;6 years. /ˈbaɾl̩/ → [baba]. ØEpenthesis (E) (segment insertion): Observed in children 3+ years. /blæk/ → [bәlæk]. ØMetathesis (M): segment reversal. basket → [bæksıt]. ØSyllable Coalescence (CO): production of a multisyllabic word with fewer syllables, with segments from both syllables being retained. /gəˈɹaʒ/ → [ˈgaʒ] ØDiphthongization: creation of a diphthong /wet/ → [weıt]; /jɔn/ [jaʊn] ØMonothongization: reduction of a diphthong to a monothong. /baɪk/ → [bɑk] SUBSTITUTION PROCESSES: changes that affect specific groups of single segments. vPlace Changes: chanɡes that affect place of articulation ØFronting: (affecting velars, palatals) § Velar Fronting (VF): Usually suppressed by 3;0 years. /kʊkɪ/→ [tʊtɪ]. § Palatal Fronting (Depalatalization) (PF): Observed in 3+ years. /ʃu/ → [su] ØBacking: (affecting labials, alveolar/dentals). § Backing to Velars (VB): (less common than Ft.) /tɔɪ/ → [kɔɪ]. § Backing to Palatals (palatalization): (less common than PF). /sup/ → [ʃup]. SUBSTITUTION PROCESSES: continued vManner Changes: ØStopping (ST): (affecting fricatives, affricates, sometimes liquids). Suppression ranges from 2;6 (e.g, /f/ to 4+ years /ð/). /soʊp/ → [toʊp]. ØAffrication (AF): creation of an affricate (typically affecting fricatives). /ʃu/ → [tʃu]. ØDeaffrication (DA): (affecting affricates, leaving fricative component). /tʃu/→[su]. ØGliding/liquid simpli=ication (GL): (affecting liquids in singletons & clusters in ONSET positions. ) Observed in children 3+ years). /tɹɪp/ → [twıp]; /laɪt/ → [jɑıt]. ØVowelization (V): (affecting affects liquids in syllable CODA position). Common in children 4+ years. /ˈteɪbl̩/ → [ˈteıboʊ]. ØGliding of Fricatives (GF): Uncommon. /ʃap/ → [jɑp]. ØLiquid replacement: replacement of one liquid for another /r/ → [l] (/ɹeɪs/ → [leɪs]; or /l/ → /r/ (/lʊk/ → [ɹʊk]). SUBSTITUTION PROCESSES: continued ØDenasalization: (affecting nasals) Uncommon. /nɛk/ → [dɛk]. ØVowel alternation (VA): replacement of one vowel for another vowel Ø Neutralization (N) replacement of a full vowel with /ә/ or /ʌ/. Uncommon. /boʊt/ → [bʌt]. ØDerhoticization: reduction of a rhotic /ə˞/to a schwa /ə/. /mʌðə˞/→ [mʌðə] ASSIMILATION PROCESSES: Phonetic context related. One sound is influenced by another and becomes similar to it. Suppression of consonant Assimilation ranges from 2;0 to 3;0 years depending upon process. ØContiguous: (adjacent) § Nasal (NA): /mæn/ → [mæ̃n]. § Labial (LA): /swɪm/ → [fwım] § Vowel Lengthening before voiced consonant: /bɛd/ → [bε:] § Voicing Assimilation: Final Devoicing (Fdv): affecting word-linal voiced obstruents. Common in children 4+ years. /bɛd/ → [bet]. Prevocalic Voicing (Pvv): affecting voiceless obstruents. Usually suppressed by 3;0 years). /tɑp/ → [dɑp] ØNon-contiguous: - non-adjacent. Can be: ØProgressive (forward) /dɔg/ → [dɔd] or ØRegressive (backward) dɔg → [gɔg]. § Velar Assimilation (VA): (affecting non-velar sounds) kæt → [kæk]; sneɪk → [ŋeɪk]. (can occur until 2;6). § Labial Assimilation (LA): /mit/ → [mip]; /læm/ → [bæm]. (present in early development). § Nasal Assimilation (NA): /mit/ → [min]; /læm/ → [næm]. Common Class Deficiencies (Hodson & Paden, 1991) Stridency Deletion (SD): affects /f, v, s, z, ʃ, ʒ, tʃ, dʒ/ Stridency feature is not maintained. Realized in many forms. E.g., FCD: /hɑʊs/ [hɑʊ] ST: /sʌn/ [tʌn] CR: /spun/ [pun] /flɪp/ [wɪp] Liquid Simplification (LS): affects /l, ɹ, ə˞, ɜ˞/. Liquid not maintained. Realized in many forms. E.g., FCD: /bɔl/ → [bɔ] Vowelization: /bɔl/ → [boʊ], /fɔə˞/ → [fɔoʊ], /bɜ˞d/ → [bɔɪ] GL: /ɹʌn/ → [wʌn] Derhoticization: /fɔə˞/ → [fɔə] Velar Deficiency (VD): affects /k, ɡ, ŋ/. Dorsal not maintained. Realized in many forms. E.g., FCD: /bʊk/ → [bʊ] CR: /ɡlʌv/ → [wʌv] Fronting: /skɪp/ → [stɪp] Model Replica Chart – Word Initial Position (onset) Model-Replica Chart – Word Medial Posi4on Model-Replica – Word-Final Position (coda) Nonlinear Phonology (Goldsmith, 1976; Lieberman, 1975) Derived from group of phonological theories: Autosegmental, Metrical and Prosodic Phonologies. Based on the notion that complex linguistic dimensions (e.g., stress, intonation, metrical and rhythmical linguistic factors) control segmental conditions, i.e., a hierarchy of factors affect segmental units. “Production of speech is more than just production of a sequence of phonemes” (BBF, p.63) There is a relationships among segments and other aspects of speech production (i.e., phrase, word, foot, syllables, segments, features) This relationship has a hierarchical organization These hierarchical levels of representation, (i.e., tiers), are described for both the prosodic and segmental units. Nonlinear Phonology (Goldsmith, 1976; Lieberman, 1975) The prosodic tier Focuses on words and the structure of words Includes the following levels/tiers: Word tier Foot tier (refers to a grouping of syllables comprised of a sequence of stressed [S] & [w] syllables) Syllable tier Onset-rime tier Skeletal tier (consists of consonant and vowel syllable shapes [i.e., CVs, CVCs] Segmental tier (phonemes) Prosodic tier: focuses on words and the structure of words BBF (2021) p. 63 Nonlinear Phonology (Goldsmith, 1976; Lieberman, 1975) The segmental tier Focuses on the production of segments (phonemes) and the features that make up those sounds BBF (2021) p. 64 NON-LINEAR (e.g., “monkey”) Word Prosodic Foot Aspects (s) s (w) s O R O R C V C C V Segmental Features Aspect m ʌ ŋ k i = = = = = +con +con +nasal +dorsal +labial 33 Terminology Tier. Separate organized levels of representaJon for prosodic and segmental units. Word. Simply denotes words Foot. Organizes syllables. A foot can contain ONLY ONE strong syllable, but can also contain other weak syllables) E.g., Banana – [wSw] = has one strong syllable à hence, 1 foot Demonstra@on – [sw.Sw] = contains two strong syllables à hence, 2 feet Helicopter – [Sw.sw] = contains two strong syllables à (i.e., 2 feet) Racoon – [s.S] = contains two strong syllables à [i.e., 2 feet] Phonology [wSww] = contains one strong syllable à (i.e., 1 foot) A foot can comprise of a strong-weak syllable (Sw) à leP prominent (i.e., trochaic) Or (wS) à right prominent (i.e., iambic) 34 Terminology con-nued: Syllable -er Syllable Tier. Represented by Onset and Rime. Rime is comprised of Nucleus and Coda. Sometimes syllable division becomes problematic in English E.g., “seven” --> would you syllabify the word as [ˈsɛv.ɪn] or [ˈsɛ.vɪn] Or “Benny” à [bɛ.ni] or [bɛn.i] Conflict: 1) Maximizing the Onset Principle States that across languages of the world (including English), onsets are preferred and codas are avoided 2) Phonotactic Constraint of English States that Stressed syllables containing a LAX vowel must contained a coda. AMBISYLLABIC CONSONANT is an intervocalic consonant that belongs to two syllables (i.e., consonant belonging to both the coda of the first syllable and the onset of the second.) A consonant is ambisyllabic when it is: (1) intervocalic AND (2) when the stress falls on the first syll. Ambisyllabicity helps resolve the conflict of the MOP and the phonotactic constraint on syllables in English Terminology continued: Syllable Tier. Ambisyllabic Consonants continued An ambisyllabic intervocalic consonant belongs to two syllables. It plays two roles: as onset and as coda. s s b ε n i 36 Terminology continued Skeletal Tier. Associates prosodic and segmental representaYons. Consists of CVs and CVCs Segmental Tier. Consonants and vowels in a parYcular syllable posiYon comprised of features. Feature Geometry. Segments are comprised of hierarchically organized features. Features can operate independently within the constraints of the hierarchy, i.e., autosegmental.  38 Manner supraglottal Voice Place 39 Nonlinear Phonology continued ApplicaJon to Typically Developing Children InteracJon seen between segmental and prosodic Jers (i.e., between segments and other speech domains) Development is progressive – maturaJonal factors taken into account ApplicaJon to SLP pracJce Nonlinear phonology principles applied to goal seTng and intervenJon for children with SSD (Bernhardt, 1992b, Bernhardt et al., 2010) E.g., using established sounds in new syllable shapes and/or developing new sounds in established syll. Shapes. Optimality Theory (Barlow & Gierut, 1999; Prince & Smolensky, 1993) Theory was originally developed to describe adult languages. Basic unit of representation are “CONSTRAINTS” (not rules) Constraints are universal and part of UG (universal to all languages); Variation across languages and in development is accounted for in terms of the relative ranking of constraints. Two types of representation: MR refers to the input representation (IR); SR refers to the output representation (OR). Optimality Theory (Barlow & Gierut, 1999; Prince & Smolensky, 1993) Two Basic Type Of AntagonisBc Constraints: Markedness Constraints - require that output forms be unmarked in structure. i.e., output is simplified by markedness constraints that are moJvated by the frequency and distribuJon of sounds in the ambient language as well as the perceptual and arJculatory characterisJcs of the sounds. Sounds that are difficult to perceive and/or produce are marked Faithfulness Constraints - require that OR resemble IR. Capture the features to be “preserved”, prohibiJng addiJon or deleJon that violates the ambient language. Little Julias’s rules (constraints) All σ s require singleton onsets, but not all target σ s have onsets. How might she produce /preɪ. ɪŋ/? ? How might she produce / εl.ə/? ? Child has syllable structure limitaMons; no consonant sequences are permiOed. But here are too many consonants in the target words. How might she produce /pleɪ/, when her system does not allow C seq.? ? How might she produce /spaɪ/, when her system allows codas, but only singleton consonants? ? Every foot must be Sw, but there are too few or too many σ s in a foot. How might she produce /eɪt/, which has only 1 syllable? ? How might she produce /də.lɪʃ.əs/. ? 43 Li9le Julias’s rules (constraints) All σ s require singleton onsets, but not all target σ s have onsets. How might she produce /preɪ. ɪŋ/? Reduce onset & insert consonant [peɪ.jɪŋ]; eliminate σ boundary [peɪ. ɪŋ] How might she produce / εl.ə/? Delete vowel [lʌ]; or move a consonant [lε.l.ə] Child has syllable structure limitations; no consonant sequences are permitted. But here are too many consonants in the target words. How might she produce /pleɪ/, when her system does not allow C seq.? Delete a consonant [peɪ]; or insert a vowel [pə.leɪ]. How might she produce /spaɪ/, when her system allows codas, but only singleton consonants? Migrate the /s/ to get [paɪs]. Every foot must be Sw, but there are too few or too many σ s in a foot. How might she produce /eɪt/, which has only 1 syllable? Insert a vowel [eɪt.ə] How might she produce /də.lɪʃ.əs/. Delete σ [ˡlɪʃ.əs/.] or [ˡdʌ. ˡlɪʃ.əs]. 44 Constraints and Repair Processes in Syllables Child’s Constraint Target Word Resulting Repair Challenge Difficulty All σs require an onset TOO FEW C’s Onsetless σ e.g., [.æ.pu] -Insert C or [.kreɪ.an] e.g. [.ʔæ.pu] or [wæ.pu] - (Delete V.) [.pu] Only one C in an onset TOO MANY C’s A C not in a σ -Delete a C e.g., [pleɪ] e..g [peɪ] -Insert a V -E.g., [pə.leɪ] 45 Constraints & Repairs Table 1.2. Constraints and Repair Processes for Stress Patterns Child's Constraint Target Word Resulting Difficulty Repair Challenge Every foot must be Sw. TOO FEW s's for a foot A foot with a single Insert V- up [ʌpǝ] s Reduplicate - bird [bɜbɚ] Every foot must be Sw. TOO MANY s's for a A s not in a foot. Delete s - banana [nænǝ] no Sww feet foot Make weak s strong (Sw, no wS feet SS). 46 Optimality Theory (Barlow & Gierut, 1999; Prince & Smolensky, 1993) ApplicaJon to typically developing children Children’s Early produc@ons are unmarked. Ini@ally MCs are high ranked so that child’s unmarked output forms are likely to surface. A child’s system gradually changes to approximate the target language. ApplicaJon to SLP PracJce James et al. (2008) applied OT principles to explain children’s development of polysyllabic words in 5 stages. Stage 1 (1;0 – 2;3) – children are faithful to the stressed syllable and the dura@on of the whole word. Stage 2 (2;4-3;11) – children are faithful to the # of syllables in a word (i.e., reduce use of WSD, CR, FCD) Stage 3 (4;0-6;11) – faithfulness to all phonemes in words and can cooccur with a period of dysprosody (i.e., word rhythm may be disrupted). State 4 (7;0-10;11) – faithfulness to word rhythm w/ accurate stressed and weak syllables Stage 5 (11;0-older) – adultlike produc@on of polysyllabic words Sonority Hypothesis The order of consonants within an onset or within a coda is typically governed by the sonority hierarchy Sonority hierarchy – is a scale based on the “sonority” of a segment Corresponds to how similar a segment is to the most prototypical segment (i.e., the vowel). Scale is charted from the most sonorant to the least sonorant Sonority Hierarchy & Sequencing Principle The vowel (nucleus) is the peak: segment with high sonority. Consonants within the onset or within the coda are sequenced from the peak from most sonorant to least sonorant. Vowel Most Sonorous Glide (planned) /plӕnd/ Liquid Nasal Voiced Fricative (clutched) /klʌʧt/ Voiceless Fricative Affricate Voiced Stop Voiceless Stop Least Sonorous (trunk) /tɹʌŋk/ 49 Sonority Hypothesis Application to Typically Developing Children According the the sonority sequencing principle (SSP), children tend to reduce onset consonant clusters in a manner that produces maximal rise in sonority & word-final clusters will be reduce to produce minimal fall in sonority. Application to SLP Practice SH used to make clinical decisions regarding target selection to analyze children’s phonemic awareness skills To analyze children’s production of clusters Psycholinguistic Models Psycholinguistic Models A PsycholinguisYc model draws on the fields of psychology and linguisYcs Tries to account for psychological processes (i.e., mental mechanisms) involved in the “percepYon, storage, planning & producYon of speech as it is produced in real Yme in real ucerances” (McCormack, 1997, p.4). Simply, this model describes the discnccon b/w UR of words and their produccon (Fee, 1995). It is used to explain interaccons b/w auditory input, underlying cognicve-linguiscc process, and speech output (Dodd, 1995a). input (speech signal child hears) Output (words spoken by child) In Sum, the “possible events that occur between input and output are the focus of psycholinguisnc models.” (BBF, 2021; p. 68).