What is tone in phonology

Car segmentals. Phonology

Transcript

1 Bielefeld University Faculty of Linguistics and Literature Phonetics k Phonology Autosegmental Phonology (Hall, Chapter 6)

2 Exercise 7 (1) Describe the following natural classes of German with as few distinctive features as possible: (a) / pbm / [-kont, LAB] (b) / mn N l / [-kont, + son] (c ) / kg / [-kont, -son, DORS] (d) / C kg N ³ / [+ cons, DORS] (e) / aa: / [+ deep] (f) / i: I y: Y / [-cons, -hint, + high]

3 Exercise 7 (2) The redemption to derive 540 from the underlying / ng / cannot yet describe all data correctly, as the following corpus shows: Ungar [UNga:] eng [EN] Ding [din] Ingo [INgo:] Rings [³ fingi6n] lung laryngal [la³ YNga: l] elongated [lenlic] slow [lanzam] ginger [INv6] Transcribe the above corpus, ...

4 Exercise 7 Determine the contexts in which [g] is deleted and set up the corresponding rule: [g] is deleted after the rear nasal (540) at the final word, before Schwa and before consonants. -son -kont + sth DORS> Ø / + nas + hint {} [+ cons] -cons + hint -tief -round #

5 Exercise 7 (3) In the following segments, change the characteristics given after them and leave all other characteristics unchanged. Enter the IPA symbol of the new segment obtained in this way. (a) / u / [± hint] / y / (b) / m / [± nas, ± son] / b / (c) / d / [± sth] / t / (d) / r / [± son, ± sibil] / z / (e) / t / [± asp] / t / (f) / 2 / [± sp, ± hint] / O /

6 Linear Phonology Generative phonology (after Chomsky & Halle, 1968) is an example of linear phonology. Representations are linear sequences of feature matrices. / t / / a: / / g / + Kons -son -sth -kont -nas KOR + ant -kon DORS + hint -high + deep -gesp + long + Kons -son + sth -kont -nas DORS

7 Linear Phonology Characteristics are viewed as properties of individual segments. The segments are arranged linearly (i.e. in a chain). Within a segment, the features are disordered and 'simultaneous'. Problem: Not all features can be restricted to segments in this way. Examples are tone (pitch), quantity (duration of the linguistic units) and intonation (speech melody). Such features are superimposed on the segments and are therefore called suprasegmental.

8 Tone In many of the world's languages, pitch is crucial in determining the meaning of a word. Such tonal languages ​​include many African languages, Indian languages, and East Asian languages. Sound is used to describe the phenomenon of distinguishing the meaning of lexical elements by pitch. Tone must be distinguished from intonation, which describes the pitch of an utterance: Peter is asleep Peter is asleep

9 tones require a carrier, a so-called Tone Bearing Unit (TBU). Usually vowels carry the tone, rarely sonorous consonants too. Phonetics & Phonology WS 2006/2007 Tone Tone languages ​​distinguish up to five pitches (1 ... 5). There are two different types of tones: Register tones stay on the same pitch, can be described by a pair of identical numbers (11, 22, 33, 44, 55). Contour tones change the pitch during articulation, can be described using a sequence of numbers (e.g. 51, 214, etc.) The specification of the pitch of a language is always relative to a language and not comparable between languages.

10 Tone IPA The IPA provides various diacritical symbols for tones. One possibility is to draw the pitch gradient to the left of a vertical line that represents the pitch range from 1 to 5. Register tones are identified in the IPA as follows: eå ó 55 extra high eæ ô 44 high eç õ 33 medium eè ö 22 low eé 11 extra low Example from the West African language Nupe: [beæ] come [beç] add [beè] are similar

11 Tone IPA contour tones are identified as follows: e <ø rising eì ù falling eí ú high rising eî û low rising eñ ü rising falling eò ý falling rising The pitch gradients can be described more precisely using other symbols: û (51), ü (52 ), ý (53), þ (54), (412), А (341), etc.

12 tone IPA example from Mandarin Chinese: [ó ma] 55 mother [ú ma] 35 hemp [ý ma] 214 horse [ù ma] 51 scold example from the Hmong: [ó po] 55 ball-like [ý po ] 53 female [õ po] 33 pancreas [à po] 24 throw [ö po] 22 thorn [ò poð] 42 grandma (paternal) example from Cantonese: [ç po?] 31 see [ó si] 55 poem [ Ý si] 21 time [õ si] 33 try [à si] 24 cause [ö si] 22 thing [ß si] 23 city

13 Tone as a distinctive feature In the system of generative phonology one could express tone using distinctive features, e.g. [+ H] high tone, [-H] low tone, / aæ / [+ S] rising tone, ... / aè / / a

14 Tone as a distinctive feature In the Chadian language Margi there are the following variations in connection with the suffix / aær è / (definiteness): (a) final consonant: [saæl] [saælaær è] man [kuèm] [kuèmaær è] meat ( b) Final stem vowel with tweeter: [? æm æ] [? æmjaær è] water [kuæ] [kwaær è] goat [taæguæ] [taægwaær è] horse (c) stem final vowel with low tone: [t è] [tja

15 Tone as a distinctive feature Observations and rules: The high tone of the suffix initial vowel changes with the lower tone in the stem-final vowel into a rising tone: R1: [+ H]> [+ S] / [-cons, -H] + stem finale [i] and [u] become (before vowel-initial suffixes) to the corresponding floating sounds [j] and [w] (see (b) and (c)), R2: / iu /> [jw] / + [- Kons]

16 Tone as a distinctive feature This leads to the following correct derivatives: / saæl + aær è / / kuæ + aær è / / huè + aær è / ZR - - â R1 saæl + aær è kuæ + aær è huè + a

17 Autosegmental Phonology Basics These rules derive the data correctly, but they explain e.g. not why the high note of the suffix initial vowel becomes a rising note after the stem-final low note. The autosegmental phonology (Goldsmith, 1976) allows a better representation of the facts. Basic idea: Characteristics are arranged in independent layers (English tiers) (Greek autós itself, own). The segmental features form the segment layer and the clay features form an independent clay layer. The elements of the two layers are connected to one another via association lines (à non-linear phonology).

18 Autosegmental Phonology Basics One assumes two features H (or [+ H]) and T (or [-H]) on the tone layer: / aæ / / r / / è / -kons + son + kont -nas DORS + hint + deep + cons + son + cont -nas KOR + ant + apik -cons + son + cont -nas DORS -hint + high -tief segment layer HT association lines clay layer

19 Autosegmental Phonology Basics The segment with which the tone is associated is called the tone-bearing element or tone-bearing unit (TBU). In the Margi language, e.g. only be vowels TBUs. Contour tones are represented by the association of a TBU with several tone characteristics: / aì / / a

20 Autosegmental Phonology Example 1: Margi This explains the Margi data better. As a reminder: / uæ? Uè + aær è /> [uæ? Wa [j w] / + [--cons] works alone on the segment layer.

21 Autosegmental Phonology Example 1: Margi The application of the sliding sound rule gives the following result: / u? w + a r i / H T H T Since a gliding sound in Margi cannot be a TBU, the circled feature loses its anchor and is gliding (floating tone). The sliding feature can, however, be associated with the following TBU: / u? w + a r i / H T H T

22 Autosegmental Phonology Example 1: Margi In addition to the segmental sliding sound rule, you only need one rule that says how free-floating tones are associated: [-kon] T H This rule says that a free-floating low tone is associated with a following high tone vowel. The example illustrates the tone stability: although a segment that acted as a TBU is deleted, the tone is preserved independently and is re-associated.

23 Autosegmental Phonology Example 2: Tone Assimilation Another example for the analysis of contour tones as a sequence of individual tone features is provided by tone assimilation in Margi. The infinitive suffix / + na / is not based on a tone, but takes it from the last tone of the stem: / saæ + na /> ​​[saænaæ] lose / ndaèl + na /> ​​[ndaèlnaè] throw> plan This assimilation (esp. In the last case can be explained if one assumes that a rising tone is analyzed as a sequence of two features, with only the last feature spreading.

24 Autosegmental Phonology Example 2: Tone assimilation / sa + na / H / ndal + na /> ​​/ sa + na / H> / ndal + na / Assimilation rule: [-cons] [--cons] [αh] TTHT> TH This example illustrates also: sound features are not segmental.

25 Autosegmental Phonology No Crowding Constraint, No Crossing Constraint and OCP In many languages ​​there is only a limited number of sound patterns that are allowed for words. In the Bantu language Etung there are four different tones: H, T, HT, TH The distribution of these tones is limited: One syllable: Two syllables: Three syllables: kpaæ, kpeè, naì, no

26 Autosegmental Phonology No Crowding Constraint, No Crossing Constraint and OCP The data can be explained if one assumes an autosegmental analysis in which words are based on certain tone patterns. These are then connected to the segments according to certain association conventions. Tone pattern for Etung: THTHHTTHTHTH Association conventions (after Goldsmith, 1976): Associate ... (a) ... tones with TBUs one-to-one from left to right, (b) ... leftover TBUs with the last tone, ( c) ... floating notes with the last TBU.

27 Autosegmental Phonology No Crowding Constraint, No Crossing Constraint and OCP Examples: / kpa / H / na / (a) (a) (c) HT In many languages ​​there are restrictions stating the maximum number of tones that can be associated with a TBU ( No crowding constraint). In the etung the maximum number is two, which excludes the following: * / na / T H T For languages ​​without contour tones: maximum number = 1

28 Autosegmental Phonology No Crowding Constraint, No Crossing Constraint and OCP Examples: / oba / / oda / (a) (b) (a) (a) / abo / (a) (a) (c) HHTHTH is implicit in the association conventions the universal No Crossing Constraint. Crossing prohibition: Association lines must not cross. * / oda / (a) H T (a) The crossing prohibition follows from the linearity with regard to the chronological sequence of the individual layers.

29 Autosegmental Phonology No Crowding Constraint, No Crossing Constraint and OCP Actually, one has the problem of having to choose between two analyzes / tone patterns with a word like / oæbaæ /: / oba / / oba / HHH That the left analysis is actually the right one , goes back to the Obligatory Contour Principle (OCP), which applies to Etung: OCP: Identical adjacent tones are ungrammatic. The OCP explains why there are only the specified tone patterns in the Etung, and not, for example, H H or T T H.

30 For audio languages, restrictions apply (language-dependent) with regard to the maximum associated audio characteristics / TBU (No Crowding Constraint) and the OCP. Phonetics & Phonology WS 2006/2007 Summary Some phenomena, e.g. Clay are suprasegmental in nature. The data can often be better explained using a non-linear, autosegmental analysis in which features are located on different independent layers. In addition to segmental rules, association conventions and rules are required to link these layers in order to describe the spread of features. The crossing ban is a universal car segmental restriction.

31 Suppose the OCP did not apply. Which false alternate derivative would one predict for / laèguè + aær è /? Phonetics & Phonology WS 2006/2007 Exercise 8 (1) Derive the tone distribution of the words eækuæeæ, b èsoæneæ, and aæd èmbaæ using the tone patterns and association conventions in the Etung on p. 28ff. (2) Consider the following date of the Margi (p. 23ff): / laèguè + aær è /> [laègwaær è] Determine the underlying shape of the trunk and derive the surface representation using the sliding sound rule (p. 23) and the association rule ( P. 25).

32 Exercise 8 (3) In the Lardil language the accusative future tense suffix / -u ± / (see 6th session, p. 27f) has the allomorphs [- ±] and [-wu ±]: non-inflected non-future tense Future tense [kentapal] [kentapal-in] [kentapal-u ±] Dugong [ketar] [ketar-in] [ketar-u ±] river [miyar] [miyar-in] [miyar-u ±] spear [mela] [ mela-n] [mela- ±] sea [kunka] [kunka-n] [kunka- ±] ledge [Nuka] [Nuku-n] [Nuku- ±] water [tþempe] [tþempe-n] [tþempe- ± ] Mother's father [ke µe] [ke µi-n] [ke µi-wu ±] wife [pape] [papi-n] [papi-wu ±] father's mother

33 Exercise 8 (3) Assume the root form of the inflected forms as the basis (e.g. / ke µi /). The suffix / - ± / can be derived from the following vowel cancellation rule. V> Ø / V Enter a rule that explains the occurrence of the suffix / -wu ± / and derive it as an example from [Ni iwu ±] (skin). How must the vowel deletion rule be arranged in relation to this rule? What do you call this order?