LNGS 401 Lecture Notes 1 & 2 (November 2024) PDF

LNGS 401 Lecture Notes 1, 2 November 05, 2024 Dr. Fusheini Hudu General intro: Today’s lecture is going to cover the following: What the course is about. We shall go through the course outline, which will give you the understanding that this c...

LNGS 401 Lecture Notes 1, 2 November 05, 2024 Dr. Fusheini Hudu General intro: Today’s lecture is going to cover the following: What the course is about. We shall go through the course outline, which will give you the understanding that this course is a continuation of what you learned last year in LNGS301. After going through the course outline, we shall do a brief review/overview of what phonology is about. The goal is to refresh your mind and get you ready for the main content of this course. What is phonology? There are three basic issues we investigate in phonology. i. Regularities that define the units of phonology in a language. ii. Patterns of distribution. iii. Alternations (morphemes) In this course, our interest is in crosslinguistic patterns/generalisations relating all three basic issues. We will then see how these patterns and generalisations are analysed using various theories. In phonological analyses, there is a focus on various units for the purpose of unearthing phonological patterns and generalisations. What are the units of phonology? There are units of grammar that constitute the core units of phonology. From the most basic upwards, these core units are features, sounds, syllables, stress, tone, and intonation. We studied the nature of each of these units in LNGS 301. Of these, syllables (and syllabic units such as the onset, nucleus, coda, rhyme, mora), stress, tone and intonation are part of a sub-filed of phonology called prosody. They are prosodic units. Prosodic units share one feature: they transcend the boundaries of individual segments. Either they consist of more than one segment put together (e.g. the syllable) or they are associated with more than one segment (e.g. stress, tone, intonation). Another name for stress, tone and intonation is suprasegmental features. In dealing with these core units of phonology for phonological analysis, we have to look into morphological units like morphemes and words, and syntactic units like phrases and sentences. In today’s class, we shall briefly go over what we said last year about segments (vowels and consonants) in language. What is phonology: The basics in brief (Note: This is a brief recap of what we did last year. Read your detailed LNGS301 notes to refresh your memory). In the course on introduction to phonology, you learned that in phonology, we study: o Sounds as mental objects in specific languages o Sounds as they are organised in the mind o The distribution of sounds in words o The sequencing or patterning of sounds with other sounds to form words and phrases. o The changes that sounds undergo and the principles behind these changes. You also learned that in doing these basic analyses, meaning is the independent Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 1: November 05, 2024) page 1 of 7  arbiter. We resort to meaning to determine if a sound exists in a language. After establishing that certain sounds exist in a language, we also need meaning to determine if two or more existing sounds are separate distinct sounds (phonemes) or different realisations of one and the same sound (allophones of one phonemes). We noted that: If the presence versus absence of a sound in a word can change the meaning of a word, that sound is part of the phonology of the language and is distinctive (e.g. [spik] versus [pik]). Thus, [s] is a phoneme (mental sound) in English. If by replacing one sound with other sounds the meaning of the word changes, the sounds being replaced are different phonological sounds (phonemes) (e.g. [taim] versus [daim] and [laim]). Thus, [t], [d], [l] are separate phonemes (mental sounds) in English. If by replacing one sound in a word with another or others, the meaning of the word remains the same, then the two sounds are not separate phonological sounds, though they are different phonetic sounds. The two or more sounds are one phoneme that changes in different contexts (e.g. [bɛtə] and [bɛɾə]). Both words have the same meaning:. Thus, [t] and [ɾ] are not different phonemes in English. It is one and the same mental sound that is either realised as [t] or [ɾ] in North American English. Notice that phonetically, [t] and [ɾ] are different sounds because they are different phonetic features. But phonologically they are not different because replacing one of them with the other in a word does not cause a meaning change. You also learned how to write rules to account for the occurrence or distribution of sounds in different contexts. If we had plenty of American English words with [t] and [ɾ], and studied the local environments of these sounds, we would realise that each occurs in a different environment. Let us illustrate rule writing using something we observe everyday. When you are thirsty, you need a liquid called “water”, to quench your thirst. When you keep a bottle containing the same water you drank into a deep freezer overnight, the water will become solid, and will be called “ice” and can no more be drank, even though it is the same content that you drank the night before. If you pour the water into a pan and place it on a burning stove, after a while, the water will change into vapour and vanish into the air. The rising vapour, solid ice, and liquid water are all originally one and the same thing. The only reason we see them in different states is the temperature it finds itself in. In a freezing temperature, it becomes solid, in a very high temperature it becomes gas, and in a normal (room) temperature, it is liquid. If I ask you which of these three states is the normal state, you will most likely say it is the liquid state. So, you could write a rule like the following to account for the different realisations of the water. [ice (solid)] / in a very low temperature /water/ [vapour (gas)] / in a very high temperature [water (liquid)] / elsewhere In level 300, we studied similar phonological rules about sounds that are not separate phonemes, using information about their distribution. We noted that when two sounds are allophones of the same phoneme, they will not be seen in the same environment. In our current example, you will realise that water, ice and vapour cannot be found in the same temperature. Where you see one, you will not see the othersother. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 1: November 05, 2024) page 2 of 7  You also learned that in writing the rule, we determine what the default realisation of the phoneme is by looking at which of them occurs in many/wider environments. In other words, which of the realisations do we see most of the time. We used the same criterion to choose the liquid state as the default, because that is what we see most of the time. If you were to ask a chemist who knows the chemical components of water to write this rule, they will most likely represent the various states differently, using their chemical properties. Instead of “water”, you would get H2O. This is similar to writing a rule using the features of the sounds, instead of the segments themselves. In our next lecture, we shall be looking at features and the two major kinds of features we have in phonology. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 1: November 05, 2024) page 3 of 7 LNGS 401 Lecture Notes 2 November 12, 2024 Dr. Fusheini Hudu Distinctiveness and redundancy in phonological features.  One of the observations we made last year is that, some segmental features are distinctive, others are redundant.  We made the point that certain features are naturally part of certain segments (vowels or consonants). For this reason, when describing these segments, we do not have to include these features before their presence is accepted. Such features are called redundant features for such segments.  To understand this better, let us use an analogy with human description. In primary school, when you were asked to write an essay titled “my best friend”, there were certain features of your best friend that you certainly did not include in describing your best friend. You did not add that your best friend has two legs, one head, two eyes, two ears, two arms, ten fingers etc.  Why did you leave these out? Because every human being normally has them, and so we assume that your best friend also has them. Even though it is true that your friend has them, it is redundant to include them in your description, as they don’t define your best friend. They are not distinctive to your best friend.  You also did not describe your best friend as a hungry person because you found him hungry after failing to eat for 12 hours. It is right to say that anyone who fails to eat for 12 hours will get hungry, but it is not right to describe that state of hunger as a defining feature of anyone.  It is the same with sounds. If you describe a vowel, a glide, a nasal or a liquid as [+voice], it is like describing your friend as having two legs. However, if the vowel, glide, nasal or liquid were to lose its natural voicing and become devoiced, that would make it unique (just as it is unique for a person to lose one leg). In that case, you can add the lack of voicing to your description.  However, if you take any vowel, you must indicate whether it is high or low, front or back. This is because not every vowel is high, not every vowel is low, not every vowel is front etc. It is like describing the height and complexion of your friend. You need to do that because not every person is tall, not every person is short, not every person is dark etc.  It means that for all vowels (but not for all consonants), voicing is a redundant feature but height and frontness are distinctive features. Similarly, for consonants in many languages of the world, the features [labial], [coronal], [dorsal] are distinctive.  In all languages of the world, changing a vowel within a word from [+voice] to [-voice], if it is possible for you to do so, will not change the meaning of the word. But most languages, changing the height feature from [+high] to [-high] will change the meaning of the word (e.g. English [bit] vs. [bɛt]). Similarly, changing the frontness feature from [+back] to [-back] can also change the meaning (e.g. English [biːt] vs. [buːt].  In English and many languages, changing the place of articulation of a voiceless consonant in a word from [+labial] to [+coronal] or [+dorsal] will change the meaning of the word (e.g. [kɔt] vs. [kɔk] vs. [kɔp]).  This means that distinctive features contribute towards meaning, redundant features do not contribute towards meaning. a. From the explanation and the examples given above, we can draw two conclusions: a. Whenever two sounds are separate phonemes in a language, there is at least one distinctive feature making them different sounds. From the above examples, /i/ and /ɛ/ are different vowels because the distinctive feature [high]. /iː/ and /uː/ are different Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 2: November 12, 2024) page 1 of 7  vowels because of the distinctive feature [front]. b. Whenever you have a minimal set of words, there is one or more distinctive features responsible for the meaning difference between the words. The English words [kɔt], [kɔk] and [kɔp] are different because the features [labial], [coronal] and [dorsal] are distinctive.  For redundant features, because they are predictable, we can always tell whether they are part of the specification of a sound or not, just as we can always tell that failure to eat for 12 hours will make one hungry. For this reason, their presence or absence does not lead to a meaning change.  For instance, when pronouncing an English word with a voiceless plosive (e.g. /k, t, p/) at the start of a syllable that is stressed, the voiceless plosive will get aspirated (e.g. ['tʰaɪm]. However, when you choose to pronounce it without the aspiration (['taɪm]), that will not change the meaning.  Because its presence in English can always be predicted, they do not contribute towards meaning, and for that matter, do not have to be specified. They can be ignored.  We have thus seen three differences between distinctive and redundant features: Distinctive features are Redundant features a. not predictable. Are predictable b. contribute towards meaning Do not contribute towards meaning c. cannot be ignored, must be specified Do not have to be specified, can be ignored  Remember that, in most cases, distinctiveness and redundancy are language-specific. Each language decides on whether to make a certain feature distinctive or not, just as each language decides on whether to make a certain sound a phoneme or not.  A feature can be distinctive in one language but redundant in another language.  Also, what is distinctive for a category of sounds in a language (e.g. vowels) may be redundant for another category of sounds in the same language (e.g. consonants). For instance, [±nasal] is distinctive for consonants but redundant for vowels in English and many languages of the world.  Below, the list of some, not all distinctive and redundant features for English vowels and consonants are listed. Study them closely and do the following: ◦ Come up with minimal sets of words similar to those shown in the examples above which demonstrate that the each of the distinctive features shown below is indeed, distinctive in English. Focus on the features that were not shown in the examples. ◦ Look for more distinctive features in English and support their distinctiveness using minimal set of words. English consonants Distinctive features Redundant features [±voice], [±sonorant], [±labial], [±coronal] [±spread glottis.], [±const. glottis], [±dorsal], [±delayed release], [±nasal] [±round] English vowels Distinctive features Redundant features [±high], [±low], [±front], [±back], [±nasal], [±long], [±lateral], [±voice] [±tense], [±round] [±sonorant] etc.  Why do we need this distinction between distinctive and redundant features. For several reasons. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 2: November 12, 2024) page 2 of 7  One has to do with rule writing using features.  When we represent sounds, there are two levels of representation: the underlying representation (UR) and the surface representation (SR).  UR = distinctive features. The underlying representation only captures the features that are distinctive for a particular language, features that are required to build meaning and thus cannot be predicted (remember being tall or short, fair or dark cannot be predicted about your friend and must be state).  SR = redundant features: The surface representation captures features that are not distinctive. These features only surface in specific and predictable environments (remember being hungry for failing to eat can be predicted and will not be part of the description of your friend).  Thus, we can write a language-specific rule to account for the distribution of a surface feature.  For instance, there is a rule that makes all English vowels nasal when these vowels appear right before a nasal sound. For this rule to apply, the vowel and the following nasal must be part of the same syllable.  There are three ways of stating this rule, as discussed in LNGS 301. 1. In words: A vowel is nasalised when it precedes a nasal within the same syllable. 2. With segments: /V/ → [Ṽ] / ____[N] ]σ 3. With features: [+syl] → [+nasal] / __[+nasal] ]σ.  As you may have realised, stating the rule with segments or features take the form: /UR/ → [SR] / environment: More analogy with human description:  In actual fact, the way we go about stating phonological rules is not different from the way we communicate and state observations on a daily basis. Let’s illustrate this with how we describe human beings.  In everyday interactions, we make statements like the following: a When I attend three lectures a day I get tired in the evening (that’s a student talking) b After talking for two consecutive hours, I get tired (that’s a lecturer talking). c Anyone who delays lunch till 3:00pm will get very hungry.  When we translate any of the above rules into features. Number (b), for instance, will look like: /lecturer/ → [tired lecturer] / two hour lecture ____ Practice Exercise: Translate the statement behind every successful man is a woman into features.  Note that rules are only possible when we are dealing with redundant features. a Phonemes are made up of contrastive features b Redundant features surface in allophones, they make certain sounds allophones of other sounds: (e.g. In English, [ĩ] is an allophone of /i/, [tʰ] is an allophone of /t/).  Just as rules in life depend on changing circumstances in which human beings find themselves, rules in phonology are due to the changing environments in which sounds find themselves.  Keep this in mind when we treat phonological processes. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 2: November 12, 2024) page 3 of 7 LNGS 401 Lecture Notes 3 November 19, 2024 Dr. Fusheini Hudu Language typology in feature specifications  What we are dealing with here is generally language-specific, but some features that surface in lexical representations are language-universal.  Language-specific features are learned as part of knowledge of the language. Thus, a child acquiring a language will acquire all the features that are distinctive and redundant in the language.  In other words, while distinctiveness and redundancy are generally language-specific phenomena, there are certain language-universal tendencies that we cannot ignore.  These language-universal tendencies come in three forms. a Certain features are distinctive in all languages of the world. For instance, ▪ In all languages, the feature [high] is distinctive because all languages have contrast between low and high vowels. ▪ In all languages, the feature coronal is distinctive because every language makes a distinction between coronal and non-coronal consonants. b Certain features are more likely to be distinctive than others in any language. c The position of a certain feature as distinctive in a particular language implies that some other feature will be distinctive in that language.  We will get back to this and discuss it in further detail below under the concept of markedness.  In phonology, as in other domains of linguistics, an important consideration is typology: the range of similarities and differences we find across languages.  When we treat phonological issues within typological perspectives, we address a number of related questions: 1. Which features/patterns/observations do we find in all languages? 2. Which features/patterns/observations are specific to only certain languages, language families, or regions of the world? 3. How do various languages treat various features?  When we focus on a specific language, we are often interested only in one question: which features are distinctive in this language, which are redundant?  When we compare one or two languages, we determine how each language treats a particular feature or a combination of features.  One language may treat a specific feature as contrastive, another may treat it as redundant.  Let us take a look at 2 features, voicing [±voice] and aspiration [±spread glottis].  Depending on which of them is contrastive, we can get 6 possible language types.  Out of the 6 possibilities, only 5 are attested in languages of the world.  Out of the 6 possibilities, only 5 are attested in languages of the world. See the discussion on markedness for an explanation on why the last one is not attested. (For the purpose of this illustration, we will consider plosives at 3 places of articulation: labial, alveolar and velar. It does not mean that there are plosives in all or only these places of articulation in the languages that will be stated below as examples. There could be plosives in other places of articulation or plosives in fewer places than the three assumed here. We are using these three only for the purpose of illustration). Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 3: November 19, 2024) page 1 of 5 Type 1: Both [±voice] and [±spread glottis] are contrastive. This will produce about 12 phonemes and no allophone in such a language. /p, t, k, b, d, ɡ, pʰ, tʰ, kʰ, bʰ, dʰ, ɡʰ/  An example of a language with such an inventory is Hindi, spoken in India. Practice Exercise 2: Come up with minimal pairs in Hindi to illustrate these phonemes. Type 2: Only Voicing is distinctive. Aspiration ([±spread glottis]) is redundant.  This produces 6 phonemes: Only /p, t, k, b, d, ɡ/  [pʰ, tʰ, kʰ, bʰ, dʰ, ɡʰ] may not even surface in the language. Even if they do, they will be predictable, redundant,  Voicing plays a role in meaning construction, aspiration does not.  Languages in this category are English, French and the and (the majority of) Ghanaian languages. Practice Exercise 3: Come up with minimal pairs in English, French, and at least two Ghanaian languages to illustrate these phonemes. Type 3: Only aspiration is distinctive. Voicing is redundant. This produces 6 phonemes: Only /p, t, k, pʰ, tʰ, kʰ/ are phonemic.  A language that falls into this category is Mandarine Chinese  Because voicing is redundant, [p, t, k] are not different sounds from [b, d, ɡ], and aspirated [bʰ, dʰ, ɡʰ] are not different from [pʰ, tʰ, kʰ].  In other words, what matters in Mandarine Chinese is to either aspirate or not aspirate a plosive. It does not matter whether the plosive is voiced or not.  [bʰ, dʰ, ɡʰ] may surface as allophones of [pʰ, tʰ, kʰ], just as [b, d, ɡ] may surface as allophones of [p, t, k]  Example of minimal pairs pà father pʰà to fear tà big tʰà to step kǔ ancient kʰǔ bitter Type 4: Neither voicing nor aspiration is distinctive. Both are redundant.  In such a language, if there are plosives at all three places of articulation, there will be only three plosives in the language.  Whether such a plosive is voiced or voiceless, aspirated or unaspirated does not matter for the purpose of meaning.  A language that falls into this category is Finnish, a language if Finland. In Finnish, there is only one alveolar stop and no contrast between [t], [d], [tʰ], [dʰ].  It also has only one labial stop and one velar stop /p, ɡ/.  In this language, replacing any of these stops with the voiced or voiceless counterpart or aspirating the stop does not change the meaning of the word. Type 5: Both voicing and aspiration are distinctive. However, aspiration is only distinctive when combined with [-voice]. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 3: November 19, 2024) page 2 of 5  In such a language, if there are plosives at all three places of articulation, only nine plosives will be distinctive in the language: /p, t, k, b, d, g, pʰ, tʰ, kʰ/.  Aspirated voiced plosives [bʰ, dʰ, gʰ] are not distinctive in the language.  A language that falls into this category is Thai, a language of Thailand.  Thai uses both aspiration and voicing, but only makes a three-way distinction: /t/, /d/, /tʰ/.  It means that both aspiration and voicing matter, but aspiration only goes with voicelessness.  Thus /dʰ/ does not exist as a distinct sound in the language. Type 6: Both voicing and aspiration are distinctive. However, aspiration is only distinctive when combined with [+voice].  In such a language, if there are plosives at all three places of articulation, only nine plosives will be distinctive in the language: /p, t, k, b, d, g, bʰ, dʰ, gʰ/.  Aspirated voiceless plosives [pʰ, tʰ, kʰ] will not be distinctive in the language.  There is no language in the world to illustrate this type.  This is where the idea of language-specificity in distinctiveness and redundancy is constrained.  In other words, there are limits to the freedom that languages have in deciding which features should be distinctive or redundant.  If there were no restriction, Type 6 languages would also exist.  The main limitation is markedness, the topic to discuss next. Practice Exercise 4: The 6 language types described above are not really the only possibilities when [±voice] and [±spread glottis] are combined. There are other possibilities not mentioned. The reason they are not included is because they are even more remote than Type 6, which is not attested. Study the 6 types described above and come up with at least two other types not included. THE CONCEPT OF MARKEDNESS  In the last section, we looked at two features: [voice], [spread glottis], each of which has two values: [+voice], [-voice]; [+spr glottis], [-spr glottis].  An important question that was not explicitly addressed is whether these feature values are of equal status or levels of preference across all languages. In other words, is one value of a feature as likely to be observed in a language as the other?  To understand this better, let's get back to our analogy of human features. Assuming there are human features such as authority, riches, intellectual etc. Each has two values: [+authority], [-authority]; [+rich], [-rich]; [+intellect], [-intellect].  Assume the following operational definitions for the values of each feature: ◦ [+authority] = the most powerful person within a certain jurisdiction (e.g. the president, vice chancellor, the paramount chief etc.). ◦ [-authority] = Anyone within a certain jurisdiction who lacks ultimate power. ◦ [+rich] = people with over 1 million Ghana Cedis; ◦ [-rich] = those with less than 1 million. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 3: November 19, 2024) page 3 of 5  ◦ [+intellect] = people with a Bachelor Degree or above, ◦ [-intellect] = those with no educational qualification or a degree lower than Bachelors (But those with Honorary PhD are out, unless they have a Bachelors degree ). ◦ [+ordinary] = people with less than 10,000 twitter followers, ◦ [-ordinary] = those with more than 10,000 followers. Now let's take each feature and look across various parts of the world. One value of each feature is going to be more prevalent than the other. To understand this, we put the feature values into two groups.  Group A has the values [-authority, -rich, -intellect, +ordinary]  Group B has [+authority, +rich, +intellect, -ordinary]. The feature values in Group A are of differ from the feature values in Group B in the following ways: 1. They are more likely to be found in different parts of the world than those in Group B 2. It is easier for a person to possess the features in Group A than those in Group B. 3. Wherever you find people with a value in Group B, you will also find people with the opposite value in Group A, but not vice versa. 4. It is extremely rare to find people possessing only the features in Group B and no feature in Group A, but it is very common to find people who lack any feature in Group B.  In the same way, phonological feature values are not of equal level of prevalence.  In phonology, and linguistics in general, the feature value that is more prevalent is called the unmarked value, the one that is less prevalent is called the marked value. The two differ in the following ways: 1. The unmarked value is more common across languages. 2. It is easier/more natural for the unmarked value to surface in sounds than the marked value. 1. The unmarked value is the default value. It is more basic or inherent in a sound. For this reason, any language that has the marked value, also has the unmarked value but not vice versa.  When markedness is applied to the feature values [+voice], [-voice]; [+spr glottis], [-spr glottis], what do we get? 1. [-voice] is unmarked, [+voice] is marked. 2. [-spr glottis] is unmarked, [+spr glottis] is marked.  This means: 1. In every language, you will find some [-voice] obstruents (e.g. [p, t, k]), but there are languages that lack any [+voice] for obstruents [b, d, g] (e.g. Finnish). 2. Lack of voicing for obstruents is fundamental/the more natural thing, voicing in obstruents involves an extra activity/measure just as being rich/having executive power involves an extra measures. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 3: November 19, 2024) page 4 of 5 Diagnostics for markedness  What do we use to determine the unmarked versus marked value of a feature?  The factors that we use to do this are called the diagnostics of markedness. We have actually been looking at these diagnostics.  We have noted the following 3 for unmarked values: ◦ more common across languages ◦ the default, basic, fundamental or natural value ◦ always there when the marked value is there. Practice Exercise 5: Find out three other diagnostics of markedness in phonology. Now, let's include more features and see what their marked and unmarked values are, using the diagnostics discussed:  Voicing in sonorants (j, w, l, r, m, n etc.): [+voice] is unmarked, [-voice] is marked.  Nasality in vowels: [-nasal] is unmarked, [+nasal] is marked.  Glottalisation in obstruents: [-constricted glottis] is unmarked, [+constr. Glottis] is marked. Practice Exercise 6: Go back to what you did in Practice Exercise 4. Using the concept of markedness, explain why Language Type 6 and the remaining two language types you identified (under Practice Exercise 4) are not attested in any language. Default Rules. This means that there are default rules universal to all languages. For each feature exhibiting a distinction between marked and unmarked values, there is a rule in Universal Grammar that assigns the unmarked value:  [-son] → [-voiced] (It is unmarked for an obstruent to be voiceless)  [-son] → [-spread glottis] (It is unmarked for an obstruent to be non-aspirated)  [+son] → [+voice] (It is unmarked for a sonorant to be voiced)  [+syll] → [-nas] (It is unmarked for a vowel to be oral)  [-son] → [-constricted glottis] (It is unmarked for an obstruent to be not glottalised)  Note that each of the above rules can be re-stated focusing on the marked value. That is, instead of saying that “it is unmarked for...” we can say “it is marked for...”. Practice exercise 7: Restate each of the above rules, focusing on the marked value.  The difference between these rules and the more familiar rules is that, there are no environments here. In other words, the observation that obstruents [-son] are voiceless is not limited to specific environments.  Phonological rules that surface in specific environments reflect phonological processes.  These phonological processes typically lead to the surfacing of marked forms. ◦ E.g vowel nasalisation results in a nasal vowel on the surface, which is marked. Practice exercise 8: Identify other phonological processes and the marked feature values they produce on the surface. Dr. F. Hudu. LNGS 401 Lecture Notes (Lecture 3: November 19, 2024) page 5 of 5 Dr. Fusheini Hudu LNGS 401 Lecture 4 November 26, 2024 Phonological processes: Assimilation  Last year, in LNGS 301, we noted that phonological rules are expressions of phonological processes.  In other words, every phonological process produces at least one phonological rule, and every phonological rule (except default rules) reflects a phonological process. You can't have one without the other  Many phonological processes also produce variants/allophones of a phoneme.  As we noted last year, the allophones are typically not noticed by the native speaker, even though they are produced. We perceive phonetically different sounds as one, we even sometimes perceive things that are not there.  Examples are illustrated below, showing some (not all) of the different realisations of English [t]. Different realisations of English /t/: Plain [t]: stem [stɛm] Aspirated [t]: ten [tʰɛn] retroflexed [t]: strip [sʈɻip] flapped [t]: atom [aɾəm] (in North American dialects) nasal flapped [t]: panty [paɾɪ]̃ (in North American dialects) glottalised [t]: hit [hitˀ] (in some British and North American dialects) glottal stop: bottle [bɒʔl] (in some North American dialects) zero: pants [pans]  There is at least, one rule to account for each of the above, but our focus today is on one category of phonological rules: assimilation. ASSIMILATION  A process by which a sound becomes more like another neighbouring sound. The assimilating sound takes on a new feature originally not part of the sound.  Thus, when sound A undergoes assimilation, it adopts a new phonological feature from a neighbouring sound B without sound B losing that feature.  Which features are involved in assimilation? Which features can a sound assimilate to? Virtually every feature we come across from our description of sounds.  For instance, from the 3-term label, we get three categories of features in our description of each of the following: /p/, /t/ /g/ etc. (place of assimilation features, manner of assimilation features, voicing features (or more generally, laryngeal features).  Any sound can assimilate to any feature belonging to any of these feature categories.1 1 Consult Chapter 4 of Hayes (2009) for details on features. The feature chart in that chapter is what we will be using in this course. Dr. F. Hudu. LNGS 401 Lecture notes Page 1 of 5 November 26, 2024   We will illustrate assimilation involving features under each of these categories using English data. Assimilation to place of assimilation features 1. Assimilation of alveolar consonants: General rule: When an alveolar consonant precedes a non-alveolar segment, a place of articulation assimilation is likely to take place. A. Dentalisation: An alveolar consonant [ t̪ , d̪, n̪, l̪ ] becomes dental before a dental consonant like /θ, ð/ /widθ/→ [wid̪θ]; /hɛlði/→ [hɛl̪ ði]; /iːt ðis/→ [iːt̪ ðis]; /ðat θɪŋ/ → [ðat̪ θɪŋ] /ɡɛt/ → [ɡɛt̪ ðoʊz]; / tɛnθ/ → [tɛn̪θ]; B. Labial articulation: In many native dialects of English, an alveolar becomes labial before a labial in casual speech. /miːt paɪ/ → [miːp paɪ]; /laɪt blu/ → [laɪp blu]; /waɪt peɪpə/ → [waɪp peɪpə] C. Velar articulation: Before velar consonants, alveolars become velar in casual speech /kʷaɪt ɡud/→ [kʷaɪk ɡud]; /braɪt kʌlə/→ [braɪk kʌlə]; /ðat ki/→ [ðak ki] D. Palatalisation: Before [ʃ] and [j], [s] and [z] become [ʃ] and [ʒ] /hɔːs ʃu/ → [hɔːʃ ʃu] /miːs ju/ → [miːʃ ju]; /ʤɔnz ʃɔts/ → [ʤɔnʒ ʃɔts] Velar and labial consonants (k, g, p, b) also become palatalised before [i]. /kiːn/ → [kʲiːn] /ɡiːv/ → [ɡʲiːv] /piːl/ → [pʲiːl] /biːt/ → [bʲiːt] E. Labialisation: Before round vowels consonants of different places of articulation, are rounded /kuːl/ → [kʷuːl], /bum/ → [bʷuːm] /tuː/ → [tʷuː] F. Nasal place assimilation/homorganic nasal assimilation: Nasals assume the same place of articulation as following consonants. ʌn-dizɛvɪn in-desɛnt im-pɔsɪbəl ʌm-peɪd iŋ-ɡɹatitud ʌŋ-ɡratefʊl iŋ-kuɹəbl  All these are examples of nasals in prefixes assimilating to the place of the following consonant. But we also see nasal place assimilation when the nasals occur in non-affixes. undə, intervʲu, umbilikəl, umbrelə, uŋkɛmpt, iŋkɹiz  Note that in some languages, nasal place assimilation or homorganic nasal assimilation takes place when the nasal comes after another consonant, unlike in English, where the nasal comes before the consonant. Dr. F. Hudu. LNGS 401 Lecture notes Page 2 of 5 November 26, 2024  MOA Assimilation Stop articulation: Sometimes a fricative is produced as a stop when following/preceding a stop, especially in casual speech. /giːv miː/ → [giːm miː]; /gɛt ðɛm/ → [gɛt dɛm]; /riːd ðiz/ → [riːd diːz] Laryngeal Assimilation. Assimilation of plural and third person suffixes.  Certain English suffixes assimilate to the voicing specification of the final sound of the word to which they are added. Two of them are the plural marker and the third person singular suffix.  Below, the data on the left column show cases where these two suffixes surface as voiced [-z]; those in the middle column show cases where these suffixes surface voiceless [-s].  The data on the right column show that sometimes these morphemes surface as [-əz], but they do not illustrate voicing assimilation.  For the purpose of analysing the various allomorphs of these morphemes, they are important. But for the purpose of analysing voicing assimilation, you can ignore them and focus on those in the first two columns. beɪð-z maʊθ-s dʒʌdʒ-əz bɔɪ-z goʊt-s mætʃ-əz skɪl-z stik-s bʊʃ- əz dɔg-z buk-s noʊs-əz hand-z map-s briz-əz hoʊm-z hoʊp-s bɛntʃ-əz dʒɔb-z seif-s loʊv-z kʌf-s weɪv-z mɔθ-s sɪŋ-z brif-s Practice exercises on assimilation 1. Describe the category of sounds that precede the various allomorphs of the plural and third person singular suffixes ([s] and [z]) shown above. Based on the sounds preceding each allomorph, which of these sounds is the main phoneme? Which of them is the allophone2. 2. Using the data on the various patterns of assimilation in English presented in the preceding sections, provide data on similar patterns of assimilation in your mother tongue or any language you can speak fluently. More on the concept of assimilation  Let us now move on to look at some key issues central to assimilation. First, there is the fundamental question regarding how assimilation takes place. 2 This exercise is meant to get you to revise your study of phonemes and allophones in Level 300. Get help from your TAs if you cannot answer this question Dr. F. Hudu. LNGS 401 Lecture notes Page 3 of 5 November 26, 2024   How does assimilation take place? The idea is that, it is a case of co-articulation between two adjacent segments. Remember that in articulating a segment, there are three fundamental phonetic components: the place of articulation, manner of articulation and voicing.  The idea is that, when two adjacent segments are different with respect to any of these phonetic components, articulating each segment using its unique place, manner of voicing specification is a tedious task. To make it easier, the place, manner or voicing specification of one sound is sacrificed, so that that sound is articulated using the specification of the other.  For instance, when the underlying representation of a nasal is [n], and the following sound is velar [k] (as in the word in-kəɹɛkt), it is difficult to articulate the nasal at the alveolar ridge and immediately reposition the tongue to make a velar articulation for the next sound. What most languages in the world do is to sacrifice the unique place of articulation of the nasal and articulate it alongside the following [k].  In the process, the two are co-articulated. In other words, the two sounds are produced with one articulatory gesture, instead of two different articulatory gestures: iŋ-kəɹɛkt. That is why assimilation is also known as (phonetic) co-articulation.3  There are four fundamental issues in assimilation. Addressing each of them is needed for a complete analysis of any assimilatory process. 1. The feature involved 2. Trigger and target 3. direction of assimilation 4. domain of assimilation 1. Features  We have already discussed features in assimilation. The only bit of information to be added is that, sometimes assimilation results in completely identical features between the two sounds involved, such that one sound becomes the exact copy of the other.  When that happens, the assimilation is said to be complete assimilation (e.g. /giːv miː/ → [giːm miː], /ʌn-mask/→ [ʌm-mask]).  In other instances, the two sounds involved do not become completely identical due to assimilation. Even though, one features takes the feature or features of another sound, they remain different sounds. Such cases are called partial assimilation (e.g. /un-peɪd/ → [um-peɪd]). 1. Triggers and targets  In every process of assimilation there is a sound which plays the role of trigger, and another which is the target of assimilation.  The segment that causes the change in the feature specification of another segment is the trigger of assimilation.  The segment that is affected or changes is the target of assimilation.  Thus, if within a sound sequence in a word, sound A and sound B are co-articulated in 3 You will see why this description is important when we come to look at harmony, which is also a form of assimilation. Dr. F. Hudu. LNGS 401 Lecture notes Page 4 of 5 November 26, 2024  such a way that sound A maintains is original features whereas sound B changes its features to look like sound A. Sound A will be described as the trigger, and sound B the target. Practice exercise:  Go back and look through the data we've already seen and determine the trigger and target in each pattern of assimilation.  Also study the data below. First determine the pattern of assimilation in each word, then identify the trigger and target in each pattern of assimilation. Note that it is not the case that every word has an assimilatory process. Some of the words are meant to guide you to understand the assimilatory patterns found in others. twelve [twɛlv] twelfth [twɛlf-θ] eight [eɪt] eighth [eɪt̪ -θ] ten [tɛn] tenth [tɛn̪-θ] 2. Directionality:  This relates to the location of the target relative to the trigger.  When the target is to the left of the trigger, the sound becomes more like the one that follows it, we get progressive. It is also known as anticipatory assimilation.  When the target is to the right of the trigger, the sound becomes more like the one that precedes it, we get regressive assimilation. 3. Domain of assimilation:  This relates to the unit within which assimilation is restricted. In some instances, assimilation can only take place when the trigger and target are part of the same syllable (e.g. nasalisation of English vowels).  In such cases, the syllable is the domain of assimilation. Beyond the boundary of the syllable, assimilation will not take place. In other words, when the potential trigger is in one syllable and the potential target sound is in a different syllable, no assimilation will take place.  In other cases, the word is the domain. Still in others, it is the phrase. Practice exercise: Identify the domain of assimilation is each of the various patterns of assimilation shown in this handout. Also do the same exercise on the assimilation data that you come up on the languages you speak. Dr. F. Hudu. LNGS 401 Lecture notes Page 5 of 5 November 26, 2024 Dr. Fusheini Hudu LNGS 401 Lecture 5 December 3, 2024 Phonological processes: Harmony  Harmony is a kind of assimilation, and all the issues central to assimilation are also central to harmony.  There are issues in harmony that do not apply in assimilation.  We will also practice analysing vowel harmony data on three languages: Kera, Dagbani and Akan.  Harmony is a process by which a sound becomes more like another sound that may not be adjacent, or a process by which non-adjacent sounds within a domain share one feature value.  As noted in the definition, the trigger and target sounds do not have to be adjacent, unlike assimilation.  The feature involved in harmony is called the harmonic feature.  Like assimilation, harmony affects both vowels and consonants, though vowel harmony is more common.  Which features can be harmonic features? Many features. ◦ All vowel features (height, backness, rounding, ATR). ◦ Some consonant features such as voicing and nasality. More on the concept of harmony  Because segments are not adjacent (next to each other), it is not phonetic co-articulation. Thus, another name for harmony is long distance assimilation. It is distinguished from assimilation, which is also known as local assimilation.  Harmony is more of phonology than assimilation, which is more phonetic. Thus, there are patterns of assimilation in every language, but not every language has harmony.  In other words, the mechanisms involved in assimilation are dictated by phonetic considerations while harmony is a choice a language makes to include in its phonology.  The phonetic mechanisms involved in sound articulation does not require that segments that are not adjacent share one feature.  Let us look at the central issues in harmony, most of which are also central to assimilation. There are four fundamental issues in harmony. 1. The harmonic feature 2. Trigger and target of harmony 3. Direction of harmony 4. Domain of harmony Dr. F. Hudu. LNGS 401 Lecture notes Page 1 of 4 December 3, 2024 1. The harmonic feature  We have already talked about the harmonic feature in the preceding discussion.  When vowel harmony takes place, it may involve one vowel sharing its height, backness, rounding or ATR feature with another vowel.  Similarly, consonant harmony may involve a nasal consonant sharing its [+nasal] feature with a [-nasal] consonant, a [+voice] consonant sharing its [+voice] feature with a [-voice] consonant etc. 2. Triggers and targets  Whatever was said about triggers and targets in assimilation also applies to harmony. However, there are some issues specific to harmony.  There is usually more than one target, unlike assimilation, where there is only one target.  The trigger and target do not have to be adjacent segments (as already noted).  There are other issues related to triggers and targets that apply in harmony systems, but not in assimilation. 1.1. Root-controlled harmony system  In many languages, trigger-target relation is conditioned by the morphology of the language. The rigger is always a root segment while the target may be another root segment or an affix segment.  In other words, a segment in the affix cannot serve as the trigger.  Such harmony systems are known as root-controlled harmony systems.  In a root-controlled harmony system, it does not matter what the value of the harmonic feature is. Whatever the value of the harmonic feature in the root segment is, determines the harmonic feature value of the entire domain  For instance, in root-controlled [ATR] vowel harmony, when the root vowel is [-ATR], all vowels in the domain become [-ATR]; when the root vowel is [+ATR], all vowels in the domain become. 1.2 Dominant-recessive harmony system  In some languages, one value of the harmonic feature is dominant, the other is recessive.  Within the harmonic domain, segments with the dominant feature value will serve as the triggers while segments with the recessive feature value are the targets.  Segments with the dominant feature value (the triggers) cause those with the recessive feature value (the targets) to change.  In such cases, it does not matter whether the triggers are located in the root or affix, and whether the targets are in the root or affix. Dr. F. Hudu. LNGS 401 Lecture notes Page 2 of 4 December 3, 2024   Such harmony systems are known as dominant-recessive harmony systems.  For instance, in a dominant-recessive system where [-ATR] is the dominant value of the [ATR] harmonic feature, a vowel with a [-ATR] feature value will cause another vowel with a [+ATR] value within the harmonic domain to become [-ATR]. In this process, there are four possibilities: o The UR [-ATR] vowel (trigger) is in the root; the UR [+ATR] vowel (target) is in the affix. o The UR [-ATR] vowel (trigger) is in the affix; the UR [+ATR] vowel (target) is in the root. o Both vowels are root vowels o Both vowels are affix vowels. 2. Directionality:  Whatever was noted regarding directionality in assimilation also applies to harmony.  The only difference is that, bidirectionality is more likely/prevalent under harmony than assimilation.  As already noted, in many languages directionality of harmony depends on morphology. In root-controlled harmony, directionality becomes irrelevant if the language has only suffixes or only prefixes.  If a language is root-controlled, and has both prefixes and suffixes, yet only prefix vowels become targets, then directionality becomes important. Similarly, in such a language, if it is only suffix vowels that get affected, directionality becomes important.  In dominant-recessive harmony system too, directionality is irrelevant if the language has only prefixes or only suffixes. In a dominant-recessive system with both prefixes and suffixes, yet only prefix vowels become targets, then directionality becomes important. Similarly, in such a language, if it is only suffix vowels that get affected, directionality becomes important. 3. Domain of harmony: The issues regarding domain of assimilation also apply (in fact more strongly) to harmony.  An example of consonant harmony is shown below, from Kera, a Chadic language.  In this language, stops and affricates show agreement for voicing within the word, both root-internally and between roots and affixes, regardless of intervening vowels and sonorant consonants.  The data show that the nominal prefix is voiced when the base begins with a voiced obstruent, as shown in (a)  The nominal prefix is voiceless when the base begins with a voiceless obstruent, or a sonorant, as shown in (b).  The data in (c) shows that the voicing affects suffixes too, and can cross sonorant consonants to do so. Dr. F. Hudu. LNGS 401 Lecture notes Page 3 of 4 December 3, 2024 Kera laryngeal harmony: the nominal prefix /k-/ Dr. F. Hudu. LNGS 401 Lecture notes Page 4 of 4 December 3, 2024 Dr. Fusheini Hudu LNGS 401 Lecture 6 December 10, 2024 The nature of phonological rules  In the past few weeks, we have looked at phonological processes, focusing on important issues in assimilation and harmony. In today’s lecture, we will focus on writing rules for phonological processes.  There are two theories we will focus on in this course that use rules to account for phonological patterns: a. Linear rule-based phonology b. Autosegmental phonology.  Today’s class will focus on linear phonology. Phonological rules in Rule-based linear phonology.  Three weeks ago, we noted the general formula for rules in phonology as follows: /UR/ → [SR] / environment:  We noted that the rules can either be stated with segments or features, as below: 1. With segments: /V/ → [Ṽ] / ____[N] ] ]σ. 2. With features: [+syl] → [+nasal] / __[+nasal] ]σ.  Rules stated this way are called linear rules. The rules are applied to a string of segments in a linear order, (everything is stated in a single line from beginning to end) and can only affect segments once.  Note that it is more informative to state rules using features than using segments. a Rules in features will show that not everything about the sound has changed, rules in segments sometimes suggest that everything has changed e.g. /z/ → [s] / [-voice]__ [-son, +cont, +voice] → [-voice] / [-voice] __ b Rules in features also show the similarity between the surface form and the environment e.g. [+syl] → [+nasal] / __[+nasal] ]σ.  In assimilation, the triggering segments and target are next to each other, in harmony, they are not, so there has to be a way of modifying the rule to reflect that.  Let’s come back to the data on Akan and Dagbani harmony and see how to state the rules.  In the first set of Dagbani data, we are dealing with a root-controlled [ATR] harmony, with only suffixes and clitics. In the rule, we don’t have to say that it is root-controlled, we only capture the directionality.  It always helps to begin by writing the rule as a statement. In that case the rule will go as follows: A: The [ATR] specification of the root vowel determines that of the following vowels. Dr. F. Hudu. LNGS 401 Lecture 6 notes Page 1 of 5 December 10, 2024  This is actually two rules stated as one. It is the two rules in (b) and (c) below that are put together to get Rule (a) above B: Affix and clitic vowels surface as [+ATR] when the root vowel is [+ATR] C. Affix and clitic vowels surface as [-ATR] when the root vowel is [-ATR]  In stating the rules as features, we will also begin with two rules and merge them into one. Rule B above is the same as D below in features. Rule C is the same as Rule E in features D. [+syl] → [+ATR] / [+ATR] C0 ___ E. [+syl] → [-ATR] / [-ATR] C0 ___  The C0 between the trigger and the environment of the target (indicated by the dashed line) is there to capture the fact that unlike assimilation, in harmony, the trigger and target are not adjacent. The subscript indicates that there is no fixed number of consonants between the trigger and target. There can be one, two or more consonants. There may also be no consonant between them.  Notice that if we were dealing with consonant harmony, in which the trigger and targets were consonants, the right symbol to place between the trigger and target would have been V0.  Finally, just as it is possible to combine rules B and C into one to get A, it is possible to express rules D and E as one in features. To do so, we make use of alpha notation, in which “+” and “-” specifications of the feature are expressed using the alpha symbol and matched in different parts of the rule. This is shown in Rule F below. F. [+syllabic] → [αATR] / [αATR] C0 ___  What the rule is saying is that, when a vowel (a [+syllabic] segment) follows a [+ATR] segment (with the possibility of one or more consonants between them), the vowel will surface as [+ATR]; if it follows a [-ATR] segment, the vowel will surface as [-ATR].  Now let’s turn to the rules for the Twi harmony pattern.  There is only one point of difference between the Twi and the Dagbani pattern: [ATR] harmony in Twi is bidirectional, unlike Dagbani. This means that the trigger has some targets to its left and other targets to its right.  This means that unlike Dagbani, in Twi, we will need two rules: one for each direction of spread. The left-to-right direction of spread will be exactly the same as the Dagbani rule stated in F above, the right-to-left direction will is shown in G below. G. [+syllabic] → [αATR] / ___ C0 [αATR] Other types of rules  So far, we have been focused on rules of assimilation and harmony, where one segment is responsible for the change in the feature specifications of another segment that is either contiguous or at a long distance.  There are other (non-assimilatory) phonological processes which, when formalised as rules, require slightly different formalisations. The following are some of them. Dr. F. Hudu. LNGS 401 Lecture 6 notes Page 2 of 5 December 10, 2024 a. Insertion/epenthesis: in insertion, we go from nothing in the UR to something in the surface representation. The symbol for nothing is “Ø”. Thus rules on insertion take the form Ø → [segment] / environment (where the segment can be a vowel or a consonant). b. Deletion: Deletion is the opposite of insertion. We begin with something in the UR and end up with nothing in the SR. Thus it takes the form /segment/ →Ø / environment. c. Rules triggered at unit boundaries. Sometimes certain rules are triggered at certain boundaries such as the syllable- or word-initial or final positions. Sometimes it may even be at the boundary of a phrase or clause. Some phonologists use the symbol “#” to mark word boundary. Others use “[” as a general symbol to mark the beginning of any unit and “]” to mark the end of a unit. In order to be specific about the relevant unit, the symbol or name of that unit is written as a subscript. Thus “word[” means the beginning of a word, “]word” means the end of a word. Refer to the rule on vowel nasalisation for an example of a syllable boundary. Deriving surface forms with rules  As already noted, the phonological analyses we are doing are part of an approach known as known as derivational phonology. What that means is that, we apply rules to underlying forms to derive the surface forms.  So we can apply vowel nasalisation and consonant devoicing rules to change /siŋz/ → [sĩŋz] and /blækz/ → [blæks]. When the rules are applied to /bɔɪz/, we get [bɔɪz] on the surface, which means the UR remains unaffected. This is illustrated below. Vowel nasalisation: [+syl] → [+nas] /__[+nas](C) ]σ Obstruent devoicing: [-son, +cont, +cor, +voice] → [-voice]/ [-voice] ___ UR /siŋ-z/ /blæk-z/ /bɔɪ-z/ Vowel nasalisation sĩŋz _____ _____ Obstruent devoicing _____ blæks _____ Surface forms [sĩŋz] [blæks] [bɔɪz]  Sometimes morphological rules may interact with phonological rules. In sĩŋ, the root vowel is nasalised. However, when the “-ing” suffix is added to get /siŋ-iŋ/, the /i/ in /siŋ/ will no more be in the same syllable as the /ŋ/, as the word is re-syllabified to get si.ŋiŋ. This means that in the suffixed form, the root vowel may not be nasalised.  However, whether we get [sĩŋiŋ] or [siŋĩŋ] depends on which of the processes takes place before the other. ◦ If nasalisation takes place before the suffixation of the progressive marker, we will get [sĩŋiŋ], as shown below. UR /siŋ/ Vowel nasalisation sĩŋ Progressive suffixation sĩŋ-iŋ Surface form [sĩŋ-iŋ] ◦ On the other hand, if suffixation precedes nasalisation, we get [siŋĩŋ]. Dr. F. Hudu. LNGS 401 Lecture 6 notes Page 3 of 5 December 10, 2024  UR /siŋ/ Progressive suffixation si.ŋ-iŋ Vowel nasalisation si.ŋ-ĩŋ Surface form [si.ŋ-ĩŋ]  The actual attested surface form is [siŋ-ĩŋ]. The reason we get [siŋĩŋ] is because the progressive marker destroys the environment needed for the vowel to get nasalised. It separates the vowel and nasal in [siŋ] into different syllables.  Sometimes both rules are phonological rules. In French, vowel nasalisation and nasal deletion are two phonological processes that interact. Nasal deletion deletes a nasal sound when it comes after another nasal sound. ◦ vowel nasalisation: [+syl] → [+nas]/__[+nas](C)]σ (/bɔn/ → [bɔn ̃ ]) ◦ Nasal deletion: [+nas] → Ø / [+nasal] ___ (/bɔn ̃ / → [bɔ]̃  When these two rules are applied to the underlying form /bɔn/, we get the surface form [bɔ]̃. The order of application is as follows: UR /bɔn/ Vowel nasalisation bɔn ̃ Nasal deletion bɔ̃ Surface form [bɔ]̃  The fact that we arrived at the right surface form means the ordering was right. The opposite ordering of the rules would have produced the wrong surface form below. UR /bɔn/ Nasal deletion _____ Vowel nasalisation bɔn ̃ Surface form [bɔñ ]  Thus in deriving surface forms from URs, the order in the application of multiple rules that may be needed to derive the surface forms matters.  There are four such interactions, but we will look at only two. An important point to bear in mind is that, you must know the UR and surface form to determine how the two or more rules were applied. FEEDING  The application of one rule creates the environment for another rule to apply. For instance, assuming Rule A needs environment X to apply, which does not exist. Thus, the rule cannot apply. However, before Rule A applies, Rule B applies and creates environment X which enables Rule A to apply and have an effect. We would say that Rule B feeds Rule A.  Vowel nasalisation feeds nasal deletion in French. When vowel nasalisation does not apply, nasal deletion does not get the chance to apply because the environment needed does not exist. It is the application of vowel nasalisation that create the environment Dr. F. Hudu. LNGS 401 Lecture 6 notes Page 4 of 5 December 10, 2024  nasal deletion needs to apply. BLEEDING:  The application of one rule destroys an existing environment in which another rule would have applied. For that reason, the second rule will no more apply.  For instance, assuming RULE A needs to apply in environment X. Environment X is present in the UR. However, before Rule A applies, Rule B applies, removing that environment. Rule B bleeds Rule A.  We will illustrate this with two rules in English phonology: obstruent devoicing and [ə]- insertion, defined below.  Last week, we looked at obstruent devoicing, which results in the plural suffix /-z/ becoming [-s] after voiceless obstruents. The rule for it is stated in (1) below. 1. RULE A: Obstruent devoicing: [-son] → [-voice]/ [-voice] ___  Below is data on [ə]-insertion, which also takes place in the affixation of the plural suffix. dʒʌdʒ-əz judges mætʃ-əz matches bʊʃ- əz bushes noʊs-əz noses briz-əz breezes bɛntʃ-əz benches  You will notice from the data that when a noun ends in a sibilant consonant (s, z, ʃ, ʒ, tʃ, dʒ), the plural suffix, which should have been /-z/, is realised as [-əz]. In other words, the schwa is inserted. The distinctive feature for sibilant consonants is [+strident]. Thus, the rule in (2) below captures the insertion. 2. RULE B: [ə]-insertion: Ø → [ə]/ [+strident] ___ [+strident] (bʊʃ → bʊʃəz)  Now let us see how Rule B bleeds Rule A. First, we start by ordering A before B. UR /bʊʃz/ /blækz/ /skilz/ Obstruent devoicing bʊʃs blæks _____ [ə]-insertion bʊʃəs _____ _____ Surface forms [bʊʃəs] [blæks] [skilz]  Since [bʊʃəs] is not the right surface form, it means that the ordering is not right. The reverse ordering will give us the right surface form, as shown below. UR /bʊʃz/ /blækz/ /skilz/ [ə]-insertion bʊʃəz _____ _____ Obstruent devoicing ____ blæks _____ Surface forms [bʊʃəz] [blæks] [skilz]  While this ordering produces the right surface forms, it comes at a cost: [ə]-insertion denies obstruent devoicing the chance to apply, unlike the preceding ordering. Thus [ə]- insertion bleeds obstruent devoicing. Dr. F. Hudu. LNGS 401 Lecture 6 notes Page 5 of 5 December 10, 2024 LNGS 401 Lecture notes 7 Jan 09 2025 Dr. Fusheini Hudu Tone and the problem with linear derivational theory  In the past few weeks, we have been looking at how linear phonological theory works in accounting for phonological processes of different kinds, including assimilation, harmony, deletion, etc.  An important aspect of phonology that has not been discussed is prosody, especially tone and tonal processes.  The attempt to use this theory to analyse tone and tonal processes revealed its weaknesses and birthed a new theory known as Autosegmental phonology.  Notice that in linear theory, distinctive features are units that are put together to form a sound, and the difference between one sound and another is attributed to the different composition of distinctive features the segments are composed of.  This raises the question of how tonal features could be captured using distinctive features. If a sound bears a high tone, how can it be distinguished from other sounds bearing low, mid, rising and falling tones?  One approach to the analysis of tones was to treat tone as a distinctive feature. Thus, in addition to the features [+syllabic, +back, +high, +round etc.] that come together the form the vowel [u], we add one or two distinctive features to show the specific tone that the vowel bears.  As part of the solution, two tone features were proposed to take care of both level and contour tones. These are [±High tone] and [±contour tone].  The following are the specifications of various tones for these two tonal features. o A high tone (H) = [+High tone, -contour] o A low tone (L) = [-High tone, -contour] o A rising tone (R) = [-High tone, +contour] o A falling tone (F) = [+High tone, +contour] Problems:  There are a number of immediate questions that this proposal cannot answer: o How about mid tones? o How about different levels of high tones (e.g. super high tones)? o How about more complex tones such as fall-rise and rise fall? Tonal contours as products of tonal assimilation  A much bigger challenge to the theory was posed by contour tones that result from tonal processes, especially tonal assimilations.  There are four such crosslinguistically common tonal rules producing surface contour tones which the theory must be able to account for. (a). H --> F/ __{L, R} (b). L --> R/ __ {H, F} (c). H --> R / {L, F}___ (d). L --> F/{H,R}___  With the proposed approach within linear phonology, Rule (a) and Rule (b) can be accounted for as follows: o Rule (a) = [+Hi, -contour] --> [+contour]/ __ [-H tone]. Dr. F. Hudu. LNGS 401 Lecture 7 notes Page 1 of 4 January 09, 2025  o Rule (b) = [-H] --> [+contour] / __ [+H].  The problem is that, Rule (c) and Rule (d) cannot be accounted for. This is because in Rule (c), low and falling tones have no distinctive feature in common.  Similarly, the triggering tones in Rule (d) have no distinctive feature in common to trigger the same tonal process.  In fact, given the nature of linear derivational phonology, rules (c) and (d) are actually predicted not to exist, since it is not possible for two different units which lack any feature in common to trigger one phonological process.  In other words, the two tones that trigger the same phonological process in each of rules (c) and (d) do not form natural classes in their respective rules, and thus are not supposed to trigger the same tonal process.  Given that these rules are attested in many languages, we can only fault the theory for not being able to analyse naturally occurring phonological patterns.  It is these and other weaknesses of linear theory that gave rise to the quest for an alternative theory, culminating in the birth of Autosegmental Phonology. Autosegmental phonology, the main idea:  In Autosegmental Phonology, tone is assumed to be independent of segments.  Tone is associated with segments, as tone must be borne by segments. A toneme cannot be seen without the segment that bears it.  Thus, tone is like a baby that cannot be seen outside alone. It is always carried by an adult. However, the fact that the baby always needs to be carried and cannot be seen without another adult does not mean that the baby is not a separate human being.  The tone and the segment are linked via association lines. This means there are several possibilities in terms of association: o (a) one-to-one, (b) one-to-many, (c) many-to-one, (d) floating tone, (e) toneless TBU. These are illustrated below. a. T T b. T c. TT d. T e. VC0V VC0V V V  In instances of one-to-one and many-to-one associations, there can be two or more different level tones. (a) H L (b) LH (c) H L (d) LH papa (pápà) papa (pàpá) pa (pâ) pa (pǎ)  A contour tone is always the result of a many-to-one association of tone to TBUs, as examples (c) and (d) show. The contour tone is not special, it is merely a sequence of two different level tones associated with one TBU. Autosegmental phonology and contour tones.  When tonal rules take place, the resulting contour tones are due to re-association between tones and TBUs. Let's the take the two rules that linear theory could not analyse: Dr. F. Hudu. LNGS 401 Lecture 7 notes Page 2 of 4 January 09, 2025  H → R/{L, F}__ L → F/{H,R}___  Each rule states two environments under which contours are formed. In the above rule, high tone (H) becomes rising (R) after low (L) and also after a falling (F).  An important point is that, both L and F end with L. Thus, any tone that comes after L or F comes after a low tone. The two tones are the same as far as their relationship with following tones is concerned.  What that means is that, the high initial pitch in a falling tone has no effect on the eventual outcome. The two UR forms are shown below. L H (H tone following a low tone) H L H (H tone following a falling tone). V V V V  To express the lack of effect of the H tone in the triggering environment, we put it in parenthesis.  The assimilatory process is expressed as a dotted association line linking the TBU to the new tone it has acquired. In the surface form, the dotted line is changed to a solid line. These are shown below. UR Tone assimilation Surface form /(H) L H/ → (H) L H → (H) L H V V VV VV  Notice that in this example, the second vowel bears the tone that is the target of the tonal assimilation, the first bears the tone that triggers the assimilatory process.  Now let us look at the other rule that could not be analysed in linear segmental phonology: L → F/{H, R}__  This also gets a straightforward analysis because H and R both end in H. Thus, whether the trigger is H or R, the target is preceded by H. UR Tone assimilation Surface Form (L) H L → (L) H L → (L) H L V V V V V V Exercise: Do an Autosegmetnal Phonology analysis of the remaining two rules we studied last week. Summary:  It is evident that Autosegmental Phonology has proven to be a better theory at analysing tone, as it is able to provide an analysis of contour tones, something linear segmental phonology failed to analyse.  There are two more tonal properties we will look at in demonstrating the relative strength of Autosegmental Phonology over Linear theory. These are tonal stability/tone preservation and floating tone. Dr. F. Hudu. LNGS 401 Lecture 7 notes Page 3 of 4 January 09, 2025 Tonal stability/tone preservation.  In tone languages, the deletion of a TBU does not necessarily result in the deletion of the tone it bears. When TBUs are deleted, the tone they bear usually remain, producing contour tones.  This is illustrated in the data on Yekhee, illustrated below. a. /ídzé élà → [ídzélà] ‘three axes’ b. /èké élà → [èkélà] ‘three rams’ c. /òké òkpá → [òkôkpá] ‘one ram’ d. /ówà ówà/ → [ówǒwà] ‘every house’  If the tone feature is part of the segment as assumed in linear derivational theory, it will not be possible for the tone to remain when the segment, along with all the features it bears, deletes.  In Autosegmental Phonology, this is not a problem, as the tones are not assumed to be part of the segments. The analysis too is a straightforward one. All what happens is that, the preserved tone re-links to another TBU.  Example with /òké òkpá → [òkôkpá] ‘one ram’ and /èké élà → [èkélà] ‘three rams’ UR V deletion and Tone Surface floating tone re-linking form L HL H → LH L H → LH L H → LHL H oke okpa ok okpa ok okpa okokpa  In /èké élà → [èkélà], the deletion of the TBU results in a sequence of two H tones. This is simplified to one H tone by the application of convention known as the twin sister convention. The twin sister convention says that there can be no adjacent identical tones on one vowel. UR V deletion and Tone Twin sister Surface floating tone re-linking convention form L H HL → L HHL → L HH L → L HL L HL eke ela ek ela ek ela ek ela ek ela  What forces the unlinked (floating) tone that results from the deletion of the vowel to re-link to another TBU?  It is the Welformedness Condition, which says that all TBUs must (eventually) bear some tone, and all tones must be borne by a TBU. Floating tone.  The final tonal pattern that supports Autosegmetnal Phonology is floating tone. A floating tone is one that at some point is not associated with any TBU. We shall discuss this in our next lecture, using data on Dialect of Ewe. Dr. F. Hudu. LNGS 401 Lecture 7 notes Page 4 of 4 January 09, 2025 LNGS401: Phonological Theory. Dr. Fusheini Hudu FLOATING TONES IN AUTOSEGMENTAL PHONOLOGY This handout explains floating tone in Anlo Dialect of Ewe, as presented in Chapter 10 of Odden (2005). If you did not understand Odden's explanation, this should make it easier to understand. Floating tone in Anlo  Take note of the following tonal minimal triplets. (1). Anlo tonal minimal pairs. a. ètò “buffalo” b. ètó “mountain” c. ētō “mortar”  To understand floating tones in Anlo, we need to understand a number of tone rules in the dialect.  The data in (2) show some phrases with ètò (buffalo). (2) “buffalo phrases” in Anlo. a. ètò “buffalo” b. ètò mè “in a buffalo” c. ētō ɸēɸlē “buffalo-buying” d. ētō mēgbé “behind a buffalo” e. ētō ʤí “on a buffalo”  QUESTION: Why is it that sometimes “buffalo” has low tones, sometimes mid tones in the above phrases?  ANSWER: ētō is the underlying representation (UR) of buffalo.  The change from /ētō/ to [ètò] is due to two tone rules of Anlo. RULE 1: /M/ → [L] / ___# (A mid tone becomes low when it occurs at the end of a phrase). When applied to /ētō/, we get [ētò].  Note that [ētò], the product of RULE 1 is going to be the UR for RULE 2, shown below. RULE 2: /M/ → [L] / ___L: (A mid tone becomes low when it precedes a low tone). When applied to /ētò/, (the surface form of RULE 1) we get [ètò].  When the mid tone of ētō does not end a phrase or followed by a low tone, it does not change to a mid tone. Try and see if you can show examples from (2).  Practice exercises: We have seen how to derive the surface form [ètò] from the UR /ētō/ using the two rules. ◦ Which other word or phrase in (2) can you derive using the rules? ◦ Try to apply the rules to it and see if you'll get the surface form in (2). ◦ Provide an autosegmental representation of the two rules. Page 1 of 3 LNGS401: Phonological Theory. Dr. Fusheini Hudu  Now let's consider the data in (3) with the word for mountain. (3). “mountain phrases” in Anlo. a. ètó “mountain” b. ètő me̋gbé “behind a mountain” c. ètó ʤí “on a mountain”  From the data in (2) on “buffalo phrases” we know that megbe has a MH sequence (mēgbé).  We also know that from (1) that mountain is ètó.  Therefore, the phrase “behind a mountain” should actually be ètó mēgbé  QUESTIONS: ◦ How come megbe has a superhigh tone in ètő me̋gbé “behind a mountain”? ◦ How come “mountain” has a superhigh tone in ètő me̋gbé “behind a mountain”?  ANSWERS: ◦ There are two tone rules responsible for these changes. One is stated in (4). (4) superhigh tone rule in Anlo. M → SH / H__H: (A mid tone becomes a superhigh tone when it occurs between two high tones). ◦ When we apply this rule to ètó mēgbé, (which is what “behind a mountain should be), we get ètó me̋gbé. The mid tone has changed to a superhigh because it occurs between two high tones. ◦ There is a 2nd rule of tone assimilation. It takes ètó me̋gbé as the UR. It is stated in (5). (5). Anlo tone assimilation to superhigh tone. H → SH /__SH (A high tone becomes superhigh when it precedes a superhigh).  We've now got to a point where we can understand where floating tones come in.  Look at the data in (6) which has phrases with the word for mortar. (6) “Mortar phrases” in Anlo. a. ētō “mortar” b. ētō mê “in a mortar” c. ētō me̋gbé “behind a mortar” d. ētō ɸéɸlē “mortar-buying” e. ētō ʤí “on a mortar” THE PUZZLES.  Compare the data in (6) to those in (2), ◦ Why is it that the final mid tone for mortar does not become L in (6), as it does in (2)? ◦ Unlike (2), when ɸēɸlē follows ētō (as in 6d) the first M for ɸēɸlē becomes H. WHY? ◦ Unlike (2), when mè follows ētō (6b), the L tone of mè becomes falling tone. WHY? ◦ Unlike (2), when mēgbé follows ētō (as in 6c) the M in mēgbé becomes SH. WHY?  To solve these puzzles, we assume that ētō “mortar” does not end in a M, but a H that is not Page 2 of 3 LNGS401: Phonological Theory. Dr. Fusheini Hudu associated with any vowel.  In other words, ētō has a floating tone at the end of it.  In autosegmental phonology, this will be represented as in (7) below: (7). Autosegmental representation of ētō (mortar) with a floating tone. M M H | | eto  With this, we can now answer all the questions raised in the puzzles.  Let's take the questions one after the other. QUESTION: Why is it that the final mid tone for mortar does not become L in (6), as it does in (2)? ANSWER: Final lowering doesn't apply because the word doesn't end in a M, it ends in a H. Final lowering affects words that end in a M tone. QUESTION: Unlike (2) when ɸēɸlē follows ētō the first M of ɸēɸlē becomes H. WHY? ANSWER: In ētō ɸéɸlē (6d), the floating H of ētō associates with the following M of ɸēɸlē, changing it from M to H. QUESTION: Unlike (2), when mè follows ētō (6a), the L of mè becomes a falling tone. WHY? ANSWER: In ētō mê (6a), the floating H of ētō associates with the following L of mè, changing it from a low to a falling. (remember when you add H to L, you get F). QUESTION: Unlike (2), when mēgbé follows ētō the M in mēgbé becomes SH. WHY? ANSWER: In ētō me̋gbé (6c), the M becomes SH b/c it's between two H (see the rule in (4)).  Thus by taking note of the falling tone in the word ētō, we are able to account for all the puzzles.  Without reference to the floating tone, there is no way we could possiblly answer the puzzles.  This is a strength of Autosegmental Phonology. In linear segmental phonology, there is no way to account for floating tone.  Given that the tone is assumed to be part of the segment, there is no way to explain how a tone can exist but does not show up in some instances while the segment, of which it is part, shows up in all instances. PRACTICE EXERCISE: Now that the Anlo floating tone has been explained (hopefully in a way that makes it easy for you to understand), read the section in your handout on Ga floating tone carefully. It should be easier for you to understand it after reading this handout). Page 3 of 3 LING 401: Phonological Theory Harmony Practice Data Dr. Fusheini Hudu Study the data on vowel harmony in Dagbani and Akan and answer the questions that follow. 1. Vowel harmony in Dagbani Note that in a word with "-" between two morphemes, the first morpheme is the root, the second morpheme is the suffix. The vowel [ɨ] is a [-ATR] high central vowel. (1) Compare the set of words and phrases on the left to those on the right. a. bɨn-ɨ ‘a thing’ pin-i ‘a gift’ b. tʊʔɨ ‘join’ diʔ-i ‘a mirror’ c. dʊʔɨ-bʊ ‘cooking’ di-hi-bu ‘feeding’ d. kpabɨ ‘carry on the back’ kpib-i ‘lice’ e. bɔhɨ tɨ ‘ask us’ lihi ti ‘look at us’ i. What harmony pattern takes place in (1)? ii. How is the harmony achieved; what is the trigger and target of harmony? iii. State a rule that accounts for the harmony process, first in words, then using features. (2) Compare the set of words and phrases on the left to those on the right. Note that the vowel [ʌ] is a low vowel. It is [+ATR], which makes it different from the [-ATR] vowel [a]. a. mɔlɨ ‘announce’ mol-o ‘an announcement’ b. kɔɾsɨ ‘scratch’ koɾ-e ‘desire’ c. kɔlɨ ‘sweep’ kote ‘snail’ d. ʧɛɾ-ɡa ‘a ladle’ ʧeʧe ‘bicycle’ e. kal-tɨ ‘doors’ kʌl-o ‘a door’ f. pal-lɨ ‘a new one’ pʌl-ó ‘a new one (animate).’ g. tadab-tɨ ‘writing inks’ tʌdʌb-o ‘a writing ink’ h. tatab-tɨ ‘look-alike-pl.’ tʌtʌb-o ‘a look-alike’ i. What harmony pattern takes place in (2)? ii. How is the harmony achieved; what is the trigger and target of harmony? iii. State a rule that accounts for the harmony process, first in words, then using features. iv. Compare the harmony patterns in (1) and (2). How are they similar? How do they differ?  2. Akan Vowel harmony (Dolphyne 1988: 15-19). (3) Study the data below and answer the questions that follow. a. wʊbɛ tʊ ‘you will throw’ g. maatɔ ‘I have bought it’ b. ɔdɪ ‘he is called...’ h. ɔbɛkʊ ‘He will fight’ c. wubetu ‘you will dig it up’ i. meedzi ‘I have eaten it’ d. ɛwʊ(ɔ) ‘honey’ j. owu(o) ‘death’ e. ɛfɪ(ɛ) ‘vomit’ k. obehu ‘he will see’ f. odi ‘he eats’ l. efie ‘home’ i. How many harmony patterns take place in (3)? Name them. ii. How is the harmony achieved; what is the trigger and target of harmony? iii. State the rule(s) that account(s) for the harmony, first in words, then using features. (4) Study the data below and answer the questions that follow. Note that the data below is unique to Fante. a. nʊkɔn ‘her neck’ e. nɪkyɛw ‘her hat’ b. mɪrɪ-kɛ-tsɪw ‘I am going to pluck it’ f. mʊrʊkɔtʊw ‘I am going to throw it’ c. nukun ‘her husband’ g. nitsir ‘her head’ d. murukotu ‘I am going to dig it up’ h. mirikedzi ‘I am going to eat’ i. How many harmony patterns take place in (4)? Name them. ii. How is the harmony achieved; what is the trigger and target of harmony? iii. State the rule(s) that account(s) for the harmony, first in words, then using features.iv. Compare the harmony patterns in (3) and (4). How are they similar? How do they differ?