The second indicator of difficulty recorded is nesting changes – creation or elimination of structures where one proposition is syntactically surrounded by the predicate and any arguments of another proposition.

This definition is deliberately conservative, counting as a nesting only a proposition which syntactically splits the predicate and arguments of its parent. If we broadened our definition of a nesting to include a proposition which splits a predicate from an adjunct, an argument from a modifier or a complementizer from the rest of its clause, we’d get much higher counts for nestings overall, especially in structurally different language pairs. This conservative approach was chosen deliberately, to avoid any suggestion that the counts for indicators of difficulty in language pairs with very different structure have been artificially inflated.

Recall that, when a proposition is syntactically split in one version of a sentence, only the main part of that proposition – the part with the predicate – is copied onto the number line for that version. So the reordering value for that version of the sentence reflects the fact that the predicate of the split proposition has been reordered, which is taken as an indication of added mental effort. However, when counting nesting values for a sentence with a split proposition, we need to look at the predicate of that proposition as well as its arguments. That’s because it’s the syntactic splitting of a predicate from any of its arguments that’s associated with added mental effort, as explained before.

This feature is recorded in the number line below each translated or interpreted version of a sentence. That’s done by placing curly brackets { } around a number for a proposition where a nesting is created – that is, a proposition that’s not nested in the original English version but is nested in translation or interpretation. A number for a proposition where a nesting is eliminated – that is, a proposition that’s nested in the original English version but not in translation or interpretation – is surrounded by crossed-out brackets { } and counted as well. A number for a proposition where a double or triple nesting is created is placed in double or triple brackets. And a number for a proposition where a double or triple nesting is eliminated is placed in double or triple crossed-out brackets. Single, double and triple nestings are recorded separately. A number for a proposition in an unchanged nesting – that is, a proposition that’s nested in the same proposition(s) in the original English version and in translation or interpretation – is surrounded by gray brackets { } and not counted. Any number for a nested proposition in the original English version of a sentence is also surrounded by gray brackets, as a reference for comparison with the other versions.

To see how this works, consider figure 36, which again shows the segmented text with parse tree and number line for the original English version of our sample sentence. One proposition in that version – number 4 – is syntactically surrounded by the predicates and arguments of other propositions – numbers 1, 2 and 3. So the number 4 in the English number line is surrounded by curly brackets, to show that it’s nested. Those brackets are gray, since they’re in the number line for the original English version of the sentence, which will be the reference for comparison with the other versions.

The Japanese translation of the same sentence, including Japanese text, English gloss, semantic parse tree and number line, is again shown in figure 37. This time the number line shows the propositions that are nested in the Japanese translation.

As can be seen in the parse tree for the Japanese translation of the sentence in figure 37, one proposition – number 6 – is syntactically surrounded by the predicate and arguments of another proposition – number 5. And proposition 5 is itself surrounded by the predicates and arguments of other propositions – numbers 1, 2 and 3. The same is true of proposition 4.

Proposition 6 isn’t nested in the original English version of the sentence in figure 36. But it’s in a double nesting in the Japanese translation in figure 37. So the number 6 in the Japanese number line in figure 37 is enclosed in double curly brackets and counts as creation of a double nesting. Proposition 5 isn’t nested in the original English version either, but it’s in a single nesting in the Japanese translation. So the number 5 in the Japanese number line in figure 37 is enclosed in single brackets and counts as creation of a single nesting. Proposition 4 is nested in the same propositions – numbers 1, 2 and 3 – in the original English version and in the Japanese translation. So the number 4 in the Japanese number line in figure 37 is enclosed in gray brackets and won’t be counted as a nesting change.

The total numbers of single, double and triple nestings created or eliminated in each translated or interpreted version of a sentence are entered in the data table on the display page for that sentence, in the row for the language in question, under ± Nestings. The pairs of surrounding brackets (solid or broken) directly adjacent to a number in the number line represent single nestings. In the Japanese translation of our sentence in figure 37, there are two such pairs of brackets (immediately surrounding the numbers 6 and 5). So we’ll enter 2 for changes in single nestings in the Japanese translation of that sentence. A double nesting contains a single nesting, so it counts as a single nesting too. The pairs of surrounding brackets one position away from a number in the number line represent double nestings. In the Japanese translation in figure 37, there’s one such pair of outer brackets (surrounding the number 6). So we’ll enter 1 for changes in double nestings in the Japanese translation of our sentence. There’s also a space in the table to record changes in triple nestings, which are rare.

Table 4
Sample data table showing counts for nesting changes in Japanese translation of sentence

Mode	Text / Speech	Sentence number			Subordinations
Legal translation	Paris Agreement	29			3
Target language	Reordering	± Nestings   { }                  {{ }}                {{{ }}}			Semantic changes
Russian
Hungarian
Turkish
Mandarin
Japanese	12	2	1	—