Mode	Text / Speech	Sentence #			Subordinations
					English 1	English 2
Legal translation	FCPA	30			⑥	④
Target language	Reordering Σi=1n-1 Σj=i+1n     I(xj<xi)	± Nestings   { }                  {{ }}                {{{ }}}			Semantic changes Δ
Russian 1	—	—	—	—	⓪
Russian 2	—	—	—	—	①

Differences in analysis:

1. In “English 1” and “Russian 1,” [4] and [5] are seen as elliptical forms of phrases providing semantically non-typical information (“for any person” and “with any person,” respectively). So they’re treated as separate propositions functioning as adjuncts to [3]. In “English 2” and “Russian 2,” [4] and [5] are seen as providing semantically typical information. So they’re treated, along with [3], as a single proposition.

2. [6] in “English 1” ([4] in “English 2”) says: “or directing business to, any person.” The equivalent phrase in Russian says: “or providing any person with opportunities for profitable business transactions.” In “Russian 1,” this phrase is treated as a single proposition, for ease of comparison with the English. In “Russian 2,” the same phrase is treated as two separate propositions, including an argument, [X], not corresponding to any proposition in the English original.

Reasons:

1. This study doesn’t treat a non-clausal phrase as a separate proposition if it provides semantically typical information like beneficiary, accompaniment, resultant, instrument, location, goal, time (except for descriptions of events), manner or measure (cf. Larson 1984, 219-223). Schauer (2000) similarly proposes treating a phrase as a separate discourse unit if it: (a) provides information other than semantically typical information like means, place or time, and (b) can be paraphrased as a clause. The distinction between non-clausal phrases providing semantically typical information and those providing non-typical information is useful for our parsing method. On the one hand, it avoids a plethora of separate segments for simple, typical phrases (like “with you” or “for me”). On the other hand, it treats phrases that feel and behave more like propositions (like “despite the awful weather”) – which tend to be marked off by a comma in a language like English, which can be rephrased as a clause and which are more likely to correspond to a clause with a verb in a different language version – as separate propositions. As useful as this distinction is, it isn’t always clear-cut. In this study, it’s applied consistently to all language versions of a given sentence. But there can be some inconsistency in how it’s applied in different sentences, or even in the same sentence as analyzed at different times.

2. The translated or interpreted versions of each sentence are segmented so as to correspond to the original English version, even though that may lead to more or fewer segments in a another language version than if the parsing method was applied directly to that version. This is done for ease of comparison between the different language versions of each sentence, unless a given translated or interpreted version seems to be significantly different in structure from the original. This is a subjective call. In “Russian 1,” [6] is treated as a single proposition, to match the segmenting in English. In “Russian 2,” the corresponding phrase is treated as two separate propositions, whose combined structural effect is felt to be significantly different from the English original.

Consequences:

1. The new analysis would lead to a complexity count of 4 rather than 6 subordinate propositions in the original English version of the sentence.

2. The new analysis would lead to a count of 1 rather than 0 for changes in semantic relations in the Russian translation.