INTD0112 Introduction to Linguistics
Announcements INTD0112
Introduction to ¾ Any questions on homework 2? Linguistics
Lecture #6 March 6tht , 2007
Speech production and coarticulation Speech production and coarticulation
¾ So far, we described segments as if they ¾ There are two reasons for the existence of are articulated in isolation. Of course, this articulatory processes: is not the case in connected speech. ¾ More efficient articulation, as we noted Sounds are typically produced while more before for vowel nasalization in English. than one articulator is active. ¾ More distinct output, as in lengthening a consonant or a vowel to make it distinct: ¾ As a result of coarticulation, sounds may get to affect other sounds in speech. “It’s Fred.” “Did you say it’s red?” ¾ These are called articulatory processes. “No, I said, ‘Ffffred.’”
Articulatory processes Assimilation: Regressive
¾ There are several types of articulatory ¾ Assimilation is an articulatory process processes in human language. We whereby a sound is made “similar” to a discuss a few here. We’ll get back to this neighboring sound. issue again, though, when we talk about ¾ Vowel nasalization in English is an historical linguistics and language change. instance of regressive assimilation: can’t [khæ)nt]
1 Assimilation: Progressive Assimilation in voicing
¾ Assimilation may also take place in voicing features. In English, liquids and glides get ¾ Assimilation can also be progressive, as “devoiced” after voiceless stops. Devoicing is in Scots Gaelic: marked by a “ • ” underneath the segment: [ne):l] “cloud” please [pl•iz] proud [pr•awd]• [mu:] “about” ¾ Similarly, voiceless sounds may become voiced ) in the neighborhood of voiced sounds, e.g., Dutch af [Af] (=“over”) is pronounced with a [v] in the words afbellen (=cancel) and afdekken (=cover).
Assimilation in place of articulation Assimilation in place of articulation
¾ Nasal consonants typically assimilate to ¾ Now, let’s look at these German data: the place of articulation of the following Careful speech Informal speech sound. From English: laden [la:dən] Æ [la:dn] “to invite” possible Æ impossible [mp] loben [lo:bən] Æ [lo:bm] “to praise” tangible Æ intangible [nt] backen [bakən] Æ [bakŋ] “to bake” complete Æ incomplete [ŋk] ¾ Question: Is this a case of regressive or ¾ What’s going on here? progressive assimilation?
Dissimilation Deletion
¾ Deletion is a process which removes a ¾ Dissimilation is an articulatory process segment from certain phonetic contexts. whereby two sounds are made less From English: similar. From English: suppose [səphoèwz] Æ [spoèwz] fifths [fIfθs] Æ [fIfts] ¾ Deletion may also occur as an alternative to dissimilation for some speakers in words like fifth: fifths [fıfθs] Æ [fıfs]
¾ Epenthesis is a process that inserts a segment within an existing string of segments. From ¾ Metathesis is a process that changes the English: order of segments. Children learning something [s√mθıŋ] Æ [s√mpθıŋ] English will typically produce metathesis length [l ŋθ] Æ [l ŋkθ] E E forms, e.g., spaghetti is typically ¾ In Turkish, a sequence of two initial consonants pronounced as pesghatti [pəskERi]. is not allowed. As a result, a vowel is epenthesized to break the consonant cluster: “train,” which is borrowed from English, is pronounced as [tiren]
Vowel reduction Phonology
¾ While phonetics studies how speech sounds are articulated, what their physical properties are, ¾ In many languages, vowels in unstressed and how they are perceived, phonology studies syllables undergo reduction, typically the organization of speech sounds in a particular appearing instead as the weak vowel [ə]: language. h Canada [k æ)nədə] ¾ As it turns out, while two or more languages may Canadian [kh)ənejdiən] have the same sounds, no two languages organize their sound inventories in the same way.
[s] and [S] in Japanese vs. English [s] and [S] in Japanese vs. English
¾ In both English and Japanese we hear the ¾ By contrast, in Japanese, we do not find sounds [s] and [S]: pairs where [s] and [S] create a difference Japanese: [Simasu] “do” in meaning. English: [slæS] “slash” ¾ Instead, if we know that a Japanese word ¾ In English, however, the two sounds can ends in [-in], we then know that it can only distinguish meaning, e.g., begin with [S], and not with [s]. [ or] “shore” vs. [sor] “sore” S ¾ The occurrence of each sound is thus ¾ The occurrence of each sound is thus predictable in Japanese. unpredictable in English.
3 Subconscious phonological Phonology knowledge
¾ Native speakers of a particular language typically treat certain sounds as being the same, ¾ Phonology addresses these questions: even when they are phonetically different, e.g., ¾ Which sounds are predictable in a the [l] in lay and play particular language? the [t] in top and stop ¾ What is the phonetic context that allows us ¾ But other sounds are considered to be distinct: to predict the occurrence of these sounds? the [l] and the [r] in light and right ¾ We discuss this today. the [p] and the [b] in pan and ban
Phonemes vs. allophones Minimal pairs
¾ Phonologists explain the difference by invoking ¾ But how do we know if two sounds are two a distinction between phonemes and separate phonemes, or are simply phonetic variants of the same phoneme? allophones. ¾ Answer: Minimal pairs. ¾ A phoneme is a meaning- distinguishing sound, ¾ A minimal pair is a set of two words that have whereas an allophone is a phonetic variant of a the same sounds in all positions except one. If, particular phoneme. as a result of the difference of these two sounds, a difference in meaning occurs, then the two ¾ As we should expect, native speakers treat sounds are phonemes. If, by contrast, no different phonemes as distinct sounds, but treat difference in meaning occurs, then the two allophones as the same sound. sounds must be allophones of the same phoneme.
Phonemes or allophones? Phonemes or allophones?
¾ So, bearing this in mind, let’s go back and ¾ Based on the minimal pair shore and sore (remember spelling is irrelevant to phonology), look at the examples we discussed earlier. are the [s] and [S] phonemes or allophones in English? ¾ Based on the minimal pair light and right, ¾ How about these further minimal pairs: are the [l] and [r] phonemes or allophones seat and sit? in English? fool and full? ¾ Based on the minimal pair pan and ban, sip and zip? are the [p] and [b] phonemes or leaf and leave? allophones in English? ¾ More minimal pairs are given in Table 3.1, p. 60, in your textbook.
4 Phonemes or allophones? Phonemes or allophones?
¾ Now, let’s consider the following minimal pair: ¾ Let’s consider another example: h play [plej] vs. [plej] top: [t Ap] vs. [tAp] • ¾ Here’s the question: Are [l•] and [l] phonemes or ¾ Now, here’s the question: Are the two sounds allophones in English? [th] and [t] phonemes or allophones in ¾ Right. Since they are in complementary English? distribution, then they are allophones of the ¾ Since [th] and [t] are not contrastive in same phoneme, which we may represent here English, they are two allophones of the same as /l/. phoneme, which we might represent here as ¾ Notice that this same allophonic variation happens with [ ] as well as the glides [j] and [w]. /t/ (notice the slash, rather than the square ® bracket, notation).
Distribution: complementary vs. Phonemes or allophones? overlapping? ¾ How about nasal vowels in English? Are ¾ From all these examples, you should’ve noticed they phonemes or allophones? that different allophones occur in different environments, that is, where one of them occurs, ¾ First, find a minimal pair: the other doesn’t, and vice versa, which is not pin [phĩn] vs. *[phIn] the case with phonemes. pit [phIt] vs. *[phĩt] ¾ We say that allophones occur in complementary distribution, whereas ¾ Is the contrast here phonemic or phonemes occur in overlapping distribution. allophonic? And this is one main distinction between a phoneme and an allophone.
So, how do languages differ, then? So, how do languages differ, then?
¾ One reason why human languages differ in ¾ Now, consider nasal vowels in French: their sound inventory is that what is considered gars [ga] “lad” gant [ga] “glove” a phoneme in one language is an allophone in ) another, and what is an allophone in one ¾ Are they phonemes or allophones? language is a phoneme in another. ¾ How about Akan, a Ghanian language? ¾ So, for example, the aspiration variation is [ka] “bite” [ka] “speak: allophonic in English, but not in Thai. Consider ) these data from Thai: [tu] “pull” [tu)] “den” [paa] “forest” [phaa] “to split” [pam] “sew” [pa)m] “confederate” [tam] “to pound” [tham] “to do” ¾ You should be able to answer this by now.
5 Phonemes are abstract entities Phonemes are abstract entities
¾ To sum up, phonemes are meaning- ¾ The psychological existence of phonemes distinguishing sounds, whereas allophones are can be noticed in native speakers’ slips of phonetic variants of the same phoneme that the tongue, e.g., key chain [ki tSejn] may occur in specific contexts. come out as [tSi kejn], but never as [ti ¾ Notice that this means that phonemes are kSejn]. actually abstract entities in your head rather than physical sounds that come out of your mouth. ¾ This shows that [tS] is stored in the mind These physical sounds are allophones, the as a single unit, just as [k] is. concrete manifestations of the abstract phonemes.
So, how are phonemes and Note on transcription allophones related? ¾ Remember the distinction between broad ¾ But how do we capture the relationship phonetic transcription and narrow phonetic between a phoneme and its allophones? transcription? We can now understand this better in terms of phonemic vs. phonetic ¾ Phonologists explain this in terms of a transcription. mapping from the abstract phoneme entity ¾ In phonemic transcription, only phonemes are to the actual physical allophones of that represented. In phonetic transcription, the phoneme. allophones of each phoneme are transcribed. ¾ To represent this mapping, we use See Table 3.15 in your textbook for examples. phonological rules.
Phonological rules Phonological rule notation
¾ Informally speaking, a phonological rule ¾ Abstractly, we can represent this in the following notation: takes a segment as input, operates on it, X Æ Y/ ___ Z and produces a segment as output. The ¾ Basic definitions: the “Æ” means “changes operation of the rule, however, is subject to”; the slash “/” means “in the environment of”; and the “___” positions the input in the to a main restriction: it has to occur in a environment (that is before or after the relevant certain phonological environment. segments that determine the phonological change). ¾ What this rules simply says is that an input X is changed to Y if it occurs before Z.
6 Phonological rule notation Phonological rule notation
¾ Suppose further we want to place a ¾ Suppose instead that we want to say that certain restriction on the occurrence of X changes to Y after (rather than before) the segment. For example, that it has to Z. How do we do that? occur “syllable-initial” or “at a word boundary”. ¾ Well, a simple change will get us the required result: ¾ Again, we can come up with two simple X Æ Y / Z ___ notations to indicate this:
Phonological rule notation Phonological rule notation
¾ As in your textbook, we will use “σ”to ¾ In some cases an element in the indicate a syllable boundary, and “#”to environment may be optional. How do indicate a word boundary. we represent that in the notation of our ¾ Now, read the following rules. Can you rules? figure out what they mean? ¾ Brackets will do the trick. Consider this X Æ Y / σ ___ rule. What does it mean? X Æ Y / ___ # X Æ Y / ___ (Z) σ
Phonological rule notation
¾ Sometimes we might have more than one context for the application of a rule. How do we indicate that using our rule notation? ¾ Ok, so why don’t we look at some ¾ Braces come to the rescue, as in this concrete examples to see how this works. rule: ⎧Z ⎫ X Æ Y / ___ ⎨ ⎬ ⎩# ⎭ ¾ The above rule simply means that “X changes to Y either before Z or at word boundary.”
7 [l]-devoicing Aspiration
¾ Let’s start with the rule for /l/ devoicing in ¾ How about aspiration of voiceless stops in English. Informally put, the rule says English? “/l/ gets devoiced when following a “Voiceless stops become aspirated in syllable-initial voiceless stop.” English when they occur syllable- ¾ How do we represent this in phonological initially.” rule notation? ¾ How do we represent that in formal rule /l/ Æ [l•] / σ [voiceless stop] ___ notation in phonology? [voiceless stop] Æ [aspirated] / σ ___
Vowel nasalization Deletion
¾ Now, vowel nasalization: ¾ How about deletion rules? “In English, vowels become nasalized ¾ For these, we use the symbol Ø in the when they are followed by a nasal output of the rule (i.e., after the arrow). consonant.” ¾ For example, English speakers delete the ¾ Rule notation: [ə] in an open syllable when it is followed V Æ [nasal] / ___ [nasal] by a stressed syllable, as in police [pl•iès]. ¾ Ok, but how about vowel nasalization in ¾ How do we represent this in rule notation?
Scots Gaelic? Remember the rule? [ə] Æ Ø/ C0 ___ σ C0 Vstressed
Epenthesis Next class agenda
¾ The Ø comes in handy for phonological rules ¾ Features. that insert sounds as well. The key difference here is that the Ø will be in the input to the rule. ¾ Using features in phonological rules. ¾ For example, in some English dialects, ¾ Syllable structure. consonant clusters of [l] and another consonant are not allowed in syllable-final position. Speakers of these dialects, therefore insert a [ə] to fix the syllable, e.g., milk [milək]. ¾ In rule notation, this would be represented as: Ø Æ [ə] / [l] ___ C σ
8