CHAPTER 4 ENGLISH BLENDING: AN OT ANALYSIS
4.2 An OT Account for Phonological Blending
and select the candidate [m] as the optimal output.
ANCHOR and IDENT-σ/Num are also used for the evaluation of phonological blending. Section 4.2 introduces further relevant constraints to examine the nature of the phonological preservation of blending types.
4.2 An OT Account for Phonological Blending
Not all blends are generated from the use of the morphological strategy. In other words, the preserving constituent of the source used to create the blend is not
necessarily a morpheme. From a phonological perspective, I divided blends into four preserving types7, as given in (78).
(78) Preserving types
In (78), the number of the syllables in the blend is identical to the number of the syllables in the second source. Once the number of the syllables in the blend is
determined, the preservation of the first source starts to compete with the preservation of the second source. In terms of the preservation of the first source, the data in example (78) can be generally separated into four types. The first type preserves the
7 The fifth type of the preservation of the entire first source is presented in The 2009 Graduate Student Workshop on Phonology in National Chengchi University, Taiwan and it is included under
morphological blending in the thesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
first onset, as in (78a). The second type retains the first syllable, as in (78b). The third type preserves the first syllable and second onset, as in (78c). The fourth type retains the first two syllables, as in (78d).
Like clipping, I make similar generalizations for blending. First, at least one of the sources is truncated. Second, the stressed syllable of the source is sometimes retained but sometimes truncated. Third, the constituents are retained contiguously.
No matter how the source is truncated, it protects the preserving constituents from being deleted internally. Finally, since the size of the blend equals the size of the second source, the number of the syllables in the blend is predictable. I propose that the nature of blending can also be examined by ANCHOR constraints on the basis of these generalizations.
In this section, I illustrate the effect of the family of ANCHOR on phonological blending and employ additional constraints relevant to the analysis. On the basis of the generalizations mentioned above, I make use of several faithfulness constraints CONTIGUITY, ANCHOR, IDENT, and MAX to evaluate four types of blending types. Different constraints have different functions. First, CONTIGUITY controls the contiguous preservation of the source. Second, ANCHOR LEFT controls the left edge preservation of the first source. Third, ANCHOR RIGHT controls the right edge preservation of the second source. Fourth, IDENT controls the size of the blend.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Finally, MAX controls the segmental preservation of the source.
As for blended outputs, they obey CONTIGUITY because of the contiguous preservation. Their first source satisfies ANCHOR LEFT due to the initial
preservation. Their second source respects ANCHOR RIGHT because of the final preservation. Few blends violate IDENT due to it being a different size from the second source. Most blends violate MAX because of the partial preservation.
Since the first type retains the first onset of the first source, it relates to
IDENT-σ/Num in (76), MAX S1B, MAX S2B and LEFT-ANCHOR-onset, as defined in (79), (80), and (81). In addition, ANCHOR RIGHT is employed to govern the
preservation of the second source.
(79) MAX (Source 1, Blend) (abbreviated as MAX S1B in tableaux)
Assign one violation mark for every segment in the first source that is not in the blend.
(80) MAX (Source 2, Blend) (abbreviated as MAX S2B in tableaux)
Assign one violation mark for every segment in the second source that is not in the blend.
(81) LEFT-ANCHOR-onset (Source 1, Blend)
Assign one violation mark for every first onset of the first source that is not the first onset of the blend.
‧
In (82), CONTIGUITY and ANCHOR RIGHT are inactive because all candidates respect them. There are three ranking arguments made in the tableau. First, we compare the winning candidate [] and the losing candidate *[]. Since
*[] is more faithful to the first source, it is more favored MAX S1B over [].
The winner preserves more segments of the second source. This makes it more harmonic in MAX S2B. For the winning [] to be optimal, MAX S2B must dominate MAX S1B. Second, we compare the winning candidate [] and the losing candidate *[]. The candidate [] satisfies IDENT-σ/Num but violates MAX S2B. The losing candidate *[] respects MAX S2B at the cost of preserving more than one syllable. This causes it to violate IDENT-σ/Num. For the winning [] to be optimal, IDENT-σ/Num must dominate MAX S2B. Finally, we compare the winning candidate [] and the losing candidate *[]. The winner []+ [] CONTIGUITY LEFT-
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
[] satisfies LEFT-ANCHOR-onset but violates MAX S2B; however, *[]
satisfies MAX S2B but violates LEFT-ANCHOR-onset. For the winning [] to be optimal, LEFT-ANCHOR-onset must dominate MAX S2B.
Tableau (82), however, fails to rule out candidate *[] that is more faithful to MAX S2B than the winner []. To select [], I use the PHONOTACTICS constraint to remove *[], as defined in (83).
(83) PHONOTACTICS
Assign one violation mark for every impossible string of the blend.
Constraint (83) disallows every blend to consist of permissible strings of segments. Its function is shown in (84).
(84) PHONOTATICS >> MAX S2B
Tableau (84) omits CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, IDENT-σ/Num, and MAX S1B since they favor neither [] nor *[]. The winning candidate [] violates MAX S2B because it does not preserve the complete second source. The losing candidate*[] violates PHONOTATICS because of the ill-formed string [mb] in onset. For [] to be optimal,
[]+ [] PHONOTATICS MAX S2B
3 1
a. (1)
b. *!()‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
PHONOTATICS must outrank MAX S2B.
The second type preserves the first syllable of the first source. Thus, the constituent that ANCHOR is concerned with syllable rather than onset segment. I propose that LEFT-ANCHOR-σ substitutes for LEFT-ANCHOR–onset, as defined in (85).
(85) LEFT-ANCHOR-σ (Source 1, Blend)
Assign one violation mark for every first syllable in the first source that is not the first syllable in the blend.
Constraint (85) prevents the first syllable of the first source from being moved to other syllabic positions in the blend. Tableau (86) evaluates the second type.
CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, IDENT-σ/Num, and PHONOTATICS are omitted because they are inactive in (86).
(86) LEFT-ANCHOR-σ >> MAX S2B
In (86), since the winning candidate [] preserves the part of the second source, it violates MAX S2B. The losing candidate *[] fulfills MAX S2B at the cost of preserving the first syllable of the first source. This incurs a fatal violation of LEFT-ANCHOR-σ. Therefore, LEFT-ANCHOR-σ must dominate MAX S2B to remove the loser and select the winning [] as the optimal output.
[] + [] LEFT- ANCHOR –σ
MAX S2B MAX S1B
3 3
a. (2) (5)b. *! (7)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The third type retains the second onset of the first source in addition to the first syllable of the first source. MAX-onset is proposed again, as in (87).
(87) MAX-onset (Source 1, Blend) (abbreviated as MAX-onset S1B in tableaux) Assign one violation mark for every onset in the source that is not in the blend.
Constraint (87) requires that every onset of the source has a correspondent in the blend. Different from LEFT-ANCHOR-onset, MAX-onset cares about the occurrence of the onset of the source rather than its location in the blend. In other words, if the onset of the source becomes the coda of the blend, it satisfies MAX-onset but violates LEFT-ANCHOR-onset. I take the same example of the second type in (86) to argue the ranking of MAX-onset and MAX S2B, as in (85). CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, LEFT- ANCHOR –σ, IDENT-σ/Num, PHONOTATICS are
omitted because they are inactive in (88).
(88) MAX S2B >> MAX-onset S1B
As tableau (88) presents, MAX S2B favors [] over *[] since the winner preserves more segments of the second source. The winning candidate
[] is less harmonic in MAX-onset because it only preserves three onsets of the first source, which has five onsets. For the winning candidate [] to be [] + [] MAX S2B MAX-onset S1B MAX S1B
3 3
a. (2) , (2) (5)b. ! (4) (1) (4)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
optimal, MAX S2B must rank above MAX-onset S1B.
Now, the ranking in (85) is employed to evaluate the third type, as shown in (89).
(89) The evaluation for the third type
Given the candidate pair (a)-(c), MAX S2B is active to rule out *[]. In addition, the winning candidate [] is more favored by MAX-onset S1B and MAX S1B over the loser (b). Thus, candidate [] emerges as the optimal output.
The fourth type retains the first two syllables of the first source. Again, LEFT-ANCHOR–Σ is used to explain this type, as in (90).
(90) LEFT-ANCHOR-Σ (Source 1, Blend) (Σ=σσ)
Assign one violation mark for every first and second syllable in the first source that is not the same as the first and second syllable in the blend.
As constraint (90) requires, the first and second syllables of the first source must correspond to the first and the second syllables of the blend. Its evaluation is
presented in tableau (91). CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, LEFT-ANCHOR -σ, IDENT-σ/Num, PHONOTATICS are omitted because they are inactive in (91).
[]+ [] MAX S2B MAX-onset S1B MAX S1B
2 2
a. ,, (3) , (2)b. ,,, (3) ! , , (3)
c. ,,,! (4) (1)
‧
Although the losing candidate *[] is more favored by MAX S2B over [], it fatally violates LEFT-ANCHOR–Σ because of the preservation of one syllable. For the winning candidate [] to be harmonic, LEFT-ANCHOR–Σ must dominate MAX S2B.
Tableau (91) cannot directly rank MAX-onset S1B and MAX S1B which stay in a stringency relation. McCarthy (2008) suggests that constraint A is more stringent than constraint B if every violation of constraint B is also a violation of constraint A, but a violation of constraint A is not necessarily a violation of constraint B. Stringency relation can be also described as the general-specific relation, and the specific constraint is usually placed above the general constraint. In this study, I employ McCarthy’s formulation to account for the relation between MAX-onset S1B and MAX S1B, and propose that MAX-onset S1B dominate MAX S1B. The ranking arguments for phonological blending are established in (92).
(92) Ranking arguments for phonological blending
a. { LEFT-ANCHOR-onset S1B, IDENT-σ/Num S2B } >> MAX S2B >> MAX S1B
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
c. LEFT-ANCHOR-σ S1B >> MAX S2B d. MAX S2B >> MAX-onset S1B e. LEFT-ANCHOR–Σ S1B >> MAX S2B f. MAX-onset S1B >> MAX S1B
In (92), rankings (d) and (e) show that MAX S2B dominates MAX-onset S1B, which is placed above MAX S1B. In addition, there is a set of unranked constraints that dominates MAX S2B, including PHONOTATICS, CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, LEFT-ANCHOR-σ, LEFT-ANCHOR–Σ, and IDENT-σ/Num S2B, as given in (93).
(93) Summary ranking for phonological blending
{ PHONOTATICS, CONTIGUITY, ANCHOR RIGHT, LEFT-ANCHOR-onset, LEFT-ANCHOR-σ, LEFT-ANCHOR–Σ, IDENT-σ/Num S2B }
MAX S2B MAX-onset S1B MAX S1B