Mark J. Jones [email protected] Historical Linguistics Lectures: week 5 SOUND CHANGE 5: THE ROLE OF THE LEXICON Introduction Today’s lecture focuses on the location of change within language: is it at the level of the sound, or the level of the word? Traditional view When a word is heard, it is stripped of information relating to speaker (identity and emotional state) to get to the distinctive sound structure, and the distinctive sounds in it are stripped of contextual information (e.g. nasalisation is factored out), sometimes erroneously (Ohala, see McMahon 1994). This process is necessary to identify words from different speakers (including new voices). Grammar vs. lexicon Grammar – where regular patterns are found (phoneme inventory, rules) Lexicon – lexical items/morphemes stored with information on category and any irregularities Neogrammarian sound change Regularity hypothesis on sound change Sound change is regular and exceptionless Occurs in all words presenting the right conditioning environment Phonetically gradual Lexically abrupt For example, Stage 1: PIE *wértō > Stage 2: Proto-Germanic *werTō Phonetically gradual – changes must not be noticed Lexically abrupt – all words affected across lexicon So change is seen as operating at some abstract level: t > T Exceptions Reformulate change – closer inspection of data Dialect borrowing – words look like they are from dialect X, but come from dialect Y after change Analogy – sound change does not apply in some cases, e.g. sword but swore (cf. swear) of which more later… Alternative changes: phonetically gradual and lexically gradual phonetically abrupt and lexically gradual phonetically abrupt and lexically abrupt Are these attested? Lexical diffusion Classic example: Cháozhōu Chinese (Chen and Wang 1975) Tonal split – 4 tones > 8 tones depending on voicing of initial consonant BUT Lent Term 2008 Middle Chinese tone 3 > 50% Cháozhōu tone 3b (as expected) and 50% tone 2b (unexpected!) No apparent phonological conditioning – homophones split English diatones (Phillips 1984, citing earlier research by Sherman 1975) segmentally ‘homophonous’ verb-noun pairs like import and import Stress placement depends on class: verb (final stress) vs. noun (initial stress) No possible phonological conditioning 90 diatones created 1880-1934 Middle English /O(˘)/ > /e(˘)/ (<eo> vs. <e>) (Phillips 1984) Ormulum data – orthographic principles adopted to indicate current pronunciation werrpenn ‘to cast’ 100% written <e> forrweorrpenn ‘to despise’ 97% <e> weorrc ‘work’ 62% <e> weorelld ‘world’ 2% <e> Not due to phonological conditioning: deor ‘animal’ <e> 0% vs. deore ‘dear’ <e> 68% þreo ‘three’ <e> 0% vs. freo ‘free’ <e> 67% leom ‘gleam’ <e> 0% vs. leof ‘pleasing’ <e> 97% Effect superficially appears random It appears that sound change can occur which is phonetically abrupt and lexically gradual, i.e. they diffuse through the lexicon one word at a time. This provides a link to traditional dialectology, which said that ‘every word has its own history’. Trivially true, but message is that abstract system (of e.g. phonemes) plays minimal role in change Neogrammarians – unit of change = sound Diffusionists – unit of change = word Finding these effects is all very well. The next question is ‘why’? What is the driving force behind lexical diffusion? There are a number of candidates: Analogy Word-frequency Word category Analogy – protects sound-meaning link. It promotes transparency. For example: English sword /swo/ > /so/, this is the regular change, but… swore /swo/ = /swo/ - this word does not change. Why not? The conventional idea is that analogy with the present tense ‘swear’ with initial /sw/, in which /sw/ does not become /s/, maintains the link between the two verb forms: */so˘®/ past tense vs. /swe®/ present tense There are different types of analogy: Analogical levelling decreases variation within a paradigm, e.g. am, are, is (dialectal am/is/are) vs. older eom, bist, is, sind Analogical extension 2 Lent Term 2008 extends pattern beyond original source to new words, e.g. Old English hūs (sg.) vs. hūs (pl.), Modern English house (sg.) vs. houses (pl.) German Hund vs. Hunde also *Hünde Analogy even applies in cases of new word formation: Word-formation smelly < smell lazy < ? must be laze But notice that analogy as typically defined applies to grammatical paradigms, so… In languages like Chinese, which lack inflectional and derivational morphology, there can be no analogy Lexical diffusion in more detail. Phillips (1984) examined not just the rate of diffusion of the Middle English change /O(˘)/ > /e(˘)/ (<eo> vs. <e>), but categorised which words it occurred in. Nouns – 2 patterns least frequent have <e> (innovate) most frequent keep <eo> (unchanged) Synchronic data Phillips (2001) 30 informants from Georgia US list of 110 words, including 16 words with initial /str-/ most informants had either conservative [st®] or innovative [St®] 2 varied – had [st®] in low frequency strait and had [St®] in high frequency straight Word-frequency more frequent words subject to ‘physiologically motivated changes’ (they ‘wear out faster’) less frequent words subject to (emerging?) constraints on phonotactics etc. Lexical diffusion also seems to operate differently depending on whether the word affected is a function word or a content word. Phillips (1984) Middle English change /O(˘)/ > /e(˘)/ (<eo> vs. <e>) Word category Function words – all <e> except heore ‘their’ Numerals – most <eo> Verbs – all <e> except geornnde ‘yearned’ Synchronic data Lavoie (2002) Looked at realisations of for and four in American English Traditional view: homophones /fç®/ Synchronic data (full > reduced) for [fo®, fç’, f´’˘, f´’, f´, f] four [fç®, fo®, fç’®, fç’] Homophones differently affected, depending on what the frequency is (strait vs. straight) and what the word category is (for vs. four). 3 Lent Term 2008 Traditional view: word is heard, stripped of non-linguistic or predictable information > abstract representation (phonemes?) Listed in lexicon tagged with necessary grammatical information Speech pays no attention to grammar or usage Exemplar models Representations more detailed than previously considered Phonetic detail (of allophones, individual voices) retained Cloud of exemplars related by form and meaning Effects of word frequency (i.e. time) can be incorporated Synchronic data Labov (1981) Neogrammarians vs. lexical diffusion “both are right, but both cannot be right.” (p. 269) “In the evolution of sound systems, is the basic unit of change the word or the sound?” (p. 268) Conclusion: s Neogrammarian change = low-level output rules, e.g. /t/ realised as [t ] Lexical diffusion change = redistribution of abstract word class into other classes Not necessarily so neat: e.g. social conditioning Summary Considerable diachronic evidence for ‘lexical diffusion’ Also for Neogrammarian type sound changes Synchronic data show that effects are also seen here: Frequency effects Category effects Evidence suggest lexicon has real role to play Reading list Chen, M. Y. and W. S-Y. Wang. (1975). “Sound change: actuation and implementation.” Language 51: 255-281. Labov, W. (1981). “Resolving the Neogrammarian controversy.” Language 57: 267-308. Lavoie, L. M. (2002). “Some influences on the realization of for and four in American English.” Journal of the International Phonetic Association 32: 175-202. McMahon, A. (1994). Understanding language change. CUP, Cambridge. Phillips, B. S. (1984). “Word frequency and the actuation of sound change.” Language 60: 320342. Phillips, B. S. (2001). “Lexical diffusion, lexical frequency, and lexical analysis.” In J. Bybee and P. Hopper (eds.). Frequency and the emergence of linguistic structure. John Benjamins, Amsterdam: 123-136 4
© Copyright 2026 Paperzz