Phonology: case studies

Lateral fronting

In addition to the dorsal stop alternation discussed in the last section, Turkish has an alternation between a velarized lateral liquid (IPA [ɫ]) and a counterpart that has no velar component (IPA [l]). Like all velar consonants, the velarized lateral has the feature [+back], while the non-velar lateral is [-back]. These two sounds are similar to the "dark l" and "clear l" (respectively) found in some North American dialects of English. Throughout this section, we have followed Clements and Sezer (1982) in representing the velarized lateral in Turkish simply as [l], and the non-velar lateral as [l^j].

Notes: Phonetic transcriptions in our other sections on Turkish do not reflect the [l] ~ [lj] alternation described here so as not to distract from the phenomena under discussion on those pages. However, the lateral alternation in Turkish is productive and pervasive, and the generalizations is this section can be extended to the examples given elsewhere. In fact, most of the examples used in this section appear in other sections in the Turkish module.

Since [l^j] and [l] are quite similar, and because the two are analyzed as allophones of a single phoneme in some languages (e.g. English), these two sounds are a "suspicious pair" (see the discussion at the beginning of the section on the dorsal alternation), and we should be especially attentive to their distributions in order to know whether or not they are phonemically related.

Examples containing the liquids [l^j] and [l] are provided in (26).

Plain [l^j] and velarized [l] in Turskish have distinctive phonetic calling cards. Spectrograms 1 and 2 show the words el^jin with [l^j] and borular with [l]. The dotted red lines in these spectrograms track the acoustic signal's formants. The property that distinguishes [l^j] from velarized [l] is the position of F2 (second formant from the bottom). Note that for [l^j] in el^jin, F2 is much higher than it is for [l] in borular.

(26)	Lateral alternation

a.	Non-back laterals
	[i]	tjiljci	'fox, nom.'		[y]	ɟylj	'laugh!'
		eljin	'hand,nom.'
	[e]	celj	'bald'
		biljet	'ticket'

b.	Velarized laterals
	[a]	palto	'overcoat'	[o]	daktilo	'typist'
		borular	'pipe nom. pl.'		jol	'road'
	[ɨ]	akɨl	'intelligence, nom.'	[u]	mutlu	'happy'
					okul	'school'

Plain [l^j] and velarized [l] in Turskish have distinctive phonetic calling cards. Spectrograms 1 and 2 show the words el^jin with [l^j] and borular with [l]. The dotted red lines in these spectrograms track the acoustic signal's formants. The property that distinguishes [l^j] from velarized [l] is the position of F2 (second formant from the bottom). Note that for [l^j] in el^jin, F2 is much higher than it is for [l] in borular.

SPECTROGRAM 1: non-back lateral liquid [l^j] in el^jin 'hand, nom.'

SPECTROGRAM 2: velarized lateral liquid [l] in borular 'pipe, nom. plur.'

In the examples given, the frequency of F2, high or low, seems to vary depending on the frequency of F2 for the flanking vowels: between front vowels, the lateral tracks the higher F2 values of ambient front vowels and similarly, tracks the lower F2 values characteristic of back vowels. This is a form of co-articulation, which is the basis for all phonological processes of assimilation. (We will say more about this later.)

Examining a broad sample of examples, the contexts in which we find [l^j] and [l] can be stated quite simply, if we remember the role of the syllable in the velar-palatal alternation discussed in the last section. It turns out that the distributions of [l^j] and [l] are exactly parallel to the distributions of palatal and velar stops: [l^j] occurs only in syllables containing a front vowel, while [l] is restricted to syllables containing back vowels. Figure (27) shows how the examples in (26) are organized into syllables, following the principles outlined in the past section (on the palatal-velar alternation).

(27)	Lateral alternation

a.	Non-back laterals
	[i]	[tjilj]σ[ci]σ			[y]	[ɟylj]σ
		[e]σ[ljin]σ
	[e]	[celj]σ
		[bi]σ[ljet]σ

b.	Velarized laterals
	[a]	[pal]σ[to]σ		[o]	[dak]σ[ti]σ[lo]σ
		[bo]σ[ru]σ[lar]SYL			[jol]σ
	[ɨ]	[a]σ[kɨl]σ		[u]	[mut]σ[lu]σ
					[o]σ[kul]σ

Note that there is are two gaps in the data cited in (26) and (27). We do not have an example showing [l] at the beginning of a syllable containing [ɨ], and we have not shown that [l^j] occurs in syllables with the mid front round vowel [ø]. However, given the way that the laterals pattern with other front and back vowels, we can reasonably assume that they will have the same distributions [i] and [ø].

In constructing a generalization, it's okay to go beyond the data you see you've identified a pattern and it would make sense for data you haven't seen to fit the pattern you've identified. In part, that's what generalizations are for Ð they provide an indication of what we should expect if our general statement is correct. Then, if we can, we confirm our generalization by looking at the missing pieces.

From the data we've seen (which is quite representative), we can infer that [l] and [l^j] are in complementary distribution, and are allophones of a single phoneme.

(25)	Turkish Lateral Fronting
	a.	/velar lateral/ → [palatal] / [ ___ [front vowel] α ___ ] SYLL

	b.	/+lateral, +back/ → [-back] / [ ___ [+syllabic, -back] α ___ ] SYLL

prev | top | next