Walkthrough

Below is a brief walkthrough of (part of) an existing analysis. The following tableaux are reproduced from [Lombardi1999].

A.	/skuːg/	Agree	IdLar	*Lar	IdOns
→	skuːg			1
	skuːk		1

B.	/vigsəl/	Agree	IdLar	*Lar	IdOns
	vigsəl	1		1
→	viksəl		1
	vigzəl		1	2	1

C.	/stekde/	Agree	IdLar	*Lar	IdOns
	stekde	1		1
	stegde		1	2
→	stekte		1		1

D.	/ägde/	Agree	IdLar	*Lar	IdOns
→	ägde			2
	äkde	1	1	1
	äkte		2		1

The VTs in show for inputs for the same language (here, one similar to Swedesh). Because the same target language is being described, each underlying form and its respective set of candidates are evaluated against the same ranking of constraints.

The constraints in these VT are briefly defined below. The first two are markedness constraints, meaning they assign violations based only on the output structure.

Assign a violation for every pair of adjacent output segments that do not agree in their laryngeal specification.

Assign a violation for every voiced output segment.

The following two are faithfulness constraints; they assign violations based on differences between the input and output segments.

Assign a violation for every input segment whose corresponding output segment has a different laryngeal specification.

Assign a violation for every input segment whose corresponding output segment has a different laryngeal specification, and that output segment is a syllable onset.

In Tableau A, the winner is [skuːg], even though this violates *Lar. The loser, [skuːk], does not violate *Lar but it does violate IdLar, which penalizes changing a laryngeal feature from the input to the output. For this language, IdLar must dominate (be more important than) *Lar.

In Tableau B, the winner is [viksəl], and it is shown here with two losers. The candidate [vigsəl] loses because it violates Agree, as the voiced [g] is adjacent to the voicecless [s]. In this ranking, Agree is undominated, so this candidate is ruled out. Both remaining candidates have one violation for IdLar, so this constraint does not make a distinction. While both remaining candidates have changed the voicing of one segment to satisfy Agree, the loser [vigzəl] does so by having a pair of voiced segments, which incurs 2 violations of *Lar, while the intended winner [viksəl] incurs none.

In Tableau C, the winner is [stekte]. This candidate unfaithfully maps an input /d/ to output [t]. The faithful candidate violates Agree, and like in Tableau B above, changing the cluster to two voiced segments would incur two violations of *Lar, so both segments become voiceless.

However, in Tableau D, the winner, [ägde], does have a sequence of two voiced consonants. However, it got there by doing nothing: even though it has two violations of *Lar, it has 0 violations for IdLar, as it is completely faithful. The losing candidate with two voiceless segments, [äkte], has 0 *Lar violations, but 2 IdLar violations. As IdLar dominates *Lar, it still loses.

OTpedia

Walkthrough